Alternative to tapped inductor in 3110a or 3115a?

[robfive]

Greetings,

I have a multitude of 2225H, 2445J (with 2380 horns), and 2404H drivers that I am trying to put to use. I would like to build something similar to a 4673 clone but the tapped inductor of the 3110a and 3115a crossovers have me stumped. Is there any way around them using multiple standard inductors?

I have seen the 3110 equivalent HP filter that gets around the tapped inductor, but I need the "a" version to go with the CD 2380 horns.

I have used the Behringer CX4300 in the past but I want to keep this a passive crossover.

My next question is how much of an issue is CD compensation if the 2404Hs are blended in at 5kHz?

Thanks for any input.
Robert

[Baron030]

Hi Robert

You cannot use multiple standard inductors simulate a tapped inductor. A tapped inductor acts like a transformer. And since, there are not separate primary and secondary winding, these devices are often referred to as Autoformers or Autotransformers. I don’t know of anyone that is manufacturing these devices, so unless you can track down some JBL crossovers. You might need to wind your own coils, which is not something that should be attempted without test equipment to measure the inductance.
And due to the necessary inductive interaction between taps, I would recommend using an iron core rather than an air core design. Rewinding an Erse Super Q coil would be a good starting point for such a project.

Here is a link to the inductance ratios that would be needed:

http://www.audioheritage.org/vbulletin/showthread.php?5376
http://www.erseaudio.com/Products/SuperQCoils16Gauge/ESQ55-16-3300


Baron030:)

[robfive]

Thanks for the info, Baron030. I am comfortable with modifying an existing inductor or creating a new one from scratch. I am also familiar with the theory of transformer windings but I need to read up on the mechanics of winding multiple taps. I have now read through many of the autoformer threads here but they seem to dwell on the theory more than the execution of the tapped inductors.

I guess my biggest question is do you just start with the primary inductance winding (say 3.3mH) and then wind the secondary coils over the primary according to their calculated ratios? Are the secondary coils connected to each other? Then I notice you mention that there are not seperate primary and secondary windings.

Thankfully I am not in a hurry to finish this project as it appears that I still have quite a bit of reading/learning to do.

[Baron030]

Hi Robert

Here is how I would attempt to make a tapped 3.3mH coil for the 3110a crossover. I would start with a 3.3mH Erse 16 gauge Super-Q coil and some additional 16 gauge magnet wire and an inductance meter. The reason for the additional magnet wire is that when you are removing windings from the Erse coil. The insulation is going to come off that wire because Erst cements the wires together with enamel or varnish. Start by removing windings and while measuring the inductance until the coil’s inductance is reduced to 2.97mH. Keep track of the length of the wire removed and the number of turns removed and record those values. Then continue removing windings and measuring inductance until the inductance is reduced to 2.7786mH. Again record the length and number of turns of wire removed. Then continue removing windings and measuring inductance until the inductance is reduced to 2.4717mH. The wire lead that is closest to the iron core is “Lead # 1” and the outer wire lead is “Lead #2 = -6db”. Attach some new magnet wire to “Lead #2”, it should be a few feet longer than the total amount removed. Start rewinding the coil until the inductance is back up to 2.7786mH and then form a loop marking it as “Lead #3 = -8db”. Continue rewinding until the inductance is 2.97mH and then form another loop, marking it as “Lead #4 = -10db”. And then continue rewinding until the inductance is back up to 3.3mH and marking it as “Lead #5”. While winding you will want to brush some varnish to cement the winding together. Needless to say this is all very labor intensive.

Baron030:)

55983

[robfive]

Wow, thanks for the explanation. That sounds very straight-forward. I'm sure it will be a very tedious process but the concept sounds simple enough.Thanks for the info.Robert

[GrooveControl]

Has anyone reworked the 3110a to have a fixed HF gain of -10db and not use L3 and S2?

[Baron030]

Well, without having an actual JBL network on hand to do an exact comparison. This is my best guess as to the correct switch for the network. I am pretty sure that JBL is using an Electroswitch brand rotary switch. But, without actual physical measurements I am not sure which Electroswitch switch series is being used. (“C4”, “D4”, or “D7”) Electrically, any of these 3 series should work and could be interchangeable. Now, there are some key words that must be followed when ordering this switch. The switch should have “Solder Lug Terminals”, “1 deck”, “2 poles per deck”, and “3 or more positions” and “Non-Shorting”. And have adjustable rotation stops. So, if the switch has more then 3 position, you can reposition some stops to limit the shafts rotation to just the required 3 positions.

Actually, for the 3110A network a “1 poles per deck” switch would work.

