Yet another 4425 DIY [Archive] - Lansing Heritage Forums

jerv

09-28-2006, 03:49 AM

Don Mascali

09-28-2006, 07:08 AM

Look here:
http://www.audioheritage.org/vbulletin/showthread.php?t=12396&highlight=2342

The L200t3 is nice. There is alot of info from the Zilch project cronicled
in the "Quick and Dirty" thread. He tested that horn amongst many others. There are many strategies for dealing with CD horns in the X-over topologies shown in that thread and many other places here.

The search fucntion is your friend. You could spend weeks reading up on this stuff.

Good hunting,

Don M

Zilch

09-28-2006, 02:12 PM

Does anyone have impedance plots (or even better: impedance files) for the 2214H and the 2416H on 2344 horn?I believe you mean 2342 horn, which is thread-on and accepts 2416H. The larger 2344 is bolt-on, and NLA.

Here's the roadmap for Q&D:

http://audioheritage.org/vbulletin/showthread.php?t=9901

Don Mascali

09-28-2006, 09:08 PM

The 2344 has about a 12.25" square mounting area where the 2342 is 8.75". JBL used 1000 HZ on 2344 and 1200 HZ on the 2342 as the nominal X-overs. Both have the infamous butt profile.

The 2344s show up now and then here and on E-bay. I even found a pair that were modified by JBL to have the ports behind the Buns. They work nice for my side H/T speakers where they need to be mounted flat to the wall.

I am about to try a 2426H driver with the bolt flange removed on a 2342 for a rear surround project. It is kind of a waste for that driver though. I have some 2416Hs too... so we shall see.

I'm glad Zilch stepped up with that link. I could never find it. I have it saved now.

jerv

09-29-2006, 02:18 AM

Yes, it is the 2342. A slip of the finger.

Cabinets are ready, units are mounted. I have already made one version of the L200t crossover (as described in varoius other threads) and one (slightly simplified) version of the original crossover. These are fun to compare.

I am also working on a crossover design of my own (with somewhat steeper crossover slopes and different eq, taking care (among other things) of the annoying bump the 2214 has at 1,9kHz) and an also an electronic version. Needs something to do when winter sets in. :)

I'll report back with measurements and listening impressions in due time.

Don Mascali

09-29-2006, 04:02 AM

Sounds like a great plan. Good listening.

Don M

Dougie

09-30-2006, 05:03 AM

I'm in the process of building a 4425 DIY. Just for the fun of it, I plan to design new crossover networks in addition to replicating the originals. Maybe the difference (?) could provide some insight and learning experience.

However, somehow I seen to have gotten into some momentarily trouble with my soundcard/measuring gear, and need some verifying. Does anyone have impedance plots (or even better: impedance files) for the 2214H and the 2416H on 2344 horn?

Thanks, Espen

I happen to have a new pair ( not broken in ) of 2214-H drivers here. I could do an impedance/phase run on them this weekend with LMS if you like :) And post it in pdf format. I could also send you the pdf file via email for you to print. Let me know.

Dougie

jerv

10-01-2006, 12:06 PM

Thanks, Dougie!

I very much appreciate your suggestion! However, I traced my problems to a faulty connector on the sound card, which is now fixed. I managed to get some good plots tonight. (Hope yours looks about the same).

Here are the frequency and impedance plots of the 2214H in my 4425-box:

jerv

10-01-2006, 12:14 PM

And here are the plots for the 2416H on the 2342 horn:

jerv

10-01-2006, 12:22 PM

I first tried with this variant of the L200t3 crossover, as in the thread http://audioheritage.org/vbulletin/showthread.php?t=7110 which was a great inspiraton for my project.

jerv

10-01-2006, 12:33 PM

Here are the simulation (top) and the real measurements (bottom) for the 2214H/2416H+2342 with the L200t3 variant crossover in 4425 cabinet. As you can see, these curves are quite similar. That means (I hope) that the simulation and measurements basically works.

jerv

10-01-2006, 12:47 PM

The L200t3 variant crossover is within 5dB up to 2000Hz, but the range 2kHz - 15kHz is about 5 dB up. Listening tests confirm this find: IMO the speaker sounds a bit too bright with this crossover.

