AR-5 Frequency Response "Problem"

[Steve F]

This is a fascinating set of curves by Julian Hirsch at Electronic World magazine. Electronic World published parallel test reports by Hirsch to the ones published in Stereo review, but the EW reviews revealed more of JH’s technical results, including a speaker’s frequency response curve, which Stereo Review never published (as a matter of policy).

Presumably SR felt that a frequency response curve was too subject to misinterpretation by SR’s emotionally-charged hobbyist readership, while their sister publication of EW had a more mature ‘technical” readership, so the publication of detailed technical specs/graphs was deemed ‘safe.”

 

This is a deep dive, so stay with me here or else I might as well be writing a private letter to Tom Tyson.

 

When the AR-5 was introduced, it generated a bit of controversy because Julian’s review in Stereo Review made note of a deep dip in the 5’s frequency response at around 2000 Hz. He said it gave the speaker an unacceptably distant sound and only with the mid-range control advanced all the way up did Julian say the speaker sounded good. (He actually used the word “superb,” but only with the mid control advanced “to the reviewer’s liking.”) 

Roy Allison, the 5’s designer, took vociferous objection to Julian’s findings. He insisted to Julian that the 3a and 5 used essentially identical MF and HF drivers and their x-o topology was nearly identical, so it would actually be impossible for the 5 to have this trait and the 3a not to. He further insisted that in all of AR’s tests, measurements and listening, the only difference between the 5 and 3a occurred below about 45hz or so. Their response and sound was identical from the mid-bass on up. There was no mid dip on either speaker.

Julian didn’t buy it, saying that he measured the 5’s dip in two completely different set-ups and the 3a did not have the dip. Julian never actually gave a reason for the 5’s apparent problem, but he insisted it was real and different from the 3a.

Julian and Roy’s disagreement was aired in public, in succeeding issues of Stereo Review in Julian’s Technical Talk column. To Julian’s credit, he gave Roy ample space to have his say, and in all candor, Julian liked AR so much he probably wanted to hear a legitimate explanation that would take AR “off the hook.” Nonetheless, like any good scientist, Julian stuck by his findings, since he measured them according to his own legitimate methods.

The AR-5 was caught from a marketing standpoint between the prestigious, highly-reviewed top-of-the-line 3a and the amazing 90%-that-anyone-would-ever-want value of the 2ax, at half the 3a’s price. So the AR-5 was already struggling from a sales standpoint. This review—a tenuous, conditional recommendation, with a serious wart—didn’t help the AR-5 at all.

Now, to the matter at hand. This is a frequency response of the four main AR speakers in early 1970s—the 4x, 2ax, 5 and 3a—superimposed on each other and normalized to 1000Hz. These curves were made by Julian for EW magazine and published, although it’s not clear if these are the actual curves made for the test reports or whether these are additional curves made at some later date, just for interest’s sake.

Here’s the really interesting thing: The 5’s and 3a’s curves are within about 1dB of each other all the way from around 90hz on up! Within 1 dB! The AR-5 has no dramatic, egregious 2000Hz “dip” in the midrange. None.

And these curves were made by Julian. Explain that, Julian: The 5 and 3a are essentially identical above the mid-bass, exactly like AR says they are. Julian? Julian?

Here’s another interesting thing: the 2ax upper mid response is a few dB lower than the 5/3a’s, which is exactly how they sounded relative to each other. I A-B’d the 3a and 2ax in my home many times and the amount of increased inner-mid detail on the 3a vs. the 2ax was apparent, just like these curves show.

Also interesting is the 2ax’s apparently stronger extreme (above 10k) on-axis treble response: This is because the 2ax was really a 2-way speaker with a super-tweeter add-on, where the 3 1/2-inch driver (remember that was the tweeter in the AR-4 and ‘early’ 2x) goes full out, as high as it can go, and the 2ax’s ¾-in tweeter is brought in as HF reinforcement. In the 2ax, the 3 ½-in driver is essentially flat and level with the woofer all the way out to about 13-14kHz, so it is combining with the ¾-in tweeter for a pretty healthy on-axis treble output.

In the 3a and 5, they are true 3-ways, where their mids are rolled off as the tweeter is brought in. The 2ax has both drivers pumping out HF above 5kHz, whereas the 3a/5 have only the relatively weak ¾-in dome doing all the HF work above 5khz. (“Weak” as in limited ultimate output level, not “weak” from a quality/performance/dispersion standpoint.)

