Jump to content

Crossover mods for the AR4x II


Recommended Posts

I also listened to both Dave's (DLS) AR4x and the eWave variant shown quite a bit over the weekend, along with another critical listener yesterday, and despite the eWave not yet being optimized, the frequency responses are similar, yet they sound significantly different. The waveguide is the same 8" square 90° x 50° used in SpitWad, so we already basically know how the directivity comparison will shake out. Listening in ZilchLab, directivity alone does not account for the clearly audible difference(s) in performance.

Is there any way to describe this audible difference in words? (I'm not talking about value judgments like "better/worse.")

Link to comment
Share on other sites

  • Replies 88
  • Created
  • Last Reply
I WAS going to take the easy path and concede defeat 'til I found the phase/forward axis thing, which I quite frankly don't fully understand as yet. To investigate that further, I'm going to generate the requisite files and load them into PCD for a look, which will likely trigger an alternative design. I'm hoping Dave will run some vertical polars to either confirm or contest my finding(s).

Not totally sure what the "phase/forward axis thing" is unless you mean the downward best-summing axis. Your result doesn't look bad to me with near 6dB summing and flat response on axis. The real comparison would be against the individual section curves and their phase responses as in my post #33

http://www.classicspeakerpages.net/IP.Boar...ost&p=83078

If your individual response and phases look the same then your system is a clone of mine and my polars must be the similarly unsymmetrical.

Looking back at my curves, the phase blend is good from 1k and up but deviates below. There is also bit too much overlap between woofer and tweeter but the deviating phase prevents much response buildup on the low side of the crossover point. As I recall I was using the 3rd order network and the response bump of the filter's corner to fill in the 1500 Hz hole of the tweeter. That tends to limit options since the crossover bump must coincide with the tweeter dip, the crossover point ended up lower than desired and thus the woofer network mods were necessary.

Still, with some extra optimizing you might be able to get good response and a little better up/down balance.

I also listened to both Dave's (DLS) AR4x and the eWave variant shown quite a bit over the weekend, along with another critical listener yesterday, and despite the eWave not yet being optimized, the frequency responses are similar, yet they sound significantly different. The waveguide is the same 8" square 90° x 50° used in SpitWad, so we already basically know how the directivity comparison will shake out. Listening in ZilchLab, directivity alone does not account for the clearly audible difference(s) in performance.

"Similar" response isn't always enough for an audible match. I did find a reasonable match between the modified AR4x and a Snell K5. The audible differences were explained by small response differences. Could that be the case for you? (Or in the end, is it really all about power response?)

David

Link to comment
Share on other sites

Is there any way to describe this audible difference in words? (I'm not talking about value judgments like "better/worse.")

Premature.

As Dave points out, the spectral balances have to be really close, and they aren't right now.

On a preliminary basis, I could AutoEQ them individually and listen, but it's not as if that won't introduce artifacts, as well. I just gotta pay the dues here, is all.

"Similar" response isn't always enough for an audible match. I did find a reasonable match between the modified AR4x and a Snell K5. The audible differences were explained by small response differences. Could that be the case for you? (Or in the end, is it really all about power response?)

We intend to find out.... ;)

Advanced directivity stuff that'll look familiar; click "Preview data program" here and step through the examples:

http://www.gedlee.com/Loudspeakers.htm

"Comments" button tells what's up with each.

[Works best in IE. ]

Link to comment
Share on other sites

Premature.

As Dave points out, the spectral balances have to be really close, and they aren't right now.

I was thinking that if the difference could be described now and again when you think they're as close as you can get them, it might provide some insight into how audible measured differences are.

Link to comment
Share on other sites

Not totally sure what the "phase/forward axis thing" is unless you mean the downward best-summing axis. Your result doesn't look bad to me with near 6dB summing and flat response on axis. The real comparison would be against the individual section curves and their phase responses as in my post #33.

We have a discrepancy.