While a N1200 network would require “2 poles per deck”

Considering cost and availability, I would have to say that the “Electroswitch D4G0205N” switch would be my best guess and it should work for this application.
Mouser Part # 690-D4G0205N

Baron030:)

[Baron030]

Hi Robert

Just to give you an idea of just much additional wire that you might need on have on hand for this project. Taking into account the DC resistance of the Erse Super Q coil is 0.187 ohms. And the fact that 16 gauge wire has 4.094 ohms per 1000 feet. This coil has total of about 46 feet of wire in it. Assuming a turn ration of 0.0512 for the -6 db tap, this would suggest that you will be removing about one half of the wire, or about 23 feet. So, for a pair of crossover networks , it might be best to buy a one pound spool of wire and have some to spare.

Link: http://www.amazon.com/Magnet-Enameled-Copper-0-0524-Diameter/dp/B000IJXXSY


Baron030:)

[4343]

Hi Robert Just to give you an idea of just much additional wire that you might need on have on hand for this project. Taking into account the DC resistance of the Erse Super Q coil is 0.187 ohms. And the fact that 16 gauge wire has 4.094 ohms per 1000 feet. This coil has total of about 46 feet of wire in it. Assuming a turn ration of 0.0512 for the -6 db tap, this would suggest that you will be removing about one half of the wire, or about 23 feet. So, for a pair of crossover networks , it might be best to buy a one pound spool of wire and have some to spare. Link: http://www.amazon.com/Magnet-Enameled-Copper-0-0524-Diameter/dp/B000IJXXSY Baron030:) Did you slip a decimal there? 0.0512 suggests one half? I think the 16 ga coil is probably overkill for the HF driver, I had a 2.5mH 18ga on hand. I had ordered it by mistake (PE 266-556) and added wire to make it a 3.3mH, so taking that wire off gave me a good idea of how much wire this would take, about 12 feet. I unwound the 2.5 down to 2.47 and spliced on some new 18 ga wire. Here's what it took: Tap 2 to Tap 3: 9.75 turns, started with 4' wound on about 3'. Tap 3 to Tap 4: 6.75 turns, started with 3' wound on about 2'. Tap 4 to Tap 5: 11.5 turns, started with 5' wound on about 4'.

Super glue holds the windings from moving. Used a whole small tube, it was just opened for another project and I knew it would go bad before I needed it again, so I used it all.

[Baron030]

Did you slip a decimal there? 0.0512 suggests one half?

Hi 4343
Oops, Yes I did slip/flip a decimal! The correct turn ratio for a -6 db tap is 0.5012. Actually, you are probably right about using a 16 gauge coil for the HF driver is over kill. An 18 gauge coil would be a lot cheaper.

Maybe this would be better starting point:
http://www.erseaudio.com/Products/IQCoils18Gauge/ELC54-18-3300
(http://www.erseaudio.com/Products/IQCoils18Gauge/ELC54-18-3300)
Baron030:)

[Baron030]

I should say this now, just in case this all blows up in my face.

The tap inductance values that I posted are based on calculations and are not based on actual measurements from a 3110A crossover network. And there is one other possible flaw in my logic. And that be that the tapped “secondary” winding of the coil must be closer to the core then the “primary” windings. This would mean that you would need to start with a 0.33mH core (-10db tap) and then add winding to you reach the full 3.3mH.

If anyone in the Chicago area has a 3310a crossover and would not mind me taking some measurements of its tapped coil. Please contact me and I would most gladly post the actually measured values here. And this offer also applies to any JBL network that has a tapped coil. I think we need to document these tapped coil networks. Because hand winding tapped coils is not rocket science and the cost of inductance meters has come way down.

4343 that sure didn’t take you very long to make that tapped coil. I sure wish you didn’t live so far away. I would sure love to CLIO test that coil to see if the all the tapped voltage drive levels are right or not.

Baron030:)

[4343]

Hi 4343
Oops, Yes I did slip/flip a decimal! The correct turn ratio for a -6 db tap is 0.5012. Actually, you are probably right about using a 16 gauge coil for the HF driver is over kill. An 18 gauge coil would be a lot cheaper.

Maybe this would be better starting point:
http://www.erseaudio.com/Products/IQCoils18Gauge/ELC54-18-3300
(http://www.erseaudio.com/Products/IQCoils18Gauge/ELC54-18-3300)
Baron030:)

I'd rather start with this one:

http://www.erseaudio.com/Products/IQCoils18Gauge/ELC54-18-2500

and add wire to get to the specified inductance for each tap. If you are going to unwind and then add new wire anyway... I suppose if you had the proper varnish, you could re-coat the wire you remove, but I have new wire, but not the varnish. (Super Glue only works to hold the wire in place, it's almost impossible to use it as a coating without getting stuck!)

BTW, those are both out of stock at present, PE sells the exact same ERSE made coils at a bit of mark-up, and they are in stock for someone in a hurry, despite the photos, they have the same mounting bracket included:

http://www.parts-express.com/pe/pshowdetl.cfm?&Partnumber=266-556

http://www.parts-express.com/pe/showdetl.cfm?Partnumber=266-560


(http://www.parts-express.com/pe/showdetl.cfm?Partnumber=266-560)

[4343]

I should say this now, just in case this all blows up in my face.