These curves show all of that quite clearly. And prove, once again, AR’s honesty, QC (the way the 3a and 5 mimic each other is pretty remarkable) and high performance.

Steve F.

[Pete B]

Notice that below 100 Hz the 5 has about 3 dB more output at 70Hz  than the 3a due to the higher

system Qtc which should have made it sound more punchy but less extended in the bass.

In fact if Julian used the low bass line as a reference, the the 5 _would_ have had less mid 

bass on a relative basis.  Good that he was able to compensate to his satisfaction by turning

up the mid level control.  I don't think that I would have liked the 5 since I prefer deep extension.

I show in this post that the 5 has peaking around (and more above) Fc, note the very high Qtc of 1.79:

http://www.classicspeakerpages.net/IP.Board/index.php?/topic/10170-ar-5-technical-details/&do=findComment&comment=120323

 

[tysontom]

Steve's message is interesting and clearly describes what happened during that testing of the AR-5 by Julian Hirsch!  We will never know exactly what caused the issue with Hirsch's AR-5 test, but once he had stated and published his findings, he could not possibly back down.  As Steve mentioned, to Hirsch's credit, he published the "back-and-forth"  correspondence he had with Roy Allison.  Once the stake was in the ground, it's hard to back off.  The anomaly in the AR-5's response is almost surely caused by some interaction in the measurement that could not be traced down; as Steve commented, the AR-3a and AR-5 shown in the subsequent Hirsch overlaid response curves show very close correspondence, so this raises questions as well.

Julian's test-measurement method was not done in anechoic space, yet he developed an in-room measurement technique that gave close correspondence to what he felt was heard from the speakers and quite similar to tests made by CBS Labs and Audio magazine and others.  Note that this was before the days of gated measurements, so the measurement microphones would detect all sorts of artifacts from various boundaries, diffraction effect, microphone positioning differences and phase anomalies.  But Hirsch was an extremely knowledgeable electrical engineer, and he knew as much about audio technology as anyone in the field, so he was able to improvise a valid test method.  Through years of experimentation, he was able to average the overall output from the microphones to get a clean and representative acoustic output into his room.   

Julian felt that neither anechoic nor reverberant measurement space was completely realistic as to how the speaker would sound, so his measurements were done in a "live" room with calibrated microphones: one capacitor microphone on the center axis close to the speaker and three microphones suspended from the ceiling at various positions that were determined by trial and error over years of testing.  The output from these four microphones were fed into a mic mixer and then to a GenRad preamp and an automatically traced GR graphic-recording system, commonly used in the day.

We'll never know exactly what caused the 2kHz dip Hirsch found in the AR-5 and whether or not it affected the sales of the AR-5.  Most likely it hampered sales somewhat, despite generally excellent reviews throughout the industry.    

—Tom  

[rl1856]

What if the issue was from the original curves JH created for the 3a ?    Looking at the curves at the top of this post, both the 3a and 5 have about a 3db decrease centered at 2khz.  What if his original measurement of the 3a did not have this dip, or if this dip was not as apparent ?   A perfectly functioning 5 would in comparison exhibit the dip he described....

"These curves were made by Julian for EW magazine and published, although it’s not clear if these are the actual curves made for the test reports or whether these are additional curves made at some later date, just for interest’s sake."    Different curves, recorded at different times *could* account for the differences cited.  Note that JH said he measured 2 sets of "5", and found an identical dip.  He did not specify having retested a 3a to verify previous measurements.    Could be key.

 

in what issue of EW did this graph appear ?    

Thanks !

[Analogman]

What's the end game debating the merits of reviews for speakers, now in some instances, pushing 60 years old?

I personally have owned "2-ax"s "5"s "3"s and "3a"s

The "5" with nothing more than fresh woofer edges was a better sounding, less dated sounding all 'rounder than any incarnation of the "3" that I have owned  - that was my experience

With a fresh set of by-passed Daytons they were holographic (I don't deal in BS hyperbole)  they threw out a sound field that went well beyond the boundaries of the speakers, I mean WAAAY out, almost like a really well implemented  Hughes AK 100 type circuit - few visitors believed they were just being powered with a modest ("vintage") old Marantz and nothing more

And I have absolutely NO idea how I can, could or would want to qualify, quantify, justify or explain my real life experience so it fits any of the narratives and rubberchickenmographs of these dusty old reviews - THEN or NOW

