I'm currently designing an high voltage DC/DC converter, from a linear technology app-note. My goal is to bias an avalanche photodiode with 200 to 300V, on a small educational satellite. I picked the circuit from figure 16, at page for that.

My issue is that I cannot find a suitable 10uF, 400V capacitor. Aluminum electrolytic are not good for space, stacked ceramic are huge (like this one). Also, I would like to have proper derating, at least 500V.

  1. Did I miss some option while picking a capacitor?
  2. Can I get away with smaller values, like 5.2uF?
  3. I can put inductors, up to 10000uH. Could I add an inductor after the diode bridge and reduce the size of the capacitor safely? Will the impact on the feedback stability be strong?

I'm also interested in any alternative design for an high voltage, low noise converter.

Thanks for your time.

  • \$\begingroup\$ So, this is large current, large battery drain? What's the max bias current for your PD? If you're just trying to generate microamps, then much of fig16 is way, overboard. \$\endgroup\$ – wbeaty May 26 '14 at 0:26
  • \$\begingroup\$ The max bias is 350V at 75 degrees C, but thermal management should bring it close to 300V in worst cases. The peak consumption is around 1 mA. The bias must be modulated to adjust the APDs gain, depending of some position parameters. Fig16 with 2 mA, low noise, and control seems adequate. The other candidate, figure 24, also include a too large and barely derated capacitor. \$\endgroup\$ – pserra May 26 '14 at 9:11

If you look on page 10 (figure 24) they are using a linear regulation circuit on the output to reduce noise and ripple - it's a tracking regulator which may be too complex for your needs but should give you some options to think about. Another method is to make a linear constant current circuit feeding into a high voltage zener diode or a known high resistance value to convert this back to a stable voltage.

An inductor after the diode bridge doesn't really help - you'll be lowering the peak voltages that are needed to generate the high voltage charging of the primary HT capacitor. Putting a larger inductor where L1 is on the circuit you referenced could help but be aware of the self-resonant frequency of larger inductors - it could make the performance really bad!!

Going for a significantly higher frequency is going to help - this ought to be considered too.

| improve this answer | |

Presumably this is a LEO application so radiation is not too bad- if it has to work in a vacuum that would rule out normal electrolytic caps, as you say, which would tend to be the most compact in that range.

You might want to consider film caps such as the PP MKP1848C61050JK2 from Vishay. It's not small at 18 x 28 x 32mm but meets the spec, can perhaps fit in your space (assuming something like a cubesat) and is probably more reliable than MLCC caps (at least that's my experience).

Without a ripple and and regulation and max current spec it's hard to speculate what you could get away with, but I'd be tempted to look at a capacitance multiplier (perhaps with a Zener) post-filter with a substantially smaller filter cap. A base-emitter diode should be added to basic circuit.


simulate this circuit – Schematic created using CircuitLab

| improve this answer | |
  • \$\begingroup\$ The MKP1848C61050JK2 look like a valid option but it's tall, way more than alla other components on the board, the transformer is at 12mm. For the specs, it will draw few nA most of the time, with pulse of less than a mA for less than a nanosecond. I would like to keep the same performances, 100uV of noise peak-to-peak. Concerning the capacitance multiplier, I'm not sure I have the level to design a reliable one a high voltage. Would a linear regulator be interesting? \$\endgroup\$ – pserra May 25 '14 at 18:07
  • 1
    \$\begingroup\$ The right hand schematic is a linear regulator with a bit of a filter. The linear regulator that Andy pointed to might work, but it may have some intrinsic noise that is too high. Capacitance multipliers are good for cutting high frequency noise at the expense of a bit of DC sloppiness. More here: wwwuser.gwdg.de/~uboehm/images/ch14.pdf \$\endgroup\$ – Spehro Pefhany May 25 '14 at 22:00
  • \$\begingroup\$ Thanks! After reading more about it, it seems that I still need a smaller cap anyway. Since I'm just using half of the max current (and even this would saturate the sensor) I'm going to make a board to try different capacitor values, and maybe higher frequencies before going for a more complex solution. \$\endgroup\$ – pserra May 26 '14 at 17:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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