The tap inductance values that I posted are based on calculations and are not based on actual measurements from a 3110A crossover network. And there is one other possible flaw in my logic. And that be that the tapped “secondary” winding of the coil must be closer to the core then the “primary” windings. This would mean that you would need to start with a 0.33mH core (-10db tap) and then add winding to you reach the full 3.3mH.

If anyone in the Chicago area has a 3310a crossover and would not mind me taking some measurements of its tapped coil. Please contact me and I would most gladly post the actually measured values here. And this offer also applies to any JBL network that has a tapped coil. I think we need to document these tapped coil networks. Because hand winding tapped coils is not rocket science and the cost of inductance meters has come way down.

4343 that sure didn’t take you very long to make that tapped coil. I sure wish you didn’t live so far away. I would sure love to CLIO test that coil to see if the all the tapped voltage drive levels are right or not.

Baron030:)

Just happened to have the right coil close at hand, and as I said, it was ordered by mistake, so I only have the one. And I just realized my 2385's have 2446H's on them, so I will need to swap them with the 2445J's on my big lenses to make use of a passive XO anyway.

PM me your address, I'll send it out ASAP.

[GrooveControl]

Would using 4 aircore inductors (2.47; 0.30; 0.20; 0.33) in series be a bad idea?

[robfive]

Actually, you are probably right about using a 16 gauge coil for the HF driver is over kill. An 18 gauge coil would be a lot cheaper

That's good news. I have an 18awg 3.5mH coil laying around that I could tryout. Now I just need to get an LC meter and I'll be good to go.

[4343]

Hi Robert

You cannot use multiple standard inductors simulate a tapped inductor. A tapped inductor acts like a transformer. And since, there are not separate primary and secondary winding, these devices are often referred to as Autoformers or Autotransformers. I don’t know of anyone that is manufacturing these devices, so unless you can track down some JBL crossovers. You might need to wind your own coils, which is not something that should be attempted without test equipment to measure the inductance.
And due to the necessary inductive interaction between taps, I would recommend using an iron core rather than an air core design. Rewinding an Erse Super Q coil would be a good starting point for such a project.

Here is a link to the inductance ratios that would be needed:

http://www.audioheritage.org/vbulletin/showthread.php?5376

...

Baron030:)


I'm not seeing the values you got.

From the chart, -6dB is supposed to be 25.1% of the original inductance or 0.8283mH. That would certainly be closer to half the wire than what I made, there's many layers on that 2.5mH I started with, and only one layer on top to get to 3.3mH. It looks like you subtracted 25.1% to get 74.9% of 3.3mH, or 2.4717mH.

I read "% of original inductance" to mean the end goal, not the amount to remove from the original inductance.

By my reading:

-8dB is 15.8% or 0.5214mH.
-10dB is 10% or 0.33mH.

>EDIT

If you look at the notes under the model, the -10dB value is in fact listed. (-10dB is 330 micro Henries = 0.33 milli Henries)

END EDIT<

Using that chart, I think it would be fun to build a 3110A with a few different steps. -6, -8, -10 could be -1, -2, -3, -6, -9, or maybe for testing, all of the them!

-1dB=79.4%=2.6202mH
-2dB=63.1%=2.0823mH
-3dB=50.1%=1.6533mH
-4dB=39.8%=1.3134mH
-5dB=31.6%=1.0428mH
-6dB=25.1%=0.8283mH
-7dB=20.0%=0.66mH
-8dB=15.8%=0.5214mH
-9dB=12.6%=0.4158mH
-10dB=10.0%=0.33mH

I'll be looking for some more wire now.

[Baron030]

Quote by 4343:
From the chart, -6dB is supposed to be 25.1% of the original inductance or 0.8283mH. That would certainly be closer to half the wire than what I made, there's many layers on that 2.5mH I started with, and only one layer on top to get to 3.3mH. It looks like you subtracted 25.1% to get 74.9% of 3.3mH, or 2.4717mH.

Hi 4343
Yes, I did simply subtract the 25.1% to get the 2.4717mH value for the -6 tap. I thought the idea would be worth trying. And in a way I am actually surprised how quickly you have been running with it. The next big question is are you getting 0.8283mH of inductance when you measure it between pins #2 and #5? If not then my logic is really flawed. And the chart’s turn ratio should be followed instead. This would mean that a 3.3mH I-Bar coil would need to be completely unwound and the total number of turns counted. And then use the bare core and a lot of fresh of wire to rewind it completely with taps. And with there is the issue of can Pin #1 be the lead closest to the iron core Or does Pin #5 needs to be closest to the iron core? The point being that does the secondary winding need to be closest to the core? Or can secondary winding be wound on top of the primary winding? If the secondary can be on the outside then a 3.3mH I-Bar when it would mean that a lot less of the coils’ original wire would need to be removed.
Baron030:blink:
56041

[Baron030]

For those of you following this thread, you can tell I have more questions than I have answers right now.

