Homebrew 4343

[Bill Shenefelt]

[quote=GordonW;12034]Well, I'd be inclined to make up a "composite" from film caps, to achieve this value.


I had tried to build the JBL 3143 crossover for my 4343 speaker sets using components from JBL and their layout. I was not initially going to use the passive part at the 300 cps point so I did not buy the 52uf and 72uf caps nor the 2.9 and 5.4 mh coils.
The inductors are small and have the iron frame like a small transformer. The 20, 13.5, and 4 uF capacitors look like a paper wrap around a "wax" core, about an inch in diameter and 2 inches long. The smaller caps are the typical polypropylene (probably metallic) caps. The power resistors seem to be about 5% tolerance at best. I was about to order some 1% Dayton metallized poly caps to be jumped with 1 uF Audiocap PPT Theta film foil caps. The foils would parallel all but the 20uF 10 inch midbass ground cap. Also I was going to order 1% tolerance non inductive resistors for the whole circuit. Does this sound like a significant improvement or am I just wasting money? I would prefer to stay with an active crossover in the 300 cps point unless it is not as good as going with the JBGL passive design. I always was of the understanding that getting the bass out of the amp that is driving the high frequencies was a good thing. (My bass amp is a 250w/ch heathkit and my upper frequency amp is an Amber 100 w/ch. I also have the potential of going with a 1200 cps crossover to a McIntosh 240 tube amp for the horn and slot radiator is that is a good thing to do.

I currently have a Marchand X1 active crossover for 24/octave at 300 cps. Despite putting out a pretty flat summed output when attached directly to a spectrum analyzer, I seem to get output sound at the 10 inch midbass that is rolling off at almost 500 cps, not 300 cps giving me a 9 dB dip between about 300 and 500 cps as read on a microphone attached to the specrum analyzer (Polarity changes do not improve this dip). Shouldn't I be "entering" the 10 inch network at the input to the 1.7mH coil when going active? This is what I am doing and it seems to not give reasonable responce at the crossover region and above. I tested continuity and got less than an ohm resistance through all the inductors, but cannot test the capacitors.
Here is what I get with the active crossover in place checking 10 inch and 15 inch separately:
10 inch: 500 cps= -3dB, 355cps= - 9dB , 250cps=-15dB
15 inch: 500 cps= -18dB, 355cps= -15dB, 250cps=-3dB


I have a Marchand X44, 24dB/octave active crossover (cards for 300 and cards for 400 to allow a little overlap testing) on order. Before I buy all new passive components for the JBL passive stages can anyone tell me if the use of an active crossover (at 300 cps) is as good or better than using the whole JBL passive network including that for the bass driver? The appearance of the JBL passive network (at least for the mid bass and horn) suggests more than a simple second order network and I am sure is highly tailored to match the drivers. I just do not know if using the Marchand active for 300 cps will in some way detract from what JBL is doing in their active network. Anyone with experience with active vs passive (for the 300 cps crossover) on this speaker?

[B&KMan]

Hello,
I understands that you reproduced that partly the original circuit of JBL 3143??

if it is the case, you will butt yourselves like me to the hollow between the 15 and the 10. However you seem to have an active network??

and does the amplifiers necessary to make a load activate then why not go directly on this side and drop completely the network 3143??

the passive and active mixture is not best idea.

Have to perhaps check itself electrically if the active network that you use really makes it possible the 10 to go down or it badly is configured or network badly conceived??

some photographs and useful diagram is appreciated for better including/understanding your step and to help people with you help.

:cheers:

Jean.

[Bill Shenefelt]

I am only using the active network for the low pass to the 15 and high pass to the 10 inch driver. The 10 inch driver still requires a high frequency network to the midrange horn and then from the horn to the slot radiator. I think I have bypassed the passive netowrk that is the high pass into the 10 inch by not using the 52 uF and 2.9 mH Hhis is what the network to the 10 inch looks like.
http://www.sheneskillies.com/4343-to-10-inch.jpg
It cannot be eliminated since it is needed to roll of fthe high end to make the crossover to the horn driver.

In your crossover photos you seem to have air coil inductors. Are they better than the iron core ones that JBL uses? Also I have the paper and wax? large capacitors in the crossover. Are metalized polypropylene better?

Hello,
I understands that you reproduced that partly the original circuit of JBL 3143??

if it is the case, you will butt yourselves like me to the hollow between the 15 and the 10. However you seem to have an active network??

and does the amplifiers necessary to make a load activate then why not go directly on this side and drop completely the network 3143??

the passive and active mixture is not best idea.

Have to perhaps check itself electrically if the active network that you use really makes it possible the 10 to go down or it badly is configured or network badly conceived??

some photographs and useful diagram is appreciated for better including/understanding your step and to help people with you help.

:cheers:

Jean.

[Bill Shenefelt]

I am considering substituting air core inductors for the OEM iron core inductors in my 3143 networks. I understand that I should be matching the DC resistance of the JBL inductors. I can only measure to the readout of my fluke which is 0.1 ohm. I get the following values for the JBL and the spec for the air cores.
JBL 16 gage ..............air core substitute
mH.. gage ..DCR............... wire gage DCR
0.16mH --18 gage -0.1 ohm .........16gage 0.09 ohm
0.25mH <20(22?)gage 0.2 ohm .....16 gage 0.12 ohm
1.0mH <20gage 0.5 ohm .............16 gage 0.30 ohm
1.7mH <20gage 0.5 ohm .............16 gage 0.45

Is the 0.1 to 0.2 ohms significant?
Should I be inserting resistors at 0.1 ohm resistance to match things?
Anyone know the JBL DCR values to better than a tenth ohm?
Would thin be an improvement or a degredation to do the replacement?

[B&KMan]

hello,

the small difference is relevant on power output. this type of crossover came with a pot jbl fine tune ouput. so normally it is fine...

:cheers:

[B&KMan]

hello again,

be shure your project to replica of crossover keep factor of the change the recone 2235 in basket of 2231 , same as 10 and horn. the original design is better with original driver and cone. :blink:

do you decide to stabilize impedance driver by addition of zobel circuit impedance?

:blink:

[Ian Mackenzie]

[quote=GordonW;12034]Does this sound like a significant improvement or am I just wasting money?


I would prefer to stay with an active crossover in the 300 cps point unless it is not as good as going with the JBGL passive design. I always was of the understanding that getting the bass out of the amp that is driving the high frequencies was a good thing. (My bass amp is a 250w/ch heathkit and my upper frequency amp is an Amber 100 w/ch. I also have the potential of going with a 1200 cps crossover to a McIntosh 240 tube amp for the horn and slot radiator is that is a good thing to do.

I currently have a Marchand X1 active crossover for 24/octave at 300 cps. Despite putting out a pretty flat summed output when attached directly to a spectrum analyzer, I seem to get output sound at the 10 inch midbass that is rolling off at almost 500 cps, not 300 cps giving me a 9 dB dip between about 300 and 500 cps as read on a microphone attached to the specrum analyzer (Polarity changes do not improve this dip). Shouldn't I be "entering" the 10 inch network at the input to the 1.7mH coil when going active? This is what I am doing and it seems to not give reasonable responce at the crossover region and above. I tested continuity and got less than an ohm resistance through all the inductors, but cannot test the capacitors.
Here is what I get with the active crossover in place checking 10 inch and 15 inch separately:
10 inch: 500 cps= -3dB, 355cps= - 9dB , 250cps=-15dB
15 inch: 500 cps= -18dB, 355cps= -15dB, 250cps=-3dB


I have a Marchand X44, 24dB/octave active crossover (cards for 300 and cards for 400 to allow a little overlap testing) on order. Before I buy all new passive components for the JBL passive stages can anyone tell me if the use of an active crossover (at 300 cps) is as good or better than using the whole JBL passive network including that for the bass driver? The appearance of the JBL passive network (at least for the mid bass and horn) suggests more than a simple second order network and I am sure is highly tailored to match the drivers. I just do not know if using the Marchand active for 300 cps will in some way detract from what JBL is doing in their active network. Anyone with experience with active vs passive (for the 300 cps crossover) on this speaker?

Hi Bill,

I have been so busy I have not had an opportunity to follow your project.

For clarity can you advise if you are modifying a stock 3143 network?

Or do you plan to build your own network?

Right off the bat messing with these 4 ways can get you in a lot more trouble than its worth unless you have a very clear understanding of the existing system and exactly how you propose to alter the stock implementation. What I am saying is unless you get it exactly right you will find yourself in a maze and the whole mess will become quite frustrating to say the least.

The stock 3143 has been arranged to provide a smooth transition between all the drivers allowing for their locations on the front baffle.

The network also provides the option of internal passive or external bi amping with a multi pole rotary switch. The switch isolates the woofer from the LF filter elements on the woofer and the HP elements on the mid cone driver.

It should be noted the voltage drive for the active crossover of the woofer and mid cone in the 4343 has been specifically tailored for these particular drivers and their location on the baffle.

You will NOT get the correct results with other than the designated voltage drive. A simple search will locate these elsewhere on the forums.

Refer to the 5234/35 Pdf manuals for the active filter RC values. It should be possible for Marchland to provide a customised card. If you get this right the active mode will be subjectively better with the Amber amp on the mid/top end. No question of that. Get it wrong and you will spend a lifetime wondering why it sounds not quite right.

I recommend you build new networks rather than attempt to upgrade the stock 3143. The reason is the bi amp switch will hinder any gains and replacing parts in the stock network will be difficult. You will also need to be very careful with all the phasing on the drivers.

If it were me I would write to Giskard nicely and arrange a pair 2122H to replace the 2121 and get new diamond surround aluminium diaphragms for the compression driver and get on the list for V 3 of the 3145 equivalent network. You stand to gain light years more from this approach than merely putting in better parts.

Ian

[Bill Shenefelt]

I thought it easiest to follow if I replied within the text belwo.

[quote=Bill Shenefelt;157040]

For clarity can you advise if you are modifying a stock 3143 network?
Not quite. The network I am rebuilding is one I built to the JBL design using "replacement parts I bought from JBL I did not buy the parts for the high pass of the 10 inch or low pass of the 15 inch so no switch was needed. I just go in where the switch would have allowed me to enter for a biamp setup.
Or do you plan to build your own network?
I plan to build the JBL 3143 but using higher grade caps and inductors and resistors, but only the components aabove the 15 inch driver circuits. It will be as if I used their switch.
I do not plan to change any values. I will be widening the baffle of the 10 inch and horns to the JBL 4343 width. I have a 6 cubic ft cabinet containing the 15 inch which lies on one side. By turning it over I will get the 18 inch spacing of 10 to 15 inch driver centers in the 4343 within a quarter inch.

The stock 3143 has been arranged to provide a smooth transition between all the drivers allowing for their locations on the front baffle.

The network also provides the option of internal passive or external bi amping with a multi pole rotary switch. The switch isolates the woofer from the LF filter elements on the woofer and the HP elements on the mid cone driver. I never installed these or the switch.

It should be noted the voltage drive for the active crossover of the woofer and mid cone in the 4343 has been specifically tailored for these particular drivers and their location on the baffle.

You will NOT get the correct results with other than the designated voltage drive. A simple search will locate these elsewhere on the forums.
I may have a problem here as I am going to use a Marchand 24/octave crossover for the 10 to 15 inch transition. I have gotten two cards set at 300 cps and 2 set at 400 cos to allow me to try different combinations including some overlap. I know JBL did some things like mating 6 with 18 dB/octave slopes. I was trying more for time alignment than necessarily purely flat responce. I miss the sound I used to get from the L300 which covered most of the bass and lower midrange with the 15 inch.

Refer to the 5234/35 Pdf manuals for the active filter RC values. It should be possible for Marchland to provide a customised card. If you get this right the active mode will be subjectively better with the Amber amp on the mid/top end. No question of that. Get it wrong and you will spend a lifetime wondering why it sounds not quite right.

I recommend you build new networks rather than attempt to upgrade the stock 3143. The reason is the bi amp switch will hinder any gains and replacing parts in the stock network will be difficult. You will also need to be very careful with all the phasing on the drivers. Again, I have no switchand never did. TIT has been biamped from the time I bought the 2121 midranges. I currently have the unit biamped at 300 cps with a Marchard X1 set (24/octave). I had tried variations of 6 and 12 db/octave with a heathkit active but the 24 seems better. It is not too bad but I wanted to be able to play a little with overlap to see if I could get it a little flatter and also wanted to upgrade resistors, caps and inductors. I already purchased the caps and resistors I think will be good, but am still looking as to how closely I need to match air core to iron core inductor DC resistance. I know it has to be an external series non inductive of 0.I ohm or less. The inductors come within a tenths of an ohm of one another for the 0.16 and 0.25uF but am off by 0.1 ohm in the 1.0 and 1.7. I only have a meter sreadout to the nearest tenth of an ohm so I cannot measure closer than about the nearest 0.05 ohm. I was trying to learn if that is significant or not. The DCR 's range from about 0.2 ohms to 0.5 ohms among the 4 iron core units from JBL but are about 0.1 ohm lower in the air core ribbon inductors I was considering for the 1.7 and 1uF inductors. Also the replacements are 16 gage which is probably a heavier gage than in the JBL, at least by micrometer diameter measurement on the JBL coil wire. What I don't know is how meaningful that diffeerence is. The ones with the big difference have padding of 2 to 4 ohms to back lower the driver efficiency to match the horns in the circuit, not just the L pads The L-pads for the horn and 10 inch are 16 ohm, again stock JBL replacement parts.

If it were me I would write to Giskard nicely and arrange a pair 2122H to replace the 2121 and get new diamond surround aluminium diaphragms for the compression driver and get on the list for V 3 of the 3145 equivalent network. You stand to gain light years more from this approach than merely putting in better parts.
As a side note, I do have a second set of horn drivers I got a while back. They are the pro designator units with the diamond surrounds, not visually similar to the LE85's. Instead of big and black cylinder shaped magnetic assemblies, they are gray and angular in the magetic region and I think they had the rubber rings on them. I have them in a second set of cabinets with another set of 15's for the TV and used the old longer horns and the L200 JBL crossovers. Should I be swapping out the magnetic compression drivers??? Rightly or wrongly I always thought the design change was for use of the ferrite magnets since alnico became tough to get due to commerce on the nickel from Africa because of so much political turmoil. Not as good, but available and with some design changes could simulate the quality of the alnicos.
I am really trying to avoid changing out the 10 inch drivers. If I do that, it will set me on another project looking for a way to use them somewhere. I am one of those nuts who does not like to see things go to waste.
What is starting to worry me is that using the 24/0ctave instead of the passive network or JBL 12/0ctave active unit may cause a problem. I know they like 12/octave with flipped polarity. Second order seems to be a standard for a lot of pro gear so I thought that is the reason they used it and 24 should be better, but if they have something else in their active circuit that is designed to overcome some impedance peaks in the drivers, using the 24 in standard polarity could be a problem. I don't know how to search the forums easily. I get moved around to a lot of different threads and wind up reading them oll only to find I learned something but not what I needed. I already have my new caps and resistors in hand (and paid for) so I would hate to start over. The caps were not like I bought Mundorf Silver gold or the like (I got Clarity Caps and 1% Dayton 0.1uF polypropllene and some 0.009uF polystyrenes for jumpers) but still I have a couple of hundred bucks in them. I really don't want to start over. I still have another Nak ZX7 and 682 ZX to get fixed as well as a neat McIntosh 240 amp to have repaired. I just retired so I am not broke but have to be a little cautious. Time for new windows this spring too! In addition, my earing in the HF range is lousy at my age. I think it is down about 80 dB above about 8k. Even have trouble hearing vocal sybillants. Stuck an 075 bullet in my center speaker for the TV so I can understand speach.

Now that I bled for a while, can you tell me two things (I know, all tell me to go a different route and I cannot seem to hear them)

1) how close do DCR's have to match if I go to air cores? Should I add some 0.1 ohm series resistors to the air cores to better match them? I cannot get much closer than around 0.05 ohms since I don't have a meter that reads past 0.1 ohm.

2) with a 24/octave between the 10 and 15, is there a phasing/time alignment problem using the stock JBL baffle layout?

As a side note I had considered inserting a second active and using the McIntosh for one or both horns. Just taking this a step at a time for now though.
And Ian, thanks for the help. If you get a chance take a peek at http://sheneskillies.com My other bigger hobby.



Ian

[Ian Mackenzie]

Hi Bill,

Nice web page.

Is it possible to post a pic of what you have there?

On your questions I dont think its overly critical on the Dcr. I wil sumulate the voltage drives when i get a moment.

The 24 db LR slope active filter might be the problem. Try fliping the phase polarity of the woofer and see what happens. The LR 24 slopes will not give you what they claim in this scenario anyway. This is because of the acoustic shifts on the phase relationships of the driver passbands where they act as a bandpass filter ie (The bandwith of the 2121.) Secondly the low pass second of the 2121 does a 180 degree phase shift.

On the proviso nothing else is wrong adding the 10 inch should do wonders for the sound not make it worse.

[Ian Mackenzie]

I have been in contact (voice) with Bill after some delay with time zones.

Sorry I got way laid on booking for the call Bill.;) .I got dragged out dinner and then to see Bobby (movie). The hour was late at nights end.

I have made some recommendations so Bill can get his 4343 project up to speed. Diy projects don't always go first pop but that is half the fun, you get to figure it out..with some help. :)

Ian

[Bill Shenefelt]

First a big thank you for a phone call half way round the world from Ian. So far in this project I have purchased Clarity Caps as replacements for the old JBL capacitors. So all know and I don't get accused of ruining a classic, I never had an enclosed set of JBL speakers. These have been assembled driver by driver and part by part from replacement parts I bought from JBL. Cabinets are all home made using high density 3/4 and 1 inch thick cabinet grade flakeboard with 2" by 2" hardwood bracing. Volumes and any porting was done to JBL specifications and I did remeasure theil small parameters for the 136A bass drivers which have thru the years been reconed as well as refoamed. Bass cabinets are currently tuned to 27 cps. The latest foams were from Orange County and were supposed to be JBL foams. Same for the 10 inch midbass 2121 drivers.
Here is where I now stand. I am assembling the passive crossovers using the new Clarity Caps (SA) with no jumpers (Per two different high end shop recommendations- something about not messing with time constannts of different caps by inserting jumpers) Some of the caps do however require parallelling of clarity caps to get the values of the 13.5 and the 4uF to be correct. I am going to purchase a BK meter to measure values for DCR in the inductors and mH in the inductors and capacitance to try to hit the JBL crossover design as best I can. I have a Marchand 3 way active 24dB/octave crossover which first I plan to use only as a 2 way at the 300 cps 15 to 10 inch driver interface (bought the 3 way for later versitility). I also plan to try to bring the 10 inch enclosure volume as close to the 0.5 cubic ft internal volume used by JBL and to somehow widen the baffle to 25 inches as it is on the 4343. Right now it is only about 14 inches wide and that is a problem with the lower output of the 10 inch. I still plan to "free air" mount the horn and slot radiator on a vertical extension of the 10 inch cone baffle.
I do have a 1/3 octave heathkit portable real time analizer with pink noise generator for measuring response. It can be set for 1 dB per led or 3 dB per led on the screen. I can input near or far field speaker output from a calibrated microphone or directly in from two channels which can be a feed of the upper and lower outputs of an active crossover. It will store and can do a comparison by difference of a stored signal reference and a new signal. This means I can compare input from the crossover with output from the drivers.

My first order of business will be to get the system up and running with the new extended baffle, box volume corrections, active 300 cps crossover and duplication of the JBL 3143 network for the 10 inch, horn and slot radiator. The switch and low pass 10 inch components (52uF and 2.8mH ) are not now, and never have been in the system-I did not buy them).
My first concern is as to tolerances for components. What tolerances should I build to for the caps, as well as the inductor mH values and DCR's for the inductors? I would not think the DCR for the 0.25 mH inductor is very significant since it is in series with a 4 ohm, 1% resistor. Right there it has a slop of nearly 0.04 ohms. I am not sure how close the values for the 1.7, 1.0 and 0.16 should be though and if I should correct them using series resistors (assuming I could find them that low in value). All the JBL inductors measure less than 0.6 ohms mesasured by my current fluke meter. The new inductors are air cors and have similar DCR (probably within 0.1 ohms) but I cannot measure more accurately than 0.1 ohm until my new meter arrives.
Second, I can either keep the slot radiator in a vertical array directly above the horn and 10 inch and 15 inch or offset it to one side (presumably the "outside" per some of the information I gathered from the forum.) If I keep the driver to driver center as it is on the JBL baffle, is it better to go vertical or horizontal to the outside? Also were I to add a second active crossover, why is the time alignment to the slot radiator from the horn in need of adjustment more than that of the midrange driver to horn which are obbviously as far apart as the 10 inch to horn? The Marchand can be used to adjust a time offset at each corssover point. To go from 2 way active to 3 way active I would also need to get an old MAc240 tube unit repaired. This is one reason I am persuing only a 2 way active at this point.

Third, There is a ton of good info on the 4343 on this forum but there are so many different forum topics and threads that I have a very difficult time refinding where I saw anything. This is probably why I initially sent private messages to Ian. I did find a great discussion of enclosure and baffle projects but now I don't know where I found it. Also I found a link to a review of different capacitors posted on the site of a gentleman from the Netherlands I think, but now I cannot figure where it was either. I think I have spent more time trying to refind things than I did reading most of the threads from beginning to end and some are pretty long.
Any secrets to storing thread locations. How does one set up the "quick links" thing on the top of the forum page?

In any case hopefully I will be able to re-find this thread to see what responses I may get.
I wish there was a way to email a response to the forum as well as cc the individual without pestering someone with private messages. I prefer private messages, but hate to bother others with them. My email is [email protected]
Again Thanks Ian!

[Robh3606]

Second, I can either keep the slot radiator in a vertical array directly above the horn and 10 inch and 15 inch or offset it to one side

Hello Bill

When I built my 4344's I faced the same decission. If you are going to a 25" wide baffle I would go side by side just like the monitors do. The vertical dispersion on both the 2405 and 2307/08 gets very narrow on the top end of both drivers. That way both drivers can be set at ear level which is something you really need to do with both to get the most HF info out of them. I have my 2405's on the outside but my cabinets are mirror imaged so I can easilly switch things around. Depends on how far apart they are and the toe in. If you mirror image you easilly go either way depending on what you like.



How does one set up the "quick links" thing on the top of the forum page?


I do what you seem to be doing. Read all you can find and then decide on a course of action. I save the threads as HTML I also decided to set-up folders for speakers I planned on building down the road. That way every time a useful tidbit comes up you just drop it into the folder. It has worked well for me so far.

Rob:)

[Bill Shenefelt]

I have been trying to improve my home built versions of the JBL 4343 speaker system. I have purchased a Marchand 24/0ctave active corssover and new decent caps for the passive crossover. I learned from the forum that I will need to widen the baffle for the 10 inch to simulate the 4343 width to prevent a rolloff of the 10 inch. Initially I had a 0.5 cu ft subenclosure(external though) for the 10 inch. When I saw the response was down at the crossover and roloff started somewhat higher than that, I used a program to see what box it liked (JBL recommended 0.25 to 0.5 cu ft and I initially used 0.5) Based on the program I reduced the box size to about 1/4 cubic ft. That did not fix the problem. I was about to take out my sawsall to increase the box volume when I got a responce from Rob.

Rob sent me a program reference today so I could look at box size versus responce for speakers. Low and behold it is the one I used way back when I reduced the box size to extend the low frequency roll off to the 300 cps crossover point.
Now what to do? I was about to increase the box size but I'm not sure why JBL used it in the 4343. Maybe to adjust the internal volume of the enclosure to better fit the 15 inch speaker desires? Maybe to somehow better match the passive part of the 4343 crossover at the 15 to 10 inch interface? Maybe for some efficiency or transient responce or phase correction? Now I don't know what to do. Any thoughts as to what is really the deal with the 2121 subenclosure size and why JBL chose to use the larger of their recommended range?

[Robh3606]

You can look at just the box response. You have the active or passive crossover slope added to that as well. You may also have mutual coupling issues between the 15 and the 10 through the crossover region. All I can say is if your intention is to build a 4343 using the JBL designed crossover you should be using the 4343 midrange box volume. I would not second quess the engineering behind why they used the larger volume however I would agree it's sure nice to understand why. At this point I would just increase the baffle width as it is basicaly painless and you don't end the day with saw dust on the drapes.

Rob:)

[Bill Shenefelt]

I am only using the low pass part (1200 down) of the 2121 JBL passive crossover and using the Marchand at the 300 cps so I guess I should let the box alone unless (or until) I later build a brand new box to replace the widened baffle?


You can look at just the box response. You have the active or passive crossover slope added to that as well. You may also have mutual coupling issues between the 15 and the 10 through the crossover region. All I can say is if your intention is to build a 4343 using the JBL designed crossover you should be using the 4343 midrange box volume. I would not second quess the engineering behind why they used the larger volume however I would agree it's sure nice to understand why. At this point I would just increase the baffle width as it is basicaly painless and you don't end the day with saw dust on the drapes.

Rob:)

[Bill Shenefelt]

It would not be a big thing to add some internal volume to the box if that in some way would help the transient response which is a concern I have. I can saw a 9 inch by 9 inch hole in the internal back and put a couple of 6 inch spacers and a new back expanding to near the 1000 cubic inches that way. Just do not want to bother if it does nothing for me. Costs no $ but a new box with new verneer wood, glue and all would be one to two hundred bucks I think.

[Robh3606]

Hello Bill

The 4343 internal volume is .5 cubic ft net so that's what you need. You don't need to add the driver volume to the .5 cubic ft. That means the driver is working into about .4 plus what ever virtual volume is added by the fiberglass.

Rob:)

[Earl K]

Hi Bill


I am only using the low pass part (1200 down) of the 2121 JBL passive crossover and using the Marchand at the 300 cps

- Please confirm; that you included ( or not ) within the bandpass portion of the circuit for the 2121H , the 16 ohm variable Lpad, as well the "T" pad ( which is comprised of a 2R series resistor followed by a 30R shunt resistor followed by another 2R series resistor ).

- Assuming that you did include this 16 ohm variable Lpad, where is it usually set, ( within its' relative rotation of; full CW & full CCW) ?

- The 52 uF cap ( 56 uF in some later systems ) and the 2.8 mH coil appear to form a fairly resonant 2 pole high pass. FWIW; These 2 poles have swapped ( inverted ) their usual relative positions to each other. This inversion is in the "classical" sense of what most of us are used to seeing ( ie; large coil & small cap / this circuit uses ; large cap & small coil ) . This inversion will dramatically effect the damping of the filter.
- A quick guesstimate of some working AC impedances suggests that this filter is fairly resonant ( higher Q or lower damping than is typical ) in the 300 to 400 hz area. This "resonance" looks to actually give a bit a lower midrange response boost. This would make sense considering the very low total "Q" of this driver ( 2121H )
- As a result , I would include these 2 passive elements in the bandpass circuit for your 10" (or else have Marchand design a custom card for use in your XM-44 crossover by recreating the transfer function of these 2 omitted passive elements ) .

- It would be handy if someone with the LEAP crossover software ( or a decent Spice program ) would figure out what is really going on in this part of the 3143 network . I see a fairly resonant circuit / but then I don't have this actual 10" to test / plus I have no SPICE program / just a calculator and some formulas .


so I guess I should let the box alone unless (or until) I later build a brand new box to replace the widened baffle?

- I blew up the photos of your old avatar , as well as your new one.
- From these 2 very grainy photos , it looks to me that your 2121H is housed in a "ported" 1.0 to 1.5 cu' enclosure .
- Am I just seeing things that don't exist / or ???
- Do you really have a .5 cu' doghouse there ( maybe it's created by a partition ) ?

Anyways, moving forwards ;
- Increase the size of the baffle board as has been discussed / as well as recreate the orginal component spacing as has been discussed . They won't fix all the problems that have resulted from your deviation from JBLs' original design / but they are a step in the right direction .
- Then do something about including the 2 missing passive elements in your crossover .


:)

[Bill Shenefelt]

[quote=Earl K;160276]Hi Bill



- Please confirm; that you included ( or not ) within the bandpass portion of the circuit for the 2121H , the 16 ohm variable Lpad, as well the "T" pad ( which is comprised of a 2R series resistor followed by a 30R shunt resistor followed by another 2R series resistor ).
YES

- Assuming that you did include this 16 ohm variable Lpad, where is it usually set, ( within the relative rotation of ; full on & full off ) ?
Midpoint to slightly higher, say "2 oclock"

- The 52 uF cap ( 56 uF in some later systems ) and the 2.8 mH coil appear to form a fairly resonant 2 pole high pass. FWIW; These 2 poles have swapped ( inverted ) their usual relative positions to each other. This inversion is in the "classical" sense of what most of us are used to seeing ( ie; large coil & small cap / this circuit uses ; large cap & small coil ) . This inversion will dramatically effect the damping of the filter.
- A quick guesstimate of some working AC impedances suggests that this filter is fairly resonant ( higher Q or lower damping than is typical ) in the 300 to 400 hz area. This "resonance" looks to actually give a bit a lower midrange response boost. This would make sense considering the very low total "Q" of this driver ( 2121H )
- As a result , I would include these 2 passive elements in the bandpass circuit for your 10" (or else have Marchand design a custom card for use in your XM-44 crossover by recreating the transfer function of these 2 omitted passive elements ) .
I could give the 15 inch a 400 cps crossover and keep the 10 inch at 300 as a means of boost there maybe

- It would be handy if someone with the LEAP crossover software ( or a decent Spice program ) would figure out what is really going on in this part of the 3143 network . I see a fairly resonant circuit / but then I don't have this actual 10" to test / plus I have no SPICE program / just a calculator and some formulas .



