Hi Francis,
Many thanks for your thoughtful, even challenging, review! You raise some very interesting points, and it's very clear to me you have a lot of experience with tube and SS audio. I commend you on your acute listening skills, and very obvious deep knowledge of topologies.
Let me offer a few thoughts on amplifier design. They are based on half a lifetime of fiddling with SS electronics; my tube experience is only about ten years old. I deliberately downplay the math (which is certainly important but tends to cloud the real issues, in my view) and so my description is based on the subjectives.
As a general rule, the scientific method used with all technologies is concerned to observe, document, measure and 'explain' natural phenomenon. Much of this is based on the human eye, which is almost 100% objective. Conventional scientific thinking therefore looks to measurement and mathematical prediction as conclusive proof of its theories, and in most cases this is entirely appropriate. Atomic power could never have been discovered were it not for the painstaking measurement from people like the Joliot-Curies, Frisch, Meitner, Rutherford, Hahn and a huge supporting army of scientists, notably Neils Bohr. But audio is slightly different because the ear is not entirely objective. There is a powerful subjective element, as we all know. We like what we like, and tastes vary enormously.
Nevertheless, high end audio has been developed along scientific lines with scrupulous care given to the math and the measurement. The engineers have done a good job and delivered amps with 0.005% distortion - absolutely commendable. But the rogue element, in my view, has been the insistence that the best systems have the lowest measured distortion, and this has divided high end audiophiles into two camps; the objectivists, like Doug Self, and the subjectivists, the 'golden ear' group, who are easy targets because they are clearly too 'stupid' to understand the real issues involved, viz, the math.
Audio sales are lucrative because people love their music, and true high end systems can easily cost many tens of thousands of dollars. From $50K speakers through $20K amps and $10K CD players, there is no shortage of available product. So how do we gauge the value of this exotica?
Two ways come to mind; marketing cachet, and sonics. We all know how seductive the marketing can be; often we see something we have to have entirely on the basis of looks alone. Take a look at the latest Shanling tube CD player - WOW!! What a great piece of industrial art!! But those who judge on sonics alone, like Francis, are worth their weight in gold because they are not fooled by good looks.......
Clearly Francis has been on a substantial journey of his own, and has already sought the golden fleece of audio.
However, there's another problem. We human units are very drawn to complexity. Who cannot admire the beauty of a DOHC, fuel injected, four valve head auto engine, particularly when compared to a conventional, OHV, carburetted, pushrod engine? Complexity is very appealing.....
And yet, the pushrod motor is probably more reliable, because it is simpler. And it's likely to be easier to drive too, because its torque characteristics are stronger at low speed. And it should last longer because of the lower speed. All round, the large displacement, simpler motor beats the complex, highly stressed, smaller engine every time, particularly for tooling around the suburbs. So how does this apply to audio? Let me give the advantages of simplicity, and they apply pretty well across all technologies.
1. Improved reliability because there are less parts and lower stresses.
2. Superior overall performance in normal operating environments.
3. Lower manufacturing costs.
4. Easier troubleshooting and repair.
5. Longer service life because of less parts and interconnections.
6. Generally more conservative ratings, giving cleaner operation across the range because of less distortive interactions.
Of course, it goes without saying that a simple design is difficult, because you are trying to achieve more with less, and that's difficult in any endeavour. I started work on power amplifiers many years ago and had huge problems keeping them simple. For starters, I loved cascodes, double diff input pairs, differential voltage amplifiers, Sziklai pairs on the output, current mirrors, current sources at every opportunity. It was difficult to fit all these modern circuit blocks into a power amplifier without a lot of components, and I had huge problems keeping the pcbs small, I can tell you. But over time, I did lots of little experiments and with golden eared friends to corroborate my assessments I generally found that none of these circuit elements improved the sound. To wit, let me tick them off one by one:
1. Double diff input stages. Great idea, but cancels even order distortion, leaving only bad sounding odd order. Also makes it difficult to stabilize the offset. More complexity, poor offset control, worse sound, out it goes......
2. Current mirrors. Very elegant, liked it a lot, gave great offset control. However, one day I pulled out the calculator and did the sums. Current mirrors lead to fundamental unbalance in the diff stage they service. This increases distortion, and when you build with and without, it's clearly audible. On the cutting floor with that one!
3. Emitter followers to precede the VAS (voltage amp stage). Again, a good idea, but doesn't bring any sonic improvement at all. Leave it out.
4. Cascoding. Tried it on both the diff stage and the VAS. No advantage I (or others) could hear. With ready availability of high voltage transistors with excellent high frequency response, couldn't see the point. No dice!
5. Sziklai Pairs (sometimes called CFP). Very, very seductive idea technically. Huge appeal - this is the Elle MacPherson of output stages. BUT, what's this - instability at crossover, particularly on the negative rail? Yes, doesn't work well at the switching transition, very prone to oscillation. OK, can we fix it with small miller caps on the drivers? Aha, yes, we can, but what about the sound? Terrible, no vitality, stone dead presentation. So, it was no good either........
Instead, I concentrated on the weak points in the conventional Bailey PP SS amplifier, and identified diff pair balance, voltage amplifier operating point and lag compensation techniques, and crossover disjunction with the conventional double emitter follower. Time spent in these areas pinned down the culprits; operating points, component choice and switching control. The result of this ruthless culling has been the AKSA.
Cheers,
Hugh