Hi 4343, Are you getting 0.8283mH of inductance when you measure it between pins #2 and #5?

I have been digging through some boxes tonight and I have turned up almost all of the parts necessary to construct the HF section of the 3110a crossover network. I have a 16 ohm load resistor to stand in for the HF driver. A 0.02mH coil and a 3uf cap for the “Max Boost” section. I have 2-8 ohm resistors that will have to stand in for a 3.9 ohm resistor. And 2-4.3uf caps that will have to stand in for the 8uf cap. As well as 2-40 ohm resistors that will stand in for the 2-39 ohms resistors. And I have one more interesting find, a pair of Jantzen 15 gauge air core inductors. I know I am not hitting all of the part values spot on. But, it might be close enough for a good test.

I am beginning to suspect that turn-ratios might be more actuate and important then relative inductances. Now, I really don’t want to damage a 3.3mH coil for this test, because I may have some uses for them later. But, if I nick the insolation in just a few outside edges, I should be able to create some taps at those points. And then I should be to test the voltage drives to see they the match mathematical formulas or not. Unfortunately, I have to make some bathroom repairs this weekend and other projects as well. So, it might take me quite a while before I can post some CLIO results and then have some real answers.

Robfive, please post the 3110 equivalent network schematic that gets around the tapped inductor. Maybe the focus can shift to just adding the 2380 CD compensation to it.

Baron030:)

[robfive]

Baron030, thanks for your help in this. I know what it's like to have several projects going on at once. Please fix your bathroom before messing with this inquiry.

Here is a link to the 3110 equivalent HP x-over posted by 4313B back in 2005:

http://www.audioheritage.org/vbulletin/showthread.php?5603-JBL-3110-crossover-question-and-aftermarket-diaphragm-report&p=52084&viewfull=1#post52084

It looks like a straight-forward second-order filter and L-pad.

Thanks,
Robert

[4343]

Decided to try the coil again, this time with 8 taps. I tried to start at -10 dB, but ran across a measurement error at that tap. (I left the wire I unwound on the floor and measured from the unwound end to the end of the wire.) It turns out my meter adds a lot of inductance to the reading with the straight wire in the circuit. I ended up winding a little past the reading I wanted and then cutting and stripping the wire to take a more accurate reading, while unwinding a little to get to the exact number required, then cutting and stripping again, before tying in a long piece and starting that whole process again for the next tap. In the end I ended up with 7 taps on my 8 position switch: -9, -6, -5, -4, -3, -2, and -1.

Later I realized that the top of the inductor would yield -0 dB, so I used it as the last position on my switch. Today I dragged out my HP Distortion Measurement set (generator and meter in one). It measured precise 1 dB steps from 0 to -6dB and exactly 3db to -9.

Just need another core and I'll wind another after I get done swapping the drivers around to put the 16 ohm 2445's on the flat fronts. I'm also thinking that the cheap wafer switch should get swapped to become a terminal strip so it does not get fried. I did that for the "HF Boost" using a 3 position strip and a wire with a lug on it to select Max, Med, or Min. For now, the "gain" switch works great, and I will be able to determine exactly which tap I want very quickly. The switch might stand up to home use, but these are SR tops, so failure is not an option once they are in service.

The 0dB setting is actually quite useful if you want to use the filter to passively bi-amp using an amp with level controls. Set to 0dB, and adjust the level with the amp. Passive bi-amp means paralleling the inputs of two amps with a full-range signal and letting the high level filters in the box do the frequency splitting. (If you do that you don't actually need the taps on the inductor at all, and as the post on the 3110 equivalent circuit says, put a 16 ohm L-pad after the output and you won't miss them, even without bi-amping...) Passive Bi-amping is not very common in SR work, but I learned long ago that having a passive in the box can sometimes save your bacon...

[4343]

Has anyone reworked the 3110a to have a fixed HF gain of -10db and not use L3 and S2?

That's easy if you just build one with a standard (non-tapped) 3.3mH inductor. I just measured mine set to 0dB, i.e. all the taps floating open and the signal drawn from the top of the 3.3 mH inductor ((the junction of R2&R3, C5 and L3 connected to the junction of S1 pin 3 and R4). Using a 16 ohm L-Pad connected to the output pins I adjusted it to read -10dB (compared to the output level when it was not there). I then measured it's resistance out of circuit. From ground to the wiper, I measured 4.5 ohms, and from the wiper to the plus output I measured 40 ohms.

With a 150W driver connected, I can see not using a 100W L-Pad! Using one to adjust the level to match the LF, then putting fixed resistors in that match the values measured on the L-Pad is a common sense way to dial in any level you might need. If you use high enough wattage resistors, it could survive...

