I've built a flyback regulator that can produce 100-400V from mains, but -- due to limited off-the-shelf transformer options -- I'm pretty sure I'm not using the best transformer for the job (unless I got super lucky).

So, I'm wondering what I'm sacrificing by not using the best transformer. I'd also like to know more about how to choose the right one. (TAoE doesn't say much about this, although I've used its other advice for mains powered power supplies.)

A few more details:

I'm using this transformer because it has a 1:1 turn ratio (all the others I've found have more turns on the secondary), it's designed for 100kHz, and it can handle more current than the regulator.

I'm using this for the regulator. It only uses an optocoupler for feedback. A comparator drives the LED in the optocoupler by comparing the output of the converter (lowered via a voltage divider) to the output of a DAC. That allows me to set the output voltage in the range I mentioned.

I have a schematic, but it's really ugly. I can post it if it'd be useful.

Again, the regulator does work, but I'm wondering what I'm missing by using this transformer (limited output power? risk of spontaneous combustion?), and I'm looking for a general source for design information for flyback converters.


1 Answer 1


Transformer calculation is not an easy task.

As a start point you need to know if its saturating (more magnetic flux stored in it that it can get). That would, along other things, result in core losses. Its hard to calculate that, so its better to study this.

Also flyback drivers are commonly for <100W application, what is the current you need, the cases of use of the SMPS, are great information.

  • \$\begingroup\$ I'm looking for 30-40W of power, although I may be able to do with less. (The datasheet for that regulator says it can't handle that much power, but I'm thinking of the next version.)The output gets turned into a ~40-50kHz square wave that drives a high impedance resonator. \$\endgroup\$
    – lnmaurer
    Commented Sep 2, 2013 at 19:39
  • \$\begingroup\$ Even if the controller can take that power, as I said in my answer you need to know if the transformer can handle the power! for basic, like I said, you need to know if the transformer is not saturating. Without some calc or an oscilloscope will be difficult to tell, but if its saturating much the core of the transformer will get hot, maybe little, don't try with the hands in this voltage :) You need much calculation about the transformer (or even in the try approach, you need and oscilloscope) to get the better performance more easily. \$\endgroup\$ Commented Sep 2, 2013 at 20:12
  • \$\begingroup\$ Look this waveforms en.wikipedia.org/wiki/… on the on time of the pwm cicle, the transformer can saturate. Not the case in the waveforms, but the primary side current will be like a line not going up, in this stage its saturated, the transformer will have a low resistance, possible damaging the primary switch (in this case the controller) and the core of the transformer heats. This can occur more easily if the PWM duty cicle is greater than 50%. \$\endgroup\$ Commented Sep 2, 2013 at 20:28
  • \$\begingroup\$ The datasheet for the transformer I'm using (previously linked to) says the saturation current is 5.7A. Is that not something that can be trusted? That seems like more than enough current to get 50W. \$\endgroup\$
    – lnmaurer
    Commented Sep 2, 2013 at 23:53
  • \$\begingroup\$ Its not that simple, I'm not an expert in magnetics too, but if your current has some DC component, and the transformer core is not reseted (that is getting the flux to near zero), the transformer will accumulate the flux and will start magnetizing till it saturates. If you just pull a DC source to the primary, even with small current, it will saturate with time (anyone correct-me if I'm wrong). What you want to do is magnetize until it is near saturation (some margin), and then reverse the current or reset the core, what in your case in doing in the off period of the PWM. \$\endgroup\$ Commented Sep 3, 2013 at 0:05

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.