Think We May Have a Keeper Here!!!

[toddalin]

After many hours of playing/experimenting with the crossover (including such things as by-assing the choke, caps, etc.), I think I have achieved a level of success!:applaud: Bands indicate a change of 2 dB.

Note the included 10-band spectrum analysis (band on far right is volume). Sorry about the poor contrast, but you get the idea. BTW, the ultimate solution was to place 16 ohms in series with the woofer that flattened the 63-250 bands while leaving the 500 Hz band fairly intact.
http://www.largescaleonline.com/eimages/lsolpics/Team_Member_Pics/toddalin/spectrum.jpg

[Zilch]

1) You should not have to attenuate the LF driver. If it's playing too loud in relation to the mid and high, that'd be because THEY are over-attenuated, most likely. There's plenty of headroom between the efficiencies of the LF versus M/H drivers. Something else is up.

2) You certainly don't want to attenuate the LF with series resistance, which will a) mess up the filter frequency and b) preclude the amplifier damping woofer cone movement.

3) Bypassing the LF filter is a bad approach. You end up with the woofer playing well into the MF range, with all of the concomitant phase interaction between drivers. I'm not able to analyze the full consequences of your unorthodox use of an L-Pad there. Having the full schematic would be helpful.

4) You know from your near-field LF measurements that the RTA is not seeing what's actually happening at the low end. Measure your LF separately to verify there's nothing anomalous going on, and then rely upon it's performance near the 800 Hz crossover range on the RTA to establish the balance. Ignore what the RTA says below that.

5) In my experience, this MUST all be done "By the book" to succeed. All of the requisite capability is already built into the crossover to accomplish the task.

6) If it sounds good without messing around with the LF, and you are confident it's working according to design, just leave it be as designed. You may be attributing more to the RTA analysis than is warranted.

7) Alternatively, send me one of the crossovers, and the schematic "as built." I'll break out an LE175 and 2402 here to run with 2235H and test/troubleshoot it....

[toddalin]

Thanks, to enlighten me, and possibly others on this forum, can we take these points one by one???


1) You should not have to attenuate the LF driver. If it's playing too loud in relation to the mid and high, that'd be because THEY are over-attenuated, most likely. There's plenty of headroom between the efficiencies of the LF versus M/H drivers. Something else is up.

While the horn can play louder than need be, the tweeter at full out is a hair down from the woofer (with the addition of a 1 mfd that was subsecquently added). However, some attenuation of the woofer results in a far smoother overall curve.

2) You certainly don't want to attenuate the LF with series resistance, which will a) mess up the filter frequency and b) preclude the amplifier damping woofer cone movement.

Mess up the filter frequency! Yes, this is what is producing a nice flat woofer response as opposed to a dip at 500 Hz without the attenuation. So if messing it up improves it, is that really messing it up???


3) Bypassing the LF filter is a bad approach. You end up with the woofer playing well into the MF range, with all of the concomitant phase interaction between drivers. I'm not able to analyze the full consequences of your unorthodox use of an L-Pad there. Having the full schematic would be helpful.

A bad approach??? I can't tell you how many JBL schematics I perused that use this approach. It also fills in the 500 Hz dip. So is smoothing the response a bad thing???


4) You know from your near-field LF measurements that the RTA is not seeing what's actually happening at the low end. Measure your LF separately to verify there's nothing anomalous going on, and then rely upon it's performance near the 800 Hz crossover range on the RTA to establish the balance. Ignore what the RTA says below that.

I'm not really concerned about what's happening at the low end and recognize that this is largely a function of room acoustics. My concern lies between the 500-1,000 Hz range. Unforetunalely with an octave eq and reads at 500 and 1,000, it's hard to get an accurate view of what's going on at 800 Hz.

5) In my experience, this MUST all be done "By the book" to succeed. All of the requisite capability is already built into the crossover to accomplish the task.

6) If it sounds good without messing around with the LF, and you are confident it's working according to design, just leave it be as designed. You may be attributing more to the RTA analysis than warranted.

Possibly, but my my ear is sensitive to the 500 Hz range for vocal intelligability. When I set it to run flat (with the 16 ohms on the woofer), there is no denying that the vocals become more pronouced and words become more intelligable.

7) Alternatively, send me one of the crossovers, and the schematic "as built." I'll break out an LE175 and 2402 here to run with 2235H and test/troubleshoot it....

Thanks for the offer.

Contrast these two spectra. The first is the crossover per the JBL L200B/N7000-8000 schematic. In this view, the 1.0 mfd cap has been removed from the tweeter (still the internal 1.5 mfd and choke per the N7000/N8000) to raise it's volume to better match the woofer. The horn is running at near full capacity. I could bump it a little more creating a broad peak at 2-4K Hz.


http://www.largescaleonline.com/eimages/lsolpics/Team_Member_Pics/toddalin/Spectrum2.jpg


This second spectrum has a 16 ohm resistor between the woofer and crossover output. Also based on the decreased volume, I can place the 1.0 mfd cap on the tweeter better tailoring its response.

http://www.largescaleonline.com/eimages/lsolpics/Team_Member_Pics/toddalin/spectrum.jpg

[hapy._.face]

X_X drunk

*_0 black eye

@_@ trippin

#_# post lasic surgury

-_- zen meditation

9_9 Look up

6_6 look down

Y_Y cryin

^_^ jolly

+_+ screen glare

=_= war paint

[Zilch]

1) You should not have to attenuate the LF driver. If it's playing too loud in relation to the mid and high, that'd be because THEY are over-attenuated, most likely. There's plenty of headroom between the efficiencies of the LF versus M/H drivers. Something else is up.

While the horn can play louder than need be, the tweeter at full out is a hair down from the woofer (with the addition of a 1 mfd that was subsecquently added). However, some attenuation of the woofer results in a far smoother overall curve.

2402 is 110 dB, 2235 is 93 dB. Something is up in the UHF? Again, we need the schematic. I recall you did not cascade the MF/UHF, which should have provided even more drive to the UHF.


2) You certainly don't want to attenuate the LF with series resistance, which will a) mess up the filter frequency and b) preclude the amplifier damping woofer cone movement.

Mess up the filter frequency! Yes, this is what is producing a nice flat woofer response as opposed to a dip at 500 Hz without the attenuation. So if messing it up improves it, is that really messing it up???

It messes it up in other ways. If it's TRULY down at 500 Hz, and you need to do your near-field measurements to verify that, then the LF should be modified in other ways to correct it. I doubt your 2235's are behaving like that; I'm bettin' room stuff. Turn off the MF and HF and put the RTA mic about 6" away from the 2235 dome center. What's that look like?


3) Bypassing the LF filter is a bad approach. You end up with the woofer playing well into the MF range, with all of the concomitant phase interaction between drivers. I'm not able to analyze the full consequences of your unorthodox use of an L-Pad there. Having the full schematic would be helpful.

A bad approach??? I can't tell you how many JBL schematics I perused that use this approach. It also fills in the 500 Hz dip. So is smoothing the response a bad thing???

Smoothing the response is not a bad thing, but letting the LF play into the MF certainly is. We're not talkin' SR or bookshelf crossover here. If the LP slope or frequency needs adjusting to smooth the response, that can be done.


4) You know from your near-field LF measurements that the RTA is not seeing what's actually happening at the low end. Measure your LF separately to verify there's nothing anomalous going on, and then rely upon it's performance near the 800 Hz crossover range on the RTA to establish the balance. Ignore what the RTA says below that.

I'm not really concerned about what's happening at the low end and recognize that this is largely a function of room acoustics. My concern lies between the 500-1,000 Hz range. Unforetunalely with an octave eq and reads at 500 and 1,000, it's hard to get an accurate view of what's going on at 800 Hz.

Use your SPL meter near field for the detail.


5) In my experience, this MUST all be done "By the book" to succeed. All of the requisite capability is already built into the crossover to accomplish the task.

Reiterated above....


6) If it sounds good without messing around with the LF, and you are confident it's working according to design, just leave it be as designed. You may be attributing more to the RTA analysis than warranted.

Possibly, but my my ear is sensitive to the 500 Hz range for vocal intelligability. When I set it to run flat (with the 16 ohms on the woofer), there is no denying that the vocals become more pronouced and words become more intelligable.

You know what you hear, of course. The L-Pad connection you described above is not 16 Ohms in series with the woofer. If 500 Hz is the frequency at issue, that has nothing to do with the MF or HF. Turn them off and see what's up.


7) Alternatively, send me one of the crossovers, and the schematic "as built." I'll break out an LE175 and 2402 here to run with 2235H and test/troubleshoot it....

Thanks for the offer.

I'll measure 2235 in L200 here. Give me 15 minutes to post it for comparison....

Edit: Near field, 4". There's a 5 dB notch at 500 Hz, a 2 dB peak at 710, and an 8 dB notch at 900 Hz. It's not the crossover. I'm guessin' it's the 4" cabinet lips all around. I'll confirm looking at the other one after dinner....

Re-Edit: Second unit similar (bottom), too much so at the same frequencies for it to be coincidence. I'm now doubting it's the room at 4", but will try a different room tomorrow.

Who else here has RTA and 2235H in L200's?

Reviewing your spectra, maybe dial the MF back to achieve a balance between the LF and UHF instead of trying to bump them both up to match the MF. How well does that work?

450 Hz. is ~30" wavelength. :hmm:

450, 900, 1350, 1800....

[toddalin]

1
You know what you hear, of course. The L-Pad connection you described above is not 16 Ohms in series with the woofer.

Actually it is. I tried two 8.2 ohm resistors in series AND I measured the resistance of the L-pad once getting the freq flat.


I'll measure 2235 in L200 here. Give me 15 minutes to post it for comparison....

Edit: Near field, 4". There's a 5 dB notch at 500 Hz, a 2 dB peak at 710, and an 8 dB notch at 900 Hz. It's not the crossover. I'm guessin' it's the 4" cabinet lips all around. I'll confirm looking at the other one after dinner....

Re-Edit: Second unit similar (bottom), too much so at the same frequencies for it to be coincidence. I'm now doubting it's the room at 4", but will try a different room tomorrow.


Who else here has RTA and 2235H in L200's?

Reviewing your spectra, maybe dial the MF back to achieve a balance between the LF and UHF instead of trying to bump them both up to match the MF. How well does that work?

450 Hz. is ~30" wavelength. http://audioheritage.org/vbulletin/images/smilies/scratchchin.gif

450, 900, 1350, 1800....


"You know what you hear, of course. The L-Pad connection you described above is not 16 Ohms in series with the woofer."

Actually it is. I tried two 8.2 ohm resistors in series AND I measured the resistance of the L-pad once getting the freq flat.

Edit: Near field, 4". There's a 5 dB notch at 500 Hz, a 2 dB peak at 710, and an 8 dB notch at 900 Hz. It's not the crossover. I'm guessin' it's the 4" cabinet lips all around. I'll confirm looking at the other one after dinner....

Re-Edit: Second unit similar (bottom), too much so at the same frequencies for it to be coincidence. I'm now doubting it's the room at 4", but will try a different room tomorrow.

