AR3a - Measurements

Zilch said:

AR was about replicating the concert hall in small rooms; that's the why and the wherefore of the white dots.... :yes:

Click to expand...

Well, Zilch, that's part of it, but they were also concerned about what the record companies were doing at that time. AR believed that records were cut with emphasis in the high frequencies to make up for the inadequate playback systems that were prevalent.

Naturally, this belief was predicated on the assumption that more records would be sold by the company who’s records sounded the best on the majority of playback devices.

Thus, AR provided a way to attenuate the highs to compensate for what the record companies were doing.

Actually, given AR beliefs the controls are a good idea until … those controls start to corrode!

Regards,
Jerry

onplane said:

Naturally, this belief was predicated on the assumption that more records would be sold by the company who’s records sounded the best on the majority of playback devices.

Thus, AR provided a way to attenuate the highs to compensate for what the record companies were doing.

Actually, given AR beliefs the controls are a good idea until … those controls start to corrode!

Click to expand...

I suspect AR was more concerned about the idea that more speakers would be sold if the speakers made the excessively screechy records of the time sound better.

The white dots were indeed a good idea, but I bet another 45-90 degrees of turn on the high side of white dot would have made a lot of modern AR lovers a lot happier.

Paratima said:

Does this include the good Doctor Geddes?

Click to expand...

Earl Geddes said:

In small rooms, the first reflections can never occur later than 10-20 ms (basically this is another way to define a small room), so the first reflections in small rooms must be thought of as a serious problem that causes coloration and image blurring.

Click to expand...

http://www.gedlee.com/downloads/Cum laude.pdf

This is from a guy who claims he never listens in the primary radiation area. I used to play pool a LOT; if you're where you don't hear the initial reflections, you're off the table! Anyway, I've got that pdf on my system and I've read it till my eyes bleed and I still don't get it.

Please forgive the interruption and continue the vastly interesting exploration of the wonders of AR3a. :tongue:

Paratima said:

This is from a guy who claims he never listens in the primary radiation area. I used to play pool a LOT; if you're where you don't hear the initial reflections, you're off the table! Anyway, I've got that pdf on my system and I've read it till my eyes bleed and I still don't get it.

Please forgive the interruption and continue the vastly interesting exploration of the wonders of AR3a. :tongue:

Click to expand...

I find myself in a very odd situation with most of the references in this thread. While I'm reading them they seem to make perfect sense to me, but five minutes after I close the files what felt like understanding vanishes and I'm just as confused as I was before I read them.

gkimeng said:

I went looking last night for copies of the old Audio or High Fidelity magazine test reports, for which I'm pretty sure the reviewers did not tear off the molding (not so sure about the grills because I don't know if the reports were done before or after AR started velcro'ing them).

Click to expand...

It seems Consumer's Union only averaged the response over a 30° listening window, resulting in "adverse comments."

It' be interesting to see what happened in the subsequent "Letters" section(s); Allison asserts the issues, whatever they were, derived from the fact that CU was listening in the direct rather than the reverberant field:

http://www.classicspeakerpages.net/...nal_ar_documents/letter_from_allison_to_hoff/

Editorial comment: Well, it's disingenuous, at best. "Our design approach is based upon in-room power response. The actual system frequency response measurements are poor, and we know the reasons, so we'll only publish data showing the individual driver responses taken without the major cabinet design features in place. That energy is going SOMEWHERE in the room, obviously, and since it's the integrated reverberant field listeners really hear, no one will be the wiser."

I haven't looked more comprehensively, but it appears as though AR did not publish frequency response measurements for any of their vintage systems, rather, their "Acoustic power output," undefined, as best I have thus far determined, consistent with this philosophy....

http://www.classicspeakerpages.net/...nal_models_other/ar-3a_series_frequency_resp/

Zilch said:

It seems Consumer's Union only averaged the response over a 30° listening window, resulting in "adverse comments."

It' be interesting to see what happened in the subsequent "Letters" section(s); Allison asserts the issues, whatever they were, derived from the fact that CU was listening in the direct rather than the reverberant field:

Click to expand...

Many years ago I went hunting through the bound CR archives at my local library to see if this generated any action in the "letters to the editor" section of subsequent issues. I could not find indication of any further exchanges between AR and CU, at least not any that CU published.

History records that this incident was the start point for the speaker-environment research that eventually led to Allison Acoustics.

BTW, how are those woofers coming along...?

gkimeng said:

History records that this incident was the start point for the speaker-environment research that eventually led to Allison Acoustics.

Click to expand...

In the Allison interview, it's clear that Small was not impressed; imaging and accuracy reign supreme today, and the objective is to minimize the influence of the room rather than attempt to exploit it. I'll know more about how these perform once I get both running complete.

gkimeng said:

BTW, how are those woofers coming along...?

Click to expand...

Woofers are done and here, and I now face the dilemma of whether to stuff all of that fiberglass back in them. I may only do that with one, to evaluate how much it actually matters.

