Home
I've been trying to create this circuit:
http://54.187.81.171/files/mjgardes-job_762.pdf We're actually using a slightly different (less explody) topology now.
You'll find the software on the references page (see below) if you want to run it yourself.
The waveforms around D1 and D2 appear fine, but the voltage across D3 never actually goes positive. I'm beginning to wonder if my selection of the 14CBH3321R0K1 will even work as a flyback transformer.
The circuit design is based on a Weinberg topology, which is sort of like a push-pull switching power supply with a flyback transformer. This was selected because it appears to have a response that will allow reasonably efficient solar power conversion during bright midday sunlight and also during twilight hours until Vin dips below Vo/2.
After failing once with a basic buck converter brain-dumped onto a whiteboard and yet again with a full-bridge topology ripped from some application notes, I did some reading and settled on the Weinberg circuit.
A Beaglebone Black was selected for the controller for convenience and cheapness. Features that would be nice to have would be automatic tuning of power supply parameters, data logging, remote control and monitoring and other fun stuff. These are low priorities.
Right now I'm just utilizing the Bone's PRUs to generate the switching waveforms because the stock debian Linux kernel doesn't give me enough control over the PWM generation subsystems (though they'd be ideal for this task (hmm, maybe try one of those new RT kernels that are coming out?)).
Thanks to Kevin, https://kevinsprojects.wordpress.com/2015/01/12/analysis-of-a-weinberg-converter-for-solar-panel-maximum-peak-power-tracking/ there is now a way to deal with the flyback voltage spikes that kill FETs. We also have working simulations in LTspice.
What I'm wondering now is why the flyback transformer doesn't seem to be doing anything...
Turns out that the added coupling caps prevent the the D1 circuit from turning on. Hmm. But the rest of the circuit works, and it's a lot simpler, so maybe i can live without the flyback.
So now this circuit is less of a Weinberg topology and more of a two-phase SEPIC. I've hacked some caps onto my last prototype board and it seems to be working. I'm having some glitches with the I2C bus but it might be okay.
Hardware works. Board layout needs to be redone because ground returns are too close to i2c lines and causing interference. It's all greenwired anyway because i ordered the wrong ltc4151 chip variant from digikey and they're expensive.
Now all i need to do is write the Software (blegh) Information I have found to be helpful