EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Rutherberg on June 27, 2016, 02:58:50 pm
-
Hi,
I'm in the design of a tube amp power supply. The idea is to have the high voltage rail variable and bias voltage tracking the HV and vary in consequence.
Attached you will find the draft schematic.
I'm concerned about the op amp configuration and type. I assume this kind of system should be carefully designed to avoid oscillation.
What would be your suggestions about op amp type and the ways to avoid oscillation ?
Any suggestion about the whole system is welcome !
Thanks
-
correct me if I'm wrong but it seems like you have +15V/GND supply to the op amp and you are feeding about 300V top to the non inverting input? I don't think she is gonna like that :popcorn:
What exactly are you doing with the op amp? it is in open loop, it cant oscillate..
-
John Broskie has lots of circuits on his site.
I think oscillation is an issue, yes.
Links:
http://glass-ware.stores.yahoo.net/pstuposu.html (http://glass-ware.stores.yahoo.net/pstuposu.html)
This is a good site and I used this compensation method for my recently completed low-noise power supply (not high-voltage, though). You need to scroll down and *read* unfortunately, to get just the power supply stuff, but yeah, it works:
http://www.synaesthesia.ca/LNschematics.html (http://www.synaesthesia.ca/LNschematics.html)
-
There is a closed loop around the OP - so it can and likely would oscillate.
In steady state and not oscillating the 300 V leading to the input of the OP is not a problem, as the other resistor will provide a negative current to bring the input to 0 V. However so protection is a very good idea, not to fry the OP just because the regulator starts oscillating to much.
I don't think the OP is rally needed, a simpler stage with just the FET (or an tranistor) should be good enough. Less gain makes oscillations less of a problem. It's often quite difficult with an OP and additional gain. It would still likely need a capacitor for compensation. Usually less capacitance
p.s. The polarity for the caps on the negative bias side is wrong. This might cause some confusion.
-
Thanks for the replies
correct me if I'm wrong but it seems like you have +15V/GND supply to the op amp and you are feeding about 300V top to the non inverting input? I don't think she is gonna like that :popcorn:
Indeed, I added zener protection, thanks.
I don't think the OP is rally needed, a simpler stage with just the FET (or an tranistor) should be good enough. Less gain makes oscillations less of a problem. It's often quite difficult with an OP and additional gain. It would still likely need a capacitor for compensation. Usually less capacitance
p.s. The polarity for the caps on the negative bias side is wrong. This might cause some confusion.
I did 2 new attempts attached, one with the op amp and one with only the transistor. If the transistor version seems ok, I would go for that.
About the transistor version, is it better to use a mosfet or a BJT in this case ? A BJT should do the job and cheaper I suppose ?
I'm a lot more hesitant here because of the HT involved...
Thanks !
PS : caps polarities corrected !
-
The OP version has way to many caps to slow down the regulator.
In the simplest case (simple dominant pole compensation) there should be 1 cap to slow down the regulation. Usually the cap should be at the part that has the highest gain and / or the lowest speed. So in this case for the circuit without the OP the best position for a compensating cap would be a miller cap at the FET or maybe a cap on the drain.
Using several low pass filters like shown will likely cause oscillation, especially if they are at comparable time constant.
It's likely OK to use a BJT instead of the FET as well. the BJT just has some bias current, so it will add a little shift in the voltage. On the other side the gate - source voltage of the FET is higher and not that well defined.
The protection can use just one zener diode - no need to alow large negative voltages. For the Version with OP even just two normal diodes are enough.
-
Add degeneration to Q4 to better control the open loop gain which otherwise is going to cause frequency compensation problems. There is no reason to use a MOSFET there but it does not really matter.
-
I tried the transistor solution and it seems to work well, thanks for your help guys !