On Tue, May 30, 2023 at 11:30 AM Christian Corti via cctalk < cctalk@classiccmp.org> wrote:
> > SMPUs don't like that, and don't even try a variable transformers. > > In the 80s, i worked for a company making just-pre-PC 8086 machines. We manufactured the SMPSU in-house, designed by a specialist company. I was involved in training the technicians to repair them and built a debug rig with the designer's advice. This may not apply to all SMPUs but worked well for that design, which was fairly conventional. There was an isolation transformer followed by a variac. The procedure was to slowly bring up the variac until the control circuit was functioning, check the waveforms around that, and then increase the supply voltage while watching the chopper, control and output signals. This allowed the chopper to work at low current limited by the circuit resistance and slowly increase with supply voltage until the control circuit backed it off (which would happen quite quickly with no output load). I agree that low input voltage with a loaded output will stress the supply but done carefully as above it should be possible. I didn't use a lamp, but the idea there is that the maximum current in the case of a short is limited to lamp current rather than fuse current. I wouldn't do this with much of a load, but it's more likely to kill the chopper due to dissipation rather than peak current since it's rated to chop the full input DC rail into the transformer. Abuse might cause higher frequency cycling or longer on-time but shouldn't increase peak current. I'm happy to learn if I'm missing another mechanism.