Recapping AR-4X

[Zilch]

Right you are, Ray, it's compensation for the HF driver response, just like we've been discussing in other threads relating to Altec and Electro-Voice crossover design. The hint above is that when the filter frequency shifts downward, there's an obvious increase in output in that region, indicating it's set where it is to knock down a peak in the response.

Indeed, a little more measurement reveals what's going on:

1) Response of the AR4x tweeters with no highpass filter. There's a 10 dB peak at the low end. That's HUGE, folks. :yes:

2) Response of the pair with the stock 20 uF in series, and 15-Ohm "pot" in parallel. Not quite flat, for other reasons, but the objective is clear.

3) Tweeter #1, before and after.

4) Tweeter #2, before and after.

5) Both shown simultaneously.

AR used a 2+ kHz electrical highpass filter to accomplish a 1.2kHz (or 1.4 kHz, depending upon the woofer lowpass filter used in combination,) acoustic crossover with this specific tweeter response.

Puzzle solved.... :thmbsp:

 

[geaugafletcher]

That's still a pretty raggedy response, no? When you say, "Not quite flat, for other reasons..." what are those other reasons?

One more question:
For anybody not interested in having several attenuations available, would he just put a 15ohm resister in parallel with the tweeter? This is for a friend, of course. Not me, I don't have a problem with collecting too many speakers. In the garage. Or office.

 

[Zilch]

That's still a pretty raggedy response, no? When you say, "Not quite flat, for other reasons..." what are those other reasons?

The 6+ dB notch at 2 kHz is pretty serious. I'd have to do more investigating to find out where all of that is coming from. Remember, though, that the woofer is also playing somewhat up in that range, and it's the acoustic sum of both that makes the total system response. See the last curve in post #16, above, for that.

These tweeters are 40 years old, so there's no knowing what the original response actually was (see below, from Classic Speaker Pages.) The deal breaker with this pair for me is in the second curve attached to post #21, above: the response of the two tweeters is so significantly different over a broad portion of the response spectrum that only separate AutoEQ for each channel could correct it. It's that, or build different filters for each one.

Another option would be to acquire more vintage AR4x tweeters, measure, and match them in pairs according to actual frequency response, IF I was dedicated to the purpose of restoring AR4xs. Remaining, however, would be their inherent beaminess. I haven't quantified that yet, but can certainly see it as I rotate the cabinet slightly off-axis. Also, again, the original FR is unknown.

A further option would be to replace the tweeters with new ones of improved design. I think everyone should be able to appreciate now that the liklihood of finding a "drop-in" replacement that would work optimally without redesigning the crossover is slim; not that it would be too difficult to redo the networks, though. Even then, the fact that the cabinets are not mirror-imaged would remain an issue, and they wouldn't be AR4xs anymore.

I'd also want to work more with the low frequencies. The impedance curves clearly indicate different performance from the different woofer genres I apparently have here. My other (presumably newer) ones have Fs down around 45 Hz, whereas these early "Crosshatched" ones are significantly higher, ~60 Hz. Is that going to matter in a closed-box alignment? Yes, most certainly, it is.

One more question: For anybody not interested in having several attenuations available, would he just put a 15-ohm resistor in parallel with the tweeter? This is for a friend, of course.

Yup, fine. Leaving it out is not a good option, though, as may be seen from the curves above.

*****

O.K., well, I hope nobody gets all bent about this. AR4xs have never been examined and analyzed to this level of detail before, to the best of my knowledge, and it's all subject to verification. I've heard them plenty over the last couple of days, and they sound just fine. My purpose here was to try out and optimize a simple switched fixed-resistor attenuation pot replacement scheme and document it, which is done, and it works.... :thmbsp:

 

[Duffinator]

Another option would be to acquire more vintage AR4x tweeters, measure, and match them in pairs, IF I was dedicated to the purpose of restoring AR4xs.

There is a discussion of replacing the stock tweets with another brand that are better and cheaper in another thread. Maybe the replacement tweets came from parts express. The bottom line of the other discussion was not to waste time with the stock tweets if they are damaged in any way and just upgrade for the same price or less of used original replacements.

 

[Zilch]

Maybe the replacement tweets came from Parts Express.

