Audio Research SP9 preamplifier Page 2

Sound Quality
On first listen, using high-level sources only, I am saddened to report that the sound was not good. In fact, it was hard, rough, thin, and both spatially and dynamically compressed. The phono preamp added more of the same of everything. The adjectives I used above—and I gave considerable thought to their choice—do not reflect merely personal value judgments on my part, but are descriptive of the changes wrought by the SP9 on signals passing through it. That is, they were audible on simple bypass tests, in which the preamp under test is compared with the signal being fed to it.

My system, straight-through from high-level source to speakers, is, if anything, a bit on the warm side, so my complaints about the SP9 are actually giving the preamp some benefit of what doubt there may be. With a more neutral system than mine, the SP9's irritations would be exacerbated, not mitigated. In addition, confirming what David Prakel had said in his review of the SP9 in the September 1987 issue of Hi-Fi Answers, crosstalk between line-level inputs was higher than I would have liked. If a tuner is operating while a disc is being played, its signal will be faintly heard in the background.

In their literature, ARC warns that the SP9 needs to see quite a high amplifier load impedance: 60k ohms in parallel with 100pF is recommended, the capacitance being very low, equivalent to no more than 2 meters or so of a typical audiophile interconnect. The worst-case loading is 20k in parallel with 1000pF, this still not unusual with many solid-state power amplifiers connected via reasonably long interconnects. Worried that I might have been mistreating the preamp's output stages, I checked my cabling: the 1m length of Monster M1000 had a total capacitance of 175pF, while the 75k input impedance of the Threshold SA-1 should not have presented the SP9 with any problems.

I was then told that according to Ken Kessler, who was reviewing the SP9 for HFN/RR, the SP9 needed "at least 72 hours of warmup before its sound reached its best." JA took our sample SP9 home and tried that (see below), then suggested I go forth and do likewise.

After four days of continuous warmup (96 hours), I was prepared to modify my initial response to the SP9's sound, but only to the extent of putting the word "somewhat" ahead of each of the adjectives I had used previously. I still found the preamp to be one of the least ingratiating I had heard in years. But why, I wondered, did it sound so completely different from the SP11, in which ARC's "patented hybrid technology" had worked such sonic wonders?

The answer may lie in the fact that the SP9 combines its tubes and FETs in a cascade, rather than cascode, configuration. The difference is that, whereas the latter allows both devices to act as a single stage, the former normally makes them behave like two separate series-connected stages. Thus, instead of canceling their inherent distortion characteristics, the two sets of distortions would be added. That still wouldn't explain why the '9 sounded mediocre; ARC used cascading in its all-tube preamps, and they sounded warm and sweet rather than thin and rough. But it might explain the differences between ARC's two hybrid designs. (If, in fact, there is some other provision in the SP9's circuitry that would make it behave differently from the way I think it does, I would be interested to learn about it.)

The SP9—my sample, anyway—is not what I would call a bad preamp. Its problems are really only moderate in degree. But most are errors of commission rather than omission, and are of a kind I personally find very hard to tolerate. But what's a better preamp, for the price? Or even as good? Well, $1700 is a popular price range for preamps, and solid-state design and available materials (like active devices) have reached a point where some transistor designs are comparable in performance to tubed units costing several times as much. Unfortunately, I haven't heard any of the current solid-state models in this price range. But I am very familiar with a $1500 tube preamp from a competing firm: the Conrad-Johnson PV-5.

A three-year-old design (although probably improved since the one we have on hand was loaned by the manufacturer), this has long been almost a cult favorite with audiophiles who dote on the classic "tube sound." To refresh my memory of it, I brought home our PV-5 from the office and gave it three days to warm up before listening to it.

The sound was much as I had remembered, but the increasing accuracy of other, more recent, preamplifier designs has rendered its previously "slight" colorations more conspicuous by comparison. Through any high-level input, its sound is warm and rich, with a soft but subtly grainy high end, a slightly forward upper midrange, an almost fat and somewhat ill-defined midbass region, and a marked lack of really deep low end. Its soundstage is wide, its rendition of depth so marked as to seem almost exaggerated. Phono-stage performance is relatively free from coloration, adding only a trace more midbass bloom and a slight high-end roughness to what was heard through the high-level section, but, of course, it will not handle MC cartridges with insufficiently low noise.

In other words, the PV-5 is a far cry from the idealized straight wire with gain. It's not a very accurate preamp—probably not as accurate as the SP9—but the nature of its imperfections is completely different. Whereas I had been actively irritated by the SP9, I had the opposite reaction to the PV-5. Yes, the PV-5 was drying up highs and losing deep bottom and making everything sound fat and rich, but these were faults I could have lived with even while acknowledging their existence.

Given the choice between these, there is no question which I would buy. A more intriguing question, though, is why my sample SP9 was apparently so different from the samples reviewed by Peter Moncrieff, and by David Prakel and Ken Kessler in the UK. I would like to think that the Stereophile sample was, in fact, not typical of average off-the-line production, although I find it hard to believe a firm as conscientious as Audio Research would allow anything sounding like this to leave the factory, let alone go to a high-end magazine for testing. The fact that they did, though, is reason enough for us to publish this review: to alert Audio Research SP9 buyers that at least some samples in stores may not be up to ARC's usual quality standards.—J. Gordon Holt