So I bought 3 of
http://www.dilas.com/pages/products.php?category=22&series=40 (30W, probably 808nm. Not sure if 200um or 400um) at auction for $500. Now I need a way to drive them.
I figured since CPUs/GPUs/etc use high amperage at low voltage I could use the same topology (sync buck) and ICs.
I ended up picking the LM25117
http://www.ti.com/product/lm25117 because it had application notes for using it as a constant current driver. I modeled it up in TINA-TI and (after much effort) got a stable output current controlled via a pot with 500mA ripple. It was at this point that I remembered that I need to be able to control the current from a micro-controller to actually use this in a laser cutter. I also need to be able to change the power over the full 50A range quickly (< 100us). My current setup can do neither of these.
Here's my current setup:
https://drive.google.com/file/d/0B3IwI7ioPGAJTkZEeWpmTjZFVEU/edit?usp=sharingI have three questions. Is this the right general approach? How would I go about Controlling the feedback with a micro-controller, preferably an analog voltage or pwm signal? And how can I improve the rise and fall times?
For the second question I have considered a digipot, but that requires an I2C (or similar) connection. Although this is probably what I'll go with if nobody has a better solution.
I'm not sure about the 3rd. All of these chips have control loops designed to keep a constant voltage with rapidly changing load demands, which seems to not work well for my needs. If I lower the comp cap too much, it just starts ringing or has a giant overshoot. If I lower the output capacitance, the ripple goes up to 1-2 amps. However the commercial laser diode drivers I've seen out there are citing 5us response times at 400mA ripple. So clearly there's a way.