I'm looking for a way to step down 3 phase, 2kV AC power to 3-12V DC. I do not have a neutral, just the 3 phases and ground. This will be entirely isolated, nobody will be anywhere near this thing. In fact, it will placed inside a thick, grounded metal case that will be essentially tamper-proof.

The components need to be rated for a 175C environment... which really messes with the available options.

I'd also like it to be able to survive and hopefully still operate if one of the phases is shorted out or grounded. :) So... kinda harsh criteria and environment.

On the plus side, the thermal sinking capability of the system is quite high.... but then I'd prefer to minimize that as much as possible. It's there if required.

I don't need that much, somewhere between 5-20mA (probably around 10mA).

I'm currently rectifying the three phases with half-bridges, voltage dividing that (burning power in the process, but limiting voltage on the smoothing cap). It seems to work, but I'm sure there's a better way.

I've looked a little into custom transformers, but it's not clear how much power will be wasted in those either.

EDIT This can weight a good bit (< 50 lbs, target is maybe 10-20ish). It should fit in say a 4" x 12" cylinder, although preferably under 8" long.

Thoughts would be appreciated.

  • 1
    \$\begingroup\$ How much space and weight can you live with? \$\endgroup\$
    – winny
    Apr 15, 2020 at 19:34
  • \$\begingroup\$ Edited with space and weight parameters. \$\endgroup\$
    – darron
    Apr 15, 2020 at 19:44
  • \$\begingroup\$ Well, yours truly designed a buck converter which would do the job for you as is, but since you have AC to start with, I would pursue to wind a custom transformer. If you are worried about no-load core hysteresis losses, you just have to sacrifice max load efficiency due to resistive losses and wind more turns with thinner wire to excite to core less. Are you allowed to pot it or immerse it in transformer oil? \$\endgroup\$
    – winny
    Apr 15, 2020 at 21:00
  • \$\begingroup\$ Is there always current on the MV phases? (Thinking three current transformers used in a non-conventional fashion.) \$\endgroup\$
    – rdtsc
    Apr 15, 2020 at 21:10
  • \$\begingroup\$ Potting is fine, oil is possible but probably not preferred (adds overall assembly complexity). As for current, this'll be in parallel to a big motor. It probably isn't practical to have consistent access to the current-carrying cables for that. \$\endgroup\$
    – darron
    Apr 15, 2020 at 21:24

2 Answers 2


How about a 0.01uf capacitor, feeding a 1n756 zener diode. Or the 3-phase version.

1 farad at 1Hz is 0.16 ohms. 0.01uf at 1Hz is 16 million ohms.

0.01uf at 60Hz is 250,000 ohms, or 4uA per volt.

0.01uf at 60Hz at 2,000 volts is 8,000 ua or 8 milliamps

Seems that 3 high voltage caps, each in series with a high voltage diode, the three diode cathodes all feeding a zener diode ----- may suffice.


Warning --- for this to work, the capacitor must be fully charging and then discharging, and then fully charged to the opposite polarity, on each cycle of the power.

That is why the zener diode works: the zener clamps to +5.6 volts on the (+) half of cycle, and then clamps to -0.7volts on the (-) half swing.

For this this to work, need 3 caps and 3 zeners and then 3 diodes from the top of each zener to a common capacitor, then to regulator.

For 1 volt of ripple at 10mA load, (given 8,333uF is needed for 1 volt at 1amp at 120Hertz recharge rate), your 360Hertz recharge rate and 1/100th the Iload allows a tiny cap ---- 33uf or 39uf.

  • \$\begingroup\$ Thanks... something like that might work. Any reason not to put the diodes on the HV side and then all tied together into the cap? It looks easier to meet specs that way. \$\endgroup\$
    – darron
    Apr 16, 2020 at 0:16

No way would I use a capacitive 'dropper' to connect to a 2kV line. Instead, consider using two transformers in stages:

Then you can use ordinary low-voltage components in the secondary - diode bridge with a simple linear regulator. Or you might even be able to find something pre-made that can survive your harsh environment (that's a hint that buy may be the better answer here.)

  • \$\begingroup\$ Your example transformer is about 2000 times larger spec'd than I need. I'm interested in transformer solutions, but I haven't found much. Custom winding sounds like it'd be the way to go if I did a transformer, but I've got a lot to learn before attempting that. \$\endgroup\$
    – darron
    Apr 16, 2020 at 3:15
  • \$\begingroup\$ So, given the small current requirements, why not a capacitor? Difficulty in maintaining creepage distances? \$\endgroup\$
    – darron
    Apr 16, 2020 at 3:16
  • \$\begingroup\$ Yes, exactly. That, and the thermal environment and getting a capacitor rated for it. \$\endgroup\$ Apr 16, 2020 at 3:27
  • \$\begingroup\$ One idea I had was to repurpose a current sensing coil on one of the legs and use that as a low-current transformer. Its output would vary with load but maybe would be in range of what you need for your small power. \$\endgroup\$ Apr 16, 2020 at 4:01
  • \$\begingroup\$ I could do two series ceramic caps... 2kV is 266 mil creepage? (7500kV/inch) ? ... route slots under them both... they're slightly expensive though. ... current sensing coil? Hmm... interesting. Probably not practical, though... I'd guess it's too easy to damage or mess up the install without something like a metal clamshell case... \$\endgroup\$
    – darron
    Apr 16, 2020 at 4:14

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.