Infinity IRS Beta loudspeaker J. Gordon Holt, January 1989

J. Gordon Holt returned to the IRS Beta in January 1989 (Vol.12 No.1):

Rave or not, my report in Vol.11 No.9 (September 1988) on this Infinity flagship system ended on a note of uncertainty, concerning an audible difference between the sound of the two mid/high-frequency–range panels which was messing up the imaging and exacerbating program grundge in one channel.

When I phoned Infinity president Arnie Nudell and reported that my upper-range panels weren't matched, he didn't believe me. "Why," he asked, "didn't any of us notice it during our visit?" I explained it was probably because no one was listening for it, and besides, the brighter, more sizzly side had been at the right, out of reach of such things as massed violins, which were most affected. At the right, it only added additional guttiness to cellos and basses. So, on the (reasonable) assumption that one of my two upper-range panels was out of whack, Infinity sent a single replacement.

The idea was to compare the new one with the two originals, and toss out (as defective) the one that didn't match. But it wasn't that simple. (Things rarely are.) Instead of two different sounds, I now had three. The new panel sounded almost exactly halfway between the other two. Of those, I put aside the most sizzly-sounding one, and phoned Arnie again. This time, he sent John Miller to Santa Fe again, with two large boxes of test equipment.

First, we listened. Yes, John could clearly hear the difference I was talking about; no, it wasn't "normal," and no, he didn't know yet what was causing it. After a full day of measuring, during which I went off and tended to other matters, John felt he had the problem pinned down. He told me he had found a very small inaccuracy in the value of a crossover capacitor, resulting in a 1dB reduction in the level of the rear-firing tweeter, which spans the 5–12kHz range. He corrected it, and we listened to the result.

Now the speakers sounded virtually identical, and I expressed disbelief that such a small change could have had such a noticeable effect on the sound. But...the panel he claimed to have found the problem in was the one whose sound I had liked the most. Now I did not care for the system's sound at all.

After John left, I continued to work over the speakers, adjusting driver levels, changing room placements, trying other electronics, all the while becoming increasingly convinced that something was drastically wrong. I could not get them to sound nearly as good as they had originally. There was now a persistent coloration—best described as a steely sizzle—in both panels, which made any loud orchestral music sound so relentlessly strident as to set my teeth on edge. (Of course, that had to be the week of Stereophile's annual reviewer convention, and everyone wanted to hear my system. Only a few of them were polite enough—or embarrassed enough—not to tell me they thought it sounded dreadful!)

Meanwhile, a couple of other things developed that put the Betas in a less than favorable light. First one, then several, then all of the loudspeaker terminals came loose. They didn't actually fall off, but they became so wobbly that I started to wonder when their electrical connections would start to become intermittent. (None has, yet.) Then the crossover module's turnover control knob started to slip on its shaft and, with continued use, finally came completely off in my hand. The reason for this then became obvious: The knob had only one set screw to lock it to the shaft, the screw was very small, and the shaft had no flatted side for the screw to seat itself against. (Worse, the set screw is recessed behind the front panel when the knob is in place; it would have been necessary to dismantle the whole case to replace the knob. I just used pliers for future adjustments.)

Granted, these are minor mechanical problems which seem to have no effect on the sound, but to my way of mind they are inexcusable in a $10,000 product. It's not as if we're dealing with frontiers of technology here; control knobs and 5-way binding posts have been around for longer than I have! I have a cheap Sears-Roebuck radio that has been in use for more than 25 years, and nothing has ever fallen off that.

I placed another call to Arnie, and learned that John hadn't just made a small crossover-part "correction," but had also replaced the EMIM and EMIT drivers on both panels, "just to be safe." Arnie declared he was sending another matched set for me to try.

The drivers were a cinch to replace, but did they solve the problem? Well, yes, no, and maybe. The speakers now sounded quite a bit more pleasant than they had, but they still lacked the gorgeous richness and ease that had attracted me so much to the sound of the original panels. (I should say, "to one of the original panels," because it was what that panel was doing to left-channel sounds that made the system so appealing.)

It seemed to me that most of the sizzle still remaining was coming from the EMITs, so, just on a hunch, I swapped out the latest pair for the previous pair. Sure enough, the problem was slightly worse.

Accordingly, we arranged for Infinity to send us yet another pair of upper-range panels. These, which I am assured are "right out of stock," are the best-sounding of any I have heard to date. The steeliness which afflicted some of the previous samples is completely absent, and nothing else of value has been lost. The system now sounds just as magnificently rich and powerful as did the first samples I reviewed, but with far better imaging than that first pair. However, the very fact that the latest pair are different from the previous pair, even if only slightly, has not helped to dispel the impression that there is some sort of quality control problem here.

The problem sounds very much like a simple frequency-response aberration, which should be easily measurable. In fact, we tried early on to find a frequency-response anomaly that would account for the perceived brightness difference between the two panels which had sounded the most dissimilar, and failed. The probe mike was not moved between comparative tests, the speaker locations were identical to within as small a fraction of an inch as we could get them, and the two people in the room during the tests were as far as possible from the soundfield and in identical poses for each response run. No consistent differences were measurable, and the inconsistent differences measured (on the order of ±1–2dB) were of insufficient amplitude to account for the audible differences. Indeed, it took almost 3dB of EQ (downward, at 5kHz) on the Accuphase G-18 equalizer to make the two panels sound fairly similar, but they then measured almost 3dB different at that frequency.

The problem for a manufacturer, of course, is that it is impractical to do QC by ear. The challenge is to find an objective substitute.—J. Gordon Holt

Footnote 1: Don't laugh. No other recording medium comes close to digital's 0Hz low-frequency limit.—J. Gordon Holt