[robfive]

4343,

So are you saying that we could get around the tapped inductor by using a standard 3.3mH inductor in its place (creating -0dB) and placing an L-pad resistor pair infront of the midrange? That sounds good to me.

[4313B]

we could get around the tapped inductor by using a standard 3.3mH inductor in its place (creating -0dB) and placing an L-pad resistor pair infront of the midrange? That sounds good to me.Nope. You cannot treat a tapped autotransformer like an inductor.

If I have time tomorrow I'll work up the 3110A equivalent network. I suspect it will be similar to what I already have along with the compensation network placed in parallel with the resistor pad and L-pad.

[4343]

Nope. You cannot treat a tapped autotransformer like an inductor.

If I have time tomorrow I'll work up the 3110A equivalent network. I suspect it will be similar to what I already have along with the compensation network placed in parallel with the resistor pad and L-pad.

Thanks!

I don't have any way to measure anything but the levels right now, so having a definitive answer would be great.:bouncy:

[4313B]

Has anyone reworked the 3110a to have a fixed HF gain of -10db and not use L3 and S2?Here you go.

Don't fret about the 316.2m value shown on the schematic for the tapped autotransformer T1. It is indeed 3.3 mH with a turns ratio of 0.3162 to 1 for the -10 dB voltage drop as shown in figure 1.


The DCR of L3 in the equivalent is 0.5 ohms.

[4313B]

Here is the standard value equivalent. I forgot to force it in the original.

[Baron030]

Wow, 4313B !!! :applaud:
You sure make it look easy. I bet that Leap Crossover shop software didn’t come cheap.

In researching this, I did turn up a 3310A variant. The high pass section is identical to the standard 3110A version.
But, the low pass section is a little different. It has no zobel network and different L1 and cap values.
Here is the schematic below. I am not sure which 3110A network version would be better for which drivers?
56095
Baron030:)

[4313B]

In researching this, I did turn up a 3310A variant. The high pass section is identical to the standard 3110A version.
But, the low pass section is a little different. It has no zobel network and different L1 and cap values.
Here is the schematic below. I am not sure which 3110A network version would be better for which drivers?
Baron030:):dont-know:

[Baron030]

Thanks again 4313B :applaud:

It looks like N4671 network is nearly identical to the standard 3110A. But without the impedance dip at 400 Hz.
From a practical point of view, low pass section of the N4671 variant might be easier for people to build because the 3.3mH (L1) and a single 30uF cap can be substituted for (C2 & C3). Both of which are stock off the shelf values. A stock 2.6mH coil might be harder to find. I did notice that the smallest coil that parts express stocks is a Jantzen 0.025mH 18ga coil (255-198). So, it is the only item that will need some custom fabrication or trimming. If you own an inductance meter and you are building the 3110A network, then please record and post here the number of turns which will need to be removed to change a stock (255-198) coil to 0.020mH.

Baron030:)

[GrooveControl]

Wow! Just what I was looking for! Thanks!

One Q. How much impact will the ab switch have on the top end? Can you add to the chart?

[4343]

Wow! Just what I was looking for! Thanks!

One Q. How much impact will the ab switch have on the top end? Can you add to the chart?

There is graph here:

http://www.audioheritage.org/vbulletin/showthread.php?5376

[robfive]

There is graph here:

http://www.audioheritage.org/vbulletin/showthread.php?5376

I believe that graph represents the different positions of the autoformer affecting the gain of the HP section of the crossover. I think GrooveControl was wondering about the two positions of the CD-compensation switch. Perhaps I'm wrong, though it would be my fist time...it the last two minutes.

[4313B]

Here's one showing the effects of the HF compensation switch. High, Low, None.

[4313B]

Nope. You cannot treat a tapped autotransformer like an inductor.
That's easy if you just build one with a standard (non-tapped) 3.3mH inductor. I just measured mine set to 0dB, i.e. all the taps floating open and the signal drawn from the top of the 3.3 mH inductor ((the junction of R2&R3, C5 and L3 connected to the junction of S1 pin 3 and R4). Using a 16 ohm L-Pad connected to the output pins I adjusted it to read -10dB (compared to the output level when it was not there). I then measured it's resistance out of circuit. From ground to the wiper, I measured 4.5 ohms, and from the wiper to the plus output I measured 40 ohms.I want to address this because I think it is important to show why.

First schematic shows substituting in a 3.3 mH inductor for the 3.3 mH tapped autotransformer as well as adding in the -10 dB point values of a 16 ohm L-Pad as mentioned above.

Second schematic shows the same but with the other stock resistors removed as well.

I also want to make sure everyone knows why there is a 20 ohm shunt resistor between the stock 8 uF capacitor and the stock 3.3 mH tapped autotransfomer. It is there to mitigate the inductive reactance of the tapped autotransformer so that the 8 uF capacitor behaves as it is intended to. This shunt resistance can be found in many JBL networks using a tapped autotransformer. It is usually high power, for example 40W in the 4331/4333/L300, or made up of two or three lower power resistors to handle the resulting heat build up. In this instance it is made up of two 39 ohm 10W. It is superfluous with a standard inductor.