The L-pad versus pot thing required considerable effort, and I'm not sure it's complete yet, even, but I've got to move on for now. The findings are in themselves quite interesting, but measuring the mid driver on axis and running my wires across the molding altered the 1.8 kHz notch, as well, as may be seen in the results. I also made a dumb mistake at one point, which turned out to be instructive; more on that in an Appendix, perhaps....

Zilch said:

Editorial comment: Well, it's disingenuous, at best. "Our design approach is based upon in-room power response. The actual system response measurements are poor, and we know the reasons, so we'll only publish data showing the individual driver responses taken without the major cabinet design features in place. That energy is going SOMEWHERE in the room, obviously, and it's the integrated reverberant field listeners really hear, so no one will be the wiser."

Click to expand...

I think at some point AR ran headfirst into a conflict between their belief that "frequency response measurements" of speaker systems were useless for anything other than QA'ing their production vs. consumer demand for them. So they stuck in test plots of drivers from the Receiving Dept. inspections, because plots of individual drivers were at least accurate and repeatable if not useful data, and then tried to devise something else to measure that they thought would be useful.

http://www.classicspeakerpages.net/...nal_ar_documents/loudspeaker_testing_and_mea/

(Someone smarter than me will need to review this and decide whether it makes sense or is just Villchur engaging in a bit of rationalizing about AR speakers' response plots.)

Reflected Impedance:

Considerable analytical effort was invested by others in the L-pad + 25 Ohms in parallel with the driver pot replacement at CSP, the objective being to simulate the variable resistance behavior of the troublesome 16-Ohm AR attenuation potentiometer. My theoretical perspective on that solution is that it amounts primarily (and perhaps merely) to lowering the impedance of the driver, and for a meaningful view of the quality of the simulation, the amount of attenuation is far less important than the impact the combination of attenuator (pot vs. L-pad vs L-pad plus resistor) might have upon the filter, i.e., the reflected impedance of the attenuator/driver combination:


Here, then, is the impedance "seen" by the midrange highpass filter in my AR3a #2. That of the driver alone is the Red curve; same as with the L-pad alone set to Max, which effectively removes itself from the circuit.

Set the L-pad to its midpoint, and the impedance increases from 5.0 Ohms at 2 kHz to 7.5 Ohms, Green, while remaining constant at the 600 Hz resonance peak, just above the filter frequency, 12 Ohms.

Install 25 Ohms across the driver, and the impedance drops some across the spectrum, Orange, but disportionately more at resonance; this technique is frequently used to mitigate resonance peaks in other designs. At Mid attenuation setting, Blue, the impedance drops to 10 Ohms at the resonance peak, and the filter frequency rises slightly.

So, how do these two compare with the stock potentiometer? With 16 Ohms across the driver at the Max setting, Yellow, the impedance drops to 4 Ohms at 2 kHz, and the resonance peak is further mitigated. However, at Mid setting, Violet, it's 11.5 Ohms there, and, more significantly, in the resonance/crossover region, higher still, and higher than either of the L-pad solutions, with a concomitantly lower filter frequency, and, because of the substantial reduction in the resonance peak, a different filter contour, as well.

Does any of this matter? See the corresponding response curves at Post #1. Click "First" at the top or bottom of this page to get there expeditiously.

Here they are overlaid for comparision under the identical measurement conditions, in the same colors. Note the addition of a 25% (4-Ohm) pot setting in both curves, Cyan:


RoyC will be quite pleased by these findings; L-pad w/25 Ohm resistor in parallel with the driver set to mid attenuation virtually overlaps the stock pot impedance, output level, and response when set at 25%.

O.K., what IS the reflected impedance of the pot/mid driver combo which we should be emulating if the result is to be the same as the stock attenuator throughout the attenuation adjustment range?


HUH? How is it possible for the impedance to vary from 4 to 16 Ohms without the filter frequency moving drastically as consequence? We're certainly not seeing that in the response measurements.

What sort of magic crossover filter could be immune to such large impedance shifts?

One that doesn't do much, apparently. It's the "response shaping" thing:


The fundamental response characteristic is built into the driver, and the filter merely trims its flanks; it doesn't give a whit, basically, subject to verification by others, of course.

On to woofers, now.... :thmbsp:

Zilch, I have to say this thread is a hell of a read. You're living up to your motto and then some! :thmbsp:

Though I still get a pang thinking about where all this is going...

winters860 said:

Zilch, I have to say this thread is a hell of a read. You're living up to your motto and then some! :thmbsp:

Though I still get a pang thinking about where all this is going...

Click to expand...

At the rate Zilch is going, by the time he's through restoring these so he can evaluate them in original configuration, they will probably be closer to factory spec than any other vintage 3a's on the planet. Maybe someone will make him an offer he can't refuse before he decides to get creative.

And, of course, if that doesn't happen there may still be spare drivers available soon...