Thanks, at least I know I'm not crazy and my analyzer is fairly accurate in that range.http://audioheritage.org/vbulletin/images/smilies/blink.gif Doesn't my testing demonstrate that it is possible to get rid of the 500 Hz notch, perhaps at the expense of ??? However, with proper crossover design/implementation the notch could be removed without too many other repercussions?

As to the tweeter being much louder than the woofer, I think the poor quality of the included microphone with the analyzer is rolling off the high end response so that there are really more highs than shown on the analyzer. When I get the highs flat, my Yamaha RX-Z9 goes through and reduces them when it goes through autoeq. Will continue to experiment/measure and will get a schematic posted.

[Earl K]

- I'd love to see a schematic for your final setup .
- I'd also like to see the impedance response for this network .
- I'd also like to know the insertian loss for this network ( it would appear to be quite high ) .

- From what I can glean from the initial descriptions ( and if I understood correct ) , it appears the woofer & horn drivers' voice coils ( maybe tweeter ) are arranged in a parallel/series topography / all the while, sharing a common earth through a fixed path to earth ( 20ohm conjugate on driver ) as well as the variable resistance represented by the Lpads' shunt path to ground ( variable R existing across, pin 2 to 1 ) . The series arrangement comes into this because ( at least ? ) 2 voice coils are tied together ( in series ) by that central terminal (#2 ) of the horns' variable LPad and the 20 ohm comp driver conjugate .
- The paralleled inductor across the horn driver line represents one element necessary ( to be kept "in place" ), to help keep this quasi "Series" network from cooking the HF drivers' voice coils. The LF current "might" prefer to flow through this coil depending on the reactive impedance of the following series cap ( also dependant on other circuit resistances. ) I also assume that the 16 ohm series resistor in line with the woofer helps limit the amount of LF current that would want to dump into the mid drivers' voice coil. The woofers' leg seems to be missing the necessary conjugate capacitor for a true series network .
- The 2 caps ( inline & in series ) with the horn drivers' circuit are electrical complications that I haven't yet fully thought through ( except they represent frequency dependant reactive impedances that are paralleled with the woofers' static load ) . I'm still assuming the woofer is operating without any LC elements on its' lines .
- To some extent, like all series networks' this circuit will be self balancing due to all the variable impedances. How effective the balance is I don't know since I generally stay away from these sort of layouts. Fun to think about though !

- ( Hmmmm, new 2410 diaphagms are now quite expensive ) .

:)

[Earl K]

Todd,

If this is your wiring setup then,

- You've got the horns' variable Lpad wired incorrectly ( it's essentially wired upside down which partially supports my initial post on series networks ) .

- Pin 1 of that variable Lpad should drain to the common ( low potential ) ground side of the circuit.

- Pin 3 , should pick up the LC "filtered" voltage from the ( high potential ) side of the horn circuit .


:)

[Ian Mackenzie]

Hey! Thanks for the visual Earl. I was starting to wonder why those two weren't talking about the whole mess over the phone.

I sent off toddalin's questions to someone asking them if they had time to write a Reader's Digest, Condensed Version-type of answer. They were actually good basic questions. We'll see what happens. I suspect they'll end up sending a book title or two instead. :p That's what I'd do. ;)

How amusing. At least its in DIY!

[Earl K]

Hey! Thanks for the visual Earl. I was starting to wonder why those two weren't talking about the whole mess over the phone.

You're Welcome !

But;
- I didn't create the posted jpg circuit ( as drawn ) . It was from an extremely brief posting of Todds' .

- Until a "fessed-up" schematic is actually presented, I can only speculate about where this went wrong .

:)

[toddalin]

You guys jumped the gun. Corrected schematic is included. All caps are Solen with Theta 0.01 mfd by-pass caps. All resistors are Mills 12 watt.


http://www.largescaleonline.com/eimages/lsolpics/Team_Member_Pics/toddalin/xover_schematic.jpg






Todd,

If this is your wiring setup then,

- You've got the horns' variable Lpad wired incorrectly ( it's essentially wired upside down ) .

- Pin 1 of that variable Lpad should drain to the common ( low potential ) ground side of the circuit.

- Pin 3 , should pick up the LC "filtered" voltage from the ( high potential ) side of the horn circuit .


:)

[Zilch]

"Thanks, at least I know I'm not crazy and my analyzer is fairly accurate in that range.http://audioheritage.org/vbulletin/images/smilies/blink.gif Doesn't my testing demonstrate that it is possible to get rid of the 500 Hz notch, perhaps at the expense of ??? However, with proper crossover design/implementation the notch could be removed without too many other repercussions?I think it's important to understand the source and nature of the problem (and whether it's real, of course,) to determine the optimum solution. Crossovers are frequently tuned for specific driver combinations and cabinet configurations, but, as Earl suggests, that tuning must be accomplished in particular ways that do not compromise the primary function. Having begun with the NL200b core, I'd expect this one to be well positioned already.

I'd suggest, at this point, you'd be significantly better off letting your EQ take out that notch at 500 Hz than adding series resistance to the LF. We need to do more to understand this better, though. If it's phase interference due to geometry, EQing by any means, i.e., externally or within the crossover itself, may produce other undesirable sonic consequences. I can guarantee that adding 16 ohms in series with the woofer is not on the list of preferred approaches....

[Earl K]

(7) Alternatively, send me one of the crossovers, and the schematic "as built." I'll break out an LE175 and 2402 here to run with 2235H and test/troubleshoot it....

- Zilch, if you are going to get into this and help Todd out , then can I suggest that it would be worth your while to also duplicate ( in a mockup ) the physical distances between the woofer & the DLH175 . Assuming the schematic presented is what was actually first wired up , then the acoustic summation between the 2235 and the horn must of just plain sucked ( to force Todd to run the woofer fullrange ).

- Likely ( maybe obviously ), different lowpass points need to be chosen for the woofer to horn transition . Perhaps leaving out the woofers' 24uf cap and then creating a "blended Zoebel" that is part low pass & part impedance equalizer is a good way ( meaning , economical ) to reclaim some lost midrange .
- Also, the le175 has less lower-midrange bloom than a le85. Therefore the 3-pole hipass on the horn-circuit might be "tailored" incorrectly for this driver/horn setup . This could easily exaccerbate any inherent FR weakness ( in the 500hz to 1K area ) .

:)

[toddalin]

-
- Likely ( maybe obviously ), different lowpass points need to be chosen for the woofer to horn transition . Perhaps leaving out the woofers' 24uf cap and then creating a "blended Zoebel" that is part low pass & part impedance equalizer is a good way ( meaning , economical ) to reclaim some lost midrange.

:)

Actually been there/done that. Disconnected the cap entirely as well as trying various resistances between the cap and gnd. Adds very little upper end boost to the woofer, such that it is audible, but the analyzer doesn't see it and the spectrum remains pretty much the same..

[Zilch]

Thanks you, Earl, for your input here. I think where I'm headed is that the 500 Hz "problem" is apparent in running the woofer just by itself, in this cabinet, with no crossover, which I'm about to further verify in a different room.

While the blending with the LE175 may also be a factor, my approach is to look at each separately, and the woofer in this box seems to be primary here....

[Earl K]

...(snip).... I think where I'm headed is that the 500 Hz "problem" is apparent in running the woofer just by itself, in this cabinet, with no crossover, which I'm about to further verify in a different room..

- Okay,, makes sense .


Actually been there/done that. Disconnected the cap entirely as well as trying various resistances between the cap and gnd. Adds very little upper end boost to the woofer, such that it is audible, but the analyzer doesn't see it and the spectrum remains pretty much the same...

- That makes sense when cross-referenced against Zilches full-range woofer measurements .

So Guys, what common measurement ( in both of your setups ) does a dimension/distance of 13" to 15" represent ? ( Floor Bounce Cancellations can create big "V" notches in a woofers' response & they are very hard to overcome )

:)

[edgewound]

How did it sound with your ears before you listened to it with your 10 band analyzer. You really can't expect to get accurate results to compare without having the same test equipment as your designer...there's gonna be band overlap that is skewed (or is that skewered?) on your 10 band that a 31 band can look at with more precision. The 10 band has to do something with all that info and put it somewhere. Maybe it's putting it where it's just screwing things up visually rather than audibly. Just my thoughts...feel free to correct me.

[Zilch]

So Guys, what common measurement ( in both of your setups ) does a dimension/distance of 13" to 15" represent ? ( Floor Bounce Cancellations can create big "V" notches in a woofers' reponse & they are very hard to overcome )Here in Berkeley, woofer is centered 18" off the floor.

I'm suspecting the culprit is the baffle being recessed 4" back from the cabinet face. The recess is 22.5" H x 29" V. The sides are "vented." The 29" correlates....

[Earl K]

Now punch it into spice and see what's up.
Or else just hook it up to the "Standard JBL Test Jig" and run a voltage drive.

***

Oh, also, go ahead and punch in the whole enclosure/woofer system into BB6P and model it. Then open a second instance and insert that 16 ohm series resistor. Compare the data if only for future reference.

- Zilch, are you able to do this ?

- I don't have BB6P or a Spice modeller.

- A voltage drive of the built network , would be illuminating . :o:

[toddalin]

How did it sound with your ears before you listened to it with your 10 band analyzer. Just my thoughts...feel free to correct me.

As I noted, I can hear the 500 Hz dip as a loss of intelligibility, especially of the male voice. That's what led to the testing.:D

[Zilch]

- Zilch, are you able to do this ?

- I don't have BB6P or a Spice modeller.

- A voltage drive of the built network , would be illuminating . :o:

Yeah, all that suff. :p

Giskard modeled it in the original thread:

http://audioheritage.org/vbulletin/showpost.php?p=84450&postcount=41

Ignore the yellow and white, which show the HF boost in the original, from which this one is derived. Todd's design LF and MF are the green and orange.

Part of the problem here is that the project is now in multiple threads.

Next time I see Todd in person, I'm gonna konk him for this.

Me, I'm the champion of the "Single-thread" aproach.... :D

Here's the history:

http://audioheritage.org/vbulletin/showthread.php?t=8507

http://audioheritage.org/vbulletin/showthread.php?t=9009

http://audioheritage.org/vbulletin/showthread.php?t=9134

http://audioheritage.org/vbulletin/showthread.php?t=9176

[Earl K]

I'm suspecting the culprit is the baffle being recessed 4" back from the cabinet face. The recess is 22.5" H x 29" V. The sides are "vented." The 29" correlates....

- Well, I'll leave it to you to suss that out .

- If you have one, put a shorter-coil 15" JBL in the box ( like a 2225 or a 2205 ). Maybe the L200 box was deliberately engineered to cancel out a 500 hz bump from one of those older type woofers which I think are hotter in this area .

- Okay, I'm just grasping at straws now, so I'm signing off .

:)

[Zilch]

Ok, then verify that the voltage drive of the actual physical network matches the model. Remove it as a suspect.

Get a bigger flashlight on it like edgewound suggests. Use CLIO.

