1970s Design Indulgence

I think I might have tracked down the problem. My mistake seems to have been because I was taken-in by thinking Dinsdale's Wireless World configuration was representative of amplifiers of the era. There is a subtle difference however - the Dinsdale voltage amplifier is DC coupled - similar commercial amplifiers are not!

It isn't the DC coupling as such, but the consequence. That being the limited voltage swing it imposes on T1.

It is not that I hadn't noticed that commercial versions are capacitor coupled - I've mentioned the Teleton (Mitsubishi) SAQ-206B on quite a few occasions - and I had noticed the Leak Stereo 30 was likewise connected (Dinsdale having accused Leak of plagiarism - it would seem falsely now).

So how does this affect things? I would guess the limited collector emitter voltage DC coupling imposes on T1 limits its transient overload capacity. Going back to Otala's AES paper he found that increasing collector-emitter voltage lessened this distortion.

Steady state and with negative feedback applied it might seem there is not a problem as the transistor only swings input signal divided by loop gain. The collector-emitter voltage can easily handle this and has a considerable margin.

However, Otala looked at what can neatly be described as propagation delay, although he never mentioned it like that, but anyone who has also worked in digital electronics will be able to see the analogy.

Should the signal race the NFB then T1 will amplify the full input signal by whatever its T2 loaded gain is. I have tried to make T1 collector-emitter voltage as large as possible but it is still only around 3.5 volts. With an input of say 900mV rms, nearly 1.3V peak, T1 doesn't need much gain to clip.

It must be understood that this will not happen steady-state with a sine wave signal, or low frequency audio program, but as we know, music is a different creature.

I have used local negative feedback to mitigate this, but I still need open-loop gain a factor of 5 more than closed-loop to lower output stage distortion.

T1 however, only needs a gain of 2.5 to reach its limit, which suggests I need to lower NFB to match, but in doing so; output stage distortion will be allowed to climb.

The commercial capacitor coupled versions operated with larger C-E voltages, and in the SAQ-206B it was 8 volts. The sensitivity was such that the input signal was lower to start with, thus the transient overload capabilities were much better.

In a DC coupled voltage amplifier stage T2's base voltage is T1's collector voltage. In a capacitor coupled voltage amplifier, the capacitor allows these two voltages to be different.

By adopting the biasing afforded by T2's emitter, which requires a greater sacrifice of available output voltage, I have at least ensured T1 collector-emitter voltage is greater than Dinsdale's paltry 0.9V, but then again, Dinsdale's input sensitivity is much higher - the input voltage lower.

So, how could T1's collector voltage be increased without changing T2's base voltage? The capacitor coupling could only be used here provided separate DC biasing is used for T1 and T2, this requiring a whole new layout, it cannot be easily tacked on.

A resistor bypassed capacitor might work. The resistor provides the "give" to allow for a DC level change whilst the capacitor acts as a low (think no) impedance connection. We must also reduce T1 collector load resistance for its collector to assume a new higher voltage.

This, if it were to work, would be the very last tweak possible before having to commit to a major redesign.

Hi Graham,I vaguely remember the "why do amps sound different when they have the same spec " but what stuck in my mind was the hifi press saying how good the NAD3020 was at driving "difficult " speakers ( low impedance dips etc )I understand the biasing problem (t1/t2)!