I'm trying to figure out the best method to get around 200V @ 15mA from a LiPo battery, and had a few questions.

  • Should I be converting down to some voltage (eg. my microcontroller supply rail) with a buck converter and then boosting up?
  • Should I go with a boost converter or a flyback converter?
  • What good off the shelf ICs exist for this type of design?
  • What kind of losses would I expect in conversion?

My application is a wearable device, and these are the design concerns I'm juggling:

  • Minimize power dissipation from losses (do I need heatsinking? assuming 75% efficiency, I'd have to drop 750mW somewhere)
  • Needs to be small form factor (how much will this decrease efficiency?)
  • How should I implement short circuit protection/current limiting?

Sorry if I missed anything and thanks for the help.

  • \$\begingroup\$ A 200V wearable device eh? Is it some kind of restraining collar for bad people? Plus heatsinking also sounds fun for a wearable device! Do you know what you are doing? \$\endgroup\$
    – Andy aka
    Oct 1, 2015 at 8:53
  • 1
    \$\begingroup\$ You have many questions but also you want 200V on a wearable device, hmm. 200V can kill people, you know that I hope ? Maybe you should start with building something simpler/safer first before venturing into this area ? \$\endgroup\$ Oct 1, 2015 at 8:57
  • \$\begingroup\$ I'm interested in making a wearable electrotactile display. I admit I'm more familiar with embedded/digital, but have some analog knowledge. Nothing about power, though. I'm hoping to scope out if this would be a viable design and what I would need to make it safe. \$\endgroup\$ Oct 1, 2015 at 8:58
  • 1
    \$\begingroup\$ 200v from 4v is a very large ratio for a boost converter, flyback would fit better. Hit the Maxim, IR, Texas sites for devices, app notes, reference designs. It's not an easy design, there are a lot of gotchas, even experienced people can look like noobs when they first meet power converters. So, the manufacturers bend over backwards with information to help you use their parts. A word to the wise, if you only need a couple of parts, Maxim tend to sample. \$\endgroup\$
    – Neil_UK
    Oct 1, 2015 at 9:00
  • 1
    \$\begingroup\$ I would probably avoid the issue of designing an efficient power supply altogether and just integrate a pre-designed module. For example, have a look at the 1364HVPS which seems to suit your needs. I've used that in the past for one of my Nixie tube projects. \$\endgroup\$
    – tangrs
    Oct 1, 2015 at 12:16

1 Answer 1


You are talking a big boost ratio.

Orthodox thinking would suggest flyback BUT that means a transformer that you will probably have to wind/design yourself.

One way to handle big boost ratios is to use a diode pump on the end of a boost convertor. The diode pump approach is good for high frequencies and low currents which is where you are.

The number of stages in the diode pump can be your choice. More stages makes normal fixed frequency, hard switched current mode perform efficiently, but it uses more pcb realestate and raises component count and reduces the peak drain volts on the N channel switching FET, which makes a low on resistance device easier and cheaper.

OK go for 4 stages if you must use the chip out of the book. 2 stages would work on a S TRAP or other switching loss reduction scheme. A big advantage of the diode pump approach is that you can use a cheap SMD off the shelf coil.

If you need 90% efficiency at a normal operating frequency you will have to start dealing to switching losses. Current limit is intrinsic on any cycle by cycle scheme.

Speaking from experience the boost diode pump is much more efficient than flyback.

  • 1
    \$\begingroup\$ Please make more of an effort to properly format your text. Proper punctuation and paragraph breaks make things much easier to read. \$\endgroup\$
    – JRE
    Oct 1, 2015 at 11:42
  • \$\begingroup\$ JRE I am trying to write better.My screen name isnt entirely innacurate.Thank you for editing it. \$\endgroup\$
    – Autistic
    Oct 1, 2015 at 19:28

Your Answer

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

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