I will investigate new crossover values next. The 3,8 mH inductor for the 2214 is too big, and will be replaced with 3.0 mH (as in the stock L200t3 network).
So far, simulations may indicate that simple a Lpad in the HF section may be difficult without messing up the HF compensation. (Tne L200t3 network is very elegant, with the 2.2 uF capacitor functions as both crossover, compensation and attenuation). Maybe we will need a topology more in line with the standard 4425 network.

Or maybe something completely different. The 2214's response peak at 1,9 kHz looks quite annoying. Maybe some filtering or higher order slopes can fix it.

This is fun! I'll keep on soldering and listening.

Espen

Zilch

10-01-2006, 01:18 PM

So far, simulations may indicate that simple a Lpad in the HF section may be difficult without messing up the HF compensation.Try it, either at the front or back end (try both).

I see two problems, both of which you've already recognized:

1) You need more HF attenuation to balance with the 2214.

2) The slope of the compensation provided by the 2.2 uF is a bit too steep.

Lowering the 2.2 uF will get you more attenuation overall, but it'll "push" more power to the driver above 20 kHz, which I think is a bad thing, sonically. I'd find a way make padding work, instead.

Reducing R3 will do it, too, but the operating frequencies of the filter will also shift. Not a terribly big deal, since there's only 4 components total to "adjust."

Try some (relatively high) resistance in parallel with the 2.2 uF to lower the slope a bit. You may have to adjust the conjugate to restore the highpass rolloff, and it'll necessitate more attenuation.

You may have to put a notch filter in the LF to tame that 2214 peak. As you suggest, 3-pole would knock it down, too, but more of a "project."

Just to be sure, how did you achieve the 7.5 Ohms for L2?

This is fun! I'll keep on soldering and listening.Yeah, it is!!

What's your measurement setup?

John W

10-01-2006, 03:30 PM

The L200t3 variant crossover is within 5dB up to 2000Hz, but the range 2kHz - 15kHz is about 5 dB up. Listening tests confirm this find: IMO the speaker sounds a too bright with this crossover.

I found them a little too bright also. I ended up with a switch on the back that lets me change between 0, 3.5 and 6 db down on the upper end. I couldn't find the exact resistor values without opening them up, but normally use them on the -3.5db level.

jerv

10-01-2006, 11:18 PM

Just to be sure, how did you achieve the 7.5 Ohms for L2?
......
What's your measurement setup?

- For the 1 mH / 7.5 ohm I used a 1 mH / 0.7 ohm inductor in series with 6.8 ohm.

- My measurement setup includes a Behringer ECM8000 condenser mic, Behringer UB1002 mixer (with phantom powering), a M-Audio Delta 1010 sound card (this card is way overkill for this application, but I use it for sound recording too) and a home made 2x75W amplifier. Software is justMLS, Speaker Workshokp and lspCAD for measuring and crossover simulations.

jerv

10-01-2006, 11:23 PM

I see two problems, both of which you've already recognized:
1) You need more HF attenuation to balance with the 2214.
2) The slope of the compensation provided by the 2.2 uF is a bit too steep.

I found them a little too bright also. I ended up with a switch on the back that lets me change between 0, 3.5 and 6 db down on the upper end. I couldn't find the exact resistor values without opening them up, but normally use them on the -3.5db level.

Thanks for the suggestions. I will try some more simulations tonight, and keep you posted!

jerv

10-01-2006, 11:36 PM

This is the simulation with L1=3.0 mH and C1=20 uF (as with the original L200t3 crossover). This increases the level 200-1000 Hz by 2-3 dB, while maintaining good phase coherence (in fact: it's even better).

But I find it almost impossible to use only L-pads to balance the 2416. Any L-pad - front or back end - or series resistance also changes frequency response and HF attenuation quite a lot.

The 2.2 uF and/or the inductor has to change as well.

jerv

10-02-2006, 12:15 AM

This is the best simulation I came up with yesterday - with (basically) the same filter topology. Further developments will (I think) change filter topology more drasticially.

