PTC Thermistor for auto fixed bias

[whitelabrat]

I like fixed bias. I don't like having to go in with a multimeter and test/adjust/repeat. So I was thinking it would be great if I could get a PTC thermistor to do the job for me.

I did find one thread that brought it up: DC hybrid folded-cascode for Schade

Anyone actually try it? Practical?

[DF96]

The problem with most simple attempts at automatic bias adjustment is that they regulate average current, when what you actually need to regulate is quiescent current. Fixed bias is best, but needs manual adjustment. Cathode bias from a resistor is a compromise. Regulating average current is worse than normal cathode bias.

This problem arises because all valves have significant second-order curvature. As well as generating second-harmonic distortion this also causes a DC shift when signal is present. This DC shift is normal, and should not be compensated away. The solution, which is rarely used, is to sample the current at zero-crossings of the signal because then you get the true quiescent current (assuming no grid current flow has charged up the grid coupling capacitor).

[whitelabrat]

Who said I was going to listen to anything?

Excellent point DF96.

So I've got all sorts of ideas, like using sensors to drive a servo controlled variable resistor, while the amp is in quiescence. Mmmm. Not simple anymore. Oh well!

[DF96]

Ah, you should have said you weren't interested in actually listening! 8-)

In that case, you can have a marvellous servo circuit with several blue status LEDs. With the wrong servo time constants and gain these will pulse with the music. The LF response will be ruined, but you can then rely on the placebo effect to convince yourself and your friends that you have an amazing system!

To be serious, I did once see a website where someone had built an auto-bias circuit which detected zero-crossings but I can't remember where I saw it. He reported that the elimination of all DC imbalance from the OPT improved low level distortion because of some magnetic effect in the iron.

I have used a combination of fixed and cathode bias for an EL34 output based on Mullard 5-20.

[Wavebourn]

Just use a keyed servo that switch off it's input when signal level is above threshold you set for the "silence". The main mistake authors of such solutions make, they decrease idle current sensing it, while the proper way is to increase it as the result of sensing, sampling, and averaging.