All I know is what happened to me and how today I am still kicking myself in the ass for selling those "5"s for, you guessed it, money to apply to a clean pair of "3a"s that couldn't hold a candle to those "5"s - why?  I do not know

The "3" is a cult of personality - I think a lot of people are as in love with the old stories and ad copy as they are with the lab results done on NEW examples 40 50 or even 60 years ago and which is absolutely meaningless today thanks to time 

For any make, not just AR, any of 'em, JBL, Kloss stuff Advent, Altec etc

Just enjoy your speakers and what's left of what they once were

Craig 

 

[owlsplace]

18 hours ago, Analogman said:

What's the end game debating the merits of reviews for speakers, now in some instances, pushing 60 years old?

I personally have owned "2-ax"s "5"s "3"s and "3a"s

The "5" with nothing more than fresh woofer edges was a better sounding, less dated sounding all 'rounder than any incarnation of the "3" that I have owned  - that was my experience

With a fresh set of by-passed Daytons they were holographic (I don't deal in BS hyperbole)  they threw out a sound field that went well beyond the boundaries of the speakers, I mean WAAAY out, almost like a really well implemented  Hughes AK 100 type circuit - few visitors believed they were just being powered with a modest ("vintage") old Marantz and nothing more

And I have absolutely NO idea how I can, could or would want to qualify, quantify, justify or explain my real life experience so it fits any of the narratives and rubberchickenmographs of these dusty old reviews - THEN or NOW

All I know is what happened to me and how today I am still kicking myself in the ass for selling those "5" for, you guess it, money to apply to a clean pair of "3a"s that couldn't hold a candle to those "5"s - why?  I do not know

The "3" is a cult of personality - I think a lot of people are as in love with the old stories and ad copy as they are with the lab results done on NEW examples 40 50 or even 60 years ago and which is absolutely meaningless today thanks to time 

For any make, not just AR, any of 'em, JBL, Kloss stuff Advent, Altec etc

Just enjoy your speakers and what's left of what they once were

Craig 

 

Good synopsis, I'll have to get my 5's out of storage ... currently listening to AR-2's from Heathkit and quite impressed but then my hearing is not what it used to be :). The 3a's here have Teledyne era drivers and reconed woofers which puts them in the AR-58 class -- an entirely different sound.

Roger

[ligs]

19 hours ago, rl1856 said:

.

"These curves were made by Julian for EW magazine and published, although it’s not clear if these are the actual curves made for the test reports or whether these are additional curves made at some later date, just for interest’s sake."    Different curves, recorded at different times *could* account for the differences cited.  Note that JH said he measured 2 sets of "5", and found an identical dip.  He did not specify having retested a 3a to verify previous measurements.    Could be key.

 

in what issue of EW did this graph appear ?    

Thanks !

It was published in September of 1970. I put that graph in a different thread. 

 

Here is the entire page.

[owlsplace]

Engaging discussion and I jumped in without reading all the posts. I seem to remember JH having written two reviews of the AR-5? In one he stated that with the AR-5 model Acoustic Research had finally "gotten it right." I believe the context had something to do with an apparent fuzziness the 3a exhibited on operatic vocals.

I did buy the 5's back in 1974. They always did seem harder to drive as in had to turn the volume control up higher. Before my recent move I aired out a pair of 5's using just an AR-SRC which is basically a passive pre- I believe and an Adcom 555 II. I played the live recording of Simon & Garfunkel's Concert in the Park and pretty much felt like I was in attendance. It is all pretty subjective for me at this point in time. Music was and is my drug of choice and AR's delivery systems were always pleased my ears.

Roger

[rl1856]

I found the Sept 70 issue of Electronics World.   Thank you.

The AR3a was reviewed in by JH in the Jan 69 issue of the same publication.

I have attached the relevant graph.  Notice anything different between this curve, and the AR3a curve included in the graph at the top of this thread ?

[Phxjohn]

I can see where Consumer Reports got their comment about the 3a having a thickening in the midrange. 

[Analogman]

On 6/14/2018 at 8:41 AM, rl1856 said:

I found the Sept 70 issue of Electronics World.   Thank you.

The AR3a was reviewed in by JH in the Jan 69 issue of the same publication.

I have attached the relevant graph.  Notice anything different between this curve, and the AR3a curve included in the graph at the top of this thread ?

Yes, like they were measuring two different boxes (systems) altogether