I recently acquired a pair of HP 6632A power supplies for a very decent price (about 100€ for the pair). These are late 90s models from HP, capable of 0-20V and 0-5A. They have a GPIB port (well actually HP-IB but it’s almost the same thing as GPIB, just an older non-standardized version from HP) so you can control them with a computer. They are 19″ in size so with mounting flanges you can mount them to a rack (gotta go buy one!) though you need to mount them from the back as well because these are pretty heavy (almost 15kg). They also have a really loud fan (gotta do something about that…) and push around 100W of heat out when idle.
They are extremely accurate; when I picked them up we did some measurements with an Agilent 34401A 6,5 digit multimeter and the power supplies were within 0.05% of the 34401A in voltage and current ranges! That’s pretty great considering I have no idea when they have been calibrated the last time (if ever). I need to buy a new multimeter since these power supplies are way more accurate than my multimeter…
Maybe the best feature of these power supplies is that they can sink current as well as source it. Which means you can use one as a programmable load. So with a pair of these power supplies I can make an automated switch mode power supply characterization setup. One of the 6632As would feed the SMPS under test and measure the current and voltage, and the other one would sink the current from the SMPS and measure the current and voltage. From this information you can make a supply voltage vs. efficiency plot of the SMPS.
The worst feature of these power supplies is that they are slow in changing mode from constant-voltage to constant-current when the CC limit is exceeded. It takes over 70ms for them to react, so for 70ms they stay in CV mode which could means tens of times more current for 70ms than you have specified!! I even tried the overcurrent protection feature but it is equally slow; it just disables the output when the CC-mode kicks in. You can see the behaviour in these waveforms:
HP/Agilent has said that this is normal behaviour for PSUs designed mainly as constant-voltage supplies. There’s a workaround to this feature; you can short out the output, turn the output on and let the CC-mode kick in, and then remove the short with the intended load (like say an LED). Still sucks though.
Now I just need to make a GPIB-adapter so I can control them with a computer… I already designed a USB->GPIB adapter which has an ATMega16U4 and driver chips for the GPIB bus:
More information to come about this project in the future. Message me if you are interested in the project and might want to buy a PCB!
Here’s some additional pictures of the power supplies: