Graham Slee Power Amplifier

Graham, Have you seen the Dynaco ST-120A power amplifier? I has a linear 72 volt regulated supply with built in over-current shut down and each channel has short-circuit and over-current shut down. Mine is now 38 years old (rebuilt in 1996 with 1% resistors and 105 temp caps and all new semiconductors). It recently took a direct short (my fault testing new speakers) on the right channel and only the zener in the amp protection circuit shorted! The other channel still worked fine and the regulator was still at 70 volts with one channel shut down! I replaced the zener, scrutinized all other semiconductors in the right channel and repaired a burned PCB land on R17, then compared all voltages with my "as-built" measurements from 1996; ALL WORKED FINE. The design description is very accurate in stating this amplifier can handle shorts, I know it would have been better if that weak solder joint on R17(right) did not open up. If you like I can send you any info needed on this unit, including specs, updated schematic, my personal upgrades.

Graham Slee wrote: Q1 offers the qualities of an inverting input in a non-inverting design (clever) and provides excellent input linearity (it is essentially a summing node or virtual earth).

I must be going blind! Didn't see C5. Without C5 it becomes a summing node or virtual earth input, but with C5 there is no local negative feedback and the open-loop input linearity of the entire amp is down to the transistor base and its emitter resistance. As the junction between R4 and Q1 is a summing node due to the negative feedback paths from R9 R10 and C15, then it's the transistor base and its intrinsic (unseen) emitter resistance (25/Ic) which determine open-loop input linearity. R1 and R2 help in a small way but linearity is only assured closed-loop by the negative feedback via R9 and R10.

So what's the problem? With a sine wave it's not a problem, but with a complex signal the speaker load and component imperfections can cause phase shifts or delays and therefore the portion of the signal fed back can be at odds with the input signal, and when it's at odds the input linearity is screwed-up.

The answer is to add emitter resistance between Q1 and the junction of R4, R9, R10 and C15. However, that reduces Q1's gain which may not be a good thing for the overall amplifier stability.

It is exactly the same with the JLH 10 watt class A amp as can be seen by examining TR4 emitter. Great idea compromised by open-loop input linearity - 60mV p-p for inputs that are much more than 60mV!








Edited by Graham Slee - 01 Mar 2011 at 6:51pm

This is very exciting!