An AR-3a response curve

[Rich W]

Given all the debate regarding reverberant field response of AR speakers, I thought it would be good to share my own test results with the forum. My post is intended to give other forum members, who may not have access to test equipment, an idea of what they can expect from a pair of restored AR-3a's in a real-world listening situation, albeit with replacement AB-Tech tweeters. I'll be happy to share any particulars regarding my restoration or how I performed the test. I have no intention, nor do I have the time or inclination to respond to what I feel is anti-AR bias that's been expressed on this forum as of late (don't bait me - I will not respond). Suffice to say, I have a musical background, am a recording engineer by profession specializing in recording live acoustic music in concert hall settings - I know what live acoustic music sounds like, and the what type of home reproduction comes closest to replicating that live event.

The Speakers

- Speakers are a pair of AR-3a's, manufactured in November of 1969. Restoration, including refoam, new capacitors, etc. largely follows what was published in the "Restoring the AR-3a Guide" available on this forum. Original tweeters suffered from some response irregularities as reported on by Roy C, so I replaced them with AB-Tech tweeters and a .07 mH shunt inductor. Response of these tweeters is very similar to the later AR-11 tweeter, as outlined in Ken Kantor's document on the AB-Tech tweeter as a viable alternative.

The Room:

- Room is 15' wide and about 45' long (a family room/kitchen combination). Ceilings are typical 8' height.

- Speakers are situated on the far end on the family room side. Room is well-damped with two sofas and an area rug.

- Speakers are close to the side walls - woofer centers are 45" (average) from the back wall, 30" from the floor and 17" from the side walls. The speakers are thus separated by about 12' or so. This placement is close to the theoretical ideal to avoid mid-bass room boundary dips, the so-called Allison effect.

- Speakers are toed in approximately 20 degrees.

- Primary listening area (from where the test was done) is approximately 17' in front of the speakers. This is where I get the least interference from room nodes.

The Test

- Software is TrueRTA used on a laptop with an Echo Indigo soundcard. A VLZ Mackie preamp was used.

- An AKG C414TLII microphone with an omnidirectional pickup pattern was used. Microphone was calibrated using a response curve provided by AKG when I purchased the mic charted specifically for that microphone.

- Average of 15 mic placements, all within a 6 square foot prime listening area.

- Tweeters and Mids were deliberately set so the speakers' response fell 3 db per octave above 2K. This agreed with my listening tests. For classical music, the speakers sounded far to hot without that rolloff.

- Pink noise feed through both speakers at once (individual curves for each speaker were quite similar).

- The anomaly at 70 hz is a room node artifact, borne out by tests I've run with other speakers in a similarly placed.

- The test was executed at 1/24th octave resolution.

I think the result of this test speaks for itself. I'm finding it hard to get motivated to restore a pair of Allison One's I got my hands on. That's how good these AR-3a's sound.

Regards to All,

Rich W

[Zilch]

Please upload using the standard forum attachment protocol; I cannot access your curve....

Edit: O.K., I see it now, and it looks like AR3a's intended design target "concert hall" presentation. Thank you for posting your measurements. I believe there is consensus afoot here.

I don't understand why members regard this as bashing ARs; they are what they are; there is value in knowing what that is and why it is liked by many. It is also important from the perspective of not presuming them to be something else, or expecting them to behave otherwise....

[Carlspeak]

Thanks Rich for sharing.

What's the difference between the 3 posted curves? They all look pretty much the same.

[Rich W]

Thanks Rich for sharing.

What's the difference between the 3 posted curves? They all look pretty much the same.

Hi Carl,

They are in fact all the same curves. Now corrected.

Rch

[genek]

- Tweeters and Mids were deliberately set so the speakers' response fell 3 db per octave above 2K. This agreed with my listening tests. For classical music, the speakers sounded far too hot without that rolloff.

Rich, what were the corresponding knob positions for these level control settings (in relation to the white dots and/or Increase/Decrease markings)?

[Steve F]

Your room curve looks just like what I would expect a good set of 3a's correctly positioned in a room to look like. As much as we can infer listening quality from such curves, it 'looks' like these sound like the 3a's so many of us admire so much.

Add to that curve the intangible quality of 'tight' low-distortion acoustic suspension bass that extends to a meaningful low frequency, and it's no mystery that 41 1/2 years after its Dec 1967 introduction, thousands of people are still discussing, arguing over, and just plain enjoying these remarkable speakers.

Very few products in any industry achieve legendary icon status by way of its abbreviated model number. The "3a" has achieved that status, and love it or hate it, it remains a cornerstone of audio history.

Steve F.

[Zilch]

Thank you, Howard for posting these high-res scans of the AR3a Tech Data publication.

There is an unresolved discrepancy between the full system on-axis frequency response shown therein versus that measured at the Harvard Acoustics Laboratory and presented to the AES by Allison and Berkovitz in 1970 which you provided earlier. It appears as though both measurements were taken with the level controls set to "Max:"

[speaker dave]

Hi Rich,

Good to see some new measurements on the AR3a. Seems like a very smooth room curve and nicely flat through the mids.