Maybe they send me a pair to try for free....

If someone would find that thread and link it here, please, that would be helpful. I'm not above buying a pair, actually, and I'll be ordering new binding posts to install in the course of upgrading this pair to the new toggle-switch attenuation scheme. We need a bit more sensitivity, as well, 'cause these tweets are apparently not capable of keeping up with the woofs, even with zero attenuation. I know lots of members have AR4xs, and it'd be good to know the crossover implications and define an upgrade path for potential pursuit by those who would like to do that.

Not sure if I'll keep the original two-pole crossover in these, but I believe I prefer the sound of it with these woofers. I used the stethoscope at 30 Hz and found one pinhole in one of the surrounds, an easy fix. :thmbsp:

Also, I'm looking for a link which someone posted a couple of months ago to an interview with a loudspeaker designer in which he describes the process of starting with driver measurements as elemental to the process. If someone remembers that and will link it here, I would appreciate that, as it goes directly to how some of what I've described above happens. :yes:

 

[RayW]

Maybe they send me a pair to try for free....

Sure, and maybe treefrogs will rule the world.

If someone would find that thread and link it here, please, that would be helpful.

Here is one talking about using an EPI tweet on the 4x. Interestingly enough, you were one of the principles involved. :screwy:

Also, I'm looking for a link which someone posted a couple of months ago to an interview with a loudspeaker designer in which he describes the process of starting with driver measurements as elemental to the process. If someone remembers that and will link it here, I would appreciate that, as it goes directly to how some of what I've described above happens. :yes:

Maybe this one?

Ray

 

[Zilch]

Interestingly enough, you were one of the principles involved. :screwy:

Aw, GEEZ!! What'd I SAY? :

Lookie what I found there. Curves are kinda familiar, huh:

http://murphyblaster.com/content.php?f=AR4x.html

The 2" cone tweeter has truly awful dispersion.

Zilch ain't makin' this stuff up, folks....

The 1 kHz peak is due almost entirely to a tweeter resonance that the simple 20 uf capacitor high pass filter can't deal with.

Didn't look hard enough, apparently, 'cause that capacitor (plus the parallel 15 Ohms) does substantially mitigate the tweeter resonance peak. The woofer bumps it up. :yes:

EPI's not a Parts Express part, right?

Not Voecks, an older interview, someone (and a company) I was not familiar with....

 

[Duffinator]

EPI's not a Parts Express part, right?

No they were brand new tweets for around $10. Might have been Gordon that mentioned it. I'll try and find it later as I have several posts from AR4X threads saved for when I was going to rebuild those speakers.

 

[ozmoid]

O.K., well, I hope nobody gets all bent about this.

Are you kidding? Lovin' every minute of it! Want to do a pair of 2ax next?

...Leaving it out is not a good option, though,...

Without any measuring equipment, I had already arrived at that conclusion. :thmbsp:

 

[Duffinator]

OZ, do you know what the inexpensive replacement tweeter option is?

 

[ozmoid]

OZ, do you know what the inexpensive replacement tweeter option is?

No, I don't recall any recent threads,but I did turn up this one. Not quite recent, though...

 

[RayW]

I don't have time to do all the crunching right now, but for a low XO point I've had good luck with the Tang Band 28-847SD. It has an Fs of ~600Hz and sounds pretty decent. I mated it with a carbon fiber 6.5" that had a massive breakup mode I had to kill of with a fairly low frequency, high order XO. You might want to check that on in your circuit.

Ray

 

[Zilch]

The response curve is certainly similar, Ray -- peak at 1K, dip at 2K, but much better damped.

If it's the 30° also shown, that's very similar as well:

http://www.partsexpress.com/pdf/264-822.pdf

Factory specs show a somewhat different curve, of course. :

http://www.partsexpress.com/pdf/264-822s.pdf

I'm looking further at the low frequency today. Seems like they're running it up to above 2 kHz to fill in the dip, and if so, there's an entire octave of intentional overlap, 1 - 2 kHz.

 

[RayW]

How about doing a vertical polar with both drivers on to see what kind of lobing you're getting around the XO. If the woofer has decent output to 2k then the list of candidates for an affordable tweeter gets much longer.

