Bad HT PSU

[jhaible]

So I wanted to design a PSU of approx. 450V 150mA with very low ripple for a pair of 5W SE KT66 Triode amps.

I thought the easiest way would be a Source Follower with heavy filtering on its gate, and I built this over the weekend:

http://jhaible.heim.at/se_kt66/se_kt66_psu_idea.pdf

It oscillates like mad. :sad:

50Vpp clipped saw wave @100hz, with nasty burst of HF oscillation sprinkled on top of the LF oscillation's slopes. nasty.

I'm quite puzzled. The only explanation i could come up is that my filter network turned the whole thing into a phase shift oscillator?

Next I'll try will be separate stages with individual MOSFETs and only 1-pole filtering on each gate ...

Any other ideas?

JH.

 

[Svart]

I did one up some time ago and it worked in simulation very well. I haven't built it yet though, but it uses BJTs instead of FETs.

http://www.theopiumdenproductions.com/PSUschemo.pdf

I attempted one with Nfets like yours and it oscillated badly too. I kinda gave up after a few weeks of messing with it.

 

[NewYorkDave]

Phase-shift oscillator? Well, it sure looks like one! It would be right at home in the tremolo circuit of a Fender amp :razz:

Reduce the number of poles and, if necessary, add a resistor in series with the gate to reduce the gain at high frequencies.

 

[VacuumVoodoo]

You might try THIS

high voltage regulator. Even though the text is in polish the circuit diagram is well annotated. Credit goes to Romekd from "Trioda" forum.

 

[jhaible]

[quote author="NewYorkDave"]Phase-shift oscillator? Well, it sure looks like one! It would be right at home in the tremolo circuit of a Fender amp :razz:
[/quote]

Sure, but if I'd design it for that, I'd add a considerably high drain resistor for enough voltage to be fed back. That's what really puzzles me: is the output impedance of the reservoir cap really so high that I get enough voltage for a loop gain >1 ? Well, apparently it is ... 32 Ohms @ 100Hz ...

Reduce the number of poles and, if necessary, add a resistor in series with the gate to reduce the gain at high frequencies.

Thanks for the hints - I'll post again when I have a circuit that works.

JH.

 

[jhaible]

[quote author="VacuumVoodoo"]You might try THIS

high voltage regulator. Even though the text is in polish the circuit diagram is well annotated. Credit goes to Romekd from "Trioda" forum.[/quote]

Nice circuit!

JH.

 

[pstamler]

What Dave said. Put maybe 1k in series with the gate. Also, might want to put 20 ohms or so, carbon comp, in series with the 1N4007 protection diode.

If all else fails, substitute a couple of TIP50s in a Darlington setup. Still use the gate resisotr (now a base resistor).

Peace,
Paul

 

[Larrchild]

The 3 pole gate filtering circuit can be changed to something with less time constant capability. Also, I'd sprinkle some .1uf's across things to decouple at high audio and low-rf =) And the resistor on the gate and diode like pstamler n dave sez.

 

[jhaible]

A tiny 1k directly at Gate, and 10R / 2W at Drain fixed it.

Bridging two of the 100k resistors (making the 3pole filter a single pole) didn't help, so it probably wasn't a phase shift oscillator after all.

Seems good now, at least at constant load. I'll try to but it under stress a little to see if it's realy stable later.

JH.