1970s Design Indulgence

"Will you ever consider this amp good enough for release Graham?"

Before I answer, let me state that in all my reading and research, not one designer has ever discussed leaving one of their designs permanently on to see if they change in any way - absolutely nobody! Therefore I wouldn't trust their opinions as far as I could throw them. And as for the twerp who said that an amplifier would be worn out after being left on 2000 hours, would that be cumulative?

Anyway, back to my answer:

All I can say is I hope so, but some "force" prevents me attaching my name to something which won't entertain. After-all, an audio amplifier is supposed to serve that purpose, no matter what the distortion freaks say.

Having been brought up in the valve era - I was an impressionable 6 year old in 1961 - I can understand why many felt the transistor was inferior.

The saturday afternoon matinees at the local flicks could never have used a transistor amp at the time: 1. they weren't powerful enough; and 2. picture house wasn't rich enough to replace its valved equipment.

It has to be admitted; the cheap Japanese transistor amps I remember from my youth did not operate flat: they had tone controls, and they were made use of in making them sound right.

In disco, the sound was always EQ'd for the best sound.

I have to admit, that I was never satisfied by a transistor amp operated flat - without tone controls - and spent more hours tweaking than listening.

The Proprius has been the only transistor amplifier I've been able to listen with without EQ, but then again, the modern switched-mode power supply and balanced input "preamp" made it "a valve amp" anyway.

You only need to frequent the forums to get the opinion that transistor amps are bright, but some are obviously forgiven for fear of offending fanboys.

It's the same in the HiFi press, where two identical designs are judged differently; and they get away with it because there are only two or three people in the country capable of reading circuit diagrams!

So why bother with doing things the "right way"? A 5% distortion valve amp is forgiven because it sounds "right", but a 1% transistor amp is condemned.

This lengthy introduction is to announce my last modification, made in frustration, which was to try and EQ out the obvious transistor brightness.

It is quite obvious that the race for wider and wider bandwidth along with staggering amounts of negative feedback at 20 kHz just takes the piss, and that transistors are about as stable as an upside-down house. Let's stop making excuses for them.

Transistors are OK for preamps. They work extremely well and can sound far better than valves, but transistors can't do power well (because it involves high voltages), even if you string hundreds together.

Stop forcing them to do what they obviously cannot do. They don't do highs. Capacitance is the reason. The valve guys knew about capacitance and made pentodes to get over the problem. Where are the pentode transistors? Correct! They couldn't be bothered.

So what's the mod? A 22 pf collector to base capacitor on T1. I'd like to go higher but we're into a tricky situation where I can't see which part is included within the NFB and which isn't. In other words it can be modelled as stable, and unstable, depending on your point of view.

Ain't it funny how the Willamson never burst into oscillation with its zero phase margin? Didn't seem to matter with valves...

This additional compensation tries to counter the drooping phase and spin it out further and in simulation it can be as high as 2GHz before it reaches 180 degrees, where gain margin is a massive 80dB.

All emitter followers are oscillators "in waiting". Output stage parasitics can and will lead to negative impedance. According to Chessman and Sokal in 1976 (https://www.eevblog.com/forum/beginners/noise-on-emitter-follower/?action=dlattach;attach=348962) the oscillation self-limits to around 1 volt, but that 1 volt might use up gain margin, and in a negative feedback amplifier (such as this) might be sufficient to give a longer burst of oscillation.

Any stimulus, which a disturbance can cause - such as the disturbance which is the audio signal itself - can cause this to happen. If the disturbance is constant - for example: playing music - then it can happen for its duration, resulting in a tinny sound.

In the first truly audio op-amp, the TDA1034 (page 105: ftp://bitsavers.informatik.uni-stuttgart.de/components/philips/_dataBooks/1976_Philips_Semiconductors_and_Integrated_Circuits_Part_5a_Professional_Analogue_ICs.pdf) we can see the compensation I refer to as C1 (top left of circuit diagram). The TDA1034 launched in 1976, and is now the NE5534.

So far not showing previous signs of divergence. Next few days should be interesting.

peterb wrote: I am curious.  Are your 'permanently on' tests with music permanently playing? If so, wouldn't the level be important, particularly for the power stage of the amp and the psu?