- I blew up the photos of your old avatar , as well as your new one.
- From these 2 very grainy photos , it looks to me that your 2121H is housed in a "ported" 1.0 to 1.5 cu' enclosure .
NO it is sealed and about 1/4 cuic ft.
http://www.sheneskillies.com/speakers/frontview.jpg
The box itself is about 16 inches deep to sit on top of the bass enxlousre but there is a rear face of the doghouse installed about 6 inches back from the front baffle limiting the size to 1/4 cu ft.
- Am I just seeing things that don't exist / or ???
- Do you really have a .5 cu' doghouse there ( maybe it's created by a partition ) ? Yes

Anyways, moving forwards ;
- Increase the size of the baffle board as has been discussed / as well as recreate the orginal component spacing as has been discussed . They won't fix all the problems that have resulted from your deviation from JBLs' original design / but they are a step in the right direction .
- Then do something about including the 2 missing passive elements in your crossover .
What do I do bypass them inserting the signal between the 52uF cap and the 1.7 ohm inductor? I will be feeding the highpass of the Marchand. Somehow I don't think using the 12/0ctave JBL active for biamp does that since they switch out the big inductor and cap. Also since it is a general active not specifically tayiored for eht 4343 I doubet they do anything unusual.

[Earl K]

What do I do bypass them inserting the signal between the 52uF cap and the 1.7 ohm inductor? I will be feeding the highpass of the Marchand.

- As I suggested, feed the eletronically crossed over signal into a Hipass section that would now include either a 52 or 56 uF cap, and 2.8 mH coil // or // have Marchand build the equivalent 2-pole HP with the proper "Q" and omit these 2 passive elements ( the 52 uF cap & 2.8 mH coil ) .


Somehow I don't think using the 12/0ctave JBL active for biamp does that since they switch out the big inductor and cap. Also since it is a general active not specifically tayiored for eht 4343 I doubet they do anything unusual.

- Doubt if you must, but ;

- Typically JBL "did fiddle" with the Q in the custom electronic filter cards that JBL sold for specific Studio Monitors ( these cards were specified for use in their 5234/5 crossovers ) .

- Even the Everest II ( DD66000 ) recommends a fairly high Q for the HiPass filter / when biAmping with an external crossover .
- Look at the included .jpg to see just what EQ ( & group delay ) JBL is adding to the horn circuit .

- FWIW ; the pictured voltage drives for the horn circuit in the Everest II, are a result of inverted relative positions for the 2 poles within the passives' High Pass Filter ( similar to what is found in midbass portion of the N3143, ie; large cap, small coil ) .


* :)

[Bill Shenefelt]

-
- Typically JBL "did fiddle" with the Q in the custom electronic filter cards that JBL sold for specific Studio Monitors ( these cards were specified for use in their 5234/5 crossovers ) .


* :)
Thanks for the additional information. What I can do is increase my baffle, stuff the enclosure and try the current crossover with the active and no 52 cap network parts and look at what I get in the room using a little 1/3 octave RTA I have. I know that Marchand can provide secondary cards for " baffle step compensation, notch/boost filters, delay sections and bass boost filter (Linkwitz transform)". In addition the cards are independant for high and low pass slopes and filters. Once I see what the response looks like I could check with him to see what might be best. Thinking about it when I first got the 10 inch drivers I bought "the JBL crossover card" for their active crossover which was to be used in the 2121 high pass network. It was a single card with a 12dB/octave slope but I did not know it ahd some other factors built into it. Also maybe some others on the Forum that have done the biamp that know what additional features the card must possess. I don't want to add the big cap and inductor since that is what everyone on the forum discussing the 4343 wants to remove by the biamp setup since it also impacts the higher drivers as they are in its circuit also. Thanks for the info.

[Earl K]

I could give the 15 inch a 400 cps crossover and keep the 10 inch at 300 as a means of boost there maybe.

- Yes, you should try that out scenario. Give it a listen & Post your impressions .
- This won't be the same as using a HP filter that has the same Q as the original passive / but it's certainly worth a listen. It may get you close enough to a pleasing result, to call it a day .
- I'd also recommend measuring this scenario with your RTA for future reference .

( If using 24 db, 4-pole network cards , start out by wiring the woofer and 10" midrange with identical polarities / this should offer the best summing throughout the crossover region )


:)

[Earl K]

I don't want to add the big cap and inductor since that is what everyone on the forum discussing the 4343 wants to remove by the biamp setup since it also impacts the higher drivers as they are in its circuit also.

- Yes, I can understand that sentiment / plus / the cost of passive componets at these sizes is quite a bit more than another 2, XM44 crossover cards .
- Still the problem remains, that the 4343 network ( & 2121H ) was designed to offer more than just simple crossover filtering . My quick analysis indicates that the high Q of the HP on the 10" is there by design. High Q also equates to more group delay in a filter. Group Delay is just like it sounds / a select area of frequencies are delayed ( due to the resonant nature of the filter ). Delaying the 10" ( in the crossover region ) relative to the horn circuit ( and maybe the woofer ) is likely a good thing .

- FWIW: You could add these 2 passives to just that portion of the network fltering the 10" midbass . This approach will maintain your existing design deviation ( which you're apparently enjoying ) and still remain true to the original design objectives of the network ( somewhat an assumption on my part ) .

- If I was a paying customer of Marchand Electronics ( like you ) I would email the principal ( Phil ?? ) and ask if he will design a custom set of 4343b cards for your XM44 crossover to work with the your fourways . Provide him with the N3143 schematic / some specific load impedance info, / & / he should be able to design a pair of 2 pole cards that recreate the same electrical transfer functions ( as found in the N3143 ). He may also be able to approximate those transfer functions into 4 pole network cards. That would be a real bonus if you could maintain absolute polarities . You'll likely need to offer to pay for his design time ..

:)

[Robh3606]

Why JBL put in the 52uf cap in series with the HF and UHF


About the 52uf capacitors I have sent and email to a higher level for qualification. From both the technical and historical viewpoint we really need to have this correctly documented on the Forums.

Apparently the 52 uF capacitor was used as protection if the system were to be switched to Bi-Amp as amplifiers from that day made a turn-on transient that could damage the compression driver diaphragms. In more recent times amps are well enough behaved that that part is no longer necessary.

Ian

Rob:)

[Earl K]

- Rob, why don't you fire up your LEAP software package and help out here ?

- Will LEAP tell you the "Q" of the resulting filter ( when one combines a 52 uF cap with a 2.8 mH coil ) ?
- Will LEAP tell you the resulting Group Delay ( in ms ) for the range of affected frequencies within the crossover region ?
- FWIW ; I use 12.5 ohms ( or vary 12R to 13R ) as the combined load impedance ( of 2121H driver & the 16 ohm Lpad & that T-pad in place ).

:)

[Earl K]

- Bill, if you closely read the underlined section you'll see it has some applicability to your initial observations & complaints .

- Your XM44 filters are likely the same transform as the quoted info for the XM1 .

- As such, I would be giving standard 24db, 4-pole Butterworth filters an audition. A pair of these HP Butterworth filters should be a no brainer for Marchand to make up. ( Butterworth filters have probably the best power transfer throughout the crossosver region / though they have their well publized drawbacks ).

- Picking a filter type is a lot like choosing your poison . :)

[Robh3606]

Hello Earl

Will it tell the Q??? Don't know at this point in the game. Simply not enough experience with it. Looking at it quickly I don't think it will however it will easilly give you Group Delay. Here's a quick look at those components with a 12.5 load.

Rob:)

[Earl K]

Thanks Rob !

- Though ( thinking outloud ) without the LP portion included into a full band-pass FR / I'm sure the overall idea one gets ( of resonant boost, in relative db ) is a bit skewed .


:)

[Robh3606]

Though ( thinking outloud ) without the LP portion included into a full band-pass FR / I'm sure the overall idea one gets ( of resonant boost, in relative db ) is a bit skewed .


:)


I think you are right. If I get a chance latter I will load up the LF portion with a resistive load sum them and see.

Rob:)

[Bill Shenefelt]

I think you are right. If I get a chance latter I will load up the LF portion with a resistive load sum them and see.

Rob:)

This is getting very interesting but a bit beyond me. Once I see what comes out, I will forward whatever you guys can recommend for him to make to Marchand to get some cards. In the interim I will reinstall the drivers on tuesday (in the existing 0.25 box with the baffle extensions) On Tuesday when the RCL meter arrives I can re size the 0.25mH coil (now a 0.27 of unknown match in resistance to my unknown resistance JBL iron cores) in the assembled passive and try the 300/300 and 300/400 active card combinations and take some photos of the RTA readings of each and post them.

[Earl K]

Thinking about it when I first got the 10 inch drivers I bought "the JBL crossover card" for their active crossover which was to be used in the 2121 high pass network. It was a single card with a 12dB/octave slope but I did not know it ahd some other factors built into it. Also maybe some others on the Forum that have done the biamp that know what additional features the card must possess.

- This 4343 specific card has the Hipass ( HP ) & Lowpass ( LP ) filters built into one card. You need 2 cards to achieve 2 channels of biamped signal . The card(s) work in JBLs' 5233 ( mono model ), 5234, 5234a, or 5235 stereo crossovers .

- Anyways, where are these cards now ?


Once I see what comes out, I will forward whatever you guys can recommend for him to make to Marchand to get some cards.

- You would be much further ahead if you could locate your old cards & let Marchand reverse engineer the voltage drives built into them. This would be simplistic for Marchand to do / what with the schematics for the just mentioned crossovers being available in pdf form, on JBLs' site . The cards' trace layout and pinouts , can be found on this site ( I just saw them recently ) .

- Waiting for a meaningful recommendation from this thread might mean a long wait . This is more complicated than you think.
- For instance, though Rob & I have shown that the HP portion for the 10" in the N3143 has some amount of "boost" around the crossover region , I doubt that the 2 passive elements actually combine to give a full 6 db of gain.
- The simulation does show 6 db of resonance ( gain ) / but a simulation is not the real world.
- Real world passive components are flawed devices. The "flaws" end up mitigating the amount of resonant boost . The gain is going to be less than 6 db / likely less than half that amount .

- So, the only way to really know how much gain comes from these passive circuits is to actually build the passive elements into a real life network and then measure the results / or / have someone with a stock N3143 network measure the voltage drives ( an 8 ohm dummy load should suffice for now ).

- Any HP boost built into a custom card must still mirror the real world voltage drives of the passive network ( which we don't know ).



:)

[Earl K]

- The simulation does show 6 db of resonance ( gain ) / but a simulation is not the real world.
- Real world passive components are flawed devices. The "flaws" end up mitigating the amount of resonant boost . The gain is going to be less than 6 db / likely less than half that amount .

- Well , there's nothing like real-world bench-testing . :p

- Here are some real world results ;

(i) I was able to achieve 3.6 db of gain ( resonance ).
(ii) A Loftech db "counter" gave me a real time, db reading. All the Loftech db readings were double-checked. This was done by using a measurement of Voltage Differentials, that were then converted into a 20 Log Scale .
(iii) Overall; a 1 volt reference , with a HP filter comprised of a 54 uF cap and a 2.6 mH coil . The load used was 11.9 ohms . The 54 uF cap was a combo of a single 10 uF MPP Solen cap combined with 2, 22 uF old PIO caps . The coil was composed of two, 16 gauge / 1.3 mH coils in series with each other ( not stacked ) .
(iv) Overall, the quality ( higher Q ) of these components vs the older JBL stuff is most likely adding some extra bit of gain .




- So, the only way to really know how much gain comes from these passive circuits is to actually build the passive elements into a real life network and then measure the results / or / have someone with a stock N3143 network measure the voltage drives ( an 8 ohm dummy load should suffice for now ).

(i) I standby those remarks / a real JBL N3143 network needs to be measured to see what gain its' components actually give ( in the 10" portion of the passband ) .

(ii) I don't have any iron core coils near the 2.8 mH value . I also don't have 52 uF +, of mylar caps around here. Therefore, I can't really duplicate JBLs' quality in vintage parts ( to see what the resonant gain might be ) . I can however, add a 1 ohm resistor inline with my 2.6 mH coil / this results in approx. 2.6 db ( vs 3.6 db ) of resonant gain .


:)

[Earl K]

Hi Bill

After sleuthing about for the last couple of days I found these ;

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=885&stc=1&d=1063675831

and Giskards' voltage drives for the special 4343 / 4350 crossover cards ;


http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=886&stc=1&d=1063676252

What does this mean to you ( Bill S. ) ?

(i) You can stop looking for your 4343 crossover cards for Marchand Electronics to reverse engineer . That's the good news .

(ii) The bad news ? After looking at the voltage dives for the card in question, it's apparent to me that the stock card will just make your situation worse ( the lousy coupling between your 15" & 10" ) . Therefore , this card shouldn't be used as a template for Marchand to copy / unless one is dealing with a bonafide 4343 cabinet and its' stock layout of components ( & all that implies ) .

The above pics came from Widgets' 4355/3155 Clone (http://audioheritage.csdco.com/vbulletin/showthread.php?p=6343&highlight=3144) thread . It's a good read that has a bunch of applicable info ( in reverse ) to your situation .

