NOT AR-11 Speaker!

Pete B

AK Member
Subscriber
It has been said that the AR-3a/AR-11 cannot be duplicated with modern drivers.
I'm calling this the NOT AR-11 since I'm sure someone will tell me that this could
never sound fully like an actual AR-11.

I fabricated this plate that mounts in place of the AR 1.5" mid but it includes a
currently available mid and tweeter closely spaced for improved off axis response.
This could serve as a replacement for the original drivers or for an entirely new
speaker, there are many woofers that would work in place of the AR but that is
for another day.

Original replacement mids and tweeters are getting either scarce and/or expensive and
this is a way around the issue.

AR boxes are showing up for sale because more money can be made parting out
the drivers and parts and this could be used to bring them back to life.

This is a fully recapped AR-11 with my mid tweeter plate installed, the box/crossover
is completely stock. Film caps were used with ESR compensation resistors, 1% Dayton
polys for the tweeters and a stack of film polys selected to be 1% of the standard value.
The woofer cap is a Bennic Electrolytic.

I used a porous dust cap to hide the mid since that is not yet for the public.

I'll be showing the one completed speaker at the Blue Mountain Audio Fest (GlennFest) in
NY tomorrow Sat Sept 15.

NOT-AR-11-FRONT-SM.jpg
 
Last edited:
It has been said that the AR-3a/AR-11 cannot be duplicated with modern drivers.
I'm calling this the NOT AR-11 since I'm sure someone will tell me that this could
never sound fully like an actual AR-11.
According to some comments I've read, that might be a good thing :)

I fabricated this plate that mounts in place of the AR 1.5" mid but it includes a
currently available mid and tweeter closely spaced for improved off axis response.
Interested in listening impressions.
 
According to some comments I've read, that might be a good thing :)


Interested in listening impressions.

Ha, yes it might be.
I actually think that the AR-11 sounds pretty good with the tweeter switch at -3dB and the mid at -6dB,
actually -3 is not enough attenuation and -6 is too much, 4 or 5 should be just about right.

@RxDx Do or did you own any AR speakers that you spent a good amount of time with?
 
Yes, many AR speakers. I started with AR-2, found CSP, and after lurking for a while,
decided to get some AR-3as. I've had two pair in the main system for about 6 years
now. I listened to AR-2a and AR-2ax for years before that. Still have four pair of the 2
series.
 
Just noticed that you are the maintainer of the Vintage AR specs - very nice!

Do you run your 2 pair of 3a's as a stack or front/rear, or ....?
 
Thanks!

I'm running each pair off its own receiver, 4.0, I guess (front and rear). I use
an Integra receiver as a glorified preamp, since it's 8 ohm only, and I can
use EQs ahead of each receiver.
 
Pete - What specific drivers did you choose, and how did you choose them? Are they wired in phase per the original configuration? Any comments on the sounds vs. stock AR11s?

(I've had AR4x, AR3a, AR11B, AR9LS(i), and have AR2ax now.)
 
I chose the drivers based on size, the same as original, response flatness, distortion, etc.
I mod one of them to adjust Fc, and impedance match them, this is just a prototype at this point.
No mods are required to the AR-11.
They are wired with the same phase as original which from what I know has the tweeter out of phase.
I did this to keep it as close to stock as possible, I prefer an in-phase connection but that would likely
require a crossover mod.
I do have a few minor mods to the AR-11 crossover that I prefer but those are for another day.
 
Last edited:
They are wired with the same phase as original which from what I know has the tweeter out of phase. I did this to keep it as close to stock as possible, I prefer an in phase connection but that would likely require a crossover mod.

Technically the tweeter is not out of phase, it is reversed polarity. This has implications for the accuracy of sound reproduction and midrange-tweeter blending in the upper frequencies.

Although often conflated, the distinction between the terms phase and polarity is crucial in electrical and audio engineering:
(a) A "Phase Shift" is delayed in time, while "Polarity Reversal" is instantaneous. While a symmetric sine wave delayed in time by 180 degrees appears to be indistinguishable from a polarity reversed sine wave—the two summing to zero—an asymmetric waveform—cf. actual music instead of idealized, pure sine waves—would have different behavior when summed, given a half-cycle (180 degree) phase shift (moved in time) vs. inversion of polarity (not moved in time). Taking a time-varying signal—i.e. music—and splitting into two signals, one phase shifted by 180 degrees and one inverted, and then summing the results will generally not yield zero; this is distortion. Generations of undergraduates have unfortunately been taught the error of a 180 degree phase shift equaling polarity.

(b) Polarity Reversal swaps the compression and rarefaction halves of the waveform while phase shift does not. Polarity is the direction of compression or rarefaction wavefronts. In short, does a positive signal move the speaker forward or backward. Somewhat simplified, the issue is a difference in how the ear and brain process the different waveform types:
Compression Wave (0 —> Vmax —> 0)​
vs.
Rarefaction Wave (0 —> Vmin —> 0)​
The reason the waveform order matters is evolution optimized our ears to prefer compression before rarefaction. This is called the "Wood Effect". See The Wood Effect: Unaccounted Contributor to Error and Confusion in Acoustics and Audio by Clark Johnsen (1988).​

The reason for polarity reversal of drivers, most commonly done with the midrange driver, is an attempt to improve transient response and phase coherence, and minimize inter-driver phase transitions. The premise is minimizing the non-linear phase shift seen across the driver's spectrum, an artifact of both the crossover (inductors and capacitors shifting the phase 90 degrees, but in opposite directions) and the driver itself.

