I picked up an old computer power supply recently and I want to convert it into a bench top variable voltage supply, say 60 V/100 W. It's easy enough to boost the 5 V rail up to 60 V design what is essentially a linear power supply using an op-amp and a FET. See the image below:
Ignore the AD822, I was just using that for simulation. Obviously I would need to use an op-amp that can support a 60 V supply.
The issue is that if I want to drive 1 V at 5 A, the FET would need to be able to dissipate nearly 300 W. Thank you, but I'm not looking to design a heater. This got me thinking that it would be really nice if I could adjust the supply to be ever so slightly higher than the desired output voltage.
Enter the LM5122. See image below:
I'm sure many of you are familiar with switching power supplies, but the controller regulates the output voltage by maintaining a constant voltage at the FB pin. So, what if I were to insert a digital potentiometer in series with Rfbb? I would need to select the input and output capacitors, and the inductor for the worst case scenario. I believe this is sometimes called a "Class H" amplifier. Since this would be a boost configuration, the lowest I could go would be 5 V, but I think I can deal with burning 25 W in a FET in a worst case scenario.
I know I could forget the linear stage altogether and output the power from the switching stage directly, however the linear stage will serve to smooth out some of the ripple and lead to a less noisy output.
Does anyone have any concerns about this design? Thoughts? Comments? Has anyone tried this before? Successes? Failures?
So I checked out the article that Peter Smith recommended (see link below). and it was a neat little circuit indeed. It wasn't exactly what I needed, but it did provide a crucial mechanism by which one could track Vout and provide a supply voltage slightly greater than it. It's very efficient, but also low noise.
So, new plan...
- Boost 12 V up to 60 V with a switching regulator.
- Buck that down to Vout + 1.4 V with a switching regulator.
- Buck that down to Vout with a amp and a FET.
This way, for a 5 A load, the FET should only need to dissipate 7 W or so, regardless of output voltage. See the pics below:
Note that I used the LT8612 as a reference switching buck converter, but any one should work providing that it can support the I and the V.
The only remaining consideration is that Vout only goes down to 400 mV. I believe that there was something in the article for this, but this is a problem for another day.