Ray

 

[Zilch]

http://www.classicspeakerpages.net/dc/user_files/2299.jpg

Genre unknown (1965,) and it looks like that's with a two-pole lowpass working.

Mate that with this, and we'd be cookin':

http://www.classicspeakerpages.net/dc/user_files/2298.jpg

Seems to be fictional, alas....

 

[RayW]

Looks like that would be the perfect pairing. Too bad that's not what is really showing up. Kind of bizarre that they did 0, 30, and 60 on the woofer and 0, 30, and 90 (?) on the tweeter.

Ray

 

[Zilch]

AR4x Woofers Revealed

1) Early cross-hatched woofers have more extended HF response (Red, Grn,) here measured with mid-generation #4 inductor, 1.4 kHz, allegedly. The two I have measure 0.959 and 0.994 mH, not the 0.88 mH "Spec." No, they're just coiled wire, so they didn't change with age.

Mid-generation fabric-surround woofers (with black mastic seal) have truncated HF and reduced efficienty, but play 10 Hz lower (Blu, Vio.) #4 is the inductor that came packaged with these woofers in my own AR4xs. :yes:

2) How to "Fix" the tweeter response anomalies using the extended-HF woofers? I put a 3 kHz two-pole highpass filter on the tweeter to bury it at the 2 kHz notch, and the woofers now play an octave higher. The tweeters certainly sound better, and the midrange is cleaner. Seems these early woofers can play as high as shown, but it's a significant departure from the stock design concept.

3) Does this scheme work with the later woofers? Not worth a poop, nope, Vio. :thumbsdn:

4) So, back to stock configuration for woofers #3 and 4, Blk. Compare to the "Fixed" early woofer scheme, Grn.

With better understanding comes more options. I'll likely get the tweeters Ray suggests and restore these to the original early design, perhaps with two-pole filters on both the LF and HF. The mounting holes are 3.313" diameter, so adapter plates will be required.

Zaph on power response and directivity:

http://www.zaphaudio.com/offaxis.html

 

[RayW]

Nice study, Zilch.

So why not just adjust the XO on top or bottom to ameliorate (love using that word in conversation!) the overlap instead of replacing the tweeter?

Ray

 

[Zilch]

Nice study, Zilch.

So why not just adjust the XO on top or bottom to ameliorate (love using that word in conversation!) the overlap instead of replacing the tweeter?

Thanks, Ray.

That works with the early cross-hatched woofers, which have response above 1.4 kHz to work with. I picked where the notch was in the tweeter response to do it and solve both problems at once. It works, but is a significant departure from the AR4x design. I think it sounds better, actually, as the tweeter is no longer operating at its lower limit, but I'd have to set up A/B to be certain.

Not gonna happen with the newer woofers, though (see below). They roll of naturally at 1.4 kHz (cursor,) so we're stuck with that bandwidth limitation, and consequently, the notch in the tweeter response.

It seems as if the lowpass filter isn't doing much of anything in the way of lowpass with them, apparently due to the rapidly rising woofer impedance. The designer(s) relied upon the rolloff of the woofer for that, and the filter is mostly just tilting the low frequency response downward to provide a more "bassy" voice and a better level match with the tweeter.

Again, I caution readers that this is just me studying the particular units I have on hand, and what I conclude from my own work with these may not be generally applicable. That'll require verification by independent investigators such as we already have for the tweeter portion.

Bottom line for me is still that this pair of tweeters is not well matched, and there's a nasty notch in the response at 2 kHz, so it's new tweeters the likely best option, and then I'll pick which woofer pair to match them with, and redo the crossover as required. I'm leaning toward using the original woofers (despite their less-extended bass response) and the original two-pole lowpass, which DID function to limit the woofer bandwidth, adjusted to minimize overlap and lobing. :thmbsp:

Note: I've also provided distortion measured at 80 dB for your review. Fun stuff, this.... :

 

[RayW]

Fun stuff, this.... :

I thought so. However, I followed realtime and in the driver analysis I forgot what the original intent was and had to backtrack through the whole thread to follow the trail and find that we were just looking to replace the pot. Go figure. :scratch2:

Ray