I am going to designing a PCB for a high voltage, 6kV and 3kV @ 10mA/200mA (660W) power supply for my Ph.D. As you will know, the equipment needed to work at this level is very expensive and as a student, the money available in my account is limited yearly.

With this in mind, what are the quintessential safety precautions that will be needed for my lab work, and possibly within the DC/DC supply itself? For example, is a short circuit or open circuit protection best, or should I have both? At the moment, all I have in my design is my isolated feedback circuit which will isolate the control board from the converter output. But I am worried that faults will destroy my equipment and they will be costly to replace.

If anyone has any recommended techniques for open/short circuit protection at such high voltages, it would be recommended. I plan on having a pulse-to-pulse current limit of the primary transformer current, but I would like to add some more features to address my concerns of faults such as short circuit protection on the secondary current level. I am thinking of having a comparator circuit driving some LEDs as a voltage level indicator also, for example. This will show me how much voltage is on the secondary side and indicate when safe working voltage is reached. I have also come across a circuit which is for overvoltage and overcurrent protection which looks like it might do what I want: https://www.qsl.net/yo8tot/overvoltage.html enter image description here

But I would also like a level indicator of sorts where maybe an LED lights when a low/safe working voltage has been detected and I am able to modify my circuit safely if that makes sense.

I would rather design in now for the most likely faults rather than worrying about it once I come to the bridge. The load is purely resistive for the purposes of my project. Any recommendations are appreciated.

I will be using digital controller, but am not against using some analog components to interface between my controller and the output voltage in case they are faster and more reliable for example.

overcurrent/overvoltage circuit

  • \$\begingroup\$ Error 404. Page not exist. \$\endgroup\$
    – Andy aka
    Commented Jul 20, 2020 at 15:51
  • \$\begingroup\$ What does 10mA/200mA mean? Is it a typo? \$\endgroup\$
    – Andy aka
    Commented Jul 20, 2020 at 15:56
  • \$\begingroup\$ "/" means "and respectively" in context here \$\endgroup\$ Commented Jul 20, 2020 at 16:13
  • \$\begingroup\$ Make a list of all concerns like a spec list. Input V operating range (with(?)without charger) Battery UV protection. Current sense (range, tolerances), Arc insulation for creepage, Current limiter, inrush current limit, physical interface, isolation or grounded, ripple voltage, load regulation error etc. Then do Make/Buy analysis. \$\endgroup\$ Commented Jul 20, 2020 at 16:21
  • \$\begingroup\$ Yes sorry it was not clear what I meant, but overall power 660W. I will attach an image of the page that apparently does not exist (unsure why it works for me!). Tony, I am unsure what that analysis is. What I am wondering is what are the most important safety features that I can build in a circuit that protect my components against overvoltage/overcurrent etc. I am not using a battery for my power source but a 270VDC+-10% DC source benchtop power supply. The battery source is just included in the picture I attached but I shall not be using it - again a seperate ~15V benchtop supply. \$\endgroup\$
    – jvnlendm
    Commented Jul 20, 2020 at 16:30

2 Answers 2


There is protecting the user/operator and also protecting the load. Will the source and load be in the same enclosure or is exposed HV a risk? Is the short/open circuit protection because the load could fail or user error? What are the power-up and power-down sequence and do those contribute to fault conditions? Will the output vary by load? Is the load DC or pulsed? Those are part of the risk analysis before determining the protection needed.

For operator protection, a HV interlock is pretty common on the control side. This ties to some input like an access panel switch, door switch or cable contact. If the input indicates a fault, the HV is shut off as quickly as possible. That could be turning off power. A fixed load resistor as others have pointed out is also useful the limit the time that stored capacitance is present but adds to the load of the supply.

If you have digital control of some sort, you can consider reading back the HV through a HV resistor divider. Ohmcraft HVD is one off-the-shelf product, you can buy HV resistors (100k-500M ohms range) to make your own as well. Set the divider to give you a voltage in your ADC range or add an op-amp follower to buffer/gain before the ADC. The same divider can also drive your comparator for an LED if a full hardware solution is required. The fault risk here is if the divider looses the ground reference and your feedback sees HV @ low current. Mitigation would involve build quality inspection, pre-operation inspection, and even a slow ramp power up check where you bring the output up to a lower voltage to check if the divider responds linearly before enabling full HV.

This divider feedback might also fold into your control loop if there is a need for it. I can't tell from your question how well regulated your supply will be. A well regulated circuit IMO don't have much open circuit risks at the output unless it is load transients, so it really depends on some risk analysis of the load. Short circuits protection can be as simple as a fuse on the low voltage side. You could also look over documentation from other HV suppliers to see their protection schemes and recommendations, common vendors could be EMCO/XP power, Ultravolt, Volgen/Kaga.


You have not stated the allowable ripple voltages.

This may be important, because higher ripple will allow you to use smaller capacitors, and that lower stored energy may save your life, or avoid destroying some of your PhD measurement setup==s.

Thus you might use 1uF at 6KV at 10 volts ripple at 10mA, at 10KHz,

What do we know? I = C * dV/dT, and I /( C * dV) = 1/dT

0.01 amp ( 1uF/10V) = frequency = 0.01/1e-5 = 10,000Hz

And us 1uf at 3KV at 10 volts ripple, 200 mA, but have a much higher chopping frequency: 200KHz will suffice.

In both cases, yhou can use a SMALL (safer) capacitor.

Just an idea.

  • \$\begingroup\$ Hi, the allowable ripple is very small. <100mV. But I am able to achieve this with a second-stage LC filter which gets me to well below this. I used I think higher voltage ripples to calculate the caps for my outputs and got 100nF and 25nF for the 3kV and 6kV outputs respectively at a switching freqeuncy of around 500kHz. So they are quite small capacitances already. \$\endgroup\$
    – jvnlendm
    Commented Jul 21, 2020 at 10:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.