Go ahead and investigate the physical aspect of the driver enclosure/relationship like you're thinking about. You might have to go with EQ or a passive notch filter to solve the problem. Don't be too afraid of the notch filter. Yeah, alot of people would scoff, but JBL used one in the Everest and it worked.Thanks, Giskard.

I've suggested that Todd might send one of the three he built up here for testing. Voltage drive will be easy to verify.

We'll be asking for your help in fixing this with a notch filter, if it comes to that.

This SHOULD be a good generic (read "no autotransformers") 800 Hz network for the common JBL compression driver/ring radiator three-way with extended-bass woofer. It's well worth working through these difficulties, I believe.

[Not to mention continuing the amusement value.... :thmbsup: ]

[edgewound]

Thanks, Giskard.

I've suggested that Todd might send one of the three he built up here for testing. Voltage drive will be easy to verify.

We'll be asking for your help in fixing this with a notch filter, if it comes to that.

This SHOULD be a good generic (read "no autotransformers") 800 Hz network for the common JBL compression driver/ring radiator three-way with extended-bass woofer. It's well worth working through these difficulties, I believe....

If I may....I think one of your variables that is not so common in your equation, is the midrange horn/driver combo being used in this setup. The 2235 variant being used was teamed up with the LE85/2420-H92/2312, rather than the 175 driver and the smaller horn. It seems to make sense to me that there will be a hole in your response by using a crossover in this range with these components....the 175 driver just doesn't have the muscle to reach into 500Hz territory, and there might be time alignment phase cancellation issues with the shorter horn. If you stick with your new crossover, maybe you'd find that an LE85/2420(2421)-H92 horn would work better....otherwise shift the crossover point up to about 1000-1200Hz.

[Zilch]

If you stick with your new crossover, maybe you'd find that an LE85/2420(2421)-H92 horn would work better....otherwise shift the crossover point up to about 1000-1200Hz.Yes, we went over that pretty thoroughly in the original thread.

See the sim curves. The MF is down 25+ dB at 500 Hz. It's in the woofer/cabinet, if it's real. Note also the LF Q is high, there, perhaps in partial compensation.

Though I'm reluctant to bring it up, we may be able to tweak it some in the Zobel.

Also, though I've been pretty adamant about the 16 Ohms series resistance thing here, I recall measuring the original LF inductors at a surprisingly high value.

Naw, they was just 0.81, 0.84 Ohms. That's not it....

[Zilch]

The 2235H simply isn't going to like a whole lot of Q added to it. It just isn't. There's no point in using a nice big wonderful transducer like a 2235H if you're going to go ahead and stick a big-ass resistor in front of it and choke it off.

We'll wait and see what else you find.Yup, agreed. I'm gonna quit typin' and go test more, now.

This is plenty good fun, and learnin' new stuff, too! :D

[Well, new HERE, at least.... :p ]

[Zilch]

1) Second unit above, in a different room. Same deal.
2) Insulation all around.
3) Ports closed.
4) Yup, a woofer in there. ;)

Standing waves inside, maybe?

[Off to Home Depot for more insulation....]

[Zilch]

Run an impedance curve a couple hundred Hz above and below your suspected frequency. Let's look for an anomoly.No problem. Gonna be a 30 ft."long wire" reading, though.

[Hand truck's at the warehouse.... :( ]

[toddalin]

While you're in there, try putting a horn on it and running up the ~500 eq bands for a flat response and see if that doesn't clear up male vocal intelligibility.;)

People been putting 2235's in L200 cabinets for ??? years and I'm the first to notice this 500 Hz suck-out phenonom???:blink:


1) Second unit above, in a different room. Same deal.
2) Insulation all around.
3) Ports closed.
4) Yup, a woofer in there. ;)

Standing waves inside, maybe?

[Off to Home Depot for more insulation....]

[Zilch]

Yup. It goes flat at 450 and 900 Hz.

What's that mean?

Cabinet's O.K.

All I heard was it got louder approaching 1 kHz.

I'll run it higher res....

Edit: 2 X res, port open, looks the same.

Note: Small world, folks. These were Giskard's 2235H's. :)

[edgewound]

While you're in there, try putting a horn on it and running up the ~500 eq bands for a flat response and see if that doesn't clear up male vocal intelligibility.;)

People been putting 2235's in L200 cabinets for ??? years and I'm the first to notice this 500 Hz suck-out phenonom???:blink:

L200's originally had an LE85 and LX16 at 1200hz

[toddalin]

Characteristic of the woofer??? If you go back to the thread where I compared the 2235 to the W15GTI in my center cabinet, you see the same phenonom but shifted up a little. While the 2235 was through the N1200 low pass during the test, the effect of the filter should have neen minimal at the frequencies we're dealing with here (as evidenced by the response above 900 Hz).


Frequency Response
1500 84.5
1450 78.9
1400 87.9
1300 89.2
1200 90.5
1100 94.2
1000 101.4
950 101.8
900 101.9
850 99.0
800 98.3
750 98.7
700 96.6
650 98.2
600 99.1
550 100.8
500 100.1
450 99.3
400 99.6
350 98.9
300 99.8
250 100.0
200 100.0
180 100.0
160 100.5
140 100.3
120 101.4
100 100.7
90 101.8
80 101.5
70 101.6
60 100.9
50 99.5
45 96.9
40 90.4
35 89.7
30 89.6
25 90.4
20 89.3

[edgewound]

I'll stay out of it...you're in good hands;). My point is that there will be somewhat of a summation between the LF and MF drivers system response as the two drivers fill in the crossover area. the L200 and L300 both had BIGGER drivers (LE85)that are better able to fill-in at that octave that's giving you pains. Hence that 500Hz suck out....but maybe I'm peeing into the wind.

[toddalin]

Egads! Don't post that stuff. Zip it in a text file or make a graph! :rotfl:

Is Todd using my 2235H's or did he recone some other cores?

I know what you are alluding to edgewound. I think Zilch is running this measurement with just the 2235H and the box, nothing else.

2205Cs remagnetized and reconed as 2235s by OCS last December 2005.

http://www.largescaleonline.com/eimages/lsolpics/Team_Member_Pics/toddalin/Woofers.jpg

[Zilch]

Ran 1000 points. Yup, it's one of your 2235H's. Todd's are recent recones, I think. He'll tell us.

Flat from 433 to 490 with a blip at 470
Hump at 808, then flat from 860 to 960

Shall I add more insulation to the box?

Edge: These are L200B-based crossovers, 800 Hz. Load was 136A (2231) and LE85 on the shorter horn.

[Zilch]

Pull the mic back and run another RTA.How far? 1M?

[Zilch]

Reconfiguring for RTA....

Sorry. Gotta reboot. WT2 messes with the camera USB connection....

[toddalin]

Run another RTA with the mic further back.

I just don't believe you guys should be getting these depressions with these drivers.

I guess I'd put the networks back in and pull the conjugates off the 2235H's, see if that adds a bit in the 500 Hz to 1 kHz range.

Thought you were the one who asked me what I was going to do between 500-1k Hz? Sounded like you expected it all along.

[Zilch]

Sounds spooky. What was it configured for?WT2. I moved everything to get the shortest possible WT2 connection without having to haul the L200 up 3 steps with no handtruck. :p

It's apparent why I never saw it before. I'd have just ignored the little notch at 900 Hz, though I believe it HAS irritated me in the past. It's nearfield measurements that show the 500 Hz notch here....


1 M RTA:

[Zilch]

Todd's sending me one of the networks to test, and I can try whatever you suggest after verifying the voltage drive. I don't have a problem with that RTA response, but it's Todd's problem I'd like to help solve. Clearly, there's something going on in the 4" nearfield, which he can hear and I measure as well.

I'll probably beef up the damping inside the box in the meantime and see if that helps. Another 6" on the rear, or 4" all around.

Todd: Your RTA - where are you measuring from? Did you get a boom and stand for your mic? I mean, it's not the pillow on the couch thing still, is it?

Let's get on the same page. Take the 16 Ohms out from the LF (it's just WRONG, trust me,) and dial down the MF to achieve the best balance. AutoEQ from there.

Perhaps Giskard can discuss the implications of the 1.0 uF you added to the HF, which I believe makes it an 18 dB crossover, and the change you made to the HF inductor as compromise between N7000 and 8000.

I'm off to dinner, as well. We'll reconvene later.... :thmbsup:

[Zilch]

Thank you, Giskard. Yes, those are the values.

[Guess I'd be removing that 1.0 uF, first. :p ]

Yes, it looks like tuning or removing the Zobel will help considerably....

[toddalin]

Todd's sending me one of the networks to test, and I can try whatever you suggest after verifying the voltage drive. I don't have a problem with that RTA response, but it's Todd's problem I'd like to help solve. Clearly, there's something going on in the 4" nearfield, which he can hear and I measure as well.

Todd: Your RTA - where are you measuring from? Did you get a boom and stand for your mic? I mean, it's not the pillow on the couch thing still, is it?

Let's get on the same page. Take the 16 Ohms out from the LF (it's just WRONG, trust me,) and dial down the MF to achieve the best balance. AutoEQ from there.

Perhaps Giskard can discuss the implications of the 1.0 uF you added to the HF, which I believe makes it an 18 dB crossover, and the change you made to the HF inductor as compromise between N7000 and 8000.

I'm off to dinner, as well. We'll reconvene later.... :thmbsup:

Mic is on a tripod with legs collapsed such that the mic is just about/under horn level. I tried both 1 and 2 meters as well as putting the stand on the couch where I sit, all to the same conclusion with only minor variation.

The 16 ohms of resistance is not in the system. In reality, the added 1 mfd cap that is after the choke in the HF section is directly on the tweeter as opposed to before the L-pad. Once I determine if it stays or goes, I will remount it before the L-pad.

The 1.0 mfd cap was not in the original configuration and is not shown on the N7000/N8000 diagrams. However, the N7000/N8000 are meant to be used downstream from a 2-way unit and that would put additional caps in the tweeter signal path.

Note that the typical JBL 3-way network uses both of these caps, albiet they put the 1.0 mfd before the choke and the 1.5 mfd after the choke. Mine are vise versa.

In one configuration, I had the L-pad on the woofer connected as a simple reostat variable from about 18 to 0 ohms. In this configuration, the L-pad serves as a "contour" control and allows me to flatten the 500 Hz band. By dialing it to 0, it is out of the system and everything is normal (with the dip).

[toddalin]

With or without the networks in place?I'm thoroughly familiar with the stock JBL networks. I even have schematics for systems that haven't been released yet. You really need a modeling program so you can see what all this is doing. At the very least you need to rig up a voltage drive measurement system if you don't want to bother with a spice backage and the time required to learn to use it. What I've been attempting to convey all along is that this is completely unnecessary. Nobody puts a variable resistor in line with a 2235H unless they are just screwing around to see what happens.

Network was in place for all testing.

L-pad (variable resistor) is out of system completely and was, just as you said, for test purposes.

On yout graph, which line represents what?

[toddalin]

Please excuse my ignorance... and the yellow and green?

As previously noted, the addition of the 1 mfd cap is possible, but based on the N7000/N8000 was not in my original design (that's why it is currently hung direcly on the tweeter as opposed to pre-L-pad where it would ultimately take up residence). Ultimately, if it stays, I can reorder them if that proves beneficial.



The cyan represents your reversal of the caps.

[toddalin]

OK, now you're confusing me... my arrangement is set up as 1.5 mF S/0.5 mH P/1.0 mF S so is the yellow line?



From post #57

1.5 uF S, 0.5 mH P, 1.0 uF S = yellow
1.5 uF S, 0.5 mH P = green

Post #63

1.0 uF S, 0.5 mH P, 1.5 uF S = cyan

[toddalin]

Huh? You lost me. When did I ask you this? And why would I ask you this?

From Post #52... Sorry, it was Widget who asked.

[Earl K]

OK, ....( snip )..... my arrangement is set up as 1.5 mF S/0.5 mH P/1.0 mF S so is the yellow line?

- You've recently pointed out that there is a variable pad between the .5mH inductor and your 1 uf cap which modifies the interactions.

- "All" these passive elements interact with each other / and therefore need to be treated as a whole / hence Giskards advice ( as I understand it ) aimed at helping you obtain an insight to these realities .



:)

[toddalin]

Of course you are right. As I noted before, at this point the 1 mfd not on the board but is hanging on the tweeter so is after the pad.

But as I also noted, if it stays, ultimately it would be before the pad and the simulation shows this so it is also of interest.:bouncy:



- You've recently pointed out that there is a variable pad between the .5mH inductor and your 1 uf cap which modifies the interactions.

- "All" these passive elements interact with each other / and therefore need to be treated as a whole / hence Giskards advice ( as I understand it ) aimed at helping you obtain an insight to these realities .



:)

[Zilch]

Viewed in the context of Mr. Widget's measurements of actual driver response, I ask, rhetorically, if there is some rationale for shaping the UHF drive like that?

Certainly not the "above 10 kHz" portion, where the uncompensated response is basically flat out to driver rolloff.

My point is, this has all got to work together, and an additional significant factor is how it meshes with what the MF driver is doing in the 5 - 10 kHz range in combination with ITS filter.

I'd say it's better to rely on "The book," i.e., what has worked successfully in the past, for starters. Innovation is great, and I have plenty good fun myself trying stuff out, but without measurement tools to asess how it's all playing together with the requisite accuracy and resolution, I have to advocate in favor of staying on the well-traveled path:

Leave out the additional cap for now....

[toddalin]

Viewed in the context of Mr. Widget's measurements of actual driver response, I ask, rhetorically, if there is some rationale for shaping the UHF drive like that?

Leave out the additional cap for now....

The steeper slope keeps the 075 from adding to the already high 2-8K band from the horn. At least that's what my analyzer indicated, for what its worth. ;) Certainly I don't want the peak, and to attain a steeper slope using just the cap coil may be nice and preclude this incidence. But that would require more $$$ that may not prove audibly beneficial. The slope for the simple cap/coil does look nice and smooth and will probably be employed in the final, as is.

[Zilch]

But it's not the 500 Hz we were looking at in the woofer nearfield.

Here with LE85 mid, same with both of two known good ones.

Not enough overlap between the HP and LP filters, apparently, and the midrange looks right.

So, I'll work with the Zobel tomorrow as Giskard suggests.

And the LE175's.

UHF plays WAY hot; gotta dial it almost off to balance like this. Connecting its filter direct to the input wasn't a very good idea, looks like.

I'll try rearranging that, too....

[toddalin]

Based on the octave resolution of my analyzer, the hole appears at about 500 Hz. Also, based on my analyzer, the tweeter is not that hot even turned up pretty high, but my mic/equipment probably has more high frequency roll-off. :blink:

Just for kicks, what does the spectrum look like with 16 ohms in series with the woofer?;)



But it's not the 500 Hz we were looking at in the woofer nearfield.

Here with LE85 mid, same with both of two known good ones.

Not enough overlap between the HP and LP filters, apparently, and the midrange looks right.

So, I'll work with the Zobel tomorrow as Giskard suggests.

And the LE175's.

UHF plays WAY hot; gotta dial it almost off to balance like this. Connecting its filter direct to the input wasn't a very good idea, looks like.

I'll try rearranging that, too....

[Zilch]

1) O.K., here's the 16 Ohms in series with the woofer. All it does is attenuate, and all the bad stuff we've already discussed, of course.

2) The tweeter wide open runs 15 dB too hot. These are NIB 2402H.

3) Before anybody yells at me about the lens, here's the midrange only without the lens.

4) And here with the lens.

5) I ain't makin' this up.

Yeah, I'm doin' CONSTRUCTIVE stuff, too, but I don't have any answers yet.... :p

[Zilch]

We're chasin' room/measurement stuff at the lower frequencies (<1 kHz) here, apparently. With two-ways, I'm usually ignoring all that.

1) If I pull back to 2M, it's all good. See D'Appolito.

2) LE175 on DLH potato masher. To borrow a JBL engineering term: sounds "Dreadful." Playin' in a coffee can, like. S/N 7242, the foilcal says "8 Ohms," DCR is 6.5 Ohms.

3) Move it to HL91, and it's a decidedly different picture.

4) At 1M, the same "problem" as LE85. The 630 Hz notch is in the woofer response measurement, in fact.

5) At 2M, all's well. Note the hump at 500 Hz. This illustrates the difficulty of system measurement. Need a "bigger flashlight."

These sound very nice. The rising response through the upper midrange confers a "crispness" to their voice. I've kept all of the settings the same for this comparative testing. I'd probably dial these down some to suit my own listening taste. A bit too much midrange, as I'm hearing them now.

I did play with the Zobel. Taking it out bumps up the response in the 800 - 1 kHz region, where the curves do overlap. The system could be fine tuned that way, but it's not bad with the existing values.

I'll look more at the UHF connection thing. It's of concern that we're not getting at all the same behavior.

Other than that, I'd say you're good with these. Let your AutoEQ tell you what's needed. The basic performance is there....

[Zilch]

Just go bury it in the ground face up and suspend your mic over it. :pI recently read where that didn't work so well with L212.

[Heh....]

[toddalin]

Interesting difference between the HL87 and HL91. Do you think this is more attributable to the difference in shape or length?

The potato masher in my center cabinet is very old (has the old decal on it), has a different number (not HL87, but don't recall what it is..., 4-digits I believe), and is the same length as an HL91 (1" longer than a typical HL87..., and yes, I held them up side-by-side-by-side).

Also, based on your analyzer, when used with the UHF, there really isn't that much difference between the LE175 and LE85 when used on the same horn and both are pretty comparable in the 1K-7K range.

The more you look, the more you notice. ;) The LE85 on th3 HL91 without the slant plate shows a slightly extended frequency toward the top end. It could then be reasoned that the removal of one/some of the "baffles" in the "masher" would have a similar effect, maybe at the expense of some dispersion. Any chance of removing a few and giving this a try to see how it impacts the highs?



We're chasin' room/measurement stuff at the lower frequencies (<1 kHz) here, apparently. With two-ways, I'm usually ignoring all that.

1) If I pull back to 2M, it's all good. See D'Appolito.

2) LE175 on DLH potato masher. To borrow a JBL engineering term: sounds "Dreadful." Playin' in a coffee can, like. S/N 7242, the foilcal says "8 Ohms," DCR is 6.5 Ohms.

3) Move it to HL91, and it's a decidedly different picture.

4) At 1M, the same "problem" as LE85. The 630 Hz notch is in the woofer response measurement, in fact.

5) At 2M, all's well. Note the hump at 500 Hz. This illustrates the difficulty of system measurement. Need a "bigger flashlight."

These sound very nice. The rising response through the upper midrange confers a "crispness" to their voice. I've kept all of the settings the same for this comparative testing. I'd probably dial these down some to suit my own listening taste. A bit too much midrange, as I'm hearing them now.

I did play with the Zobel. Taking it out bumps up the response in the 800 - 1 kHz region, where the curves do overlap. The system could be fine tuned that way, but it's not bad with the existing values.

I'll look more at the UHF connection thing. It's of concern that we're not getting at all the same behavior.

Other than that, I'd say you're good with these. Let your AutoEQ tell you what's needed. The basic performance is there....

[Zilch]

Interesting difference between the HL87 and HL91. Do you think this is more attributable to the difference in shape or length?I know zip about horn theory, but there's an article on lenses in the site Library. I think it's the lens. Lookin' through it, it's a wonder anything passes.


Also, based on your analyzer, when used with the UHF, there really isn't that much difference between the LE175 and LE85 when used on the same horn and both are pretty comparable in the 1K-7K range.?So it seems. Remember, though, we shouldn't generalize from just this one sample of unknown history here. I yanked it from a C56 cabinet. There should be another here, somewhere. Hamilton wants to know how they play on CD horns, so I need to find it for a Q&D "Somethin'-'r-other."


Any chance of removing a few and giving this a try to see how it impacts the highs?Sure, whenever YOU get around to doin' that little inquest.... :p

[Zilch does CD, not lenses. You know the drill.... :thmbsup:]

[Zilch]

Where is your CLIO system?It's here, waitin' for me to solve the second computer problem, or drill some holes through the wall.

Mr. Widget suggests I set up a separate testing lab AND a listening room.... :D

[I ain't doin' no billion-dollar anechoic doo-dah here. Nope. :no: ]


You aren't the one I got a complaint about are you?I've only bugged 'em about:

1) Whether 1.5" OA horns like 2352 are operationally compatible with 243x drivers (they are,) and,

2) Gettin' 1.5" PT-F waveguides, which I KNOW they have because they've published pictures of them, recommended them, and gave 'em a part number.... :mad:

[If they sent me a pair, it might shut me up for a coupla days evaluatin'.... :thmbsup:]

[Zilch]

Found the second LE175. The horns are 1217-1290. The performance is the same. Call it the "Ski-boot" curve. :p

In the full system, as Todd observes, the overall frequency response looks good:

[Zilch]

I tried the crossover and drivers (LE175 and 2402) with a bunch of different horns here. Surprisingly, many of the CD horns are "plug 'n' play" with the appropriate adapters, since they aren't being asked to do VHF.

1) $10 PT-F95

2) PT-H1010

3) PT-H1010 mounting

4) Front view

[Zilch]

Regarding the UHF sensitivity issue, I had to pad the 2402H down 10 dB to get it into the L-Pad adjustment range.

First try at remedy was to connect the HF after the fixed (2.5, 5-Ohm) attenuation in the MF circuit. That only got a couple of dB, though, so I reconfigured the sample per the standard 310x cascaded configuration.

That works very well; the full MF attenuation is also applied to the UHF. Once the MF/UHF balance is established with the UHF L-Pad, then the overall balance is adjusted with the MF control. It's intuitive, and easy to get it all properly balanced. There's still plenty of sensitivity left to overdrive the UHF for "air" or "sizzle," if desired.

I did observe one characteristic which may explain the different performance results Todd and I are obtaining: the 2402 is quite "beamy" in the higher frequencies. The two RTA plots below show the response on-axis versus 30° off-axis.