I am amazed by the bass extension and smoothness. Is that a "magic" spot in the room that you found after much trial and error, or just a lucky happinstance? Flat to 25 Hz is pretty good when most of the anechoic curves show the woofer rolling off from about 45. (The room can extend this some with luck with the standing waves.)

Regards,

David

[Rich W]

Rich, what were the corresponding knob positions for these level control settings (in relation to the white dots and/or Increase/Decrease markings)?

Hi Genek,

The original "white dot" settings don't have much meaning for my restored pair of 3a's. I was a bit sloppy reinstalling the pots after I cleaned them, and did not take care to align the slot on the pot with the indentation on the masonite - so they may or may not have been positioned properly (I claim ignorance as my defense!).

To get a good mental image of their settings, maximum on both MID pots is between 10 and 11 PM (seen as a clock face). Mine are set to 6 PM (which actually turns out to be on the white dot, as I reset the plastic knobs to align to my own 3a's "normal"). Tweeters are set 1/4 of a turn down from maximum.

Keep in mind that nonwithstanding the lack of true "white dot" alignment, there are other factors that make my particular settings unique. I recapped with Solens and added some additional series resistance in a theoretical effort to match the old caps ESR. That would be .33 ohm for the tweeter and .27 ohm for the mids (thanks John O and Roy C). In addition, the new AB Tech tweeters are much more sensitive that the originals, which had to be run with the pots at max to get any sound out of them at all.

Best Regards,

Rich W

[Rich W]

Hi Rich,

Good to see some new measurements on the AR3a. Seems like a very smooth room curve and nicely flat through the mids.

I am amazed by the bass extension and smoothness. Is that a "magic" spot in the room that you found after much trial and error, or just a lucky happinstance? Flat to 25 Hz is pretty good when most of the anechoic curves show the woofer rolling off from about 45. (The room can extend this some with luck with the standing waves.)

Regards,

David

Hi David,

Over the years, I more or less gravitated to that general area for critical listening, just because it did sound better than other listening perches. Once I started using the RTA, I was able to refine the position further. Sadly, this sweet spot does not coincide with where my living room couchs are. In fact, it falls right at my kitchen table (makes meals and writing out the bills great though).

I've read that to avoid setting off room nodes, the best spot for listener to be is 35% of the distance from the front to the back of the room, assuming the speakers are situtated at the front of the room. Since my room is about 45' long, and I'm about 17' back, that makes sense. I noticed that when I get closer in, I start experiencing a more roughness in the mid- bass. Going further back augments the extreme low-end even more, but overall still sounds very good (the sound stage is superb anywhere further back).

The other important factor is avoiding boundary cancellation by ensuring that the speakers are not equal distances from the adjacent floor, side wall and back wall. Roy Allison did a lot of research on this, and there's a document of a 1976 Stereo Review article in this website's Allison archives which was my holy grail for speaker placement since I first read it.

I've been using a 1/3 octave equalizer to even the response further. Unfortunately, it does little to correct my 70 hz dip (adjacent bands are at 63 and 80 hz). For that I'll need a parametric with a really narrow Q setting. I will say that for overall naturalness of sound, I'll take my 70 hz dip over anomalies in the 100 to 400 hz range anytime.

Before I got the 3a's, I was using a pair of restored AR-7's in the same location. The overall sound and bass extension were superb. I know the room was playing its part to help, but can you imagine a pair of AR-7's reproducing a 32 hz organ note that could be felt as well as heard?

Best Regards,

Rich W

[Steve F]

Re: The discrepancy between the curves—

In the Allison-Berkovitz curve, the 3a is in a 4 Pi environment and Allison himself says that the most interesting thing about that curve is how the woofer’s output decreases as it transitions from 2 Pi to 4 Pi with decreasing frequency.

In the Technical Data brochure, the on-axis curve is a composite, apparently spliced together from an outdoor measurement below 200 Hz and an anechoic one above 200 Hz, so this 2 Pi-4 Pi effect isn’t seen in that curve. (As Speaker Dave would say, “Marketing drew that curve!”)

I’d think that the Technical Data curve is similar to what one might expect as “first arrival on-axis sound” in a good 2 Pi placement situation: In a bookshelf unit, with books right alongside the speaker, the speaker’s front face flush with the forward edge of the bookcase. This was not an unusual placement scheme in the 1960’s-‘70’s, when ‘bookshelf’ stereo speakers were at the height of their popularity. Several photos in AR’s brochures show just such placement.

AR also sold their popular “X” floorstands, and if 3a’s were mounted on those well away from room boundaries, then the 1970 Paper’s curves might be expected.

The “discrepancy” between the curves is probably due more to the different measurement conditions (which would affect any conventional forward-facing box speaker to more or less the same degree) than any sinister/nefarious intent on AR’s part.

I’m sure Speaker Dave or Ken could expound more on the 2 Pi-4 Pi effect on woofer radiation in an actual listening room if need be.

Steve F.