Here are the plots presented with both normal and minimum phase, both 1.0 mH and 1.5 mH LF inductors, and as a final check, on axis response with normal vs. inverted tweeter polarity. We agree on the response, but not the phase. Did you recheck the phase alignment after installing the conjugate?

post-102716-1277784723.jpg

post-102716-1277784747.jpg

post-102716-1277784777.jpg

post-102716-1277784795.jpg

post-102716-1277784809.jpg

Link to comment
Share on other sites

I was thinking that if the difference could be described now and again when you think they're as close as you can get them, it might provide some insight into how audible measured differences are.

These are certainly audible:

post-102716-1277788711.jpg

Link to comment
Share on other sites

These are certainly audible:

E-wave appears to be lower at 2k and higher at 3-5k relative to the 4x. One might expect an oboe to sound a little depressed and a piccolo a little elevated. Hear anything like that?

Link to comment
Share on other sites

We have a discrepancy.

Here are the plots presented with both normal and minimum phase, both 1.0 mH and 1.5 mH LF inductors, and as a final check, on axis response with normal vs. inverted tweeter polarity. We agree on the response, but not the phase. Did you recheck the phase alignment after installing the conjugate?

The conjugate, or quasi second order is included in the curves. (See previous post in that thread.) It didn't change the phase too much.

Its a little hard for me to follow your curves. Your minimum phase curves have less wrap around but wouldn't be the same time delay? In Holm you can set the time delay for aproximate minimum phase for the tweeter and then use exactly the same window for the woofer. Any minimum phase autosetting by the program will kill the relative phase difference that you need to see.

Looks to me like you have a different bottom corner on the tweeter: less overlap and a rising phase because of it. Take a long look at the curve from #33 at how the tweeter response doesn't turn a corner at 1500, more at 900-1k. That is what got the phase overlap just right. Third order networks are sensitive, you might try varying the inductor to see if you can get a more similar response, or just go for your own values for the network. (not sure how your tweeter matches the one I used.)

As Laurie Fincham (KEF) used to say, anyone can design one perfect speaker, it's building multiples (in production) that is the challenge.

Link to comment
Share on other sites

The conjugate, or quasi second order is included in the curves. (See previous post in that thread.)

O.K., I do see that, now.

Looks to me like you have a different bottom corner on the tweeter: less overlap and a rising phase because of it. Take a long look at the curve from #33 at how the tweeter response doesn't turn a corner at 1500, more at 900-1k. That is what got the phase overlap just right. Third order networks are sensitive, you might try varying the inductor to see if you can get a more similar response, or just go for your own values for the network. (not sure how your tweeter matches the one I used.)

Old school, I guess, I call that the "knee," and yes, they are quite different:

http://www.classicspeakerpages.net/IP.Boar...ost&id=5425

http://www.classicspeakerpages.net/IP.Boar...ost&id=5774

Its a little hard for me to follow your curves. Your minimum phase curves have less wrap around but wouldn't be the same time delay? In Holm you can set the time delay for aproximate minimum phase for the tweeter and then use exactly the same window for the woofer. Any minimum phase autosetting by the program will kill the relative phase difference that you need to see.

Yes, which is why I provided the normal phase with all of its attendant wraps, where you can see that we have the same driver phase relationship in the region of interest, but alignment occurs at a higher frequency, above crossover, from the axis midway between the drivers. Where they DO align at crossover is at the axis of the maximal inverse vertical cancellation, which occurs well below the midpoint between drivers, indicating that the forward lobe is skewed downward, and it would be best to listen to the speakers upside-down with this crossover. The implications for horizontal deployment are obvious, though I suppose it could be argued that they must be toed-in to avoid the HF rolloff, anyway, but even then, tweeters outboard mirror-imaging would be preferred.