The difference is substantial (6+ dB). Note that response above 10 kHz falls off decidedly more (2-3X).

"Yeah, but 30° is a LOT, too!"

Well, no. It's the difference between being at 12:00 and 1:00, only, right? So, for comparable measurements, all must be taken precisely on UHF driver axis....

[Zilch]

O.K., so here's the final schematic and a proposed low-cost parts list.

For those who haven't followed this project from the beginning, what's important here is that it's a generic solution to building a DIY 800 Hz MF, 7 or 8 kHz UHF 3-way crossover for a variety of JBL woofers, 16-Ohm compression-driver midranges, and 07x/240x UHF drivers without using tapped inductors (autotransformers).

It is derived from combining N200B (LX200B) and 3105/6 (N7/8000), in cascade. I've tested it with 2235H LF, LE175, LE85 (2420), and 2446J MF, a bunch of different horns/waveguides, and 2402H (075), 2404H-1, 2406H, and 2407H UHF. It may also be suitable for use with 222x woofers.

Connect LF and HF inputs in parallel for single amp, or biamp passive, as shown. For triamp, separate the UHF section (bottom), as well.

Disclaimer: As with all DIY projects, mileage may vary. :p

[Let us know how it works for you, please, and any suggestions for improvement....]

[toddalin]

Looks good! You left off mention of the six 0.01 mfd by-pass caps.




O.K., so here's the final schematic and a proposed low-cost parts list.

For those who haven't followed this project from the beginning, what's important here is that this is a generic solution to building a DIY 800 Hz MF, 7 or 8 kHz UHF 3-way crossover for a variety of JBL woofers, 16-Ohm compression-driver midrange, and 07x/240x UHF drivers without having to mess with tapped inductors (autotransformers).

It is derived from combining N200B (LX200B) and 3105/6 (N7/8000), in cascade. I've tested it with 2235H LF, LE175, LE85 (2420), and 2446J MF, a bunch of different horns/waveguides, and 2402H (075), 2404H-1, 2406H, and 2407H UHF. It may also be suitable for use with 222x woofers.

Connect LF and HF inputs in parallel for single amp, or biamp passive, as shown. For triamp, separate the UHF section (bottom), as well.

Disclaimer: As with all DIY projects, mileage may vary. :p

[Let us know how it works for you, please, and any suggestions for improvement....]

[Zilch]

Looks good! You left off mention of the six 0.01 mfd by-pass caps.Thanks, Todd.

Allegedly, Solens don't need bypass caps, so they're cheaper than Daytons + Thetas. Like you, I'd probably bypass, anyway.

Tomorrow's a holiday, so I get another day to play here. :p

I'll send your prototype back on Tuesday leaving in the changes I made for you to try. It's easily restored to your original topology, if you decide to do that, tho.

May have to build myself up a pair of these; they seem to work with just about anything here.... ;)

[johnaec]

Zilch - your crossover looks great to me - perfect for PA use with 2226H, 2446H/2380A, and 2405H. Or would it need to be modified for the flat-front 2380A constant directivity bi-radial horn?

And no - I haven't forgotten about the cabinet stuff we're planning. I actually finally got the 2214H's I want to work with, for L100T3 components/crossovers in your (what'd you call them - Z1??) special angled cabinets.

John

[Zilch]

Hi, John!

While, in theory, some compensation would be required between 3.5 and 7 kHz, I'm able to dial in 2446J on 2380A with 2402H fairly well (RTA below).

With 2445, the specs say it needs a little help between 800 and 1000 Hz, which can be provided by adjusting the Zobel, depending upon which woofer is used.

http://www.jblpro.com/pages/pub/components/2380a.pdf

Compare to two-way with 3110A results here:

http://audioheritage.org/vbulletin/showpost.php?p=81644&postcount=75

Alas, no 2405H here to try. :(

[We're gonna build both Z1 and Z2 cabinets, remember? :p ]

[johnaec]

Alas, no 2405H here to try. http://audioheritage.org/vbulletin/images/smilies/frown.gif I can let you borrow a NIB 2405H if you want to experiment, (don't do it for me - I'm OK with my needs). I also have a couple 7K crossover cards for the 5235's coming. 'Gonna have to start putting all this stuff together sometime... http://audioheritage.org/vbulletin/images/smilies/wink.gif

John

[Zilch]

If you can get down here tomorrow, John, you can hear it for yourself!

Gotta send this crossover back to Todd on Tuesday for him to complete his system. I've delayed him with all my tweaking around with it here.

Not that I mind storing your 2446J and 2380A horn, mind you.


[Listening to it right now. Be nice to have TWO of them, of course.... :p ]

[toddalin]

Actually, if you were to stick with one brand cap (excepting the bypass caps) like I did, I priced it out and the Solens were only slightly more than the Daytons. This is because Solen had the 33 mfd value (Dayton doesn't) and these are cheaper than two Dayton 16.5 mfds. (May mave also been the case for the Solen 24 mfd vs two 12 mfd Daytons).

On the other hand Dayton makes a 16.5 that could have replaced the two Solen 8.2 mfds and would have saved some money.

If this was done for PA use, or even just slightly less discriminating tastes, the $3.5 Mills resistors could be replaced by $<1 resistors. If cap types were mixed and lessor resistors specified, the crossovers would probably come in at under $100 each.

BTW, when I put this on the scope over the weekend, I found the actual crossover points (-3 dB) to be maybe a little under 700 Hz (on the bench). There is really no reason that couldn't even be used with the 375 drivers. I'll look at the one in the cabinet driving its actual load today.

Regarding cascading the tweeter. If the L-pad can attenuate the tweeter to the proper balance level with the tweeter signal only going through the cap/coil (not cascaded), wouldn't this more direct hook-up result in improved sound simply by virtue that the signal goes through fewer components that may each leave their own sonic signiture??? :blink: (The minimalist approach.)


Thanks, Todd.

Allegedly, Solens don't need bypass caps, so they're cheaper than Daytons + Thetas. Like you, I'd probably bypass, anyway.

Tomorrow's a holiday, so I get another day to play here. :p

I'll send your prototype back on Tuesday leaving in the changes I made for you to try. It's easily restored to your original topology, if you decide to do that, tho.

May have to build myself up a pair of these; they seem to work with just about anything here.... ;)

[toddalin]

If you want to hold on to it for another week, no problem. May as well get all the mileage out of it that you can.



Thanks, Todd.

Allegedly, Solens don't need bypass caps, so they're cheaper than Daytons + Thetas. Like you, I'd probably bypass, anyway.

Tomorrow's a holiday, so I get another day to play here. :p

I'll send your prototype back on Tuesday leaving in the changes I made for you to try. It's easily restored to your original topology, if you decide to do that, tho.

May have to build myself up a pair of these; they seem to work with just about anything here.... ;)


Yes, but compared to your others and personal experience, how do they sound???

[Earl K]

Hi Todd


If this was done for PA use, or even just slightly less discriminating tastes, the $3.5 Mills resistors could be replaced by $<1 resistors. If cap types were mixed and lessor resistors specified, the crossovers would probably come in at under $100 each.

- I use 50¢, 10 watt power resistors & quite like them . I never blame the resistors for all the different textures my various cap recipes conjure up .


BTW, when I put this on the scope over the weekend, I found the actual crossover points (-3 dB) to be maybe a little under 700 Hz (on the bench). There is really no reason that couldn't even be used with the 375 drivers. I'll look at the one in the cabinet driving its actual load today.

- I built up this networks' band-pass portion about 10 days ago. I measured an electrical 3 db down point ( with actual horn load in place ) of @ 780 hz ( if my memory is correct .) I used the "db counter" function on my Loftech/Goldline generator to measure the electrical drop . I did need to approximate the .8mH coil with 3 .56mH coils arranged in a parallel/series topology / so that would introduce a margin of error in my findings .


Regarding cascading the tweeter. If the L-pad can attenuate the tweeter to the proper balance level with the tweeter signal only going through the cap/coil (not cascaded), wouldn't this more direct hook-up result in improved sound simply by virtue that the signal goes through fewer components that may each leave their own sonic signiture??? (The minimalist approach.)

- My own preference ( for sonic concerns ) would be to avoid running the UHF signal through the entire HF portion . I don't feel that getting the benefit of the padding from that 16 ohm Lpad is worth putting the UHF signal through the 2 extra inline caps.

- I'd deliver signal directly to the UHF section and spend the minimal $$ on a two resistor "fixed" pad ( placed just before/"source-side of" the variable 8 ohm pad ) . This mid circuit pad will allow one to rotate into the more normal operational area of the 8 ohm variable pad . Assuming the variable Lpad still "reads" around 8 ohms ( with load in place & attenuated @ 9 db ) I would create a simple 6 db fixed pad with 4 ohm & 8 ohm resistors ( the 4 ohm is the build-out resistor ) .



:)

[toddalin]

As noted, I did some more testing today. Using a sine wave generator through a working speaker system (under the actual load it is to drive), and using an a/c volt meter on the speaker input terminals the electrical minus 3 dB points are approximately:

Woofer LP is electrically (1/2 voltage) down 3 dB at about 700 Hz.

Mid HP (from woofer) is down 3 dB at about 650 Hz.

Mid LP (to tweeter) is down 3 dB at about 7,500 Hz (50 mH coil).

Tweeter HP (from mid) is down 3 dB at about 5,500 Hz (50 mH coil).

These pretty much confirm what the scope saw on the bench.

[Zilch]

Your remark was pointed out to me in an email and I felt compelled to respond.

You're insufferable! :rotfl:

UT, oh -

SYNTAX ERROR!!!

[How 'bout "Presumably," then.... ;) ]

Todd built with Theta bypassed Solens, his option.

I merely affirmed my understanding that was not necessary with the parts list above.

But ALSO, that one might try it both ways.

I stated I've never built with Solens, and don't know.

I ain't rippin' the Thetas outta Todd's crossover to find out, nope.

It sounds fine.... :p


BTW - It's a real sad day when you guys start sending me emails and PM's asking me about Zero's recommendations...

I hate riding herd on his bullshit and I don't have time for it. :banghead:Please don't be buggin' Giskard about what I post, folks. When I do really stupid stuff, he's not above calling me on it right here.

I don't mind that a bit. In fact, I appreciate it. It's usually illuminating. ;)

["Allegedly" may raise issues, but I don't suspect it warranted a bunch of background correspondence....]

[Zilch]

- I'd deliver signal directly to the UHF section and spend the minimal $$ on a two resistor "fixed" pad ( placed just before/"source-side of" the variable 8 ohm pad ) . This mid circuit pad will allow one to rotate into the more normal operational area of the 8 ohm variable pad . Assuming the variable Lpad still "reads" around 8 ohms ( with load in place & attenuated @ 9 db ) I would create a simple 6 db fixed pad with 4 ohm & 8 ohm resistors ( the 4 ohm is the build-out resistor ).Yep. Several options since I moved the MF L-Pad back into its original position per 310x topology.