Zilch said:

Considerable analytical effort was invested by others in the L-pad + 25 Ohms in parallel with the driver pot replacement at CSP, the objective being to simulate the variable resistance behavior of the troublesome 16-Ohm AR attenuation potentiometer. My theoretical perspective on that solution is that it amounts to lowering the impedance of the driver, and for a meaningful view of the quality of the simulation, the amount of attenuation is far less important than the impact the combination of attenuator (pot vs. L-pad vs L-pad plus resistor) might have upon the filter, i.e., the reflected impedance of the attenuator/driver combination:

Here, then, is the impedance "seen" by the midrange highpass filter in my AR3a #2. That of the driver alone is the Red curve; same as with the L-pad alone set to Max, which effectively removes it from the circuit.

Set the L-pad to its midpoint, and the impedance increases from 5.0 Ohms at 2 kHz to 7.25 Ohms, Green, while remaining constant at the 600 Hz resonance peak, just above the filter frequency, 12 Ohms.

Install 25 Ohms across the driver, and the impedance drops some across the spectrum, Green, but disportionately more at resonance; this technique is frequently used to mitigate resonance peaks in other designs. At Mid attenuation setting, Blue, the impedance drops to 10 Ohms at the resonance peak, and the filter frequency rises slightly.

So, how do these two compare with the stock potentiometer? With 16 Ohms across the driver at the Max setting, Yellow, the impedance drops to 4 Ohms at 2 kHz, and the resonance peak is further mitigated. However, at Mid setting, Violet, it's 11.5 Ohms there, and, more significantly, in the resonance/crossover region, higher still, and higher than either of the L-pad solutions, with a concomitantly lower filter frequency, and, because of the substantial reduction in the resonance peak, a different filter contour, as well.

Does any of this matter? See the corresponding response curves at Post #1. Click "First" at the top or bottom of this page to get there expeditiously.

O.K., what IS the reflected impedance of the pot/mid driver combo which we should be emulating?

HUH? How is it possible for the impedance to vary from 4 to 16 Ohms without the filter frequency moving drastically as consequence? We're certainly not seeing that in the response measurements.

What sort of magic crossover filter could be immune to such impedance shifts?

One that doesn't do much, apparently. It's the "response shaping" thing:

The fundamental response characteristic is built into the driver, and the filter merely trims its edges; it doesn't give a whit, basically, subject to verification by others, of course.

On to woofers, now.... :thmbsp:

Click to expand...

More 3a midrange measurements (power resp. and imp/phase) can be found here for you data hungry lurkers.....
http://www.classicspeakerpages.net/IP.Board/index.php?showtopic=4521&hl=midrange

spkrdtr said:

More 3a midrange measurements (power resp. and imp/phase) can be found here for you data hungry lurkers.....
http://www.classicspeakerpages.net/IP.Board/index.php?showtopic=4521&hl=midrange

Click to expand...

Thanks, Carl, you've been here before, obviously. :thmbsp:

Now that I have worked out how, I should do the same attenuation/impedance/response study with the tweeter, as well, I suppose, but it appears the result will be very much the same with the same design approach already in evidence there. Maybe I'll come back to that; for now, tho, I most want to move on to evaluating the complete system with the woofers.... :yes:

Zilch said:

Thanks, Carl, you've been here before, obviously. :thmbsp:

I should do the same attenuation/impedance/response study with the tweeter, as well, I suppose, but it appears the result will be very much the same with the same design approach already in evidence there. Maybe I'll come back to that; for now, tho, I want to move on to evaluating the complete system with the woofers.... :yes:

Click to expand...

Excellent work here, Zilch!!

I'd like to make a qualitative observation about the impedance curves showing the original pot and L-pad + 25 ohms. In the full ON position they are almost identical. Yes, there is a slight difference, but it's really, really small.

Then when you compare at 50%, it's clear that the original pots provide significantly more attenuation. Not sure that's of much interest to most AR folks as most perfer the "dot" position which is no where near the 50% mark.

Again, very nice work, Zilch. I'm looking forward to your measurements with the woofers installed. I believe Carl has done some measurements looking at the impact of the amount and type of fill.

Regards,
Jerry

onplane said:

I'd like to make a qualitative observation about the impedance curves showing the original pot and L-pad + 25 ohms. In the full ON position they are almost identical. Yes, there is a slight difference, but it's really, really small.

Then when you compare at 50%, it's clear that the original pots provide significantly more attenuation. Not sure that's of much interest to most AR folks as most perfer the "dot" position which is no where near the 50% mark.

Click to expand...

I have updated post #51 with a remeasure of the reflected impedance, added the corresponding frequency responses, and included a new measurement with the stock pot at 25% rotation per your suggestion.... :thmbsp:

I managed to dig up a higher-res scan of the 3a driver response curves from the old AR brochure.

Again, they don't show the response in the cabinet; we know why not.

The "Overlap" notation is conceptually interesting, consistent with my earlier observation that the midrange is "propping up" the tweeter on-axis.

Power response ruled, so they weren't afraid to compromise the on-axis response to optimize that....

Zilch said:

Again, they don't show the response in the cabinet; we know why not.

Click to expand...

Right. This isn't any different from the other one, just easier to read and with some additional notes.

I'm curious to see how superimposing these on each other compare with measurements of the entire system.