Issues with measuring high voltage signals with high probe attenuation (1000x)

Thread Starter

SiCEngineer

Joined May 22, 2019
392
Hello all,

Been a while since I posted in this forum but I have since managed to create a DC/DC converter that generates about 1.5kV from a 40V input. However, I am having (what I believe to be) significant issues with measuring the output voltage accurately. I will attach some pictures that may better demonstrate my point, but the high voltage probe that has a 1000x attenuation is then amplified in the oscilloscope by the same 1000x ratio.

Depending on the voltage fine scale, my oscilloscope will measure very different values, even if there is no voltage present on the output of the converter (verified with analogue and digital multimeters). Since with no output voltage on the secondary whatsoever, the probe can sometimes read a large peak-to-peak value due to the 1000x amplification, I am unclear whether this is EMI or actual output voltage ripple. It is much higher frequency than one would expect from the DC/DC converter. I am most likely not probing the output in the most efficient manner and there may be some significant noise coupling from elsewhere in my switching circuits being amplified by the scope - but how do I verify and then protect against/avoid this issue from happening?

I have an AC coupled ripple extraction circuit in my high voltage board but I wanted to verify the voltage ripple on the high voltage probe first before I send it to low voltage operational amplifier circuits to get a more accurate representation of what kind of AC variations I am seeing on the output.

The waveforms are a VDS MOSFET waveform and the output voltage measurement (green). The ripple is measured as quite high. But from the rest of the measurements it can be seen that the amplified noise can be as high as the measured ripple when I have certain voltage division ratios set up in the scope. Note the different green waveforms show the measured quantities for different voltage and time scales with zero voltage on the output. As the fine scale decreases we can see the actual voltage (mV range), but at higher voltage scales the quantities can be much higher than zero, or at least the oscilloscope reads them that way.

Any information on how to improve these measurements or where the noise could be coming from is appreciated.

Best regards,
Joel
 

Attachments

Top