For example, using 2406/7H as a tweeter on a CD horn uses most of the available headroom of the cascaded connection, because it's taking more energy to distribute the UHF more broadly as opposed to focusing it. I'm thinking I might connect the UHF filter for that combination ahead of the MF L-Pad with additional separate attenuation, as you describe, or direct, per Todd's original modification.

All of those choices for connecting the UHF are easily available, now.... :thmbsup:

[Johnaec just dropped off another 2446J/2380A and a pair of 2405H UHF drivers to try.... :bouncy: ]

[Zilch]

As noted, I did some more testing today. Using a sine wave generator through a working speaker system (under the actual load it is to drive), and using an a/c volt meter on the speaker input terminals the electrical minus 3 dB points are....It's interesting to review JBL's own published measurements on the crossover points:

http://www.jblproservice.com/pdf/Network%20Schematics/3105%20Network.pdf

http://www.jblproservice.com/pdf/Network%20Schematics/3106%20Network.pdf

5 kHz for 3105 (I THINK that's a "5" there,) and 7 kHz for 3106? :hmm:

If I understand correctly, that's electrical. The acoustical performance will vary with the drivers used....

[Zilch]

Have we not already gone over these old networks sufficiently? You just keep going around in circles dude.I dunno. I'm merely pointing to additional information for Todd (and others, perhaps,) to consider in understanding their testing of the performance of these networks. Seems like several members may already be working with them....

[Zilch]

I think that is what you aren't getting Zilch. People actually read this freakin' forum and then go out and do all this shit. They actually take all this crap posted all over this forum and apply it.There are few subjects that have been more thoroughly massaged on these forums than biasing and bypassing crossover capacitors. With respect to this particular project, we discussed it in the initial thread. You recently posted in another thread regarding JBL's use of Solens in their latest designs. In my view, this is all good an useful information for everyone interested to consider.

I believe even the earliest information posted here indicated that biased crossovers using Solens don't require bypass capacitors. It's not my intent or responsibility to undertake a comprehensive exposition of all of the relevant design issues in the course of participating in a DIY project, though I believe I do a creditable job with the fundamental considerations known to me.

You seem to be arguing that it would be better not to do it at all. Here, again, this is not about bypass capacitors, but rather, the larger question of whether such endeavors belong in these forums, and I believe the vote is long since in on that one....

[briang]

I hesitated to write this at first...it might be unwelcome.:blink:

Thanks to Zilch for posting all this work.

[toddalin]

Boy, I'm one heck of a trouble maker!:p

When I initially asked for a design to build the crossovers supplemental to replacing the 130A's with 2235's in my L200 cabinets, I knew nothing about biasing and by-pass caps in crossovers. This was all learned on the fly.

Zilch stepped up to the plate and suggested the N200B as a viable option. He suggested the use of Dayton caps to save costs and even revised the circuit for biasing at my request. When the circuit diagrams were presented for consideration, there was talk of by-pass caps for use with the Daytons.

When I ordered the components, I decided to stay with one brand for the big caps that didn't cost an arm and a leg. Solen had the values I wanted, was only a little more expensive than the Daytons, and I had heard good things about them. The PE website also talks of by-pass caps and no mention is made that they are not necessary with Solens.

After the parts order was in, it came up that the Solens didn't need the by-pass caps. I already had them on order so figured WTF, may as well use them..., couldn't hoit, and as long as I was going to the trouble to build these things, what's another $20 each in the grand scheme of things?

Do by-pass caps help the sound of the Solens? Did JBL leave them off the Solens because there was no difference in sound or was it ultimately $$$ consideration of the bean counters? Would they have used them if $$$ was no object?

If the world wants to know if they make a difference in this instance, Zilch feel free to remove them for trial (no easy task I know as they are all tucked up tight under the bigger components). However, I am of the impression that if they do make a difference, it will be extremely subtle and unless you can instantaneously A/B the crossovers without and with the Theta's in place, you would never know. Not everything the ear hears can be quantified using our rudimentary instrumentation.

[Zilch]

Boy, I'm one heck of a trouble maker!:p

If the world wants to know if they make a difference in this instance, Zilch feel free to remove them for trial (no easy task I know as they are all tucked up tight under the bigger components). However, I am of the impression that if they do make a difference, it will be extremely subtle and unless you can instantaneously A/B the crossovers without and with the Theta's in place, you would never know. Not everything the ear hears can be quantified using our rudimentary instrumentation.'Zactly.

I'm gonna build up a pair with spaces for bypass caps. I'll put Thetas in one and none in the other. We'll A/B 'em at the Lansing Heritage Weekend Expo! :p

[Earl K]

As noted, I did some more testing today. Using a sine wave generator through a working speaker system (under the actual load it is to drive), and using an a/c volt meter on the speaker input terminals the electrical minus 3 dB points are approximately:

Woofer LP is electrically (1/2 voltage) down 3 dB at about 700 Hz.

Mid HP (from woofer) is down 3 dB at about 650 Hz.

Mid LP (to tweeter) is down 3 dB at about 7,500 Hz (50 mH coil).

Tweeter HP (from mid) is down 3 dB at about 5,500 Hz (50 mH coil).
- ( I'll assume these 50 mH coils are actually .5 mH )
These pretty much confirm what the scope saw on the bench.



- Well, okay, I'm baffled . :o:

- I just used the 1/2 voltage method ( on a partially reconstructed horn circuit / ie ; minus its' low pass section , but the rest of the horn circuit is present ).
- Capacitors are Solen ( I have boxes of them ) .
- All the test equipment signal grounds are referenced back to the amps output ground / so there is no floating ground .
- I built a woofer circuit & loaded it with an 8 ohm dummy load ( looking for interaction between the 2 network branches ) I didn't find any, though I didn't build up the Zobel . Maybe that's next .
- My reference frequency is 3000 hz ( to stay above this circuits' "bump-filter" ).
- Volt Meter is my old AC powered B&K 290 . Useful for its' huge metering . All the AC powered stuff is "warmed-up" .

And still:
- I get a 3 db down point for the horns' hipass at @ 760 hz ( with the 16 ohm Lpad fully open ) .
- With that same Lpad attenuated ( 10db, measured acoustically ) the new 3 db down-point has risen slightly to @ 775 hz .

So :
- Anyone have any ideas where the discrepencies might be arising between Todds' and my measurements ?

- Anyone want to comment on my testing methods and offer up some insights ?

Thanks ! :)

[toddalin]

'Zactly.

I'm gonna build up a pair with spaces for bypass caps. I'll put Thetas in one and none in the other. We'll A/B 'em at the Lansing Heritage Weekend Expo! :p

Actually..., what would provide the best representation would be a single crossover with a multi-pole relay on-board that could instantaneously kick the BP caps in/out at the flip of a switch. Alternatively, the relay could kick between two crossovers, but that introduces other differences (e.g., no two caps/coils are exactly identical so the crossovers are not identical and we are looking for what may be very small changes here). 6+pdt relays are common and cheap.

[toddalin]

Sorry, but I am a scientist (show me) and work in sound and noise and recognize that at least my ear has a poor acoustic memory and for me an A/B/X test would be the acid test.



I think you will find that it is easier to distinguish the difference by listening to the bypassed version for awhile until you become acclimated and then removing the bypass capacitors.

[Zilch]

I hesitated to write this at first...it might be unwelcome.:blink:Naw, no problem. We're just havin' plenty good fun with JBL stuff here, is all. :thmbsup:

[None of this is fatal, or even disfiguring, for that matter. Just keep the SPL within reason.... ;) ]


Yes, but compared to your others and personal experience, how do they sound???Well, I'm gonna say your crossover is smoother sounding that the vintage 3110/3105 combination I have used here recently.

Johnaec dropped off a second 2446J/2380A today. I'll set up a system with that and the old crossovers for A/B listening comparison.

Here it's running on your crossover with a 2405H slot tweeter, which he also brought to try. This combo works, too. Michael's (Nestawasright) system is gonna sound mighty good, if this is an indication. These 2" drivers and larger horns seem to have an inherent effortlessness and transparency to their sound, and the slot, well, everybody knows.... :D

[Also, with 2407H on $10 PT-F95 horn, bottom pics.]

[hapy._.face]

I think the consensus amongst those who have limited knowledge of this sort of thing (myself included) is that the technical information on this forum should be chased with a pound of salt. Two major variables influence my decision to spend any time or money on anything discussed herein:

1. The SOURCE of the information

2. The supporting data found elsewhere

One quickly realizes "who knows what" on this forum- even when you, yourself, may not know much at all. Anyone jumping into the whimsical suggestions of just any ol' member is taking the kinds of chances that (likely) he or she has taken many times before (JBL aside). That's their thing. OTOH, if people want to have fun tinkering with something that others think is a waste of time, so be it- so long as the 'tinkering' isn't a sermon on the new world order of doing things.

Taking all this into account, my best lessons learned were from actual mistakes- it's a beautiful thing, sometimes! This forum is interesting because you oftentimes get to watch others make them and work through them. It can also yield an amazing new discovery. I'm gonna learn something either way! :)

[toddalin]

Potentially my sine wave generator is off. I'll see how close it comes to A-880 on my digital keyboard tomorrow.




- Well, okay, I'm baffled . :o:

- I just used the 1/2 voltage method ( on a partially reconstructed horn circuit / ie ; minus its' low pass section , but the rest of the horn circuit is present ).
- Capacitors are Solen ( I have boxes of them ) .
- All the test equipment signal grounds are referenced back to the amps output ground / so there is no floating ground .
- I built a woofer circuit & loaded it with an 8 ohm dummy load ( looking for interaction between the 2 network branches ) I didn't find any, though I didn't build up the Zobel . Maybe that's next .
- My reference frequency is 3000 hz ( to stay above this circuits' "bump-filter" ).
- Volt Meter is my old AC powered B&K 290 . Useful for its' huge metering . All the AC powered stuff is "warmed-up" .

And still:
- I get a 3 db down point for the horns' hipass at @ 760 hz ( with the 16 ohm Lpad fully open ) .
- With that same Lpad attenuated ( 10db, measured acoustically ) the new 3 db down-point has risen slightly to @ 775 hz .

So :
- Anyone have any ideas where the discrepencies might be arising between Todds' and my measurements ?

- Anyone want to comment on my testing methods and offer up some insights ?

Thanks ! :)

[JuniorJBL]

my best lessons learned were from actual mistakes- it's a beautiful thing, sometimes!

Sometimes EXPENSIVE!! but it can be a wonderful thing:D

[briang]

Sometimes EXPENSIVE!! but it can be a wonderful thing:DThat describes nearly all my mistakes...:duck: ...

[Zilch]

I built up this networks' band-pass portion about 10 days ago. I measured an electrical 3 db down point ( with actual horn load in place ) of @ 780 hz ( if my memory is correct .) I used the "db counter" function on my Loftech/Goldline generator to measure the electrical drop . I did need to approximate the .8mH coil with 3 .56mH coils arranged in a parallel/series topology / so that would introduce a margin of error in my findings.Yeah, but do you LIKE it? :p

18 dB/octave HP and 12 dB/octave LP, no? And then 12 dB/octave HP to the UHF?

Is the 800 Hz crossing to the LF 12 dB/octave LP "asymmetric," then?

