Temperature changes and AC128's

[Crackle]

I discovered something strange the other day that I cant explain, the effect temperature had on an AC128.
I had been working on an transistor amplifier with a pair of AC128 transistors as the output pair. There was a fair amount of "sssssss" noise on the output. I tried changing both the OC75 preamp and OC75 driver transistors and it hardly made any difference. I then set about changing all of the resistors in the preamp section, very slight difference in the noise.
I then discovered by using a freezer spray that one of the AC128's appeared to be the cause of the noise.
With a DVM connected to the speaker, when the amp was first powered on there was about 5mv of noise, this increased to about 16mv over the next 15 minutes. When I sprayed tx3 a AC128 the noise dropped to less than 2mv after a very short while, then started to rise slowly again.
I replaced the transistor with 3 alternative AC128's and even tried an AC153k, they all responded the same, when frozen the noise dropped to almost zero.

Please can anyone explain this for me.

Thanks
Mike

[Mark Hennessy]

Could you post the schematic?

[Crackle]

Hi
Yes, I should have thought to do that, here it is.    

thanks
Mike

[boater sam]

Perhaps you have created superconductivity? The reduced temperature will reduce the conductivity of the junction? Quiescent current decreased?
The reverse of thermal run away.

[Crackle]

Hi Sam
It only happens when tx3 is frozen, I am wondering if the noise could be the diode associated with that transistor, and freezing tx3 actually causes it to work LESS efficiently and hence the ?noise? from the diode is reduced.

Mike

[boater sam]

That's what I was implying, have you checked if the quiescent current falls when you cool it? If I does, it would seem that we are correct.
Funny things these cold emitter fuses.

[Mike Watterson]

All semiconductors are noisy. It's worse at higher temperature. Zeners are often worst. Germanium are noisier than Silicon. Construction, contamination, current and voltage affect the noise too.
Cooling will always reduce noise. It's possible the the noise source is something near the AC128 if the noise level is higher than it should be, since you have tried different parts.

[Mark Hennessy]

This is most curious

Generally, it's unusual for output transistors to be noisy. At the output stage, any noise they make won't be amplified by further stages, so for their noise contribution to be heard on the loudspeaker would require them to be very noisy indeed. Obviously, the converse is true for transistors in the earlier parts of the circuit, where - for example - the noise generated in TX1 will be amplified by everything following on from it.

It is possible for germanium transistors to fail with gross noise. I've seen this at least half a dozen times in the Hacker sets, where an AC128 in the role of driver (or "VAS", more correctly) intermittently breaks down between collector and emitter, producing large spikes at the output. I've yet to see this with the output devices - perhaps the breakdown mechanism is "weak", so is unable to pass enough current to affect the output (remembering that the opposing output transistor will "fight" it via negative feedback). Sadly, in each case, I simply throw away the offending transistor and move on (time is money on a repair job). Next time, I'll investigate further.

Anyway, while that might be the same mechanism that you're experiencing, what are the chances of it affecting 4 separate transistors? It's not impossible, but it does seem highly unlikely.

What happens when you swap TX3 with TX4? Could it be the diode? Or even the pre-set resistor? I'd want to monitor the DC voltage at the mid-point to see what changes with temperature occur, because it could be - as you imply - that freezing the transistor is effectively turning it off. For silicon, the Vbe temperature coefficient is around -2mV/C - without looking it up, I know that germanium is much greater than that. Certainly, I'd expect that voltage to fall as TX3 only is frozen.

[Mike Watterson]

I suspect the output transistor or amp is amplifying other noise. Four obviously noisy output transistors sounds mad. Note an output stage could be 20 times more noisy than the first audio amp and you might not notice. The first stage noise is amplified by entire amp gain. Which is why an RF pre amp needs to be much less noisy than IF amp, but recieved noise level is dramatically lower at VHF, lower still UHF and much lower at microwave, so while 1dB NF amp is useful at 11GHz, such is irrelevant on LW/MW/SW and pretty much on VHF.

[Crackle]

Hi Guys
Thanks for your replies.
I have just experimented with swapping the 2 diodes round, no difference to the "noise" but it does now seem to improve when either and both OP transistors are frozen, so my previous experiment may have been flawed.



Mike