[Baron030]

Thanks, 4313B for taking the extra time in showing us why.
Wow, do those curves look really messed up.
I guess that really proves that you can’t treat a tapped autotransformer like an inductor.
So, in true MythBusters fashion, you know how to call it…
56115

[GrooveControl]

Here's one showing the effects of the HF compensation switch. High, Low, None.

Thanks 4313B, that's what I meant.

I think I'm going to use the N4671 lo pass filter as I'm using the 2035HPL which doesn't seem to need the zobel. I measured the imp of mine a while back and they are 8ohms up to 5khz.

http://techtalk.parts-express.com/showthread.php?231384-JBL-IMP-Measurements

[4343]

I believe that graph represents the different positions of the autoformer affecting the gain of the HP section of the crossover. I think GrooveControl was wondering about the two positions of the CD-compensation switch. Perhaps I'm wrong, though it would be my fist time...it the last two minutes.

The map portion of that graph is labeled.
(RED) -6dB tap, MAX HF
(GRN) -8dB tap, MAX HF
(BLU) -6dB tap, MED HF

Interesting that the different attenuation taps don't affect the 20KHz level on MAX HF, so the curve appears different on the two.:dont-know:

[4343]

I want to address this because I think it is important to show why.

First schematic shows substituting in a 3.3 mH inductor for the 3.3 mH tapped autotranformer as well as adding in the -10 dB point values of a 16 ohm L-Pad as mentioned above.

Second schematic shows the same but with the other stock resistors removed as well.

I also want to make sure everyone knows why there is a 20 ohm shunt resistor between the stock 8 uF capacitor and the stock 3.3 mH tapped autotransfomer. It is there to mitigate the inductive reactance of the tapped autotransformer so that the 8 uF capacitor behaves as it is intended to. This shunt resistor can be found in nearly every network JBL made using a tapped autotransformer. It is superfluous with a standard inductor.

Thanks for clearing this up!

My next faux pas involves discovering a huge inductor in my bass cabs. I've been collecting 2226J's to populate them, and I finally have 4 of them. Today I took out the pair of K-140's and what looked to be 2225's. (On closer inspection one was a 2205A and the other one had a sticker on it saying it was a 2205A, but it covered the real model 2220B...)
The inductor is 18mH 16ga feeding just the bottom woofer. It seems I should have been looking for 2226H. Or I can look at changing the LP filter to work at 16 ohms.

[Baron030]

Interesting that the different attenuation taps don't affect the 20KHz level on MAX HF, so the curve appears different on the two.


Hi 4343
There is a simple explanation why the different attenuation taps don’t affect the response curve at 20KHz on the MAX HF setting.
The 3.0uf cap and the 0.02mH coil form a RLC series resonant circuit and it’s resonating at 20,551Hz, which means the impedance drops to nearly zero.
So, you get no voltage or signal drop.

Here is a link to a RLC series resonant circuit calculator and some other handy calculators as well:

http://www.lautsprechershop.de/index_hifi_en.htm?http://www.lautsprechershop.de/tools/t_ps_schwingkreis_en.htm

Baron030:)

[GrooveControl]

Here is the standard value equivalent. I forgot to force it in the original.

Not sure if I'm reading this correctly. Does R6 represent the driver? Or is that a real resistor?

Also, are 10 watt resistors suitable for application? Or should I double the value and run two in parallel for each?

Thanks again.

[4313B]

Yes, R6 is the 16 ohm driver load. It's easier than putting in a driver model along with a 16 ohm objective curve. :D
An actual driver impedance curve is rather pointless since this is a "generic" JBL network.

I would use double, it keeps them more linear at any rate.

[GrooveControl]

How about Cap ratings, Solen offers 250V, 400v and 630v. I usually use the 250v, but I'm thinking 400v for this application.

Is 400v enough for the lo pass section given the woofer can take 600w?

[4313B]

JBL uses lots of 100V components. I use 250V. I've never had any issues. However, I have come to the realization over the years that perhaps I don't flog my systems like some other forum members do.

[Mr. Widget]

How about Cap ratings, Solen offers 250V, 400v and 630v. I usually use the 250v, but I'm thinking 400v for this application.

Is 400v enough for the lo pass section given the woofer can take 600w?400V?

100V is rated at 625W at 16 ohms... 1250W at 8 ohms. 250V at 16 ohms gives you almost 4KW.


Widget

[4313B]

Some guys swear the higher voltage capacitors sound better, especially on the top end. That's why they use 400V capacitors on their tweeters.

The higher the voltage rating the more expensive and physically larger the capacitor for any given capacitance. JBL uses 100V because they are less expensive, physically small so the whole network isn't overly large, and perfectly adequate for the intended application.

