What I'd like to do is to automagically manage the control resistance based on this indicator voltage.
eg. As the indicator starts to exceed 2.5v, increase the control resistance to bring the output back down.
The thing is, I can think of ways to reduce the resistance as a result of increased voltage (eg. an LDR/optoisolator) but I don't know how to do the opposite.
Actually you've have got a very good idea and it might work, it just need a little twist in your thinking. Simply put, the power supply output voltage is sampled via a voltage divider, consists of a top and bottom resistor in series. You've already known how to change the sampled voltage by changing the value of the bottom resistor element. However, you can also do the same with the top element too. By reducing the value of the top element, you can mimic an increase in output voltage as the sampled voltage is now larger. The control circuit will then works to reduce the power supply output voltage as a result.
However, will it actually works? Well one won't know if one don't try. I rigged up the few parts needed and tried it on my supply. In the schematic below, the current signal from edge connector pin#34 is fed to an emitter follower made up of NPN 2N3904, with the 470R trim POT as load. It is advisable to make sure the wiper is at the NPN emitter position before testing the circuit. This current limit can be totally disabled by setting the wiper to the fully clockwise(i.e. to GND) position. The parts like NPN, optocoupler and POT values are not critical. Mine works with a 5K POT but the control range is small.
Yes, I can limit the current to 40A as the output voltage reduces from 12.2V down to 9.8V, with the POT wiper set at the fully anti-clockwise(i.e. at the NPN emiter) position. However, the current limit is gradual and it is difficult to conduct a proper test without an electronic load.
This is BRILLIANT!
All the bits arrived a couple of days ago, and I've put them together on a prototype board, so I could mess around with it a bit and maybe even learn something.
The only changes I made: I used a PC817 rather than a PC817A as that's what I could get quickly. Also I swapped the 22k out for a 17k to get the transistor to trigger a little earlier in the range.
You say you got it working at 40A but with the 22k I wasn't seeing any life until >80A.
So, I have just run a full charge cycle: that's 285AH of charge put into the batteries in just 3hrs without
any manual input from me
Kicking off at just over 100A, current very slowly dropped toward 90A as the voltage rose. But this is OK because critically, the power output remained consistent - within the range of 1212-1250w the whole way. This is the optimal result
I had tried to set the range to max out just before full and so at about 97-98% charge, the optocoupler went fully open and current started tailing off over the last minute or so, dropping to about 50A by the time I hit 100%.
This is such a fantastic mod! And I think we're the first to have done it. Now I just need to do the final build - if only my work was as neat as yours!
Respect for your skills Sir and
thanks for your efforts putting the circuit together.
I should probably stop here as I've achieved the objective but... you know how mission creep goes, right?
So meanwhile, one of these just arrived today:
eBay auction: #https://www.ebay.co.uk/itm/HP-704604-001-ProLiant-DL580-G8-G9-1500W-Common-Slot-Hot-Swap-PSU-684532-B21/401795623783?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649(They accepted £30 as an offer)
It's platinum rated so 94% efficient and has a double fan to dissipate more heat. Of course the Pinout is the same.
I liked the idea of 1500w instead of 1200w / 120A instead of 90A. Am I being greedy?
I hope I can figure out how to adapt the HSTNS-PL11 mods we've done here over to the HSTNS-PL33. I mean, the model number's almost the same, so how hard could it be, eh?