Measurement of Low Voltage Perturbation on HV Power Supply

Discussion in 'The Projects Forum' started by etdzi1, Sep 30, 2015.

  1. etdzi1

    Thread Starter New Member

    Sep 30, 2015
    I have a 2400V, 1kW power supply being switched (Willamette HV, solid state, 10ns rise time) between an open load and the equivalent of 100pF. When the load is pulsed, the rail voltage of the power supply drops by approximately 5V and it recovers over ~500ms (100ms time constant). In order to solve a problem we are facing, I need to do the following:
    1) Measure the voltage perturbation applied to the load and convert it to low voltage for signal processing. Essentially, I need to convert the 2395V -> 2400V exponential rise observed on the output of the power supply to a -5V -> 0V exponential rise.
    2) Invert the low voltage signal with an option for variable gain such that it becomes an exponential fall from 5V -> 0V
    3) Offset the 5V -> 0V exponential fall by 2400V and apply that voltage to the opposite electrode

    Step two is trivial... I really need help with steps 1 and 3.
    I was thinking of using 1 or 2 AD210 3 port isolation amplifiers (see attached data sheet) to accomplish steps 1 and 3. As there is no pspice model, I am unable to model this in OrCAD and test the part from a theoretical perspective. I am seeking your advice. Will the AD210 be able to accomplish steps 1 and 3? If not, does anyone have any thought on how to do this?

  2. ifixit

    Well-Known Member

    Nov 20, 2008
    Hi Eric,

    It sounds like you are trying to improve the regulation, or step response, of the 2400V PS. An opamp is not the answer.

    The instance you apply the load, the output drops 5 volts. This means the DC impedance of the load at that instance must be approximately 480 times the impedance of the PS output. The load impedance then goes very high over the next 0.5 seconds.

    You cannot eliminate this effect, but maybe you can reduce it to an acceptable level by lowering the PS impedance. Without more details it is hard to advise what to do, but I can suggest adding more output filtering to the PS right near where the load is applied. Try 1000pF (low ESR), which is ten times your load capacitance. You should see a significant reduction is the dip when the load is applied.

    It may be hard to find a suitable capacitor with a 2500V rating.

    Good Luck,
  3. crutschow


    Mar 14, 2008
    The AD210 amp should work, but if you only need the AC component of the voltage then you might be able to use a high-voltage series capacitor to block the DC component, leaving only the AC part. Then you could use standard op amps to process the signal.
    You probably want the time-constant of the DC block capacitor and its load resistance to be at least 10 times the 100ms time-constant of the signal or >1s.
    You'd need to add zener clamps at the output of the capacitor to limit the voltage and prevent damage to the amp from to the large voltage swing when the supply is turned on and off.

    If you think that would work for you, I can supply more details if you need them.
  4. DickCappels


    Aug 21, 2008