[GrooveControl]

EDIT: I used the wrong value. Looks like it should be .02mh. Ignore blue line and question. thx.


With the exception of the coils, I found the exact value components of the high pass filter in my spare parts drawer (for the -10db pad). I also found a .24mh coil that I was able to wind down to .20mh (measured it in speakerworkshop).

Then assembled the network and took freq resp measurements, without the big coil, here are the results. 2445J on 2380A.

Green = no hf boost
Red = 2uf bipass
Blue = .20mh + 3uf bipass

Something unexpected is happening with that coil I unwound. It's a Solen S18.24 that I unwound from the outside.

I'd like to get the response above 7khz up by about 3db. Any thoughts?

56132

[Baron030]

Hi GrooveControl

Yes, the correct value is 0.02mH. By using a 0.20mH coil instead, you can clearly see how the 0.2mH coil and the 0.3uf cap is resonating at about 6,499Hz. Call it one of those D’Oh moments . We have all had them at one time or another. When you finish un-wounding that coil down to 0.02mH, you will see just how tiny that thing really is.

Baron030:)

[Baron030]

Well, as you all know 4313B has already come up with a non-tapped autotransformer version of the 3110A network. But, I thought I would do some tests and see if I could come up with some suggestions for constructing autotransformers for use in other crossover network designs for which a non-tapped autotransformer version has yet to be created. And also possibly bust a few autotransformer myths in the process. Pictured below is the high pass section of 3110A network made from some stuff that I just had lying around. And since, I did not want to destroy a perfectly good Jantzen 15 gauge Air Core inductor for these tests, I decided to nick the insulation in just 2 places and then solder some leads to create taps. There are 3 things that I wanted to determine in my tests.
When using air-core coils so the voltage drives come close to the mathematical formulas or not?
And which formula appears to be more accurate? Turn Ration or Percentage of Inductance?
And does it make a difference where the secondary is? Should it be near the core or can it be near the outside of the coil?

56372

[Baron030]

The first big surprise was measuring the tapped coil’s inductance. It would appear that when it comes to inductance the total is much greater than the sum of its parts. So, any myth of stringing together a bunch of inductors together to create an autotransformer is totally “BUSTED”. The values just don’t add up right.
Since I wanted to determine if there were any differences in secondary winding locations, I purposely tried to created two separate tap points with as close to same turn rations as I could.
Starting with the secondary windings closest to the core, that is to say that pin #1 is the primary input and pin #5 is the secondary output and pins #2 & #6 are the common leads. The turn ratio formula calculates the output to be -6.7db. And the percentage of inductance formula calculates the output to be approximately -8.2db.
Alternatively, with the secondary windings being closest to the outside of the coil, that is to say that pin #2 is the primary input and pin #4 is the secondary output and pins #1 & #3 are the common leads. The turn ratio formula, it calculates the output to be -7.5db. And the percentage of inductance formula calculates the output to be approximately -6.5db.
56373

[Baron030]

Now we come to some test results. I should point out that in the 3110A network, when the HF Boost is set to its minimum setting that a 3.9 ohms resistor (R4) is bypassed. And so it would appear that the output level is actually higher than when the HF Boost is set its other positions. And that I used a 0.025mH coil instead of the required 0.020mH value. So, the “MAX” HF Boost comes in at a little lower frequency then it should be.
Inside Secondary:
56374
Outside Secondary:
56375

[Baron030]

Admittedly, looking at the test results, it does have me scratching my head a lot. There are a few things that I kind of expected like the measured output voltage drives being lower than their calculated values. And if you think about it, when was the last time you ever saw an air-core power transformer? Or a tube amp with an air-core output transformer? So, it’s not surprising that measured values are lower because the mathematical formulas are not taking into account the greater core losses associated with air-core coils.
Since, the curves are fairly flat at 2511Hz, I will use that frequency for direct comparisons between measured and calculated values.

Inside Secondary:
Calculated: Turn Ratio Formula: -6.7db, % of Inductance Formula: -8.2db
Measured: No Boost: -9.34dbV, Mid Boost: - 12.26dbV Max Boost: -12.29dbV

Outside Secondary:
Calculated: Turn Ratio Formula: -7.5db, % of Inductance Formula: -6.5db
Measured: No Boost: -7.4dbV, Mid Boost: - 10.24dbV Max Boost: -10.46dbV

It would appear to me that the “Percentage of Inductance” formula might be far more accurate than a “Turn Ration” formula when it comes to constructing coils. And in the case of an air-core coil, the location of the secondary does not appear to make much of a difference. And yet it is strange, how 146 turns near the outside can generate more inductance and voltage drive then 160 turns wound near the center. So, I am thinking of creating another tap point near the core with hopefully something close to the outer tap’s 0.726mH of inductance and I will run some additional tests.

