Hi,

I am building a regulated high voltage power supply for photomultipliers. Idea is that I generate fixed clean 880V DC and then I can regulate it down to whatever voltage I want via software. For that I am using one high voltage MOSFET and I turn it on just as much, so its source's voltage match my target voltage - MOSFET is essentially used as a voltage controlled resistor. This approach is suitable for me, since average opearating current is bellow 1 mA, and switching characteristics of OSFET are irrelevant (transitions seconds long are acceptable). I have an appropriate high voltage resistive divider across drain source junction to sense, how much voltage is being dropped across MOSFET itself. That sense pin is connected to MCP6001 opamp, another input is hooked to digital potentiometer to set wanted reference level (control MOSFET operation). It's very important to mention that MCP6001, AD5241 and the whole sensing network is not referenced to true system ground, but rather a MOSFET's source - I essentially have a floating gate drive network. You can see everything in the schematic diagram I attach bellow.

This circuit works partially. I do get corresponding high voltage regulation according to reference voltage that I program. I get clean DC ramp, clean MCP6001 ouput, both inputs of the opamp are driven equally etc. The fundamental problem I face is that there is a voltage ceiling around 780V (Source to true GND) that I bump into when trying to turn the MOSFET on entirely. Myopamp output is verified to be clean 3.8V DC (Gate to source) in such case when digital potentiometer is set to its boundary value. At first I thought that 3.8V is not enough to fully turn on my MOSFET, but after plugging a standard DC lab supply to gate source junction, I verified that even 3.3V was enough to fully turn on my MOSFET - source potential became exactly 880V (with negligible losses). Since I am sure that my opamp ouput reaches 3.8V while driving the gate, I do not understand the problem why is the opamp not turning my MOSFET on hard enough? We are talking about low currents and effectively no switching, so I can not believe that MCP6001 output lacks current to charge the MOSFET's junction even during multiple seconds. I hope I provided enough information and details about this strange problem. Any help or observations are highly appreciated. I thank you in advance.

 

How about you redraw it using different symbols for the different grounds?
As it stands, you seem to have 3V in the op-amp output, but you need 803V on the MOSFET gate to get 800V output on the source.

 

At the problem point, (780V out) what is the voltage from the Pot? I am guessing 0.3V. If so, I know the problem.

It looks like you are regulating the voltage across the MOSFET not the output voltage? But then I might also have lost track of which ground in which.

 

At the problem point, (780V out) what is the voltage from the Pot? I am guessing 0.3V. If so, I know the problem.

It looks like you are regulating the voltage across the MOSFET not the output voltage? But then I might also have lost track of which ground in which.

Voltage from the pot might be couple or maximum 10 mV, but definitely not 0.3V. In general its 0V. Anyway, the fact is that I am indeed getting solid 3.8V from the opamp while it is driving the MOSFET, so that voltage can not come from nowhere and can not just be ignored by MOSFET. I agree that floating gate network might seem complicated, but that is the easiest way to drive the gate with adequate voltage (otherwise I would need another HV supply only for the gate). My MOSFET's source and the HVGND are the same node (I call it floating GND in the diagram, red color). This ,,Red floating GND" is the positive regulated HV output for my photomultiplier, at the same time being reference GND for the sensing network. It should not be very complicated taking a second look at the diagram.

 

How about you redraw it using different symbols for the different grounds?
As it stands, you seem to have 3V in the op-amp output, but you need 803V on the MOSFET gate to get 800V output on the source.

I have an isolated 5V supply, whose ground is tied to MOSFET's source. Whatever voltage the source reaches, MCP6001, AD5241 will be supplied with +5V referenced to source. Since my gate voltage has to be higher than source, and source is already at around 800V, it would be very inconvenient to design a gate drive network thats swinging lets say 800 to 815V. My approach provides 5V headroom for gate network regardless of source's voltage. I hope it's clear now.

 

I tried to get your circuit to work and almost got it but..........
Here is my try at it. V1=800V, V2 is the 5V supply. V3 is the voltage from the Pot. which makes a voltage that hangs down below the top of the 5V supply.
R6,R7 watches the (output+5V)
It is not right but close.

------edited--------
In my schematic the MOSFET needs to be a 1000V MOSFET, low current, low Gate turn on voltage. You only have 5V of gate drive. Don't use the MOSFET I used. I just picked one from a list.
I removed the 10meg resistor in the negative feedback of the op-amp. That increases the gain of the circuit.
I chose R5 to be about the same value as R6 for a reason.
R1=load.
I chose R7 to be 8meg because it helps when doing the math in my head. 8,000k resistor with 800 volts across it gives me the value of R6 simply as 5V across R6 for 50k. The current is the same in both resistors.
You have a Pot and I used V3. Same thing.
-----edited----
There are two "grounds". one is at that bottom of the 800V supply. The other is at the top of the 5V supply. The op-amp, the pot (V3) and the voltage divider all use this node as ground.
The op-amp needs to be the "r-r input and r-r output" type.