Tony asks what engineering parameters are factors in the reduction of grain and brightness?
Good question, there is no simply answer we are aware of other than doing our best to design circuits that improve linearity, even if we cannot measure the changes with our test equipment.
In our big hybrid power amplifiers, I can kind of give you a time line of the things that have been improved over the years. Each made easily audible improvements in musicality and reduction of grain and listening fatigue, but almost none can be observed with our test equipment.
Probably the first important improvement was going from the original Hitachi power mos-fet output devices to Exicon devices. This change was forced upon us as the Hitachi devices went out of production. The only circuit changes necessary were an increase in the value of the gate resistors from 100 ohm to 220 ohm to insure high frequency stability. The musical differences were obvious, not only in our hybrid amplifiers, but in our solid state amplifiers too. The most notable audible effect was a sense of "sparkle" and life to the high end previously missing. That and a feeling of overall clarity too, with no negative effects at all. The necessary changes in gate resistor values told us the Exicons were faster devices, but nothing at all different about them we could see up to the three meg limit of my generator and scope.
Another fairly early change was done to eliminate a hump in the ultrasonic frequency response around 200K Hz. Way up there beyond the hearing of bats, the square waves were not perfect. We finally traced the issue down to the outputs being slightly loaded by the protect zeners installed to make sure that the allowable gate voltages were never exceeded. Changing one set of 1000 ohm resistors to 100 ohms fixed this issue. It also made another useful reduction in grain and brightness in the hybrid amplifiers, of course with no measurable effect other than that now ultrasonic square wave performance was perfect.
Much later, Dean Kayser, our resident EE expert, suggested that the same active regulated power supplies that we had installed in the hybrid and tube preamplifiers would also be useful in the hybrid amplifiers. This required a new multi-stage power supply board, with active regulators for the tube and for the mos-fet in the small signal transimpedance loop. We did regulated power supplies for each tube heater at the same time. This provided a major improvement in overall clarity and musical involvement. Again, we could hear it and knew that it was better engineering, but could not measure it within the capabilities of our test equipment.
That next step was finding a replacement for the Hitachi J79 mos-fet being used in the audio circuits. This was a tough one because there seemed to be no suitable replacements out there. We finally decided to give the Exicon 10P20 plastic case mos-fet a try as it had the voltage and current requirements, but somewhat higher gate capacitance. In spite of this, it made a major improvement musically. Evidently the Hitachi had some unpublished characteristics that was limiting its value in a music amplification application, again something we cannot verify on the test bench but certainly can hear.
More recently, we had the entire audio circuit reanalyzed by an outside EE and he came up with only one useful new suggestion, that the capacitor across the bias pot was limiting internal bandwidth, the effect being masked by the feedback. That part went from one micro-farad to 10,000 micro-farads, with another improvement in overall musicality and purity.
Finally, in the release before the current new production Fet Valve 400R and 600R amplifiers, we redesigned the audio boards to allow one more active regulated power supply per channel, so that the tubes no longer had to share one supply per channel, now each tube plate and the accompanying mos-fet each had their own independent active regulated power supply. Again an engineering improvement we knew would be worthwhile, but as usual, not one we can measure, except with our ears.
The final breakthrough has been providing active regulated power supplies for the output transistors themselves. This is a tough one, as since all the current supplied to the loudspeakers has to first pass through the power supplies, each active regulated power supply must have at least the same current and thermal capacity as the output mos-fets do themselves. We were able to meet this goal with a clever use of the new Exicon double die mos-fets. Two in parallel per channel with a zener reference, provides the amazingly stable active regulated power supply we were looking for. An added benefit is that the amp runs cooler as the output devices no longer see the whole hear 200V raw power supply rails, just the appropriate lower regulated rails. The regulator devices run very safely too. At idle, they see high voltage but no current draw. At clipping, they must provide very high current, but as the raw supply pulls down from demand, the voltage across the regulators substantially decrease, keeping them always in a secure safe operating area.
As usual, nothing new to see on the test bench. The bandwidth (3 dB down at 350KHz and 4Hz) remains the same and square wave performance is still perfect and the slew factor remains infinite. However, as all of you have reported, the overall musically of the new Fet Valve 400R and 600R is quite amazing.
So, what engineering affects grain and brightness? All of the above. Just lots over very hard work applying improvements that we know are good engineering efforts even though the test bench does not give us much help. I think the Fet Valve amplifiers are a done deal now, unless, of course, the future reveals new concepts to try. Right now I am just enjoying the music.
Best regards, and Merry Christmas,