Baron030:)

[Baron030]

Since, the results of my first tests just did not seem to me as being very conclusive as to which secondary location came closest to fitting the mathematical model. I decided to create another tap and rerun my tests. And then I soon realized that I did count the total windings correctly. The correct total count is 386 turns and this changes all of the “Turn Ratio” formula calculated values in my previous test.
Pictured below are the correct values and the new Clio Results:

56383
56380
56381

Since, the curves are fairly flat at 2511Hz, I will use that frequency again for direct comparisons between measured and calculated values.

0.785mH - Inside Secondary: Calculated Turn Ratio: -5.7db, % of Inductance: -6.8db
Measured: Min Boost (R4 shorted): -7.12dbV, Min Boost (with R4): -8.90dbV, Mid Boost: - 9.90dbV Max Boost: -10.12dbV

0.726mH - Outside Secondary: Calculated Turn Ratio: -8.4db, % of Inductance: -6.5db
Measured: Min Boost (R4 shorted): -7.37dbV, Min Boost (with R4): -9.16dbV, Mid Boost: - 10.20dbV Max Boost: -10.43dbV

Well, I think the differences between the two formulas and the measured results appear closer together in the case where the secondary winding is wound closest to the core.

So, my recommendations for anyone wanting to make an autotransformer would be to use an iron core from an Erse I-Bar coil and follow the percentage of Inductance formula, and wind the secondary windings (taps) closest to the core.
Baron030:)

[4313B]

So, my recommendations for anyone wanting to make an autotransformer...Mine would be to try and live life a little, travel a bit, see the world, enjoy a cold beer, volunteer in your community... something, anything, besides sitting around making tapped autotransformers. Hell, worst case, fire up a game of World of Warcraft for fuck's sake. :rotfl:

Of course, if someone has already done all that other shit and the only thing left to do on the bucket list is wind an autotransformer then by all means...

[Baron030]

But, what about the people that want that real “Autotransformer” sound? :blink:

Whatever that is?:dont-know:
I am kidding of course…:lol_fit:

No, I only did this research just to prove or disprove a few ideas and to kill a few myths.

Baron030:)

[4313B]

I think it's great. :)

Thank you for taking the time.

[Elac310]

EDIT: I used the wrong value. Looks like it should be .02mh. Ignore blue line and question. thx.


With the exception of the coils, I found the exact value components of the high pass filter in my spare parts drawer (for the -10db pad). I also found a .24mh coil that I was able to wind down to .20mh (measured it in speakerworkshop).

Then assembled the network and took freq resp measurements, without the big coil, here are the results. 2445J on 2380A.

Green = no hf boost
Red = 2uf bipass
Blue = .20mh + 3uf bipass

Something unexpected is happening with that coil I unwound. It's a Solen S18.24 that I unwound from the outside.

I'd like to get the response above 7khz up by about 3db. Any thoughts?

56132


Hi GrooveControl,
Interesting tests. There is not much on the inernet about the actual, measured impact of the two HF boost circuits (the bridging 2uF capacitor on the one hand, and then the bridging 3 uF and 0.02mH coil. So I was pleased to find this thread.
Did you eventually manage to get it right with the 0,02mH coil ?
If so, would you by any chance have some measurements of the end result with the correct coil value ?

Thanks in advance.

[GrooveControl]

Hi Folks. It's been a few years since I acquired the parts for my next speaker, but still have not started the cabinets.

Anyway, along come a pair of LX13s. The cost of the parts to build the xo in this thread isn't much cheaper, and these come in a nice jbl box, so I thought why not? I'll just add the CD compensation to it.

So off I go to take some measurements. I thought I would share these with you guys, and I have a few questions at the end.

The first chart shows the response of the 2445j on a 2380a horn with three different cap values in parallel with the hi-pass section of the xo. What I found interesting is how they all behaved. Note 3rd octave smoothing applied to make it easier to see.

The blue line is a 3.0uf cap, which provides the most boost centred around 20khz.
The red line is a 6.2uf cap, which provides the most boost centred around 15khz.
The green line is a 9.1uf cap, which provides the most boost centred around 10khz.

To my aging ears, the 9.1uf cap provided the most pleasing overall sound, even though it rolls off at the top end quicker than the other caps.

Chart A


The next chart shows the difference between having the 9.1uf cap in and out of the circuit. Again, 3rd octave smoothing applied.

Chart B


Finally, here is the second chart again, but without smoothing (just gating). Except for the dips around 8khz, I think it looks pretty good.

Chart C


Now I have a few questions:

1 - Is there any way to keep the response from falling when using the 9.1uf cap? Ideally I would like the boost at 10khz offered by the 9.1uf cap and the boost at 20khz of the 3.0uf cap combined.
2 - Looks like the biggest boost in chart B is about 7db at 11khz, but the HF gain is in the min position which should be a 10db reduction. Any ideas why I'm not getting a 10db boost?

Thanks.