:)

[Bill Shenefelt]

Thanks Earl. I have not yet enlarged the boxes but am prepared to do so if I get a hump, rather than the current dip in response after adding the extra baffle width. That box enlargement is supposed to cause a dropoff of the output per the WinISD program peredictions of box response which is why I made them smaller way back when I first encountered the dropoff. I have made a couple of wings to widen the baffle, raised the cabinets up from the floor a little and turned the bass cabinets 180 degrees to get the 15 closer to the 10 (within a quarter inch of that on the JBL baffle) and further from the floor. Tomorrow I receive my B&K RCL meter and once I set the coils and check the DCR I will be able to start measuring with the RTA. I still have the option to overlap the crossovers or spread them about 100 cps with on hand cards. (300 lhigh with 400 low or 400 low with 300 hi pass or same on both sides). This should give me an idea as to which direction is best to proceed or if I get lucky I will have a very close pair of cards already. Also the Marchand I have can be modified with a secondary card(S) in each channel to give boost, cut or delay if that appears desirable. I think the wider baffle alone, assuming it does give me back a 3 dB loss from 300cps tapering to flat at 700 , may fit the bill with little modification. Before I go and enlarge boxes I need to see where I am. Maybe a partial enlargement rather than going the whole way to 1/2 cubic ft will be indicated. I doubt I will get done tomorrow but should at least have some measurements later this week and post what I learned. Not sure about driver phase in the region but the 24/octave there should be ok for using no polarity reversal. I did remember that the JBL card was designed to give a 12dB/octave slope on both sides and that makes it clear why they reverse polarity between the 15 and 10 inch. I think JBL also liked to be down 6, not 3 db at the crossover point. Something about power versus voltage summing. (I did not have their crossover but bought only one card to see what it might be doing. Looked like a simple two pole.) I'm not sure what they do with the horn versus the 10 inch offset in the passive though. With the 10 inch being in a different detached box it can be moved back slightly to align the voice coils with the 15s. The horns move right with the 10s so if JBL did do something in the passive network to adjust for the horn to 10 inch voice coil offset I won't be disturbing what they did by moving the upper box. Were it desirable to play with phase there, I bought the three way crossover to fiddle with driver offset if need be. Would rather not do that since I would need to get another amp up and running. Messing with that area is more or less acedemic since my hearing is not very good above about 5K and I actually have a tough time with speach due to not hearing sibalants well. It's the principle of the thing. Just like getting my Nakamichi decks up and running when I use DAT for most recording anyway. Hate to see things not right. I will be 64 this summer and just retired recently and helath is not great but I'm at least having fun. As a physics/mechanical engineer major I know just enough to get me in trouble I guess. Love my tropical fish, auto mechanics and audio video stuff. Spend more time messing than using though I'm afraid. Just got to have fun. Again thanks for all the help. If you ever get into tropical fish just let me know and maybe I can help you.

[Ian Mackenzie]

Bill,

It might be an idea to model witht the software what you appear to be measuring on the RTA. Floor relfections and other issies can easily distort measurement data from being useful to meaning less if you are not able top verify this accuracy of the test proceedure.

[Earl K]

Hi Bill


I have not yet enlarged the boxes but am prepared to do so if I get a hump, rather than the current dip in response after adding the extra baffle width. That box enlargement is supposed to cause a dropoff of the output per the WinISD program peredictions of box response which is why I made them smaller way back when I first encountered the dropoff.

- Sometimes I have a difficult time following the flow of peoples' postings' ( & keeping facts straight, so please bear with me ) but your above statement does somewhat confuse me .

(a) Are you using .25 cu' subenclosures ( for the 2121 ) because you obtain a more linear level in the 250 to 1000 hz range ?
(b) If this is the existing situation, have you measured ( RTAed ) the performance of the 2121 in the .25 cu' vs .5 cu' ( subenclosures ) to verify any simulation that you might be working from ?
(c) Remember, JBL always had the option of sticking the 2121 in a smaller subenclosure if they felt that was the best approach to deal with transition issues ( in the stock 4343 ) . Since they didn't do this, you need to be asking yourself why they weren't precient enough to follow this path . ;)



I have made a couple of wings to widen the baffle, raised the cabinets up from the floor a little and turned the bass cabinets 180 degrees to get the 15 closer to the 10 (within a quarter inch of that on the JBL baffle) and further from the floor.

(d) Something that I don't believe I previously mentioned are; destructive diffraction effects. These could be occuring around that symmetrically dimensioned doghouse.
-To test for them, simply place couch cushions on either side of the mids baffle ( sitting on the woofer box ). If you can balance the cushions on their ends and flush them up to both enclosures baffle surfaces / all the better / if not, simply stack more either side until they start to fall off the woofer boxes .
- Large 18" x 18" sofa cushions can become a quasi-standin for your new, fixed wood baffle extensions .
- Offhand I don't know a workable diffraction formula that would help predict the deleterious diffraction effects ( though I could probably invent one to support my bias :p )
- If severe diffraction cancellations are falling within the 100 to 400 hz range / then they'll be also causing a hole in response ( throughout that crossover region for the 15" to 10" ) .

:)

[Bill Shenefelt]

I got my LCR meter today. Seems my new inductors are actually closer to the JBL crossover schematic than JBL sold me as replacement parts. Their 0.25 inductors were 0.32 and 0.31 My new ones are 0.27 without doing any unwinding. So even with the DCR a hair off ( 0.306 on theirs and 0.276 ohms on the new ones) they should work well I guess. The 0.16 mH inductors were a closer match in DCR and to the slightly undersize JBL parts (0.156 JBL 0.154mH new ones). The 13.5 uF cap of JBL was slightly over 14 uF but maybe age is some sort of factor or just tolerance.
I should get all assembled tonight for some trial runs tomorrow with and without the baffles. I can also use some big sofa pillows as suggested to minimize room interaction. Also moving around a bit should give me some better ranges. I did make some baffle extensions but out of simple pine boards. Hopefully not being solidly fixed in place ro as stiff as I would use for a real baffle they will at least let me know how things might work. They extend upward and include additional width alongside the horns. Just do not reach the slot radiators.
The material I always used for speaker cabinets was called cabinet grade high density particle board. Heavier than sin. Really eats carbide blades. I used 7/8 inch mostly and some 1 1/8 on an 18 inch JBL sub enclosure. All braced with hardwood 2" by 2" glued and screwed. I may have to look around to find new high density particle board like that though. I did find that my local verneer guy is stilll in business (or at least the business is still listed) and that is a plus. Already have a plunge router but no circle fixtures. I just got close to a drawn circle and sanded to fit from there. Had to get pretty close though as I was using the JBL speaker mounting holes and the little threaded push in sleeve nuts. They come awfully close to the driver hole so you cannot be far off. Even have to grind the base of them to clear a round JBL driver as the base extends to the holes.
I will see how things look and how it sounds tomorrow and maybe open up the back of one the cabinets for more volume expanding to the 0.5 cubic ft till I see what it produces. If all goes well and I decide to go with new cabinets I will only build an upper cabinet though to hold and baffle the 2121 in a 0.5 cu ft enclosure and mount the horn and slot radiator. My 15 inch cabinets are too good to trash. (too heavy to move also). I did raise the main cabinets off the floor so the 15 inch driver is much higher. I feel sure that will get me back toward the sound I was able to get from the L300. Flipping the bass cabinet to bring the 15 closer to the 10 and adding a 2x4 box frame under the cabinet raises it from 2.5 inches to 10.0 inches from the floor. (add 7.5 inches for driver center location). It does put the port closer to the floor however. My face driver locations are not exactly those of the 4343. The baffle I see posted had a 10 to 15 distance center to center of about 14 inches. Mine will be 15 7/8 inches. I cannot get closer due to two boxes with internal bracing restricting how close to the edge a driver will mount to the edbge of a box. Th horn to 2121 distance is right on though.

[jandregg]

When the 10 inch and the 15 inch are in seperate boxes mutual coupling between the two speakers can be broken by physically offsetting the speakers cones front to rear. Move the box containing the 10 inch back by one to three inches. This will break the mutual coupling and help time align the voice coils.

John

[Bill Shenefelt]

That is partially why I put them in separate boxes. To see if I could get better "time alignment" It also let me shorten the height of the 15 inch driver enclosure and keep the 5.5 cu ft volume. Besides I like the looks of the horn and slot radiator mounted in the open. I have the passive crossovers done so now I can start to test things.

[Earl K]

Hi John,


When the 10 inch and the 15 inch are in seperate boxes mutual coupling between the two speakers can be broken by physically offsetting the speakers cones front to rear. Move the box containing the 10 inch back by one to three inches. This will break the mutual coupling and help time align the voice coils.

- Some of us believe that Bills' 4-way effort suffers from having "not enough coupling" between the 15" & 10 " .
- Due to your above post, I'm not sure where your opinion sits on this question, do you have a contrary POV that you want to present ?
- Bill Shenefelts' first post (http://audioheritage.csdco.com/vbulletin/showpost.php?p=156479&postcount=5)


:)

[Bill Shenefelt]

So far looks promising. Have a nasty 60 cps hum that I have to find but with the wings it looks good. Bass is a little high on this shot but at least gives an idea as to what I am seeing. The scale is one LED is 3dB. Ignore the missing stuff at 1.4 and 2k. Just assume a dropoff to the next column. Sometimes contacts inside get loose and you lose a coulum. http://www.sheneskillies.com/speakers/with%20wings.jpg

[Bill Shenefelt]

Finally got one of the wto crossovers done. I kept the pots rather than go to fixed resistors since I really like the versitility. Not sure if I should have gone with a better grade wiring, but with my other equipment I thought plain old 16 gage monster cable would do the job. Tough to fit all the stuff in a "logical layout, but here is what I came up with.

http://www.sheneskillies.com/speakers/crossover.jpg

[boputnam]

Have a nasty 60 cps hum that I have to find...I'm not seeing it on that RTA shot...


Ignore the missing stuff at 1.4 and 2k. Just assume a dropoff to the next column. Sometimes contacts inside get loose and you lose a coulum. :hmm:

[Earl K]

Hi Bill ,

So far looks promising.

And now ? ( what with all the recent changes , etc. ) how are things sounding ?

:)

[Bill Shenefelt]

Hi Bill ,


And now ? ( what with all the recent changes , etc. ) how are things sounding ?

:)

http://www.sheneskillies.com/speakers/response.jpg
REsponse is still down in the 3-400 cps region. Yet they add up alright if you look at separates.
\http://www.sheneskillies.com/speakers/2121rolloff.jpg
\
\http://www.sheneskillies.com/speakers/Marchand--15-inchrolloff.jpg
\
\2121 with no Marchand the 2121 seems to go relatively low see below
\http://www.sheneskillies.com/speakers/no-Marchand.jpghttp://www.sheneskillies.com/speakers/no-marchand.jpg
\The wings seem to do nothing really. The 15 inch can go much higher (was 800 in the L300) and the 2121 is good to about 200 with no crossover. Should work. I checked it at different points in the room to look for maybe some standing null but all were roughly the same. I wrote to Marchand to ask if I should raise the 15 inch driver crossover to maybe 400 cps or maybe add a 3 to 5 dB bost on the HF between maybe 300 and 500 ond if there was an add on card that might do this. Have not heard back yet. The 2121 just seems to quit with the 15 in there. I tried reversing polarity but get even worse with the feeds out of phase. What is the wavelength of the 300 cps wave. could I be just "slightly " out of phase and getting non summing. I can get a card to delay the 2121 up if that would help or I could move the upper box back. The sound in phase is decent but out of phase does not sound coherent but like two separate sources so I think I mam doing that right. More apparent audibly than o the RTA. I have a card set I can build to roll off the 15 inch at 400 cps. Maybe that will fill the hole. I only have one crossover built (got thrown off the diningroom table) so cannot do any serious listening. I want to feed the crossover directly to the RTA to see how that sums. The wavelenght at 300 cps is about 3.75 ft so anything less than about 6 inches out of line should not be major, right? Voice coil planes should not be different enough to do this should they?

[Earl K]

Hi Bill,

- I'm assuming you're still using the 3db/per led scale on this RTA .

- The "hole" at @ 350 hz is very deep & quite wide . 9 db down at 355 hz and 6 db down at 500 hz. Yes, I'd agree that this looks bad and can't sound right .

- Have you checked to make sure this hole is not from reflected, acoustic "floor bounce" ( coming back out of phase to the mic ) ? Floor bounce , is smoothness/hardness effected . The acoustic wave reflected from any hard surface , returns at a pure 180° phase difference . Measured cancellation is quite specific to wavelength to the measuring mic and the distance of woofer cone from the reflective surface . To predict the effects entails triangulation of these specifics.

(i) Concrete ( and some hardwood floors ) give very pronounced artifacts.
(ii) Move your measuring mic down to within a 1/4" of the floor and see what you sort of FR you get down there.
(iii) Alternately, try piling up a few of sofa pillows on the floor between you and the measuring mic.

- Good speaker designers take some "reflected bounce" into consideration when deciding on the initial layout of the components .


The wings seem to do nothing really. The 15 inch can go much higher (was 800 in the L300) and the 2121 is good to about 200 with no crossover. Should work.
- If the problem is floor cancellation, the wings won't help / you'll need to find other solutions .


I checked it at different points in the room to look for maybe some standing null but all were roughly the same.

- If the measuring mic was at a constant height from the floor / the problem ( as you are measuring it ) could still be explained by destructive floor bounce .


I wrote to Marchand to ask if I should raise the 15 inch driver crossover to maybe 400 cps or maybe add a 3 to 5 dB bost on the HF between maybe 300 and 500 ond if there was an add on card that might do this. Have not heard back yet.

- You need to eliminate all the acoustic variables first . ie; Adding boost EQ won't properly overcome a destructive floor bounce scenario .


The 2121 just seems to quit with the 15 in there.

One photo of an RTA measurement that you didn't include was that of the 15" running fullrange. A shot of this would help determine if the 15" is fully capable of contributing its range of needed frequencies through the crossover region.


I tried reversing polarity but get even worse with the feeds out of phase.
- It's good you tried that.



What is the wavelength of the 300 cps wave. could I be just "slightly " out of phase and getting non summing.

One can get proper summing if the wavelengths being summing are within a 0-120° spread. At 350 hz , a 90° offset works out to be roughly 9.7 inches. 120° = 12.9 inches . So you can safely say a voice coil off set of 4" is definately in the safe summing area .



I can get a card to delay the 2121 up if that would help or I could move the upper box back. The sound in phase is decent but out of phase does not sound coherent but like two separate sources so I think I mam doing that right. More apparent audibly than o the RTA.I have a card set I can build to roll off the 15 inch at 400 cps. Maybe that will fill the hole.

- I'm as curious as anyone as to how the overlap will work out if you take the woofer out to 400 hz .



I only have one crossover built (got thrown off the diningroom table) so cannot do any serious listening. I want to feed the crossover directly to the RTA to see how that sums. The wavelenght at 300 cps is about 3.75 ft so anything less than about 6 inches out of line should not be major, right? Voice coil planes should not be different enough to do this should they?
- You're thinking on this is correct .

:)

[Bill Shenefelt]

I got the second crossover done but will have to take tomorrow off for company/family dinner. I did try the 15 inch 136A just now and here is the graph. Yes they are all 3dB/led. the 1dB scale is too eratic to see much. Usually around a 2 to 3 db fluctuation with the pink noese source.
http://sheneskillies.com/speakers/136Afullrange.jpg

Remember, JBL used this up to an 800-12dB/octave crossover so it is pretty decent. I tried several altitudes and there was some dip in the 350 region but not as big as seen in the other shots. The meter does jump up and down about 3 dB during reading the pink noise. I can also let it "sum max readings" for a minute or two to try to get some more stable reading but that is not always as good as just freezing a reasonable average looking image. This reading was from the same location as the others and the dip is not as pronounced. I think that sort of rules out floor cancellation. Tried different "altitudes, includning on the floor but down there the bass dies somewhat in the lowest ovtave.
Got family coming for dinner tomorros so maybe on Monday I'll have a chance to get the second sp=eaker up and running. Got bot h crossovers done now I just need to remount the 2121 and wire in the speakers.