I have loaded the data and design into PCD, which confirms my findings. A simple means of verification is to measure the system response on the design axis, midway between the drivers, with both normal and inverse tweeter polarity. Does a deep inverse notch appear at crossover? If not, at what vertical angle does that occur? I don't believe that tweeter variability accounts for this, but I'll check that out by locating the forward axis with each of these six samples.

http://www.classicspeakerpages.net/IP.Boar...ost&id=5806

PCD also reveals another issue - measure the system impedance to see it. The second pole of the highpass is the culprit, and frankly, I'm somewhat stymied as to how to resolve it; I'm trying to make this happen with a tank filter, instead, maybe. Perhaps it's time to move the challenge over to Tech Talk.... ;)

Link to comment
Share on other sites

Perhaps it's time to move this challenge over to Tech Talk.... ;)

As long as the ultimate goal is still to "mod, tweak or upgrade" a classic New England speaker and there's no personal potshots being taken, you can get as technical as you want. Complicated on-topic subject matter discussed in a civil manner was never the problem.

Link to comment
Share on other sites

O.K., I do see that, now.

Yes, which is why I provided the normal phase with all of its attendant wraps, where you can see that we have the same driver phase relationship in the region of interest, but alignment occurs at a higher frequency, above crossover, from the axis midway between the drivers. Where they DO align at crossover is indicated by the axis of the maximal inverse vertical cancellation, which occurs well below the midpoint between drivers, indicating that the forward lobe is skewed downward, and it would be best to listen to the speakers upside-down with this crossover. The implications for horizontal deployment are obvious, though I suppose it could be argued that they must be toed-in to avoid the HF rolloff, anyway, but even then, tweeters outboard mirror-imaging would be preferred.

I have loaded the data and design into PCD, which confirms my findings. A simple means of verification is to measure the system response on the design axis, midway between the drivers, with both normal and inverse tweeter polarity. Does a deep inverse notch appear at crossover? If not, at what vertical angle does that occur? I don't believe that tweeter variability accounts for this, but I'll check that out by locating the forward axis with each of these six samples.

http://www.classicspeakerpages.net/IP.Boar...ost&id=5806

PCD also reveals another issue - measure the system impedance to see it. The second pole of the highpass is the culprit, and frankly, I'm somewhat stymied as to how to resolve it; I'm trying to make it happen with a tank filter, instead, maybe. Perhaps it's time to move this challenge over to Tech Talk.... ;)

Wow, you guys are really getting into perfecting these things. Fun stuff it all seems, but I have to wonder how beneficial all this effort is going to be to the average end user.

Remember, these are 4x's - not 3a's or 7's or 9's or LST's, etc. I note that because I view the 4x as more of a utility speaker tucked somewhere on a shelf or stand in a corner where not much critical listening is done. See the pic at this link showing an AR brochure ad pic http://www.classicspeakerpages.net/library..._1970_pg12.html

From the ad, it appears even AR felt the 4x was not intended for any real critical listening.

Keep up the good work. I truly hope your efforts help those doing restorations in the future with the intent of making them the centerpiece of a two channel system.

Perhaps for a few owners, this is the only speaker they have and they will benefit from your work. At least I hope so.

Link to comment
Share on other sites

Wow, you guys are really getting into perfecting these things. Fun stuff it all seems, but I have to wonder how beneficial all this effort is going to be to the average end user.

Remember, these are 4x's - not 3a's or 7's or 9's or LST's, etc. I note that because I view the 4x as more of a utility speaker tucked somewhere on a shelf or stand in a corner where not much critical listening is done. See the pic at this link showing an AR brochure ad pic http://www.classicspeakerpages.net/library..._1970_pg12.html

From the ad, it appears even AR felt the 4x was not intended for any real critical listening.

Keep up the good work. I truly hope your efforts help those doing restorations in the future with the intent of making them the centerpiece of a two channel system.

Perhaps for a few owners, this is the only speaker they have and they will benefit from your work. At least I hope so.

I grew up with 4x's (that's as expensive as my Dad could afford), and trust me, they were used for super-critical listening. My Dad loved music, and I was a fledgling jazz drummer and I hung on every stroke of Joe Morello with the Dave Brubeck Quartet or Tony Williams with Miles Davis. We listened for hours on end.

