VAC - any comments? Good or bad.

Larry. I threw out ALL my files save for those pertinent to gear I acually own, BUT I did keep some of the sets of measurements pertinent here, buried in a file of N803 vs Fidelio measurements, as it was the Fidelios that I used, not the Parsifal Encores, which weren't purchased until a month later, as I see my graphs are dated April 22, 2000....
I don't know how to plot curves in this space, nor how how to post and refer, so I'll just list calculated DIFFERENCE data between two pairs of test runs. The first string lists the differences in dB when going from the Audio Refinement Complete Integrated to the AVATAR, adjusted for equal pinks, rounded to nearest easy fraction (when read visually), and then to nearest typable decimal for your reading....The second string of data is similar, except showing a second test run REPEATING the AVATAR again, but with an NAD 7400 amp in comparison. The index is frequency.
If you plot the two curves, you will see quite remarkable agreement....If you then take the "difference of the differences" data to evaluate method imprecision among other
info note the otherwise randomness and smallness of the noise (indicating the relatively high confidence level of
my conclusions), and perhaps the wildness shown in the top octave. However, looking back at my raw data, I see that the original SPL levels of ALL data entry pairs for both test runs for the ARC and the NAD amps were within +/-1dB across the whole band! So it's clear that the variability is due to some inherent behavior of the Avatar....
Once you plot the curves (a picture's worth a thousand words here, guys 'n gals, so PLEASE...), you'll notice an interesting old "west coast" u-shaped curve, with a pronounced, wide dip starting at 100Hz, bottoming at 250Hz, and then starting a long climb back to reference at about 2kHz. The bigger problem is that this climb continues into a sharp spike at 4-5kHz, then tumbling fiercely into the ravine -15dB down at 12k! That it wildly rebounds past unity at 20kHz is also of interest....That this behavior was repeated in a separate test when compared to another amplifier later is of course important. (If the acoustic results were simply a u-shaped curve then of course one could just increase the gain a bit to bring the mids up, and thus enjoy a plumper bottom and bright top, referring to the amp as one with a warm bottom, recessed mids and a bright top. But the trampolining between 1k and up belies categorization nor acceptablility.)
I also ran several additional sweeps more casually, noting that other than a gain difference of 1.16dB, the triode and ultralinear modes acted similarly. It should also be noted that I did NOT plot response below 40Hz, as it was not of interest, and the Fidelios were set up too far out into the room to have sufficient boundary support to have adequate output below 50Hz........................................................Freq:ARC to VAC diff/NAD to VAC diff: 50Hz:-1dB/-1dB 63:-1/0 80:-0.67/0 100:-1.67/-1.33 125:-2/-2.33 160:-2.33/-2.67 200:-2/-2.33 250:-2.67/-3.33 315:-1.33/-2.5 400:-2.25/-1.67 500:-1.5/-2.5 630:-2/-2 800:-1.25/-1.5 1k:-1/-1.25 1.25k:-0.5/-0.5 1.6k:0/-0.25 2k:+1/0 2.5k:+1.25/+0.5 3.15k:+1.5/+1.25 4k:+2.67/3 5k:+3.25/2.33 6.3k:+1.25/+1 8k:-2.67/-4.67 10k:-7.67/-10 12.5k:-10.33/-12 16k:-5.33/-3 20k: +6.33/+2.67 (Phew!)
Difference data: test 1 minus test 2:
0 -1 -0.67 -0.33 +0.33 +0.33 +0.33 +0.67 +1.17 +0.6 +1 0 +0.25 +0.25 0 +0.25 1 +0.75 +0.25 -0.33 +0.9 +0.25 +2 +2.67 +1.67 -2.67 +3.67!
Further, the very high Q of the data above 6kHz suggests that a finer source of tests signals (1/6 or 1/10 octave warbles) would have been most beneficial to assess this apparent unruly behavior..............................................................................................................
So what do you think's going on guys? VAC had no additional response after I sent them the actual data and curves three years ago. The responses below 3kHz can almost be lived with in the old pulled-back-mids style, but I've never seen anybody's "simulated performance with a real speaker load" curve look this bad. And yes, BOTH channels driven simultaneously, so I doubt that it wasn't one side oscillating wildly or anything obvious. All tubes glowed equally in pairs, etc. And again, triode = ultralinear except for 1+dB gain. So I'm stuck here, hearing only wonderful things about VAC and its pricipals, but having only experienced the behavior noted above with the one product I spent much time with. Again, they redesigned this Avatar, right?

I think everyone who reads this can draw their own conclusions as to what was actually measured here.
The amplifier's real output is not measurable through a non calibrated loudspeaker, with further room calibrations made, to compensate for room, frequency anomolies.
If this measurement of the "amp" were taken with a million speakers, we would see a million different results.
I thought, when I read this that the, so called, 'amplifier measurements' were suspect, now I, and everyone can see what has happened. Poor Kevin, he must be pulling his hair out.
Good listening,
Larry

The ARC and NAD were shown to have only +/- 1dB across the whole band? PUULEASE.....this whole set of test measurements is just plain FLAWED. I say you stick to just listening to your system. ; )

Well, it's clear that Lrsky and Balekan don't understand the test. Once again, a reference test was devised using the same software (CD of 1/3 octave warble tones), CDP, cables, speaker/positioning/room, microphone/meter/battery/tripod/positioning/decade-selection, and operator eyes, with the ONLY VARIABLE being the choice of amplifier. Three ss amps and one VAC AVATAR were tested. The three ss amps agreed with each other so closely in ALL results that it is easily said that they grouped as a "standard" curve. The VAC AVATAR's responses were extraordinarily different. I posted the results of two of the test runs in a previous post above, noting only the DIFFERENCE DATA from the ss amp to the AVATAR in each of the two test runs.
LRSKY, Everyone CANNOT "draw there own conclusions as to actually what was measured here"! What were measured were in-room responses of total systems, identical except for the substitutions of amplifiers. The difference-data was then plotted as a curve. It should appear as a straight line of zero slope, as it does between any two of the ss amps. When any of the ss amps is compared with the AVATAR the difference surve is as provided above. What the original data curves are doesn't matter, and of course WOULD differ dramatically would a change in ANY item in the reference system chain...especially a transducer (speaker or the test microphone) and/or its positioning. That's why I took GREAT pains to assure matched tested conditions for each pair of sessions. (Even test meter battery strength was assured). So now you understand that NO claims of linearity can be made for ANY amp in this test, as there is no absolute reference for linearity. Granted. Of course. I don't care. Nobody does. My only attemt was to validate what I heard empirically: the VAC significantly changed the sound of the Verity Audio speakers, and much for the worse (IMO), compared to other amps I had on hand. I wanted to see if I could bear this out spectrally, and indeed the results are all too obvious.
So Larry, I trust you understand that I wasn't attempting to measure amplifier output on an absolute basis. It's unnecessary in this case. Differential testing is much easier and at least equally valid, as common biases are eliminated. My background is in test methodology, so give me a nod here....
Balekan, you too misunderstand. The ss amps agree with EACH OTHER within a dB across the band, not with any other reference axis. Again, the actual curves plotted are pretty wild messes that combine the speakers in-room with the pretty serious errors of the SPL meter. But I've gotten good enough at REPLICATING test runs with this setup to be able to tweak crossover designs in 1/3dB increments in the past, so it's been a pretty subtle differential tool.
As others who use similar setups know head-movements account for the greatest amount of the imprecision of the noise-floor. The ss amps all agreed well within +/- 1dB of EACH OTHER as a group, which allows a conclusion ofan "ss standard" for this test, if you will. The two VAC curves agreed quite well with each other, also (I didn't test p level, as it's only two sets of data), and the differences between this data and the ss "standard" is so large as to be very significant, and remarkably audible.
Again, if you plot out this "difference" curve as provided above (either one...they look the same), you can get a sense of the acoustic analogue (no pun).
L&B, I'm trying to think of a good example to explain to you the differential testing method I used. If you and your buddy are having a race it's unnecessary to know far or how fast both of you have run to know who's ahead; you just have to measure the distance between you.
This setup has been used innumerable times by speakerbuilders, especially, to tweak their designs. Again, there's almost no interest in the ACTUAL reference level against 0dB, but only in what happens when you tweak a cap or resistor on that tweeter, for example. You rerun the curve, and look for a subtle shift in the results, and plot the difference data. Same thing applies for testing amps...especially those with high enough output impedences that could actually change speaker output spectrally. Again, the amplitude of the distortion here is what is so shocking, and I imagine said non-linearity was probably one of the reasons for the redesign of this amp. I didn't mean for this chapter to ambush the thread, but it's clear that
L & B didn't understand the test methodology nor the validity of the results. Thanks.

I think they are beautifull to look at. I love the fact that by all accounts they have excellent customer service. Unfortunately, in my experience they sounded like the models of everything that's reputed to be wrong with tubes. To each his own, and definetly not for me.