[Zilch]

The “discrepancy” between the curves is probably due more to the different measurement conditions (which would affect any conventional forward-facing box speaker to more or less the same degree) than any sinister/nefarious intent on AR’s part.

I'm looking more at the higher frequencies, and the 2-Pi measurement without the grille molding in place appears to be the best match, if also smoothed, Fig. 4 in Allison and Berkovitz (1970).

2-Pi doesn't mean backed up to the wall, rather, embedded flush within it, according to the description, and I'd say that few, if any, owners used them that way; I don't believe I see the bookcase scenario you suggest might have been common at the time represented in any of the "typical" listening rooms shown in the paper.

I don't mean to imply any deceptive intent, but clearly, AR was showing AR3a in the "best" light it could muster. If our desire is to know how these speakers actually perform, I'm going with what Rich W and others have recently presented here....

[genek]

2-Pi doesn't mean backed up to the wall, rather, embedded flush within it, according to the description, and I'd say that few, if any, owners used them that way

I've had my 2ax's (same tweeters as 3a but 8 ohm instead of 4) "embedded" into wall-sized banks of "entertainment center" cabinets for the past 30 years. We moved a couple of years ago, and while we waited for the cabinets and the rest of our big furniture to catch up with us, the speakers spent several months on a waist-high credenza "backed up to the wall." You could probably have measured a difference between that arrangement and what we have now that the speakers are back "in the wall," but I couldn't hear any.

[Shacky]

Wow! That bass extension to 30 Hz is impressive. I love the bass from my 3a's as well as the midrange. Very smooth and non-fatiguing.

Thanks so much for posting that curve. I hope the pair I dropped off with Peter fairs as well. Mine have original tweeters but they seem to be playing very well for 40+ year old tweeters. I didn't add the slight resistors to caps based on feedback I received. I used Solens on the tweeters and Dayton on the mid and woofer. I replaced my original pots with Memcor 15 OHM units.

[Steve F]

I'm looking more at the higher frequencies, and the 2-Pi measurement without the grille molding in place appears to be the best match, if also smoothed, Fig. 4 in Allison and Berkovitz (1970).

2-Pi doesn't mean backed up to the wall, rather, embedded flush within it, according to the description, and I'd say that few, if any, owners used them that way; I don't believe I see the bookcase scenario you suggest might have been common at the time represented in any of the "typical" listening rooms shown in the paper.

I don't mean to imply any deceptive intent, but clearly, AR was showing AR3a in the "best" light it could muster. If our desire is to know how these speakers actually perform, I'm going with what Rich W and others have recently presented here....

I'm aware of 2 Pi vs. 4 Pi locations and measurements. I was simply pointing out that any conventional box-type speaker of roughly 2 cu. ft. external volume with forward-facing drivers would show similar differences between in-room stand-mounting and flush in-cabinet mounting. The 3a on stands in a room vs. a 3a under more controlled measurement conditions will show these types of discrepancies, as will an Advent, a JBL L-100, a Dyna A-25, etc. The sloping woofer response characteristic as it goes from 2 Pi to 4 Pi will be evident in the relative comparative responses of all box-type loudspeakers. That's all I was saying.

As far as "best light," I'd like to see ANY manufacturer's literature in ANY industry that doesn't present their product in the best light. Cars, cameras, software, speakers, skis, restaurants---what do expect them to say: "Our stuff sounds lousy, the food will make you sick, and you'll drive into a ditch on the way home"?

Good lit is reasonably informative and believable. Bad lit is an outright lie. We can tell the difference pretty easily.

Steve F.

[Rich W]

Yes, the system curve is designed to show how the speaker will input a direct output under ideal conditions. Later on (after he left AR and started his own company) Allison designed speakers that were configured in terms of cabinet design and woofer locations to both handle the 2 Pi to 4 Pi shift (as the front cabinet baffle became smaller than the wavelengths) and signals reflected back from room boundaries just in time to generate a null in the mid-bass output. While he remains proud of the remarkable midrange and tweeter drivers he designed after he left AR, he is probably most proud of his research and design work in the realm of mid-bass smoothing. In nearly all cases, if one locates a pair of conventional-cabinet speakers the way the manufacturer suggests (and some do not even go far enough to suggest much of anything) it would be a good idea to deal with the mid-bass anomalies with a good equalizer and analyzer. With the Allison product line, placement according to manufacturer suggestions usually did the trick.

Roy Allison's paper "The Speaker and the Listener" should be required reading for anyone using a conventional bookshelf speaker. It's here on the forum: http://www.classicspeakerpages.net/library...icles_by_roy_f/, pages 14 though 19. I think many listeners will be surprised how much better their ARs will sound once any boundary effect issues are dealt with.

Rich W

[Zilch]

Roy Allison's paper "The Speaker and the Listener" should be required reading for anyone using a conventional bookshelf speaker.

Figure 1 is a power-output vs. frequency curve (measured in an anechoic chamber; a near-field response measurement would yield the same measurement) for a hypothetical "standard" speaker.

Oh? That's a 2-Pi anechoic chamber, then, right?

Are we looking at power or SPL? He's saying, "Power."