AR built the 4x to a price point, and its main attraction was the 8" scaled-down version of the 10"/12" AS systems into a true 'bookshelf' size. The 2 1/2" tweeter and simple x-o was all that AR was going to spend on the M-HF portion of that product, and considering the cost/marketing constraints (as well as the limitations of x-o knowledge in those days), the original 4x did more than pretty well for itself, both from a sound/performance standpoint and certainly from a sales/marketing standpoint.

I think these Z-SD mods to the original 4x take it to another level of refinement and performance, but AR would not have been able to do this in 1966.

I'd love to hear it though, especially against a 'stock' 4x.

Steve F.

Link to comment
Share on other sites

Wow, you guys are really getting into perfecting these things. Fun stuff it all seems, but I have to wonder how beneficial all this effort is going to be to the average end user.

Mods, Tweaks and Upgrades is separate from the main speaker brand forums because "average end users" of classic New England loudspeakers probably aren't planning on doing any mods, tweaks or upgrades...at least not to the first pair they own. But how many people who frequent CSP only have one pair of speakers...?

Link to comment
Share on other sites

I grew up with 4x's (that's as expensive as my Dad could afford), and trust me, they were used for super-critical listening. My Dad loved music, and I was a fledgling jazz drummer and I hung on every stroke of Joe Morello with the Dave Brubeck Quartet or Tony Williams with Miles Davis. We listened for hours on end.

AR built the 4x to a price point, and its main attraction was the 8" scaled-down version of the 10"/12" AS systems into a true 'bookshelf' size. The 2 1/2" tweeter and simple x-o was all that AR was going to spend on the M-HF portion of that product, and considering the cost/marketing constraints (as well as the limitations of x-o knowledge in those days), the original 4x did more than pretty well for itself, both from a sound/performance standpoint and certainly from a sales/marketing standpoint.

I think these Z-SD mods to the original 4x take it to another level of refinement and performance, but AR would not have been able to do this in 1966.

I'd love to hear it though, especially against a 'stock' 4x.

Steve F.

I kind of expected this response to come from someone reminiscing on the good old days.

That was then. However, this is now. If you could have a super-tweaked pair of 4x's, would you use them for critical listening (other than as nearfield monitors) with your main system?.

Yes, Z's and SD's mods will indeed take the 4x to another level of refinment. I just feel their enthusiasm could/should? be channeled towards a more worthwhile loudspeaker that is used primarily for critical listening, is all (as Z would say).

Link to comment
Share on other sites

I just feel their enthusiasm could/should? be channeled towards a more worthwhile loudspeaker that is used primarily for critical listening, is all (as Z would say).

What about AR4x renders them inherently less worthwhile?

What did Allison and Berkovitz say about this?

Would we agree with it today?

Link to comment
Share on other sites

I just feel their enthusiasm could/should? be channeled towards a more worthwhile loudspeaker that is used primarily for critical listening, is all

AR-4's have the virtue of being relatively plentiful and less expensive than the higher end models. If this effort works out it could be the proof of concept for a more ambitious effort. Zilch does have a pair of 3a's...

Link to comment
Share on other sites

What about AR4x renders them inherently less worthwhile?

What did Allison and Berkovitz say about this?

Would we agree with it today?

1) The background music they put out will sound a bit better - for what it's worth... ;)

2) Huh? Can't answer and don't care

3) diddo

Link to comment
Share on other sites

AR-4's have the virtue of being relatively plentiful and less expensive than the higher end models. If this effort works out it could be the proof of concept for a more ambitious effort. Zilch does have a pair of 3a's...

Aaahhh, a much more worthwhile effort - but a WHOLE NEW BALL GAME AS WELL. ;)

Link to comment
Share on other sites

I'd love to hear it though, especially against a 'stock' 4x.

Steve F.

Well, for 5 new network components, you can hear it. Well within your capabilities.