A reversed midrange, relative to the woofer and tweeter, selects and accepts compression/rarefaction distortion in the midrange. This choice notes that the woofer and tweeter each have a single overlapping region where distortion may arise from phase shifts (woofer-to-midrange and tweeter-to-midrange) while the midrange has two overlapping regions and thus more potential for a noticeable transition. Others invert the woofer and tweeter polarity with respect to the midrange so that the midrange, where humans and most animals (cf bats, dogs, and cats) best hear (I believe Klipsch wrote, "we live in the midrange"), is not polarity reversed and the other frequencies, where we comparatively poorly hear, better blend with the midrange.

Tweeters tend to have peculiar phase shifting behavior, so reversed polarity is may be restricted to the tweeter to make it appear to be "in phase" with the other drivers, although, by way of the foregoing, this can be seen to be a polarity error which is significant from the standpoint of physics and engineering. It isn't "in phase" at all, it is reversed to simulate a phase delay.

Modifying the crossover to resolve the issue is impossible. Using a higher-order crossover with greater phase shift would only further delay the tweeter in time relative to the midrange and woofer. This does not align the phase of the driver, because the time-varying signal is not uniform. The result is group delay, where certain frequencies arrive before others. This is why the higher-order crossovers (cf. Linkwitz-Riley) are often described as having inferior sound to first-order ones. I'm not opening that can of worms, merely noting the issue.

Acoustics and physics are messy.
 
Oh boy! To be honest it is clear to me that you are not an electrical engineer, but you like to tell
others "the facts" not going to go into a lot of detail with you.

What I was saying is that the tweeter is wired out of phase, you should get what I was saying.
What matters is the electro-acoustical response of the crossover cascaded with the acoustical
transfer function of the driver that is the response that must sum with the electro-acoustical
response of the other driver(s).
Generally, if a speaker sums well with reverse phase on the driver the electroacoustical response
is odd order and then, in fact, the opposite polarity also provides proper summing. However, it is
often difficult to hit a target exactly so one polarity works better than the other. If the crossover has
what we call "in phase" electro-acoustical summation then the system response will notch at the
crossover point when wired in reverse.
 
Let me clear up a few things.
Midranges are wired in reverse, either when they attempt to follow "old school" 2nd order
networks that notch when wired in phase, the reversal removes the notches. This is classical
crossover design. Edit: "old school" is pre Linkwitz-Riley where they did not have a solution
that provided a flat summed response. Classical 2nd order designs notched when wired in-phase
and had a 3 dB bump when wired out of phase. Most lived with the bumps or tweaked
them to do better.

Tweeters do not have "peculiar phase shifting behavior" rather the majority of dome and cone
types have a closed box 2nd order high pass electro-acoustic transfer function. If you read
Linkwitz, Leach, and many others you'd understand this.

It is clear that you do not understand crossover design.
 
Last edited:
Acoustics and physics are messy.

Actually, those and crossover design are messy when you don't understand them.

Read the AES there ARE higher order linear phase crossover designs but they are
complex and you have a lot of reading to do.
 
Last edited:
Polarity reversal of the midrange driver alters the midrange's sound because of the manner in which the ear processes sound. This isn't math, per se, it's biology. That's why acoustics is hard. It isn't pure math, it's how physics is perceived by the ear and brain. Compression and rarefaction are not identical. Actual instruments, as opposed to synthesized waveforms which are likely identical in both halves, have this artifact.
 
You are mixing linear and non-linear analysis, generally when discussing crossover design
we do most of the work in the linear world.
Yes, human hearing is non-linear, look up the wood effect.
This has little to do with why midranges are sometimes wired out of phase.
If you do research on the wood effect you'll learn that it is VERY difficult to hear the
difference especially above the low frequency range where excursions are greater,
enough to cause the asymmetrical non-linearity to be audible.

Electro-acoustics is well understood and involves basic engineering, how we perceive the
sound is another factor both should be considered but designing a crossover is college level math.
 
I will be showing these at Frankenfest this Saturday Oct. 20 in case anyone is interested.
 
Thanks Scottie,

I've been a member of the AES since about 1977 when I was a student member and
have tried to read every important paper on crossover design since then. Certainly
all the classic papers, and material from Linkwitz and probably a few others I've
forgotten about.
 
If anyone interested looks at the picture in the first post, the plate was made out of
wood. Since then, my nephew, a mechanical design expert in Solid Works, designed
the plate and produced an .stl file for 3D printing. I've also since gotten a 3D printer
and have printed a few of them. Everything fits perfectly.
 
Back
Top Bottom