I am trying to design an on board DC DC converter that generates 150V from a 12V input. I'd like to design it to handle at least a 30-50mA load, but in reality the load will most likely be smaller. The output does not need to be isolated - it can share a common ground with the input. The other circuitry on the PCB is relatively sensitive to noise, so conducted and radiated emissions are a concern with whatever design I end up choosing.

My first idea was to use a simple boost convert topology, but the step-up ratio seems to be close to the upper end of what most off the shelf boost controller ICs can handle. For a boost converter, DutyCycle = 1 - (Vin/Vout) = 1 - (12V/150V) = 92%. Some controllers might be able to produce a 92% duty cycle, but I'd like to have some more margin in the design, so a simple boost probably won't work.

I've looked into some alternatives, but I don't have enough experience with any of the more complicated switcher designs to really understand the pros and cons. Here's a list of potential options:

-Flyback: Flyback seems like the most straight forward way to generate the 150V. I've designed a 24V to 70V flyback using a dedicated flyback controller from Linear Tech with an off the shelf transformer designed for flyback applications. I would imagine I could pretty easily design a flyback circuit for my application assuming I could find a transformer with a high enough winding ratio. My concern with using a flyback is the noise that it generates. I know there are ways to suppress some of the noise in a flyback, but, from what I understand, a flyback is inherently pretty noisy.

-Coupled-Inductor Boost Converter: There are some application notes out there that detail using a coupled inductor in a boost setup, which allows higher step-up ratios with smaller duty cycles (https://www.onsemi.com/pub/collateral/an-5081.pdf). I am also concerned with noise in this topology, since it seems like any leakage inductance would cause unwanted emissions that would be difficult to contain. The switch and diode would need to be rated for a relatively high voltage, but that's not really much of a concern for me.

-SEPIC: To be honest, I don't know much about SEPIC converters, so I can't really speak to the potential pros and cons. I don't even know if it can product the step-up ratio that I need, I just wanted to get it on the list anyway in case someone has more insight.

-SEPIC Multiplied Boost: I found an app note from Analog Devices that describes a topology that they call "SEPIC Multiplied Boost Converter" (http://www.analog.com/media/en/technical-documentation/application-notes/AN-1126.pdf). I could not find any other information on the internet about this topology, so it's unclear if it's been widely adopted, or still just exists as a curiosity in some random app note, but it looks like a good candidate. It would not suffer the effects of leakage inductance, so there would be less noise. I am a little hesitant however, since it seems like a moderately novel and complex design with not a lot of documentation out there.


My question is, for someone with more experience in designing switchers and power electronics, how would you approach this problem? What topologies would you use? Are my concerns over noise and high duty cycles valid?

  • \$\begingroup\$ First tell us what kind of noise the surrounding circuitry is sensitive to. e.g. you could select a high frequency switcher (>1MHz) and easily filter out low frequency ripple. slow switcher with small currents would have high ripple at switching frequency but have little parasitic noise.. \$\endgroup\$
    – gommer
    Commented Feb 5, 2018 at 15:49
  • 1
    \$\begingroup\$ Personally, I would go for flyback topology because of its simplicity and ease of design. Since the input voltage is relatively low, spikes across the primary switch won't be a problem. The biggest problem can be ringings but they can be suppressed by snubbers. Care should be taken on PCB design because it directly affects CE and RE performance. Please note that the cost will be relatively high. IMO, SEPIC has the highest cost due to the use of more than 1 switches and diodes. \$\endgroup\$ Commented Feb 5, 2018 at 15:51

2 Answers 2


As you've already stated yourself, you'll have a hard time finding an OTS flyback transformer. Custom made and I'd prefer this solution too.

Another approach that I would personally pursue, is a 2-stage boost converter. Duty factors are within reason and first step will be easy. Second step in principle is just as easy, but the trouble will be to find a controller that can handle the high output voltages. Little chance of finding one with integrated FET. This approach can use shielded inductors. Small if high switching frequency is used.

  • \$\begingroup\$ Thanks guys. I'll try out both flyback and 2-stage boost on my test PCB. It won't be much of an issue to use an external switch for the 2-stage boost since space isn't super tight. The final board will contain several low noise RF signals, and well as some ~100kHz rep rate analog signals, which should be free of noise and cross talk (though we have really characterized how much), so I'm working on the assumption of "as low noise as possible". \$\endgroup\$ Commented Feb 7, 2018 at 15:50

A simple boost convertor with a 200V mosfet would work .Finding a chip at these voltages would be too expensive and hard.The mosfet should have a on resistance of 100miliohm or less to keep conduction losses reasonable .Your switching losses will be high so choose a low frequency if you can tolarate a larger inductor or do what I have done which is to employ a switching loss reduction scheme to get good efficiency at normal frequencies .I have done 48V in to 400V out which is not far away from your needs .On other jobs where the output current was low I used a simple diode pump made from BAV21 and 1 microfarad ceramic caps .Sure I used discretes for my boost converter to reduce switching losses ,If you use the diode pump approach you could find a cheap chip with a lower voltage rating .


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.