Hi Bill,

- I'm assuming you're still using the 3db/per led scale on this RTA .

- The "hole" at @ 350 hz is very deep & quite wide . 9 db down at 355 hz and 6 db down at 500 hz. Yes, I'd agree that this looks bad and can't sound right .

- Have you checked to make sure this hole is not from reflected, acoustic "floor bounce" ( coming back out of phase to the mic ) ? Floor bounce , is smoothness/hardness effected . The acoustic wave reflected from any hard surface , returns at a pure 180° phase difference . Measured cancellation is quite specific to wavelength to the measuring mic and the distance of woofer cone from the reflective surface . To predict the effects entails triangulation of these specifics.

(i) Concrete ( and some hardwood floors ) give very pronounced artifacts.
(ii) Move your measuring mic down to within a 1/4" of the floor and see what you sort of FR you get down there.
(iii) Alternately, try piling up a few of sofa pillows on the floor between you and the measuring mic.

- Good speaker designers take some "reflected bounce" into consideration when deciding on the initial layout of the components .


- If the problem is floor cancellation, the wings won't help / you'll need to find other solutions .



- If the measuring mic was at a constant height from the floor / the problem ( as you are measuring it ) could still be explained by destructive floor bounce .



- You need to eliminate all the acoustic variables first . ie; Adding boost EQ won't properly overcome a destructive floor bounce scenario .



One photo of an RTA measurement that you didn't include was that of the 15" running fullrange. A shot of this would help determine if the 15" is fully capable of contributing its range of needed frequencies through the crossover region.


- It's good you tried that.



One can get proper summing if the wavelengths being summing are within a 0-120° spread. At 350 hz , a 90° offset works out to be roughly 9.7 inches. 120° = 12.9 inches . So you can safely say a voice coil off set of 4" is definately in the safe summing area .



- I'm as curious as anyone as to how the overlap will work out if you take the woofer out to 400 hz .


- You're thinking on this is correct .

:)

[boputnam]

- Have you checked to make sure this hole is not from reflected, acoustic "floor bounce" ( coming back out of phase to the mic ) ? Floor bounce , is smoothness/hardness effected . The acoustic wave reflected from any hard surface , returns at a pure 180° phase difference . Measured cancellation is quite specific to wavelength to the measuring mic and the distance of woofer cone from the reflective surface . To predict the effects entails triangulation of these specifics.

(i) Concrete ( and some hardwood floors ) give very pronounced artifacts.
(ii) Move your measuring mic down to within a 1/4" of the floor and see what you sort of FR you get down there.
(iii) Alternately, try piling up a few of sofa pillows on the floor between you and the measuring mic.
If you've done all the data collection/measurements properly, and answered Earl's questions, then...

http://www.sheneskillies.com/speakers/response.jpg
Response is still down in the 3-400 cps region. That looks like a phasing issue at the crossover point, to me.


I tried reversing polarity On which element - the 2121? Try reversing the connections to the 15" 136A...


...but get(s) even worse with the feeds out of phase. ... The sound in phase is decent but out of phase does not sound coherent but like two separate sources... Please describe what "out-of-phase" sounds like. I wonder if the characteristic you are hearing relates to the 2121, and might be less agreable if you reverse the 15" driver, instead...

A quick glance at the 3143 shows the LF is alone, and out-of-phase (electrically) with the other drivers. Is this what you are doing? The 2121 should be phased with the HF and UHF...

But AGAIN, as Earl asks, take a moment and describe the physical conditions under which you are measuring the speaker(s) and crossover performance, the position of the measurement mic, what it is, etc.

(btw - how do you guys imbed images inside your post (rather than as attacments)? I have no idea how you do that...)

[Earl K]

(btw - how do you guys imbed images inside your post (rather than as attacments)? I have no idea how you do that...)

(A) Simple Answer : paste that specific images' URL into your message editor as part of the posting process .

(B) ( the ) How to do This : as an Answer ;

- Find an existing image within an existing thread ( at LH ) that you want to refer to ( actually can be anywhere on the net ) , then ;
- For a PC guy , right click on any existing attachment and open it up in a separate window . A Mac guy can just keep the mouse clicked down without relaesing and a dialog box gives one the option of opening the image in a new window.
- Copy the URL of this new page ( containing the single image ) into the clipboard ( or is "clipboard" just a Mac term ? ).
- Open the "Reply" editor .
- Then within this "Reply to Thread" editor :
- Find the little icon that looks like a "framed mountain with a postage stamp in the top right corner".
- Just as in referring to a URL that you want to make into a link . / Click on this "Image Insertion" icon & paste the URL within the popup window . Once one clicks "OK" ( or whatever ) the URL is pasted into your message .
- Post your message ( or preview it ) .
- Once the URL for the image is present in your response / the image magically appears / no matter where it's hosted ( more or less ) .


:)

[johnaec]

(btw - how do you guys imbed images inside your post (rather than as attacments)? I have no idea how you do that...)They've used the image html function to link to an actual location where the image is on a webpage, (it's the little yellow icon with mountains on it in the editor). Unfortunately, if the webpage later disappears, (as often happens), so does the image. That's why it's recommended to use the "Manage Attachments" to upload them here, so they stay...

Edit: I see Earl beat me to this. :)

John

[boputnam]

Ah, that is what I thought. Thanks guys!

I know on other forums when you upload an image there is an option to imbed it, rather than have it merely attached. This is different...


Unfortunately, if the webpage later disappears, (as often happens), so does the image. That's why it's recommended to use the "Manage Attachments" to upload them here, so they stay...
Ah, then I will maintain the "hard" upload approach. Elsewise the threads get pretty weird when the discussed image goes missing...

[Earl K]

Ah, then I will maintain the "hard" upload approach. Elsewise the threads get pretty weird when the discussed image goes missing...

- My policy is to just link to this sites' existing images .
- To do this it's obvious that one needs to be very familiar with where things are most likely buried / or at least / competent with the use of the Search Engine .


:)

[Bill Shenefelt]

[quote=boputnam;161491]If you've done all the data On which element - the 2121? Try reversing the connections to the 15" 136A...
THe normal use of the stock passive crossover attaches the 136A out of phase and uses a 12dB/octave crossover at 290 or 300 cps. Mine is active andd at 24dB/octive so is in phase. Out of phase it sounds like two separate speakers, but when in phase it appears tough to tell what is coming from the 15 or from the 10. Much more "coherent" sound when in phase.

Please describe what "out-of-phase" sounds like.
A quick glance at the 3143 shows the LF is alone, and out-of-phase (electrically) with the other drivers. Is this what you are doing? The 2121 should be phased with the HF and UHF...All are in phase due to the 24/octave active. I'm not sure what overdlap does though.

But AGAIN, as Earl asks, take a moment and describe the physical conditions under which you are measuring the speaker(s) and crossover performance, the position of the measurement mic, what it is, etc.
About 8 ft away at about the level of the horn to slot radiator. About 4 ft elevation, carpet, drywall and paneling in part of room, sonex on wall behind speakers. The mike is for the analyzer. It can measure and record a direct input pink noise source, store it and give a difference. Pretty flat looking on the 1dB scale with pink noise generated on the computer in Adobe Audition and cut to CD. The RTA unit is only about a $500 unit but that was in pure kit form. A ton of vertical mounted resistors on a couple of levels for filters at 1/3 octave intervals plus IC's for storage and difference comparisons to a stored reference. Took as long to build as my 25 inch color TV back when I got it. Can take line input from left, right, left+right and left- right as well as the microphone. Can store three signals in memory and stroe a fourth as a reference to yeild a difference plot. Has 1/3 or 1 octave scales with 1dB/led or 3dB/led. CAn take slow responce, fast responce or stored peaks over any time interval. Can do a or c weighting or no weighting. Nice little toy.

I got the low pass at 400 cps built today. Initial trials show it helps a good bit. I can get the dip down pretty small. The "wings" seem to help a bit but I may need to "chop them off" lower than the horn since they seem to give it too much boost in certain areas. I got things good enough to move to the 1dB scale. On the 3dB scale I am flat almost within the 3dB. I have to assemble the 2121 in the other cabinet tomorrorow and see what that brings. It is "in a corner" so I don't know what that will do to the midrange. My wall covered with Sonex should help cut the HF reflection down above the 200 to 400 cps region but not much help below that as it is the 3 inch thick accoustical treatment not the 6 inch thick stuff used in anocheic chambers. Meanwhile the 400 cps low pass combined with the 300 cps high pass giving me some overlap looks very promising. Notch is not completely gone but is reduced a lot.

[boputnam]

- My policy is to just link to this sites' existing images .Yea, that works so long as the images are already here...

- - To do this it's obvious that one needs to be ... competent with the use of the Search Engine.Boy-howdy - that would be SO cool...


The normal use of the stock passive crossover attaches the 136A out of phase Yup...

...and uses a 12dB/octave crossover at 290 ... cps. Mine is active and at 24dB/octive so is in phase. Out of phase it sounds like two separate speakers, but when in phase it appears tough to tell what is coming from the 15 or from the 10. Much more "coherent" sound when in phase. Odd - that is not what the RTA suggests, accepting that you are getting reliable acoustic measurements.


I'm not sure what overlap does though.

If out-of-phase, it creates that dip at crossover your RTA is showing you.


The RTA unit is only about a $500 unit but that was in pure kit form. A ton of vertical mounted resistors on a couple of levels for filters at 1/3 octave intervals plus IC's for storage and difference comparisons to a stored reference. Took as long to build as my 25 inch color TV back when I got it. :rotfl: I don't think the RTA is the issue. And while I am not entirely convinced you do NOT have boundary issues, since the RTA dip is at/near the crossover point, my inclination is that you are struggling with a phasing issue at/near the crossover point, and not with a boundary issue.

I cannot/will not argue with what your ears are telling you, but your response curve tells me something quite different. I would not be "moving" crossover points to mitigate the symptom you describe - by that you are minimizing the symptom but perhaps not addressing the problem...

[Ian Mackenzie]

Bill,

If your alternative cards are LR and 24 db slopes they are in theory in phase and the drivers will sum flat at a particular plane and location in space.

This would be at a point centrally between the woofer and mid cone centre poles. (near field measurement)

However, this is all relative to the location of the mic on your RTA.

If you move the mic vertically in front the the woofer and mid cone aside from the dispersion of the drivers which will also effect your readings the polar response will show the vertical dispersion lobes.

These vertical dispersion patterns or lobes are shaped like a clover leaf and will vary with the kind of filter , the number of poles and the filter Q and location of the driver accoustic centres.

Depending on the particular position of the mic you could be reading a null, a peak or a sum flat at the crossover point.

My reference previously to using an alternative measurement system is that you will find it very difficult to define the actual source of the dip you have noticed with what you have available.

If you were to use MLS type measurement with FFtransformation you can create a graphical picture of what is happending from small slice of the magnitude of the sample length. The sample can also be gated to remove the impact of the room in near field measurements.

Here is a link that you may find helpful and interesting
http://pcbunn.cacr.caltech.edu/jjb/airr.html

Julian is a scientist involved with CALTECH. (ie he is not a dumby) This is a simple and inexpensive software program that will run a a home pc with a duplex sound card.

Ian

[Bill Shenefelt]

Hi Bill,

One can get proper summing if the wavelengths being summing are within a 0-120° spread. At 350 hz , a 90° offset works out to be roughly 9.7 inches. 120° = 12.9 inches . So you can safely say a voice coil off set of 4" is definately in the safe summing area .

- I'm as curious as anyone as to how the overlap will work out if you take the woofer out to 400 hz . (good to near 1k cps)

:)

If I am thinking right, a 180 degree phase change at 355 cps would amount to an 18 inch offset. So if I am out of phase (180) at the terminals, moving the upper box back 9 inches should surely tell me within that move if I am at the correct side of the 180 degree phase change. That is, at some point within the 9 inch range it must be either be exactly in or exactly out of phase. With the front flush it is a little better with one terminal connection than with the other but the difference is not so drastic as I think it should be. My upper box is shorter in depth than the woofer box so I would have the ability to shift the upper box back around 4 inches to align the voice coils if I decided to do so. I will do some messing with things today and see what I can find out. Although room effects seem to not be the deciding factor (heavy carpet and similar dip results at the floor) I can try a 355 cps tone to see what it shows with swapping terminals. Considering I am using a 24dB/octave slope, I could be masking phase mismatch with the broadness of the 1/3 octave LED band. I'll get back with some more data this evening.

[Bill Shenefelt]

I did a less than comprehensive read of the referenced article. I don't right now have the hardware but may have a very viable pair of alternatives. I have a Yamaha surround system It uses a pulse ping and its own microphone to check the phase among and within the 5 surround speakers. This sounds very much like what this program is doing. For this I need about 2 times 35 ft of cable to reach the main speakers (both the 15 and 10 inch) with the yamaha. Also I have a nephew who owns some sort of (call it a black box) that works in a similar manner designed solely for checking phase. Pending my doing what I just posted, I can try one or both of these things, the Yamaha, and/or my nephew's phase checking device. Access to either is more of a hassle than what I have proposed and would be a while off, but in the interim I can at least try box shift and single frequency check as posted. I can also check in several verticle locations to see the impact of altitude of the mike. My first thought is that I should be offsetting the 10 inch front to rear with respect to the 15 to get a sum at my listening altitude.

Bill,

If your alternative cards are LR and 24 db slopes they are in theory in phase and the drivers will sum flat at a particular plane and location in space.

This would be at a point centrally between the woofer and mid cone centre poles. (near field measurement)

However, this is all relative to the location of the mic on your RTA.

If you move the mic vertically in front the the woofer and mid cone aside from the dispersion of the drivers which will also effect your readings the polar response will show the vertical dispersion lobes.

These vertical dispersion patterns or lobes are shaped like a clover leaf and will vary with the kind of filter , the number of poles and the filter Q and location of the driver accoustic centres.

Depending on the particular position of the mic you could be reading a null, a peak or a sum flat at the crossover point.

My reference previously to using an alternative measurement system is that you will find it very difficult to define the actual source of the dip you have noticed with what you have available.

If you were to use MLS type measurement with FFtransformation you can create a graphical picture of what is happending from small slice of the magnitude of the sample length. The sample can also be gated to remove the impact of the room in near field measurements.

Here is a link that you may find helpful and interesting
http://pcbunn.cacr.caltech.edu/jjb/airr.html

Julian is a scientist involved with CALTECH. (ie he is not a dumby) This is a simple and inexpensive software program that will run a a home pc with a duplex sound card.

Ian

[Ian Mackenzie]

If your pc has a Sound Blaster card thats about it

The sort of phase relationships you appear to be talking about in terms of absolute phase only come into relevance at higher frequencies where the wavelength is measured in inches, not feet.

The point that I am making is regards polar lobes in the crossover region.

The attached image is very simplified. You need to understand the principle of that in order to appreciate what you are dealing with.

Essentially JBL tweaked the high and low pass filters so that they sum flat on axis vertically in the crossover region. Otherwise you have the sort of problems you are talking about where there is a tilt in the vertically polar response because the accoustic centres (not to be confused with the location of the voice coil ot apex of the cone) are not aligned.

Without giving the game away, based on a first cut of the 3143 voltage drives (ref 8 ohms Lpads full) the spread of 307 hertz low pass and 428 hertz high pass (referenced to -3 db) with 12 db butterworth slopes should sum flat on axis. The -6 db point is 320 as per JBL data. The curves are very close to the target Butterworth slopes for these crossover points meaning there is no rocket science here. They just spread the crossover points to get a flat on axis result.

Therefore you could emulate these voltage drives with tweaking of your active cards and it would work provided the drives are position in a similar manner to the stock 4343.. I have not had time to check but the JBL specified cards probably perform to a similar end result.


The phase of the woofer and mid cone should be reversed.

[Robh3606]

My first thought is that I should be offsetting the 10 inch front to rear with respect to the 15 to get a sum at my listening altitude.

What you need to do is emulate the stock component layout if you are going to use the stock crossover. There was no baffle offset in the 4343 so you should be fine as well without one.

Rob:)

[Earl K]

Hi


I had tried to build the JBL 3143 crossover for my 4343 speaker sets using components from JBL and their layout. I was not initially going to use the passive part at the 300 cps point so I did not buy the 52uf and 72uf caps nor the 2.9 and 5.4 mh coils.

- edit .
:)

[Ian Mackenzie]

This recent endeavour reminds me of Jean's situation.

If someone local went over for a look see it would be quite worthwhile:) .

Ian

[boputnam]

This recent endeavour reminds me of Jean's situation.Funny - I had the same thought. He went quiet. I hope he ended pleased :D and not merely confused... :blink:

[Ian Mackenzie]

I think Bill is talking to Marchland about some custom cards.

This has been posted numerous times but here are the JBL specified voltage drives for the 52-5140 (JBL 4343) card values. Here the summation is shown with the polarity inverted. (The acoustic summation of the actual drivers is no doubt different.)

At any rate there will always been a need for some adjustment.

Ian

[Hofmannhp]

Hi

(i) I'd like to see all posts from 511 onwards broken off from this thread and put into a new one destined for the DIY forum . ....
- Having a new thread ( such as; "A 4343 inspired 4-way" ) in the proper DIY forum , will be a good first step in helping the author re-jig perspective so that he can move forward with his project .
....:)

agree

HP

[boputnam]

agreeHi, HP... :wave:

It is done - the threads are severed. Bill's DIY project continues, here!!

So, where were we...? :hmm:

[Robh3606]

Cool Thanks Bo

I didn't get home till about 8pm my time from work. This is definately a better idea.

Rob:)

[boputnam]

This is definately a better idea.Yea, it was Earl's suggestion. Trouble is, we lost the patient!
Bill's gone quiet for almost 24-hrs. Maybe he's just enjoying his homebrewed 4343's...?

[Earl K]

Hi Bill,

- I set aside some time today and played about trying my best to blend a JBL 15 ( ME150H ) in a test box / with a 2123H ( which resides in one of my tiny custom SR boxes ) .
- The 2123 doesn't have nearly the bottom end extension as does a 2121 or a 2122 / but mine do live within a tuned cabinet ( @ 115 Hz ) which does ( somewhat ) popup the response above that tuing point . I added a passive 1K lowpass onto the 2123 so I wouldn't get distracted by it's odd top end signature .
- Taken together, these two components form the bottom 2 components in a JBL 4344 mkII .

So ;

- I tried measuring vertical arrangements . I have to lay my speakers side ways & fill the spaces with foam props ( the speakers are trapezoidal shaped ). I never could get decent summing when measuring out in the room .
- I eventually put these boxes side by side which allowed a centre of cone ( to centre of cone ) spacing of about 15" . ( Off hand I'm not sure what the stock 4343 is.) The horizontal centre line ( drawn between the cone centres ) was 13" off the floor .
- Using a tunable 24 db/octave crossover ( with LR filters ) I was eventually able to create & measure some very flat RTAs ( using mostly pink noise )
.
- These flat responses were only possible when measuring close in ( as in a couple of inches ) from the baffle board & equa-distant between 2 woofers frame edges . Height, from floor to mic was about 13". This "closein" method got rid of all the floor bounce cancellations that where ruining my all attempts at a flat line ( till that point ) .

What I found ;

- Getting these two drivers to blend & sound like one "taller ribbon" is incredibly hard. I never felt I had achieved a very convincing blend . Certainly nothing as homogenous as a single speaker. I was always staring at the 10" ( my eyes would just follow my ears ). Only with a sine wave, set to @ 270 hz, did they actually blend into one ( and only over a pretty small "side to side" distance ).

- I'd suggest that you try these blending experiments , using the approach I just took. Since your setup is a two box system, this horizontal arrangement will be easy for you to execute .
- You might find that a standard 24 db LR arrangement still sounds too disjointed. If so, you'll need to play with filter types ( try Butterworth ) and also try gentler slopes .
- Once that upper box is beside ( on wheels ) the woofer box / it's also easy enough to findout if time-alignment is an issue . I also spent some time with this / coming to no firm conclusions since the RTA display barely changed though moving the 2123H backwards was the equivalent of turning it down some .

- Ian Mac has a thread somewhere that laid out a nice simplistic way to balance the 4 parts of a 43xx big box. Right now I don't know where its' located. I'll eventually find it, if someone doesn't beat me to it .

:)

[Ian Mackenzie]

I agree getting the levels right is the fundermental requirement as it the correct phase. Perhaps in all the excitment we forget to tell Bill that......:blah:

[Earl K]

Hi,

- Since advice is now going in a circuitous fashion, I thought I would take the focus back to where it all started .
- The following is Bills' first post at LHF ( I've added formatting changes & underscored some areas ).
.


- I have a home built cabinet system. It started with the drivers and the crossovers(design and parts from JBL) for the L300's. A 15 inch 136A bass, an LE85 horn driver and a slot radiator.
- Since then I added the 10 inch midbass driver and changed the horns on the LE 85 compression drivers from the 800 cps to the 1200 cps horns.
- I have JBL design and parts incorporated for the passive 10 inch cone to horn and horn to slot radiator but have been trying to find the right active crossover point and slope for the 136A to the 10 inch midrange.
- I had thought that an active crossover would be superior to the passive.
- Initially the addition of the midrange was nice for midrange "non horny" sound but I seem to have lost the great punch from the system that was present with the straight active 800 cps crossover without the midbass cone .
- Any suggestions?

Well, so far there have been plenty of suggestions . A recap of two of those thoughts :

(i) Mimic the original 4343 design as close as possible. This includes adhering to cabinet size and component layout within the baffle board, as well as duplicating the original crossovers' transfer functions .

- Bill has done his best in this area by flipping the bottom box to get the 15" & 10" closer together plus has tried adding fake wing extensions either side of the top box ( to create a larger baffle surface ) .
- He has shown that his 2121 mids have pretty good extension down to @ 200 hz in his .25 cu ft dog house ( though JBL does recommend .5 cu' ) .
- FWIW, I also did a voltage drive foronly the lowpass portion of the N3143 bandpass filter ( since this is what Bill is using when he is biamping ). This 2-pole LC arrangement does ring some ( or resonate ) and does in fact, add @ 2 db of EQ boost. It mirrors ( close enough ) the overall voltage drive of the full blown N3143 bandpass . Seen here courtesy of Giskard . This is with a dummy load .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=859&stc=1&d=1063230965



- I selected (in white) the frequecy range of driver programme network response
- any big horizontal is 5 DB scale in according 40 DB full scale

- Well acoustical test is appear relatively normal test... but You touch my curiousity. I check . But remember I keep mesure in load all driver in contact at network and work... so my my probe is parallele in standart of load driver. In other side the amp is standart nominal load 8 ohms so do you ahe othre idea where the load is possible not good ???

This is Jeans' voltage drive for the 2121 portion of the N3143 with original JBL parts / and / as stated above , it is measured at the woofers' terminals with the 2121 speaker acting as the load. It mimics Giskards' simulation quite well .
http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=6919&stc=1&d=1113853634

One can see the acoustic effect of this voltage drive in the following pic from Jean . The upper portion of the bandpass is boosted a good 4 db vs the lower portion .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=6877&stc=1&d=1113797385

Now here is the voltage drive for the 2121 portion of the N3143 after Jean installed new Solen parts ( 2nd trace ) . He used higher quality parts and therefore obtained a higher "Q" . This portion of the N3143 network is a fairly resonant circuit, so changing the Q of the LC components will of course change the amount of resonance ( or boost ) .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=6998&stc=1&d=1114098343

Note how the lowend portion of 2121s' bandpass is significantly adding in its' "ringing/resonance" energy to the overall bandpass curve . Now the lower peak more closely matches the max db of upper mid peak. IMO, this "could" help make a case for leaving the 52 uH cap & 2.9 mH coil in the circuit ( when biamping a crossover with newish parts and if there is a measurable hole in the 290 hz region ) .

- The 2121 looks to be linear enough that it'll reflect anything fed into it. Ie; if the voltage drive "boosts" its' upper range / then we see that translated into a real acoustic boost .

So ;
- Q: Why is that boost there in the original N3143 ?
- A: I can only speculate that it was designed to fill in the missing lowend of the 2420/2307 combo or maybe just add a bit gain to the 2121 ( somewhere I saw that the 2121 is only a 91 db/w speaker / living in a higher ? efficiency design ) .
- The designer of the 4343 was up to something tricky and since we haven't been told what that intention was, we can only speculate .
- see Giskards' V-drive to understand my speculations .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=7671&stc=1&d=1116430646

Here are the 3144/5 Voltage drives ( courtesy of Giskard ) .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=6978&stc=1&d=1114035663
It should be quite apparent why Giskard ( & anyone else who prefers working with a flatter response ) prefers the 3144/5 network. .



Conclusion One :


- Personally, I'd be experimenting with the 3145 lowmid bandpass on the 2121( upper portion only ) to see what I can get out of that combo . Of course, this would likely cascade into changing the bandpass on the le85/2307 combo, and on and on etc.

Conclusion Two :

- Bill will be soon be trying out a 400hz lowpass card on his 136a ( with the 2121 set to 300 hz as a hipass ) . Therefore there is going to be more overlap ( than is normal ) in the crossover area .
- All the above posted info ( from my readings of the above pics ) would indicate that this solution has a good chance of succeeding for him .



(ii) Bill has been told to : Pay attention to phase & polarity issues .

- Bill has indicated that he understands the polarity issues which arise when using 4 pole filters vs the original 2 pole filters ( between the woof & midbass ) .


:)

[Ian Mackenzie]

To make any worthwhile sense of it all you really need to model and measure the drivers on the actual baffles.

Bill's baffles are NOT the same. Diffraction distortion adds ripple of +-2 db or more in the passband. These bandpass filters and rather messy and reactive things at the best of times.

[Earl K]

To make any worthwhile sense of it all you really need to model and measure the drivers on the actual baffles. Bill's baffles are NOT the same.

- I agree with that sentiment. Designing a usable filter first entails measuring the performance of the actual components as they exist on the actual baffles.
- Bills' measuring capabilites are a bit primitive / but / by using a close miking technique / they should suffice .


Diffraction distortion adds ripple of +-2 db or more in the passband.

- This has been brought up to Bill previously. But it's worth reinforcing the point that diffraction cancellations can be very destructive.
- I suggested that he look & test for them / and then pile up foam pillows beside his top box to mitigate any potential effects .



These bandpass filters and rather messy and reactive things at the best of times.

- Yeh, even the "best of the class" bandpasses have some interaction between the hipass & lowpass portions.

- In case it's not obvious, IMO the bandpasses in N3143 are not "best of class" ( from the POV of being non-resonant ). Far from it / that is also one reason for my previous post with its' comparison to the voltage drive of a 3145 .

- I was able to achieve measurably flat summing between a 15" and my 2123 / using electronic 24 db LR filters . This ought to be somewhat telling.
- The 2123 isn't as linear in the bottom end as its' foam edged counterparts yet I was able to achieve a quite linear bandpass ( using a passive lowpass on the upper end of the 10" ).
- Since Bill can't seem to achieve this ( or so his far field measurements indicate ) it's time to scrutinize the net effect of using only 1/2 of the passive elements within the bandpass for his 10" .

- As I mentioned in a previous post , the 1/2 he is using constitutes a pretty resonant 2-pole circuit, which creates a bump.
- Since his integration problems seem to all occur below this bump, the inference is obvious to me ;
- Get rid of the bump ( using a different filter type ) and then simply turn up the 10" ( since he is biamping ) .

--------------------------------------------------- / or /


Meanwhile the 400 cps low pass combined with the 300 cps high pass giving me some overlap looks very promising. Notch is not completely gone but is reduced a lot.

- Bill has a solution in hand . It may seem unorthodox / but / whatever works . :p
- This is afterall ; DIY ;)
- Also, with a tunable crossover ( such as the Ashly 1001, driving just the woofer circuit ) one could sweep its' lowpass up to maybe 500 hz, to see if this adds in enough of the missing mid information, to fill the gap .


:)

[Ian Mackenzie]

Here is a measurement of the 2121. I dont know how reliable the data.

[Earl K]

Here is a measurement of the 2121.

- Interesting plot . Thanks !
- It would be nice to see the official JBL info for this "obsolete" midbass .


I dont know how reliable the data.

- Yeh , it flaunts a few laws of physics by implying a sensitivety of 97.5 db with that much bass extension.
- ( Having a bass curve similar to many 15" woofs, along with comparable sensitivety to 15(s), just can't happen ) .
- Mind you, the plot doesn't claim that this is a sensitivity figure / so I suppose / the FR might be accurate enough .


:)

[boputnam]

- It would be nice to see the official JBL info for this "obsolete" midbass .It would also be nice to see ol' Bill return to this thread! He started this question!!

Hellooooooo....? :dont-know

[Ian Mackenzie]

There has been some work going on behind the scenes.

The large delta (dip) @355 hertz is apparent with only a single driver (the woofer) from several mic positions. This makes much of the earlier speculation moote. But I would like to see calibration of the RTA with a known source and perhaps a ground plane measurement.

Other customised cards are also being investigated.