David

Link to comment
Share on other sites

Yes, which is why I provided the normal phase with all of its attendant wraps, where you can see that we have the same driver phase relationship in the region of interest, but alignment occurs at a higher frequency, above crossover, from the axis midway between the drivers. Where they DO align at crossover is at the axis of the maximal inverse vertical cancellation, which occurs well below the midpoint between drivers, indicating that the forward lobe is skewed downward, and it would be best to listen to the speakers upside-down with this crossover. The implications for horizontal deployment are obvious, though I suppose it could be argued that they must be toed-in to avoid the HF rolloff, anyway, but even then, tweeters outboard mirror-imaging would be preferred.

Forward lobe of your system angles down, but that is not the case with mine. You need to adjust your network values to get the knee and phase of the tweeter to match, otherwise you will be stuck with the downward lobe.

I have loaded the data and design into PCD, which confirms my findings. A simple means of verification is to measure the system response on the design axis, midway between the drivers, with both normal and inverse tweeter polarity. Does a deep inverse notch appear at crossover? If not, at what vertical angle does that occur? I don't believe that tweeter variability accounts for this, but I'll check that out by locating the forward axis with each of these six samples.

Of course, maximum cancelation occurs when driver levels are exactly the same and the reverse connected phase split is exactly 180. Cancelling vectors, you know. My phase curves are on top of each other at the crossover point so I believe the design axis was good. I would agree that tweeter variation is less likely and suspect that one of us is using different enough crossover values. Try increasing you tweeter inductor to see if it moves things in the right direction.

What is PCD?

PCD also reveals another issue - measure the system impedance to see it. The second pole of the highpass is the culprit, and frankly, I'm somewhat stymied as to how to resolve it; I'm trying to make this happen with a tank filter, instead, maybe. Perhaps it's time to move the challenge over to Tech Talk.... ;)

Yeah, I always had a suspicion that the impedance curve would be less than perfect. Since the network is filling in the on-axis tweeter hole (electrical bump) the impedance will have to dip accordingly. If it is excessive then a less bumped compromise could be had. I've done that before: you decide how low of an impedance you can stomach and get the corresponding amount of EQ to go with it. On the other hand, my crappy old Scott amp was happy to drive it...

David

Link to comment
Share on other sites

Wow, you guys are really getting into perfecting these things. Fun stuff it all seems, but I have to wonder how beneficial all this effort is going to be to the average end user.

Sure, the effort is beyond what may be warranted by a humble pair of AR4x. Its been more of an academic excercise, and hopefully educational to others curious about the ins and outs of system design.

Still, if we can work out the best nominal values, anybody that can recap a system would be able to build their own pair. Makes more sense than replacing the original network with zillion dollar stock value parts.

David

Link to comment
Share on other sites

Here are the measured forward axes of the system with each of the six sample tweeters installed, as determined by an inverse notch 40 dB or more deep, varying between -15.10° and -20.41°, with three clustered at -20°:

post-102716-1277867590.jpg

Link to comment
Share on other sites

Sure, the effort is beyond what may be warranted by a humble pair of AR4x. Its been more of an academic excercise, and hopefully educational to others curious about the ins and outs of system design.

Still, if we can work out the best nominal values, anybody that can recap a system would be able to build their own pair. Makes more sense than replacing the original network with zillion dollar stock value parts.

David

I agree...

Actually, there are not many non-hobbyists who will bother to tackle capacitor replacement, much less anything else. The results of this exercise can prove interesting to hobbyists, however, who are MUCH more likely to run across a pair of 4xs to play with than a pair of 3as. There were around 400,000 4xs manufactured compared to around 100,000 3as, and the 4x does not have a "collector's value" tag attached to it. I doubt there are many 3a owners (audio hobbyists or otherwise) who would be inclined to change anything, regardless of any sonic improvements discovered in an internet thread.

At around 300,000 manufactured, the AR-2ax would be another good candidate for experimentation, imo.

I'm looking forward to the outcome of this thread.

Roy

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...