[Earl K]

- Well, okay, I'm baffled .
...........snip, snip, snip,...............
And still:
..........snip, snip, snip,...............
- With that same Lpad attenuated ( 10db, measured acoustically ) the new 3 db down-point has risen slightly to @ 775 hz .

- Okay, my best guess ( for Tues. ) is that the "tight-pack" layout of Todds' construction is forcing the .8mH coil into being a larger "virtual" value. I'll have to try some parts substitutions ( Wed. ) to see what value coil would cause this lower reading in F3 . ( Coils interact with other coils when placed in close proximity )

- In the meantime Zilch, you should verify Todds F3 numbers . Once finished ; lift the .8 mH coil from its' present location / extend with 12" of wire on either end of the coil/ replace the "new coil" into the circuit & remeasure the horns' F3 for its' high-pass .


- A couple of notes ; the .6 mH coil in the 3105 omits the use of both 20 ohm load resistors ( compared to the 3106 ) . Todds use of .5 mH coils begs the question ; "What about the usage of load resistors ? Impedance plots should be run with & without the resistors inplace . Then full voltage runs should be worked-up for each scenario so one can study the consequences of ad hac parts substitution .


Yeah, but do you LIKE it?

- My opinion, on the whole is pretty well beside the point / except to say / both Todds' and your declarations, that this "workedup design" is a done deal and ready to implement ,,, is vastly overstated - to put it mildly.



18 dB/octave HP and 12 dB/octave LP, no? And then 12 dB/octave HP to the UHF?

- A study of the actual derived "acoustic" slopes is enlightening .

regards :)

[toddalin]

"Okay, my best guess ( for Tues. ) is that the "tight-pack" layout of Todds' construction is forcing the .8mH coil into being a larger "virtual" value. I'll have to try some parts substitutions ( Wed. ) to see what value coil would cause this lower reading in F3 . ( Coils interact with other coils when placed in close proximity )

- In the meantime Zilch, you should verify Todds F3 numbers ."

Yes Zilch, if you can, please do. My equipment here is less than steller.

"Once finished ; lift the .8 mH coil from its' present location / extend with 12" of wire on either end of the coil/ replace the "new coil" into the circuit & remeasure the horns' F3 for its' high-pass ."

Earl, this is no easy task as the coils are glued in place.

"- A couple of notes ; the .6 mH coil in the 3105 omits the use of both 20 ohm load resistors ( compared to the 3106 ) . Todds use of .5 mH coils begs the question ; "What about the usage of load resistors ? Impedance plots should be run with & without the resistors inplace . Then full voltage runs should be worked-up for each scenario so one can study the consequences of ad hac parts substitution ."

I think that the 20 ohm resistors may have been a later addition after the 3105 specs were published (continued on-going development/specifications subject to change). Even Zilch's N200B has them and they are not included in that schematic either.

"- My opinion, on the whole is pretty well beside the point / except to say / both Todds' and your declarations, that this "workedup design" is a done deal and ready to implement ,,, is vastly overstated - to put it mildly."

I've never said it was a "done deal" and sent the crossover to Zilch to see if it could be improved (especially in the 500-800 Hz area).

"- A study of the actual derived "acoustic" slopes is enlightening ."

Agreed.

regards http://audioheritage.org/vbulletin/images/smilies/smile.gif

[toddalin]

Earl, as I noted, one possibility is that my function generator is off and I said I would check it against my digital keyboard.:o:

I think that your measurement is valid and these do crossover close to 800 Hz as they are supposed to. :applaud: When I play an A880 on the keyboard and listen to the "beat frequency", the function generator shows this to be at about 750 Hz on it's dial. So the display on the generator is at least 100 Hz low in this range.

I guess I'll have to find a way to recalibrate the function generator using the keyboard as a reference.:blink:


- Okay, my best guess ( for Tues. ) is that the "tight-pack" layout of Todds' construction is forcing the .8mH coil into being a larger "virtual" value. I'll have to try some parts substitutions ( Wed. ) to see what value coil would cause this lower reading in F3 . ( Coils interact with other coils when placed in close proximity )


- In the meantime Zilch, you should verify Todds F3 numbers . Once finished ; lift the .8 mH coil from its' present location / extend with 12" of wire on either end of the coil/ replace the "new coil" into the circuit & remeasure the horns' F3 for its' high-pass .


- A couple of notes ; the .6 mH coil in the 3105 omits the use of both 20 ohm load resistors ( compared to the 3106 ) . Todds use of .5 mH coils begs the question ; "What about the usage of load resistors ? Impedance plots should be run with & without the resistors inplace . Then full voltage runs should be worked-up for each scenario so one can study the consequences of ad hac parts substitution .



- My opinion, on the whole is pretty well beside the point / except to say / both Todds' and your declarations, that this "workedup design" is a done deal and ready to implement ,,, is vastly overstated - to put it mildly.



- A study of the actual derived "acoustic" slopes is enlightening .

regards :)

[Zilch]

I compared Todd's crossover to 3110/3105 using "Big Boy" drivers (2446J) and 2380A horns, 2407H tweeters on PT-F95 to match. :p

Had to pad down the 3110/3105 midrange output 6 dB to achieve balance with 2235H woofer, as frequently before, with mid control set to "Min."

Voicing is remarkably similar; I get strong imaging.

Todd's crossover is perceptibly cleaner sounding. It'd take Mr. "Golden Ear" Widget to be more definitive, probably.

I find myself wishing I had damped diaphragms to try. N/A in 16 Ohms, looks like, unless that'd be 2451SLJ?

Also now see y'all made a little "To do" list for me. ;)

"Migration"....]

[Zilch]

SIM with MF L-Pad set -6 dB, HF wide open, then actuals from system above:

LF taken 12" nearfield.

MF SIM HP -3 dB = 877.5 Hz per cursors.

[Earl K]

Hi Zilch

Some Observations ;

- What are the loads ? Are they dummy resistors or the actual transducers ?
- It's hard to actually read this plot .
- I'd suggest using a linear scale in the horizontal ( frequency axis ) .
- If continuing to use a log scale , then 10 hz to 100 hz to 1000 hz and so on , is more the norm .

More Observations on the Band-Pass Circuit ;

- The F3 , in the horns' hipass reads to me as @ 540 hz ( ??? ) . I just don't understand the cursors declaration of a 877.5 hz point . That seems to be some sort of anticipated Fx point or a point in a specified "XY" window where the woofers curve crosses a 3 db down point .

More Observations on the Woofer Circuit ;

- The electrical F3 , for the woofers' low pass appears to read; 360 hz . I doubt that is right / so I'm confused .

In Closing ;

- Is it only my eyes / giving me these numbers ?
- All in all, this plot creates more questions in my mind than it answers .
- I'd like to hear from others as to what they get from this FR plot .


:)

[Zilch]

What are the loads ? Are they dummy resistors or the actual transducers?The black SIM plot uses dummy resistors, nominal, at 8, 16 and 8 Ohms, respectively. If I use the actual impedances, nothing changes much.

The RTA plots are the actual sonic performance of the drivers connected to the crossover, with dummy loads on the sections not illustrated in each, 8, 15, and 8 Ohms. These illustrate the resultant acoustic slopes of each section.

The SIM scale is set to match the RTA scale, and the RTA plots enlarged so you can see how they relate graphically. It's easy. Think "200, 400, 600," etc.

It appears to me that it's all behaving according to design. The woofer rolloff is making the acoustic LF LP steeper than the electrical slope illustrated, matching the MF HP's 18 dB/octave slope. The acoustic crossover point is therefore somewhat lower. Cool.


The F3 , in the horns' hipass reads to me as @ 540 hz ( ??? ) . I just don't understand the cursors declaration of a 877.5 hz point . That seems to be some sort of anticipated Fx point or a point in a specified "XY" window where the woofers curve crosses a 3 db down point .You guys are measuring the F3 down point of the MF HP. I set cursor (a) to show what the SIM says it is (c,d). That does not vary with attenuation, but the "crossover point" does, of course.

Cursor (b) shows the M/H crossover point with L3,4 at 3105's 0.6 mH being 6.581 kHz with no UHF attenuation. That point shifts as the balance between MF and UHF is altered, one argument in favor of cascading: it holds that point stable when LF/HF balance is adjusted. It's easily seen that attenuating the UHF raises the MF/UHF crossover frequency as it "slides" down the MF LP slope.


The electrical F3 , for the woofers' low pass appears to read; 360 hz . I doubt that is right / so I'm confused.No, the curve is correct. See Giskard's earlier SIM of the N200B here:

http://audioheritage.org/vbulletin/showpost.php?p=84450&postcount=41

That particular part can be tweaked with the Zobel, if desired.

For better or worse, it's one reason we have confidence in this design: It's assembled from production models using similar drivers....

[Earl K]

- The sim scale is set to match the RTA scale so you can see how they relate graphically. It's easy. Think "200, 400, 600," etc.

Ahh, okay ! Thanks for that explanation .

- Why didn't you test the actual network , as contructed by Todd ? We'd all like to know the actual F3 points that he has achieved with real loads .

- Since this is a "SIM" of an actual physical network , could you please post a "screen-capture" of your softwares' data-entry page ? This will show the components ( with values ) that were used to create the "black sim plot".

- This way, we will know what band-pass configuration you are using for the components . ( Moving the LC components one side or the other of the circuits' Lpads does change the crossover points ) . The last orientation I saw was the cascading type with the LC elements, driver-side of the 16 ohm variable Lpad .

Thanks !







- Originally Posted by Earl K
The electrical F3 , for the woofers' low pass appears to read; 360 hz . I doubt that is right / so I'm confused.



No, that is right. See Giskard's sim of the N200B here:

- Wouldn't the F3 points be closer to 560 hz , considering the horizontal scale that's in use for both Giskards' and your just published sim ?


:)

[Zilch]

Since this is a "SIM" of an actual physical network , could you please post a "screen-capture" of your softwares' data-entry page ? This will show the components ( with values ) that were used to create the "black sim plot".It's per the construction schematic, above, i.e., cascade, with the MF HP after the L-Pad.

I'm reluctant to post the input schematic as it incorporates additional components to properly model the inductors, and fixed resistors in lieu of L-Pads, serving largely to confuse those who might want to build these.

So, if you'd PM me your eMail address, I'll be happy to send it along to you.... :thmbsup:


( Moving the LC components one side or the other of the circuits' Lpads does change the crossover points.)Yes, and we don't have the autotransformer helping us out with that.


Why didn't you test the actual network , as contructed by Todd ? We'd all like to know the actual F3 points that he has achieved with real loads.Yeah, I can do that, as well. Just gotta rig up an attenuated and isolated XLR input cable. I'll try to get that together this evening....


Wouldn't the F3 points be closer to 560 hz , considering the horizontal scale that's in use for both Giskards' and your just published sim ?Yes. It's certainly not 360 Hz.

[But, it's not 800 Hz, either.... ;) ]

[toddalin]

I assume that like me, you are using silver solder exclusively.:D



It's per the construction schematic, above, i.e., cascade, with the MF HP after the L-Pad.