No doubt Bill will post more details when he has a chance.

This is a link to a "The inside the studio monitor" paper in the Library spelling out the details of the driver transitions and other useful information.
http://www.lansingheritage.org/images/jbl/reference/technical/inside-monitor/page02.jpg




Ian

[Ian Mackenzie]

If anyone has a fair dinkum JBL data file on the 2121 I could attempt modelling it in the crossover (3143) per the baffle layout.

As I recall the factor unsmoothed response of the actual JBL 4343 was bit of a rollercoaster.

The placement of the drivers central to the baffle does not help matter in that regard.

I guess Bill has lot of fish to feed.

Ian

[boputnam]

Bill and I have been pm'ing about his project. I'm hopeful he posts his recent work, here...

[Ian Mackenzie]

Good to see everyone is having a shot at offering some assistance.

Have you told him about your theory on Oval Ports?:bouncy:

[grumpy]

are we talking big-block chevy heads here? :D

:duck: -grumpy

[Bill Shenefelt]

Bill and I have been pm'ing about his project. I'm hopeful he posts his recent work, here...


I have done a couple of things and have more in progress. First, I raised my main cabinets about 6 inches and of course inverted the bass cabinets to move the woofer closer to the 10 inch 2121. Although it does not mach the proximity in the posted baffle fab drawings, it sure matches the JBL photo of a 4343 when calculated using ratio of hieight to driver position.
http://www.sheneskillies.com/speakers/4343photo.jpg
I measure 18 inches center to center in my boxes. I calculate 17.9 inches using measurements from the JBL photo. I must assume that JBL used several baffle designs over the production span.

Second I got a much better measurement system. A program called TrueRTA which measures in 1/24 octave. Also a Behr 8000 measurement microphone and Tascam USB microphone preamp to feed the computer for the TrueRTA program. Ya, there went another $250. My measurements still show the slope of the 2121 rising about 6 db over most of its range. To try to fix this, I ordered several additional cards for the Marchand active corssover. One set is to try the 12dB/octave slope with woofer reversed polarity as recommended to me by Ian. Second, I ordered two sets of bass boost cards for the 2121 crossover region. Each should provide about 5 dB of maximum boost in the 325cps range tapering off to about 1 dB boost at about 500 cps on one set and at 700 cps on the other set. They should arrive late this week or worst case early next week. Third to tame the peaks at 40 and 150 and lull at 70 and 210cps, I ordered some Sonex Bass Traps. The upper curve is just a smoothed version of the lower response plot. It clearly shows the slope.http://www.sheneskillies.com/speakers/response%20at%20listening%20position.jpg.
The bass traps should help smooth the whole region somewhat. Not sure how much but worth a try. That traps should arrive early next week. (oops, there went another $100) I only ordered four, two ft sections. Each is a corner wedge about a foot on a side.
http://www.sheneskillies.com/speakers/BassTrap.jpg
I can say the sonex on the wall behind my speakers did wonders for the room. Even highly noticible in conversation in the room. It has deteriorated over the past 25 years so I ordered some replacement material. I doubt it will improve things acoustically but will esthetically.
http://www.sheneskillies.com/speakers/treated-end.jpg
I did find a nice informational page on reducing room resonances. Very informative. It is at http://www.ethanwiner.com/acoustics.html#bass%20traps It also has a link to a program for plugging in room dimensions to see where resonances will occur. It does predict the two peaks in the bass region so the traps should help. I have always believed, improvement of room acoustics, not equalization, is the route to take if possible. From all the info I have read though, the 2121 does have a slight rise over its range so a bass boost card should be applicable there. I found very little if any change in the response using the "wings" to widen the baffle at the 10 inch. Until I try the cards and the bass traps I am not going to cut the rear baffle of the 2121 chambers to enlarge them. Need to see where things stand first. I did try reversing the polarity of the bass driver and in the graph below the blue is the reverse polarity. It ia audibly more apparent than on the graph, although it does show a wider low output range in the 200 to 500 cps realm. The reversed polarity (in green on the graph) just sounds wrong. Sort of separates the drivers. With normal polarity they sound more like a single source. The use of the bass traps may also improve this region. I will just have to wait to try them.
http://www.sheneskillies.com/speakers/blue-is-reversed.jpg
Yep fish to feed, eggs to pick slowed me down a bit. I actually hesitated to post since I have more to do before I will really know what is going on. I can say that I got a lot better feel for the overall responce with the RTA program and 1/24 octave measurements. Real nice toy. Clearly room acoustics are playing a big part overall. Placement of the microphone is difficult since movement can show a relatively big change. Still, some of the 350 cps dip may be room, but a lot is the slope in the 2121. It stays there but in very slightly different magnitudes depending on mike placement. That was not very apparent with only the 1/2 octave RTA meter. It is there with the 2121 alone and in about any position so it is not a nulling effect of the 15 inch output or room acoustics alone. The cards should help there and if the bass traps do much they should add a little overall assistance beyond just the 20 to 200 range. Ideally I should put a false wall bass trap at the far end of the room, but I think the wife might just lose it if I did. I think buying in on the bass traps is about her tolernace limit.

When I get things on hand and can try some different bass trap locations and do some measuring I'll post the results.

[Bill Shenefelt]

Oh, check out the new orientation and compare it to the placement in the JBL photohttp://www.sheneskillies.com/speakers/4343photo.jpghttp://www.sheneskillies.com/speakers/speaker.jpg

[Ian Mackenzie]

Bill,

I have done a quick model using actual drivers tonight and (on full the passive crossover shematic) and its flat as a tack.

The driver polarity :
woofer +
mid -
HF-
UHF-

I can only suggest you check and trace the wiring to the L Pads on the 2121 and the 2121 filter wiring. The 2121 (2420 and 2405) driver should be out of phase with the 2231 . The arm labeled 2 on the L pad should go the the 2121 black terminal and 1 should go to red terminal as I understand it.


The signal polarity entering the mid filter should NOT be inverted (polrity inversion is after the L Pad) . Hot from the amplifier should go the 1.7 mh choke input if you are biamping

The 2121 L pad adjustment may also alter the response shape and you need to play around with this.

[Bill Shenefelt]

Bill,

I have done a quick model using actual drivers tonight and (on full the passive crossover shematic) and its flat as a tack.

The driver polarity :
woofer +
mid -
HF-
UHF-

I can only suggest you check and trace the wiring to the L Pads on the 2121 and the 2121 filter wiring. The 2121 (2420 and 2405) driver should be out of phase with the 2231 . The arm labeled 2 on the L pad should go the the 2121 black terminal and 1 should go to red terminal as I understand it. That is how I have it. The red goes to the ground connected the whole way staight back to the amp output.
The pot is simply a 16 ohm pot. One end goes to "hot"(from the crossover) the other end terminal to ground (crossover ground). The speaker should connect to the center tap and the ground of the pot. In this case the center goes to black on the speaker and ground to red. 1 and 3 can be interchanged on the pot, but the speaker always goes to the center and to the end that is ground in the crossover. As for reversing polarity, I do not yet have the 12/0ctave crossovers which need a polarity reversal. The 24 does not. The 3 way passive has ground to ground to ground. Input is hot to hot to hot all in parallel. The 077 has a 8 ohm pot and a parallel 10 ohm resistor(terminal 1 to 3) to cut the output in half.


The signal polarity entering the mid filter should NOT be inverted (polrity inversion is after the L Pad) . Hot from the amplifier should go the 1.7 mh choke input if you are biamping.
Since all three speaker grounds are in parallel (input to input to input and all connected at the same pole at the output pots and all three inputs to the network are in parallel within the passive network, all get the same polartiy at the speakers (black hot) then I can reverse pollarity to all three just by swapping the amp output into the network (or inverting the input to the bass driver speaker terminals. Easier to reach the 3 way passive for quick trials though so I check things there. No DC and no polarized caps so no problem. When all is balanced out I can give the initial push to the black terminal in the upper network and which ever is applicable to the bass driver(depend on if I use the 12 or the 24dB/octave.

The 2121 L pad adjustment may also alter the response shape and you need to play around with this.
What I have to get fixed are the bass standing waves. Once I can tame them it will be easier to match the bass level with the midrange level. Should know more on Tuesday after the bass traps get here. Once all is running with traps and balanced out, I may try increasing the volume of the 2121 chamber. I want to keep things the same as much as I can and not go varying too many parameters at once.

[Ian Mackenzie]

Quote: "The pot is simply a 16 ohm pot."

For the benefit what others who maybe reading this thread the above is statement is seriously incorrect.

Simple application of ohms Law will prove this. See below

http://www.sengpielaudio.com/calculator-Lpad.htm

Play with this on line calculator to see the effect of changes in attentuation in Db

If Bill is in fact using a 16 ohm pot it might explain a lot of what he had being attempting to understand....

The L pad in the 3143 schematic is not a 16 ohm pot. Its constant impediance attentuation device. In this instance its a little more complicated than that because there is a T section fixed impediance attenuator and a 16 ohm L pad is used to improve power handling of the variable L pad on the 2121 filter circuit.

The connections are not interchangable. 1 is always common (ground to the schematic), 3 is in and 2 is out. To ensure the outgoing signal from the crossover filter is in fact attenuated correctly the Red connection from the amp should always be referenced to the positive input via the filters. Black should always go to common or ground.

If you have a true L pad its worth opening it up to see just how they work. What you will find in there is two nichrome wire elements with a wiper that makes contact along the edge of the element.

One element represents a series resistance Rs in series ( total resisistance 16 ohms for a 16 ohm L pad ) with the loudspeaker. The other element respresents Rp, a parellel resistance (total resistance about 65 ohms for 16 ohm L pad ) across the loudspeaker (after the series resistance)

They tend to have two problems one being they are sloppy and unreliable. Secondly they are not a linear constant impediance attenuator as per the calculator. Often a filter voltage drive curve will vary depending the position of the L pad.

[Bill Shenefelt]

My error Ian;

The pots I used were those supplied by JBL for the crossover replacement parts. I could measure no difference in the use of the pin one or the pin three for the input or ground as long as I used the ground for one speaker terminal and the center pin 2 for the other speaker connection. I guess I incorrectly referred to them as simple 16 ohm pots incorrectly as opposed to L pads. I did know that it kept a constant load despite pot wiper position. I just meant that pin one was the equivalent of pin 3 and could be used interchangibly as long as the speaker was across pin 2 and the outside pin connected to the crossover common ground. This particular pot from JBL is used both for the 2121 and for the LE85 horn. A different one (marked 8 ohm by jbl) is used on the 077 slot radiator. It has a 10 ohm power wire wound resistor across pins one and three. I have none on the 136A bass driver as it is direct to my amp in the active crossover.

As a side item, I measured the on monitor distance from the 2121 rim to the 15 inch driver rim on a photo on line of the nameplate for the crossover in another thread on the 4343. The real distance was about 5.25 inches. Mine is only about 3.75 inches so my 2121 is actually a bit closer to the 136A than is the one in the 4343. (I used the ruler shown on the photo which was intended to acquire new nameplates. ) http://audioheritage.org/vbulletin/attachment.php?attachmentid=289&stc=1&d=1058109730 I just measured the distance between drivers using my ruler and then measured that distance on the on screen ruler. I can change this distance easily to the exact value on the 4343 by either raising my 2121 enclosure or just moving it slightly to one side rather than purely vertical.
I did do a mike measurement of responce with the mike on the floor. Reversing leads on the 136A in this case slightly improves the 325cps summing but it does the opposite at the listening height. Not sure if that is room effect or is the change in driver distances to the two positions.

Is there a real perferred distance from the 10 to 15 inch driver among the various JBL baffles? Seems to vary among the different models and yet the crossover schematic is not changed.

[quote=Ian Mackenzie;164413]

[Bill Shenefelt]

Quote: "The pot is simply a 16 ohm pot."

For the benefit what others who maybe reading this thread the above is statement is seriously incorrect.

Simple application of ohms Law will prove this. See below

http://www.sengpielaudio.com/calculator-Lpad.htm

Play with this on line calculator to see the effect of changes in attentuation in Db

If Bill is in fact using a 16 ohm pot it might explain a lot of what he had being attempting to understand....

The L pad in the 3143 schematic is not a 16 ohm pot. Its constant impediance attentuation device. In this instance its a little more complicated than that because there is a T section fixed impediance attenuator and a 16 ohm L pad is used to improve power handling of the variable L pad on the 2121 filter circuit.

The connections are not interchangable. 1 is always common (ground to the schematic), 3 is in and 2 is out. To ensure the outgoing signal from the crossover filter is in fact attenuated correctly the Red connection from the amp should always be referenced to the positive input via the filters. Black should always go to common or ground.

If you have a true L pad its worth opening it up to see just how they work. What you will find in there is two nichrome wire elements with a wiper that makes contact along the edge of the element.

One element represents a series resistance Rs in series ( total resisistance 16 ohms for a 16 ohm L pad ) with the loudspeaker. The other element respresents Rp, a parellel resistance (total resistance about 65 ohms for 16 ohm L pad ) across the loudspeaker (after the series resistance)

They tend to have two problems one being they are sloppy and unreliable. Secondly they are not a linear constant impediance attenuator as per the calculator. Often a filter voltage drive curve will vary depending the position of the L pad.
I am starting to realize pin 1 is not equal to pin 3. Problem is, which is which. Are they 1,2,3 cloclwise viewing from the adjustment knob face or the back ceramic face?

[Bill Shenefelt]

I am starting to realize pin 1 is not equal to pin 3. Problem is, which is which. Are they 1,2,3 cloclwise viewing from the adjustment knob face or the back ceramic face?

If I am understanding correctly the correct use of pin 1 vs pin 3 is the one that if I measure DCR across 1 to 3 and vary the position of the attenuator knob, I get near 10 to 11 ohms no matter the position, correct? When 1 and 3 are connected in that way I would assume also that the output of the attenuator to speaker would be connected to the attenuator input (1 or 3) that is connected to passive ground (and red speaker post) and the pin 2 (black speaker post). This is the way it is connected now at least. Were I to reverse the connections of pin 1 and 3 I would not get a constant impedance shown to the network. I had assumed it did not matter but got lucky I guess.

[Earl K]

I am starting to realize pin 1 is not equal to pin 3. Problem is, which is which. Are they 1,2,3 cloclwise viewing from the adjustment knob face or the back ceramic face?

With the shaft of the Variable Lpad facing towards you :

- the 3 terminals are then configured as ; 3, 2 & 1 .


:)

[Zilch]

Pin 3 goes to the source, Pin 2 is the output, and Pin 1 is the common.

Measure between Pin 2 and the others, and you'll see the difference.

Pin 3 varies between the nominal impedance and zero, going clockwise.

Pin 1 vaies between zero and some relatively high resistance, 35 ohms or more, then infinity, typically, also going clockwise.