L1 reduced even furter to 2.2 mH and C1=22 uF. This increases the level 200-1000 Hz even more.

In the HF section the 2.2 uF is parallelled with 16 ohm. Then, to compensate for the crossover rolloff at 1200 Hz the inductor has to be reduced all the way down to 0,27 mH / 0,3 ohm. R3 is now 8 ohm.

All is now within 5 dB 100Hz-20kHz, buth phase coherency has suffered somewhat. I will build this network, measure it to see if it corresponds to the simulations and then compare it sonicially with the L200t3.:)

Dougie

10-02-2006, 02:54 AM

Hi Jerv;

I decided to do an impedance / phase measurement on my 2214H drivers anyway as I've had them here in boxes for a couple of years and have been wanting to do something with them ( or sell them ) LOL!!!

The first plot is for the first driver (a) Fs is 29.474 Hz and reached a max Z of 49.989 ohms at resonance. The Fs is a bit higher than JBL's specs of 23 Hz, but my drivers are not broken in yet and may account for the higher Fs !!

Dougie

Dougie

10-02-2006, 03:02 AM

I measured the second 2214H just out of curiousity. Fs measured out at 28.274 Hz and the max inpedance reached was 46.637 ohms, a bit lower than the first driver.

I overlaid the 2 measured curves to show you the comparison. See attached pdf file......

Dougie

Dougie

10-02-2006, 03:36 AM

Jerv;

The color differences in the overlaid plots didn't come out all that well on the pdf's posted here, seems to have lost something in transit.

You can PM me with your email address and I would be glad to send you both pdf's that way. The plot lines are much wider and the color differences in the two plots are of much better quality in the origonals.

Let me know;

Dougie

jerv

10-03-2006, 12:28 AM

Thanks for sharing your results, Dougie!

Your impedance plots looks basically the same as mine - accounting for that I measured mine in 4425 cabinets and you measured your (as I understand) in free-air.

Very nice to have results verified by some other's measurements.

From 200 Hz up the curves look very similar - including the small anomaly at 1,9 kHz (which corresponds with the annoying peak in the frequency response).

BTW: What measurement gear and SW do you use?

Espen

jerv

10-03-2006, 12:55 PM

I made up new a new filter as described some posts ago.
For the LF the components are: L1 = 2,2 mH 0,2 ohm DCR, C1 = 22 uF, R1 = 47 OHM, no R2.
For the HF, the components are: C3 = 2,13 uF, L2 = 0,27 mH 0,31 ohm DCR, R3 = 8,2 ohm.

Here are the simulation for the filter (top) and the measured result (bottom). Almost, but not quite similar. The HF is about 2 dB up compared to the simulation. The HF is easily damped by taking R3 down to 4-5 ohm, but anyway, I have to investigate this.

Meanwhile, I listen to music and compare them sonically.

Dougie

10-05-2006, 04:33 AM

Dougie

10-07-2006, 01:09 PM

Jerv;

I don't see an anomaly at 9 Khz in my impedance plot:) Dosent matter as the 2214H wont be used way up there anyway.

The small " hump " in my plot around the 2 Khz area is the result of my measurement system pausing and switching decades, then continuing on...

The roughness between 1 and 2 Khz is the result of some cone breakup in the midrange region.

Dougie

jerv

10-08-2006, 12:15 PM

:o:
Jerv;

I don't see an anomaly at 9 Khz in my impedance plot:) Dosent matter as the 2214H wont be used way up there anyway.

The small " hump " in my plot around the 2 Khz area is the result of my measurement system pausing and switching decades, then continuing on...

The roughness between 1 and 2 Khz is the result of some cone breakup in the midrange region.

Dougie

I meant 1.9 kHz. (In my country, the "," is used as the decimal delimiter - not the "." I forgot.:o: ).

My thought was that the impedance anomaly at 1.9 kHz was caused by cone breakup in that region. The 1.9 kHz resonance is quite pronounced.

