Electronics > Projects, Designs, and Technical Stuff

Wide *Output* Range Off-line Flyback Advice

(1/2) > >>

TimNJ:
Hi everyone,

I'm working on a wide output range off-line flyback design. It's a USB PD design, 3-20V, around 100W. Most books I own are concerned with optimizing efficiency around one load/output voltage condition. In this case, there are several points which need reasonably good efficiency. My current strategy is to start by calculating transformer inductance at the low output voltage condition, and then calculate for the high voltage condition, and perhaps take the average. Or maybe push the design point to somewhere around 2/3 maximum power. I don't know if that makes any sense. I know this is a super vague question, but am struggling to find a good starting point!

Thanks,
Tim

Nominal Animal:
Write a numerical simulator, brute-forcing the component values over all input-output voltages?

TimNJ:
Thanks. Yes, something like that, though I won't pretend I'm remotely qualified in "numerical simulation". There's a lot to consider here, especially because the DCM/QR controller is not really a "linear" control element (if you will) because it has a few different modes of operation. I suppose I need to come up with a big spreadsheet and run through a bunch of iterations..I also won't pretend I really know how to do that either. Let's see...

Nominal Animal:
Do you have links to the mathematical equations that describe your circuit?

If I had those, it would be easy for me to write the simulator/solver in say Python (it being available for all OSes, me using Linux), and describe how it works.  It might be useful for others on this forum too, you see.

TimNJ:
Thanks for your offer and interest! Yes, there are equations, lots of them. It's definitely some sort of multi-dimensional gradient descent problem, trying to minimize power loss due to 3-5 loss equations, which all share variables. That's hard enough for one load condition, which is why I'm trying to figure out how to "get in the ballpark" of a solution, because if you're too far away then you'll never "converge", whether or not I'm doing real numerical analysis or just fiddling around in the lab.

I need to get my act together and see what a reasonably accurate model would have to include.

Thanks.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod