Grid Current Un-Limiting?

[R.G.]

A lot of FET based ciruits I've seen have 12V zeners added for safety. Does that not really apply in this situation, does the IRF820 have the zener built into it?

No, the IRF parts do not have built-in zener protection. Be sure to use a zener to prevent the gate being driven more than it can tolerate.

I'll pick up some IRF820's and other bits soon and try it out.

There are better MOSFETs for this these days. The problem with MOSFETs is the high gate capacitance. There are some newer MOSFETs with much lower gate capacitances if you look. The trick is to find a device with a drain voltage spec comfortably larger than the B+ if you use B+ to drive it, or larger than the MOSFET supply if you use a separate drain+source supply for it, a reasonable current rating (this is easy) and enough power rating for the job. In practice, any device in a TO-220 package will have enough power and current rating. There are some TO-92s that might work if you are very careful with power calculations and possibly heat sinking.

[defaced]

I'm glad you found that helpful.

Does that not really apply in this situation, does the IRF820 have the zener built into it?

RG already answered the question, but the reason why they're not in that schematic is because I was a MOSFET newbie and didn't know better. Since this circuit is still in the prototype stage, these will need to be added in the real version if I decide to keep it in the amp. Good eye! I kinda forgot I didn't include those, and probably wouldn't have remembered until I had a board layout 90% done and was comparing it to some other MOSFET layous I've done.

[exclamationmark]

So I actually tried sticking a mosfet in front of a 12ax7, albeit only with a B+ of only 100v (for fear of melting something - maybe I already did :P) and got some pretty strange results. With the tube approximately centre biased I almost achieved rail to rail clipping within about 5v of the supply voltage. That's not the weird part though... The tube actually exhibited some sort of hysteresis effect and had some sort of memory of it's past behaviour. For example, I turned my signal generator up to the max (I think it's something like 10v pk-pk voltage) and it generated a clipped waveform. Normal enough. I then turned my waveform generator down and the scoped signal was initially miniscule - it then began to grow to "full size" over the course of perhaps 5 seconds. I initially thought it might of been the coupling capacitor - so I removed it from the circuit and DC coupled the plate to my scope, and still got the same result. There were a couple things that changed/stopped the strange behaviour though. One was changing the bias point. If I stuck a 10k+ bias resistor in there, it stopped exhibiting this effect (perhaps because of the low 100v B+, the quiescient bias point at 'centre bias' was actually drawing some grid current). And the second thing that stopped it was adding a grid stop resistor of some variety.

As to why this occurs... I theorize the rail to rail clipping would indicate that the cathode is running pretty much flat out and there are very little electrons left. When the signal is abruptly changed to a smaller signal, the heater has some time to replenish the electron cloud and conditions of the tube slowly change back to their original operating conditions. This would explain the ridiculously long 5+ seconds for the signal to "grow" back to it's maximum size. I didn't actually play anything through my little experiment, though.

[Enzo]

Well anything is possible, but i doubt the tube ran out of electrons and took 5 minutes to recover.

ANy chance you had some RF oscillations?

Did you watch teh DC voltages at grid and plate when this 5 minute recovery was underway?

ANy chance your MOSFET heated up then needed to cool back down?

[R.G.]

Those would be my guesses too, Enzo.

There is some dark-horse possibility that something funny happened with the oxide coatings on the cathode and messed with emissions for a while till they recovered, but I would look outside the tube first. Normally you would expect emissions to go UP when the cathode gets hotter.

Electron cloud replenishment is pretty fast. I would expect that even if the electron cloud was wiped out, it would recover on the signal halves when the tube was turned in the "off" direction. That pushes my guesses into the things Enzo mentions.