I've been dragooned to help a friend I've known since the second grade. Jimmy designs and builds ultra high end audio systems, he builds everything up to the speakers.

?

Recently he's ran into some odd problems where "identical" transistors behave differently in high end audio. He purchases directly from the manufacturer so the chances of counterfeits is pretty low.

?

It's been a life time [1968~1973] since I designed any serious audio amps. I realized my talents lay elsewhere and it was a better value to trade money for decent amps then spend nearly unlimited hours to obtain similar results.

?

Note: I do not consider myself an audiophile, I damaged my hearing riding a motorcycle for 35 years and had tinnitus as a young child.

?

Sooo, I'm refamiliarizing myself with the issues of high end audio design, theory and reality.

?

No amplifier is perfect, they all have artifacts. One thing that is very different today from way back when is moderate priced USB audio I/O 'cards' allow meaningful FFT analysis that actually matches reality, within limits.

?

I downloaded this book to start with....

?

Bob Metzler - Audio Measurement Handbook

https://convexoptimization.com/TOOLS/Metzler.pdf

?

While copyrighted it appears in quiet a few respectable locations, I'm guessing the author has waived copyright issues.

?

Rod Elliot of ESP audio has quite a few articles on design and measurement of distortion.

?

https://sound-au.com/articles/distortion.htm

?

https://sound-au.com/articles/intermodulation.htm#ref

?

https://sound-au.com/articles/intermodulation.htm

?

https://sound-au.com/project52.htm

?

Uses a sound card to measure distortion

https://sound-au.com/project232.htm

?

Perhaps the single most useful trick shown by Mr. Elliot is his "distortion adder." It allows sanity checks, which, when dealing with audio distortion is necessary. It is way too easy ending up chasing your tail like a puppy or kitten.

?

That's why I gave up on amplifier design in 1973. A quasi-comp Sansui AU-555A sounded much better then my best efforts at about 1/2 the cost in material and 1/gazillionth the wasted time. I could have collected sodapop bottle along the highway, turned them in for the deposit and been time ahead. I'd reached a point where I'd lost confidence in my ability to measure differences in slight design variations.

?

?

One important issue, I've been playing with several USB sound cards for several years and realizee "The 5V power rail in most USB systems simply suck!" Modify your USB cable and add your own 5V power supply. I went with a LM7805 with the standard design with an added 1uF tantalum (I hate and distrust tantalums but they have the best filtering capacity for size and a 100uF and 1000uF on the 5V out. I'm not convinced the added capacitors were worthwhile but the PCB I used had the pads for them so I decided to toss them in.

?

[Be certain to tie your 5V power supply V-/ground to PC USB V-/ground! I spent an hour late one night wondering "Why doesn't this damn thing work!"]

?

The residual noise floor was just under 10dB lower with the external supply versus the laptop USB 5V power.

?

You mileage will vary because every maker of USB almost certainly has different noise issues..

?

I suspect Jimmy is running into the hard wall of physics, where each device is inherently different even if made at the same time. This is something I have no idea how to verify.

?

They don't call it the bleeding edge of technology for nothing.

?

Oh, and don't bother trying to test the modulator in your SSB/AM/FM ham rig, they all suck.