I'm reluctant to post the input schematic as it incorporates additional components to properly model the inductors, and fixed resistors in lieu of L-Pads, serving largely to confuse those who might want to build these.

So, if you'd PM me your eMail address, I'll be happy to send it along to you.... :thmbsup:

Yeah, I can do that, as well. Just gotta rig up an attenuated and isolated XLR input cable. I'll try to get that together this evening....

[Zilch]

I assume that like me, you are using silver solder exclusively.:D


Pfffttt!!!



[At least it says "Rosin Core."]

:rotfl:

Two full system adjustment curves:

[Earl K]

- I'm reluctant to post the input schematic as it incorporates additional components to properly model the inductors, and fixed resistors in lieu of L-Pads, serving largely to confuse those who might want to build these.

- Don't be shy ,,, publish it .

- To address the concerns of the easily confused, dump it into a photo editor like "Paint-Box" . Then paint big red circles around those areas of parts' that represent substitutions or a deviation from the preferred circuit ( ie; in the case of the variable Lpads ).



:)

[Zilch]

All drivers connected and running

Low res = 5 dB/dotted line:

[Reposting the SIM so y'all don't have to scroll....]

[Zilch]

Hi res = 2.5 dB/dotted line:

[Zilch]

Well, lookie there!

All working according to design, apparently.

-6 dB at 800 Hz and 6 kHz, looks like.... :p

Unless somebody needs something else quickly, I'm sending it back to Todd today.

I'll build up a pair to investigate optimum UHF configuration further.... :thmbsup:

[Zilch]

What's going on here?

Three approaches to analyzing the crossover are shown above:

1) SIM (the black plot) is the theoretical performance. The circuit and component values are entered into a Spice simulation program which "runs" it in a mathematical model to display what it does.

2) RTA uses a microphone to "listen" to the actual performance and show it. Run each driver on the crossover individually with appropriate value resistors in place of the other drivers as "dummy" loads. There's an entire literature on getting accurate RTA (and other) measurements of loudspeaker performance.

3) "Voltage Drive" uses RTA as display device, but in place of the microphone, the actual crossover outputs (with all of the drivers connected and running) are each used as input. As with RTA, I'm using pink noise is the program source, attenuation as required, and an input isolation transformer to protect the RTA. White noise and sinusoidal sweep are additional source options.

Taken together, the three approaches provide means to assess crossover performance: Is it working according to design? Is the design appropriate for the task? Is it optimum for these drivers in this alignment?

[4313B]

While I've covered all this many times before I'm glad to see you are finally getting it Zilch. I'm not real keen on your regurgitation of my past posts but at least you read them all in addition to all the books and AES journals I've suggested. And your last post presentation here is pretty damn decent. You're also coming across better I think. You seem to have killed off those "we" people that were loitering around in your shop which is a welcome relief. No more quick and dirty shenanigans it seems. One more request?

Get CLIO up and running so we never have to see the orange RTA thingie again ok? ;) There has to be someone out there who can whip a CPU together for you. Sell an LE15B to fund it. Those things are boat anchors.

Or do you have a boat...




The above post is the kind of post I've been looking for from you Zilch. I'm serious. You've towed the line dude.


>

[toddalin]

What's going on here?

Taken together, the three approaches provide a means to assess crossover performance: Is it working according to design? Is the design appropriate for the task? Is it optimum?

Appears to.

Possibly but, see below.

Is it?, see below

It would appear to me that in the analysis, the woofer could still be crossed over a little higher. But I'm no expert.

[Zilch]

You've towed the line dude.It's taken no small measure of dedication on your part, for sure. Thank you for that....

BTW, what system are we lookin' at there?

[Heh, heh....]

[Zilch]

It would appear to me that in the analysis, the woofer could still be crossed over a little higher. But I'm no expert.Let me get mine built and CLIO running. The Zobel might be adjusted, if required, to accomplish that. I just don't have sufficient measurement resolution presently to make a determination.

In general, I'd say you're good to go for now. It's working, it sounds decent; it's time for some serious listening. :thmbsup:

Maybe think about upgrading your measurement capabilities if you want to do more of this. It's all good clean fun....;)

[toddalin]

Maybe think about upgrading your measurement capabilities if you want to do more of this. It's all good clean fun....;)

Your 64-band Barringer seems nice. Do they offer just the analyzer portion at a cheaper cost? I don't really need an eq and actually took my two Yamaha 31-band units out of the system to reduce background noise.

[Zilch]

The Behringer UltraCurve Pro DEQ2496 has certainly carried me a long way during the past year. It's not avalilable as RTA only, but the AutoEQ function is very useful as an analytical tool, even if you don't use it for actual EQ in your system. It tells you right away what (and how much) adjustment is needed to achieve any desired response curve.

I consider it test equipment, not a system component.

If you just want RTA, there's other options suggested in these forums.

Note: The UltraCurve RTA is 61 bands, but its EQ is 31 (1/3 octave)....

[toddalin]

The Behringer UltraCurve Pro DEQ2496 has certainly carried me a long way during the past year. It's not avalilable as RTA only, but the AutoEQ function is very useful as an analytical tool, even if you don't use it for actual EQ in your system. It tells you right away what (and how much) adjustment is needed to achieve any desired response curve.

I consider it test equipment, not a system component.

If you just want RTA, there's other options suggested in these forums.

Note: The UltraCurve RTA is 61 bands, but its EQ is 31 (1/3 octave)....

61 bands means that there is a band for every other note of the diatonic scale. Does anyone offer 122 bands. I wouldn't want any of my notes to feel left out. ;)

[Zilch]

I wouldn't want any of my notes to feel left out. ;)There's an inherent smoothing in the averaging it does to display with 1/6 octave resolution. For looking at the general trends, it's plenty. As you can see in some frequency ranges, it's anybody's guess what's going on due to measurement methodology artifacts.

The manual advises against attempting AutoEQ below 100 Hz, for example. It's just not giving real measurements below there in a standard setup, as you know from your nearfield testing experiments.

In certain respects, I believe your 10-band RTA might be effectively used as a low-frequency measurement device. The bottom 6 bands are all below 1 kHz, no? Maybe try "in-box" testing per Small.... :yes:

[Zilch]

Completed per the parts list above, first tests on my system look quite good. This is in cascade configuration, but HF is connected before the MF variable L-Pad, i.e., after MF high pass and fixed L-Pad. Call it "cascode," maybe.

http://audioheritage.org/vbulletin/showpost.php?p=93836&postcount=62

Note separate LF and MF/HF input (bridged here,) for potential biamp operation later, as desired.

2407H on PT-F95
2450JSL on 2380A
2235H in 4507, two ports closed

Per RTA used for balancing, it produces flat response (within +/- 2.5 dB) with HF at max (red), but that's too "hot" sounding. Both controls at mid postion shown gives more listenable green curve, HF -4 dB at 20 kHz.

[Sounds mighty nice, BTW.... :) ]

[Zilch]

O.K., I need some guidance now on a preferred method for measuring voltage drives with CLIO. I have a dedicated amp set up with it now, so I'll also be able to generate MLS response curves using Todd's driver configuration as well as this one.

I want to compare cascade, cascode, and standalone HF connections, about four different configurations for Todd and others. There are interactions here to document and analyze....

[toddalin]

Completed per the parts list above, first tests on my system look quite good. This is in cascade configuration, but HF is connected before the MF variable L-Pad, i.e., after MF high pass and fixed L-Pad.

2407H on PT-F95
2450JSL on 2380A
2235H in 4507, two ports closed

Per RTA, it produces flat response with HF at max (red), but that's too "hot" sounding. Both controls at mid postion gives more listenable green curve, HF -4 dB.

[Sounds mighty nice, BTW.... :) ]

OK, stupid question off-topic:

Why do you plug the top ports as opposed to the bottom ports? Wouldn't leaving the top ports open bring some bass up closer to the other drivers possibly resulting in a more coherent sound? Ever try it and listen for a difference?

I recognize that bass frequencies are fairly omni-directional.

[Zilch]

Why do you plug the top ports as opposed to the bottom ports?I believe having the ports closer to the floor takes better advantage of boundary effect to enhance the low bass they produce, but I have NO (ZIP, ZERO, ZILCH) substantiation of that to cite.

Coherence would not come at the port frequencies; as you state, they are omni-directional. Generally speaking, having the ports below the woofer lets the radiation that matters in this respect originate closer to the mid/high drivers as they can be closer together.

I can't do much about that with these boxes, but I am using them here because the L200 woofer is even further separated, down at the bottom of the box. The c/c distance presently is 17" to the 2380A; sitting on top of L200, it was 25" or more. These 4507's are more coherent sounding.

[4313B]

I recently read where that didn't work so well with L212.

[Heh....]Yeah but why?

[Zilch]

Yeah but why?Elimination of boundary effect, as I recall, resulting in incorporating an over-emphasized mid-bass. Let me go check before holding me to it, tho.... ;)

Edit:

http://www.audioheritage.org/html/perspectives/smith.htm

Saw confirmation of it somewhere else, though I forget where right now. D'Appolito, maybe....

[4313B]

The crossover was designed to give a highly flat system under those conditions.

Terry showed me a 3 element passive notch circuit that was developed to fix the response and so it sounded quite good.

This was one of the main incentives for developing the ground plane measuring technique.

[Zilch]

Accurately measuring system bass response is certainly challenging, in my experience. For the most part, I've just ignored it. Some of my stumbling attempts at near-field appear earlier in this thread.

I think I'm gonna punt and try in-box per Small, but I still can't get Andre interested in selling me a second CLIO mic to do it with. He just won't respond to my phone calls or eMails. :(

Ground-plane also looks interesting; gotta read up more on that. :thmbsup:

For now, it's near-field and its shortcomings, though I'm getting better results with it lately....

[Zilch]

Higher XO frequency (~8.5 kHz) for 077/2405:

http://audioheritage.org/vbulletin/showthread.php?p=130246#post130246

Two-way version with adjustable HF:

http://audioheritage.org/vbulletin/showthread.php?p=127396#post127396

[toddalin]

Higher XO frequency (~8.5 kHz) for 077/2405:

http://audioheritage.org/vbulletin/showthread.php?p=130246#post130246

Two-way version with adjustable HF:

http://audioheritage.org/vbulletin/showthread.php?p=127396#post127396

You recognize that the 1.0 mfd/coil/1.5 mfd cap in the HF path adds that unnatural peak just before the slope (either Giskard or Widget posted the results). I would suggest that one try using just the 1.0 mfd and the 0.16 coil for a more gentle slope, but alleviate the peak. May like the results better, but if not, it's simple enough to add the 1.5 mfd afterwards.

[Zilch]

(either Giskard or Widget posted the results)Find it and link it here please, Todd. :thmbsup:

[I'm attempting to get this all referenced in one place....]

[toddalin]

Find it and link it here please, Todd. :thmbsup:

[I'm attempting to get this all referenced in one place....]

It was Giskard, and of course he deleted all of that stuff way back when. Go to Post #45 of this thread and you'll still find reference to this effect.