Thus, it may be seen that, at full CCW, there's the nominal (8 or 16-Ohm) resistance in series with the driver, and the driver itself is shorted.

At full CW, there's zero resistance in series with the driver, and very high or infinite resistance across it.

[Ian Mackenzie]

Bill,

If what you say in your last post is in fact the case you have (been supplied ) the wrong part. ie If you asked for a pot that is what they would have sent you.

It just won't work properly at all with a 16 ohm pot in there.

Please contact Parts Express or look at their online catalogue.Someone might have the part no, I don't have time to look it up right now The correct L pads are only a few dollars.

As Lord Zilch Von Zilchen-Curvemeister says with shaft facing you connect one probe of the multimeter to the far left tab and measure the Resistance to the centre tab. At full clockwise rotation it should read close to zero ohms. Start turning anti clockwise and it will increase slowly to about 16 ohms. (8 ohms with an 8 ohm L pad)

Now connect the probe to the centre tab and the other probe to the far right tab.

At full clockwise postion it will appear open ciruit (infinite ohms). Turn slowly anticlockwise and it will measure a high resistance (35-60 ohms depending on 8 or 16 ohm L pad) and start to decrease and will be zero ohms at full anitclockwise rotation.

As a further test, connect the probe to the fat left tab and the other probe to the far right tab. At full clockwise rotation it will measure infinite ohms. Then turn slowly anti clockwise and it will measure a high resistance (over 30 ohms) and will decrease slowly to about 16 ohms ( 8 ohms with an 8 ohm L pad).

(The L pad shouuld not be connected to the crossover during the above procedure)


If you are unable to get the above range of measurements then without doubt you have the wrong part.

Ian

[Bill Shenefelt]

[quote=Ian Mackenzie;164488]Bill,

If what you say in your last post is in fact the case you have (been supplied ) the wrong part. ie If you asked for a pot that is what they would have sent you.
I asked for the components used in the crossover for a 4343. The following measurements were made on one of the bigger units used for the 2121 and horn, not the slot radiator which is much smaller in size. It is however connected to give the same type results. The attenuator is not connected to either the speaker or crossover for these measurements. I should note that with the knob full at the ends of travel, it sometimes will give infinite resistance or 0 resistance in the last few degrees of travel. The values I state are thru the full rotatioon but may not include theis "off scale" value at a very end.


Left versus right depend on if the tabs are on the down or up side of the unit. If they are "up" then the speaker is connected across the middle and right (common) tab-the common being the furthest clockwise looking from the stem side. I will continue to use both left and right conotations but will add clockwise or counterclockwise to make it consistant no matter the orientation of tabs.

Center to common measures 0 to 60 ohms as I turn the pot clockwise (this would be in parallel with the speaker -speaker is not connected).


The input from the crossover is on the left tab (furthest counterclockwise, stem up) . I measure about 14 dropping to 0 ohms from the input tab to the speaker center tab as I turn the dial clockwise (this is the resistance that is put in series with the speaker from the crossover. Input to center tab.)

The far right tab is common. From the far left to the far right (input to common) tab I measure from about 14 ohms rising to 60 ohms then infinity just at the very end as I turn the knob clockwise.

Not sure if this matches your description or not due to the tab numbering. Does this sound correct? For these variable units it is the only way they would work to allow proper attenuation of the output. Either they are the correct units or incorrect units wired to give variable output but improper load to the corssover. I did specifically ask JBL for the parts of the 4343 crossover, not specifying pots, caps, inductors or resistors, just all parts except those for the 15 inch 300 cps crossover components as I was going to biamp the 300 cps crossover. They sent me all the inductors, caps, resistors and attenuators but no connections binding posts or wires or terminal strips. I really think I got the correect attenuators.


Yesterday I bit the bullet and enlarged the chambers. They are now 10" wide by 9" high by 11 1/2 inches deep. I almost doubled the depth from before. In the full rear of the chamber I put a piece of 3 inch thick sonex. The chamber toward the front is filled with a loose stuff of some white fiber speaker fiberfill material that seems to have the feel of wool. It is pushed against the upper , lower and side walls but not held tight there. Just forms sort of a basket for the speaker basket.

[Bill Shenefelt]

[quote=Ian Mackenzie;164488]Bill,


As Lord Zilch Von Zilchen-Curvemeister says with shaft facing you connect one probe of the multimeter to the far left tab and measure the Resistance to the centre tab. At full clockwise rotation it should read close to zero ohms. Start turning anti clockwise and it will increase slowly to about 16 ohms. (8 ohms with an 8 ohm L pad) Far right not wleft as it is a rear view though. L pad connections
http://www.sheneskillies.com/speakers/midandhornpads.jpg

This is an "inverted view" as it is from the back. I think it matches what you say above.

[Ian Mackenzie]

Bill,

The first graph is a Model using the JBL specified active card (5235) for the 4343. These are spread 12 db butterworth filters. The mid , horn and slot are connected out of phase to the woofer.

The 2nd graph is the LR 24 db slopes are 320hz. All drivers are in phase.

As you can see there is a slight depression centred at 500 hz (91db) and rises to 93 db at 980 herts.

Some notes
In your system you have to appreciate your 2121 may well be peaking out more at 1000 hz as this is a characteristic of the driver and your room acoustic may well be causing a null oin the 300-500 hz region. The dimensions of the baffle will also contribute to ripples in the 2121 passband. Typically mid range and HF driver produce a smoother response with asymetric postioning (off centre) on a baffle.

What you may wish to try is taking the 2121 and measuring the 2121 without any crossover connected (in several different locations) in an attempt to understand its characteristics. Do this for both drivers, then relocate back where you normally have the 2121 on the woofer box with the horn and measure again without the crossover connected to try and determine the effect of the room/ baffle layout on the response.

If you want to to keep the vertical array and the other cards still dont help the falling response you can try and modify the low pass passive filter on the 2121. The problem with actively boosting the low end of the 2121 toio much is it will stress the amps dynamic headroom and the Xmax of the 2121. Just to show you how clever I really am...Muhhahaha the third graph shows the modelled effect of increasing the 1.7 mH inductor in increments of 1.9 mH, 2.15 mH and 2.38 mH. This has the effect of taming the rising response on the 2121 driver. This is simple enough to try by adding a few turns to the inductor (if you have an inductance meter) and doing some measurements with Smart Live to confirm what actually happens. You will of course need to need to modify the gain of the HF amp and adjust the mid, horn and slot levels to shelve the response overall.

The Doctor.

[Bill Shenefelt]

Well here is what I can get with room responce. I canot say the bass traps did anything, at least on the RTA responce. The following RTA plots are with the speakers flat against the wall, about 6 inches off the floor, the 2121 close to the 136A on the baffle, the 2121 enclosure at 1/2 cubic ft now, the 136A slope at 400 cps and the 2121 slope at 300 cps with 5 db boost at about 350cps. I have not received the 12dB/octave cards from Marchand yet so I have not tried Ian's recommended crossover configuration. I ran scans with both pink oise and with a feature called Quick Sweep which gives a fast pink noise sweep.

left channel sweep
http://sheneskillies.com/speakers/sweep-left.jpg

Right channel sweep
http://sheneskillies.com/speakers/sweep-right.jpg

Both channels driven sweep

http://sheneskillies.com/speakers/sweep-both.jpg

Pink noise left
http://sheneskillies.com/speakers/pinknoise-left.jpg

pink noise right
http://sheneskillies.com/speakers/pinknoise-right.jpg

Pink noise both driven

http://sheneskillies.com/speakers/pinknoise-both.jpg

Other than the 60 cps dip which is room for sure, it looks pretty good with about a +- 5dB range. (the scale is marked each 10 dB on the left). I guess that is not bad for room response.

[Bill Shenefelt]

I was trying to measure voltage drive but did not get anywhere with it. I tried conneting my speaker terminal pair instead of the microphone for input to the USB unit. I kept getting the whole drive range from the crossover, not the drive range after the passive crossover. I had the speaker connected and the leads to the + and - terminals on the speaker. I know the speaker is only getting the correct range of frequency but measuring a voltage drive at the terminals shows the full range coming from the active crossover. Not sure if I am doing something wrong trying to measure it this way or not.

[Earl K]

Hi Bill,

Thanks for the update !

FWIW;
- Larger pics would be quite a bit more useful for the rest of us to peruse.
- If TrueRTA doesn't have a "save to jpg" feature then capture a "screen-shot" from the computer ( I don't know the keyboard commands for the PC ) / edit & save the resulting .bmp file ?? as a jpg / & then upload the .jpg into the thread as an attachment .


,,,snip , ,,(the scale is marked each 10 dB on the left).

- The 10 db scale isn't very helpful for determining the "hills & dales" ( it's a bit too flattering ). Try 5 db instead .

:)

[Ian Mackenzie]

Learn how to attach images too.

That way more people will be able to follow and take an interest in what you are up to.

Links are such a temporary thing.....

[Bill Shenefelt]

24772

Learn how to attach images too.

That way more people will be able to follow and take an interest in what you are up to.

Links are such a temporary thing.....

[Bill Shenefelt]

24773
I tried saving from paint screenshot and it was terrible so photos will have to do.

[Bill Shenefelt]

24774This is with the microphone on the floor rug as suggested by Ian.

[Earl K]

I tried saving from paint screenshot and it was terrible so photos will have to do.

- That's unfortunate .

- Here's an FR shot, done by LHF member " dmtp " . This is done with his copy of TrueRTA .

http://www.audioheritage.org/vbulletin/attachment.php?attachmentid=20188&d=1162677117

- Ya might want to shoot him a PM and ask for his advice on how to create this sort of uploadable file .

:)

[Bill Shenefelt]

24776If you select 1/3 octave for a view like in the JBL publication fot the 4343 it looks a lot better. This is quick sweep both channels at listening position. Each vertical step is one dB. I don't expect it to match JBL since I do not have an anocheic chamber so I think it looks decent with no equalization. There is not much I can do with the peak at 40 and the dip at 60 cps without somehow changing the ceiling ht and room dimensions. Looks like other than the 60 and 40 cps realms I am pretty much within +-3dB.

[Robh3606]

Things are getting better. The 2121 level looks a bit up. How do they sound??

Rob:)

[Bill Shenefelt]

Pretty good I think. I still have to get my turntable set up but with a couple of CDs and tape deck they are nice. I will need to do a little level tweaking amongh the drivers but need to listen to more things to get it right. It is sort of a balancing act since my high frequency hearing is down a good bit above about 7K. To get it good for me and for others is a little tricky. I still am waiting for the 12dB/octave cards to come in to try. Also have my old McIntosh 240 tube amp to get back from repair to see if I want to go 3 way active or maybe use it instead of the Amber amp for the 300+ range. If it works well it may be a better unit on the horns at least. My Apt Holman preamp died so I am using a Nakamichi preamp right now until the Apt and my McIntosh are fixed. Additionally I need to see if a new Grado cartridge is better than my Dynavector diamond(old and may need replaced due to age, not wear though).

[4313B]

I don't expect it to match JBL since I do not have an anocheic chamber so I think it looks decent with no equalization.JBL measured the 4343 on its back flush with the ground facing up into the sky so don't fret.

[Bill Shenefelt]

Bill,
If you want to to keep the vertical array and the other cards still dont help the falling response you can try and modify the low pass passive filter on the 2121. The problem with actively boosting the low end of the 2121 toio much is it will stress the amps dynamic headroom and the Xmax of the 2121. Just to show you how clever I really am...Muhhahaha the third graph shows the modelled effect of increasing the 1.7 mH inductor in increments of 1.9 mH, 2.15 mH and 2.38 mH. This has the effect of taming the rising response on the 2121 driver. This is simple enough to try by adding a few turns to the inductor (if you have an inductance meter) and doing some measurements with Smart Live to confirm what actually happens. You will of course need to need to modify the gain of the HF amp and adjust the mid, horn and slot levels to shelve the response overall.

The Doctor.
Is the 1.7 just a very slow slope to minimize the responce rise or is it part of the crossover at 300, or a phase thing? Should it be there at all with the active xover present? I'm using the JBL 1.7 mH so I cannot increase turns easily unless I just buy a bigger one and remove turns. I do have a meter for inducatnce now and an ohm meter capable of 0.001 ohm readout so that is not a problem. I really do not run it that hard so I doubt x max is a problem, especially with the 24/0ctave active. .

[Earl K]

Is the 1.7 just a very slow slope to minimize the responce rise or is it part of the crossover at 300, or a phase thing? Should it be there at all with the active xover present? I'm using the JBL 1.7 mH so I cannot increase turns easily unless I just buy a bigger one and remove turns. I do have a meter for inducatnce now and an ohm meter capable of 0.001 ohm readout so that is not a problem. I really do not run it that hard so I doubt x max is a problem, especially with the 24/0ctave active. .


- The 1.7 mH series coil together with the 20 uF conjugate capacitor, form a resonant 2-pole lowpass ( with an eventual slope of 12 db per octave ) on the 2121.
- It's essentially a 2-pole bump filter / that before it achieves the 12 db/octave lowpass / first "boosts" a selective range of frequencies .
- ie ; It is a significant contributor to the "rise" that you see in the 2121 .
- This midrange "boost" is quite apparent in the following "simmed" voltage drive ( courtesy of Giskard ).
- This extra bit of boost consistently shows up in your posted FR snap-shots ( & once you level the top to the surrounding topography you are left with a "hole" below the knowle ).
- At some point, you'll need to ask yourself if you want to keep this midrange rise on your 2121 / or / flatten it all out, by implemeting a different lowpass on that 10" ( by using a different LC design as IanM has started to work out for you ) .

http://audioheritage.csdco.com/vbulletin/attachment.php?attachmentid=859&stc=1&d=1063230965

- It's the green graph that represents the N3143 voltage drive . This was mentioned previously here ! (http://audioheritage.csdco.com/vbulletin/showpost.php?p=162016&postcount=70)

:)

[Ian Mackenzie]

24774This is with the microphone on the floor rug as suggested by Ian.

Its not quite that simple. There are references on the www regards ground plan measurements.

Overall your plots are reasonable (for what we are dealing with)

Adjust the L pads over a period of weeks and see how you go.

I would not turn into a curve junky (tweaking) without doing a lot more listening on a variety of program material at this point. (Modification if the stock network is ill advised unless you are detouring off the road map...because you have an odd ball active filter at the moment I would leave as it.)

[4313B]

Yes, basically the L-Pads were kind of a bad idea. They've managed to cause alot of end users considerable angst over the years. They don't behave like perfect level controls and they shift crossover frequencies arbitrarily. The 250Ti is a classic example of going to the significant expense of proper shelving that has carried through in various systems right up to the Everest II. Where the 250Ti has a rather large range of shelving available the Everest II is quite subtle.