Espen

jerv

10-13-2006, 09:24 AM

After much experimentation I have ended up with 3 filters to compare sonically. Filter A is based on the L200t topology, but with different component values. L1=3.0 mH/ 0,2 ohm, C1=15,6 uF, R1=47 ohm.
C3=2,2 uF, L2=1mH / 7,5 ohm, R3=8 ohm.
Filter slopes approximate 4th order. The revese null is deep. Sound is very good and lively - but somewhat bright. Measurements confirm this. The curves are roughly the same as my sumulations some posts ago, with the HF 2-3 dB more down.

Snapshot here:

jerv

10-13-2006, 09:36 AM

..has a somewhat different topology, with a more traditional 4th order layout in the HF section to attenuate the 2416 without messing up the filter slope. Slopes are still 4th order, reverse null is deeper and narrower and phase traching better.
In the LF section L1=3.0 mH / 0.2 ohm, C1=15,7 uF
In the HF section C1=8,2 uF, L1= 3.3mH / 0.25 ohm, R1=16 ohm, C2=2.7 uF, L2=0,2 mH / 1.66 ohm and R2=4,7 ohm. Sound is pleasant and warm - a little subdued compared to filter A.

Schematics and measurements:

jerv

10-13-2006, 09:42 AM

The last one is an 8th order slope experment. This was the only way I could get rid of the 1.9 kHz 2214 peak. This filter measures the best. Reverse null and phase tracking is very good. I haven't listened to it much yet.

Schematics and measurements and pict of filter A and B:

Zilch

10-13-2006, 11:13 AM

RLC 1031 is a notch filter, no?

Looks like you've still got the 1.9 kHz peak, tho.

Or am I not quite understanding yet?

I think I'd have gone after the peak with a series notch filter to common....

jerv

10-14-2006, 06:26 AM

Yes, the RLC 1031 is a form of notch filter. This filter topology is called an elliptic or Cauer filter. The added capacitor (C1031) in addition to the standard 4th order topology creates a notch with increases the filter rolloff substantially. It is possible to get up to 60dB/oct rolloff with as little as 5 components. This filter is optimised for 48dB/oct rolloff slopes. For more on Cauer filters, see http://ldsg.snippets.org/FILTERS/Cuadra/elliptic.php3 or varoius forum postings (do a search) on www.htguide.com (http://www.htguide.com).

These two drivers are not really well behaved in the crossover range. The 2214 has a pronounced peak at 1.9kHz, the 2416 has its resonance frequency near (with two ugly impedance peaks, one at 1350 Hz).

With 4th order filters I could not get rid of the 1.9kHz 2214 peak with a conventional LRC series notch filter in parallell with the driver without messing up filter slopes - probably because the notch is so close to the crossover frequency. Only an 8th order filter makes it (though you can still see it down at -27 db).

The broad 2 dB peak at 2 kHz in the summed frequency response is because of the tweeter. (See simulations below with individual curves. The simulation looks a little bit better that the measurement). The impedance peak at 1350 Hz at a point where the response is at its strongest makes a good design using passive components difficult. (An electronic filter might be easier, and is a future design option). Anyway, I fully agree that the 8th order maybe needs some more work.

northwood

10-20-2006, 04:27 PM

northwood

10-20-2006, 04:28 PM

northwood

10-20-2006, 04:29 PM

hope it is useful

John W

01-10-2007, 03:09 PM

They have some 4425 networks in the tent sale (3rd page in).

http://www.jblpro.com/commerce/tent_sale/catalog/index.php?cPath=23

I just picked up a pair for my DIY project.

Titanium Dome

01-10-2007, 06:05 PM

Nice shots.

jerv

01-11-2007, 08:33 AM

I have done some additional work on the 8th order filter. Frequency response is now within plus/minus 2 dB from 200-20000 Hz, and phase tracking is very good. See attached graph - version 1 is black, version 2 is the red curve.

Unfortunately, the number of components has gone up somewhat. To get rid of the 1.3 kHz peak in the 2416, I used a parallell notch filter in series.

To my ears, the new crossover sounds very good - as dynamic and engaging as the original L200t, but with less perceived high-frequency harshness.

4313B

01-11-2007, 08:42 AM

Unfortunately, the number of components has gone up somewhat.Yeah, that happens. :p

Nice job!

jerv

09-28-2007, 01:57 PM

This is my last endeavour with my (strongly) modified 4425 crossover. The topology is now 6th order. The component count is way down compared to the 8th order version, at the expense of some slight frequency response unlinearity at 2 kHz. But it's all still more or less within +/- 2dB 200Hz-15kHz, - and the phase tracking is better then the 8th-order version.

When I built the network, I charge-coupled (thanks, forum!;)) all capacitors in the HF section and used foil inductors in the LF section. I think this version is a keeper.

4313B

09-28-2007, 02:21 PM

Very nice! :)

Earl K

02-27-2018, 06:16 AM

jerv

03-01-2018, 01:37 PM

Thought I'd bump this up from the depths of the LHF glacier ( for Heather's interest ).

Jerv, this was some mighty fine network design ( plus, all accomplished over a decade ago ! )
Do you still have and listen to your special 4425 efforts ?
Any chance ( if you still have them ) of posting frd & zma files for the drivers used ?
thanks :)

Wow! This was a really unexpected and nice surprise: seeing my old 4425-thread resurrected like that!

Thanks for your questions - which I, unfortunately, cannot answer.
The 4425-DIY's are long gone.
They have, in turn, been superseded by a long string of JBL-DIY's (all with special crossovers, of course): 4430, 4333, 250ti, 4313, 4345 variants, and so on. Even DIY M2 monitors.
Had to try them all.
And a pair of very special "250XPL" (as I call them): with LE14H-3, 2108A, 093ti and 046ti (they became keepers, and are still in my house).

Now I'm mostly into large horn systems, with JBL and TAD drivers.
And, alas, the 4425 frd & zma files disappeared long ago in a computer crash.

Earl K

03-01-2018, 01:58 PM

Ahhh then, thanks for the update Jerv.

It's most appreciated .

:)

314carpenter

03-01-2018, 09:11 PM

Earl K

03-02-2018, 05:34 AM

I just pulled four L200T3's out of the graveyard, and would be of limited use as a source of information presently. Maybe you wanted a photo of some internals or something before they go under the knife?

I appreciate the offer, but, I don't think there's anything ( visually oriented ) that I need .

I was simply trying to match-up my network reworks to Jervs results ( of his crossover designs ).

My problem is I don't actually own any of these drivers/horns which makes it tough to get reliable results ( when using extrapolated frd & zma files ).

Thanks for the offer!

:)

johnlcnm

03-02-2018, 10:24 AM

I found a pair of 4425s a few years ago. These were untouched speakers in basically new condition. I never liked the sound of them. Bass was heavy and uncontrolled. There was a resonance in the upper midrange that was irritating and the mids were muddy at best. The first thing that I replaced were the 2416 horn drivers with 2426 drivers. Also replaced the diaphragms with Radian aluminum units. This did not tame the midrange muddiness or the obnoxious resonance. Looked for replacement woofers. The TAD/Pioneers and JBL 2206H were contenders. Got a deal on a pair of 2206Hs new in the box. Removed one connection to the LF inductor on the existing crossover to disable the Low Pass section. I built three 2nd. order Bessel Low Pass circuits at center and +/- crossover freqs. Tested each outside the boxes with the woofer wires running out the port. I don't recall which one was chosen. I also used an inductance compensating section for the new 2206H. It has a higher inductive reactance then the 2214. No box retuning. This did correct the upper midrange resonance and the midrange muddiness. Actually, these speakers sound very respectable now.

In hindsite, I don't think the 2214 was a very good choice for the high crossover frequency that the 4425 needs. That being said, I am sure GTs feet were held to the fire by management imposed cost constants on this system. Also remember, the pair of 4425s that I have are the only ones I have ever heard. The 2214s are now hanging out in the old 2206 boxes in the back room. Oh, I think the 2206H is the right driver for these speakers. What an improvement!!