HP 6622A repair

[Cyberdragon]

New guy here.

Just got my first digital power supply, an hp 6622A, for $10 at an old electronics auction.

When I set the voltage the actual output (measured with an external multi-meter) is less. I set 5.000 volts and it gives me 4.8 no load (both outputs). But the display reads 5.00 (4.99 flickering to 5.00). And all voltages set do have a drop, it's less the lower the voltage but still an issue. The sense connections are jumpered to the outputs (externally). The outputs do work, they will power things, just the voltage is low. And the two outputs are separate boards so they must both have the same problem. I'll check the solder joints, but the connectors don't feel loose. Maybe just the sense solder joints went bad?

Output 2 is also reading 1ma just turning it on when output 1 reads 0.

[KJDS]

find the service manual and read what it says about calibration.

[tautech]

New guy here.

Just got my first digital power supply, an hp 6622A, for $10 at an old electronics auction.

When I set the voltage the actual output (measured with an external multi-meter) is less. I set 5.000 volts and it gives me 4.8 no load (both outputs). But the display reads 5.00 (4.99 flickering to 5.00). And all voltages set do have a drop, it's less the lower the voltage but still an issue. The sense connections are jumpered to the outputs (externally). The outputs do work, they will power things, just the voltage is low. And the two outputs are separate boards so they must both have the same problem. I'll check the solder joints, but the connectors don't feel loose. Maybe just the sense solder joints went bad?

Output 2 is also reading 1ma just turning it on when output 1 reads 0.

And you have checked your meter?
X checked against another/others?
Simple stuff I know, but easily overlooked.

[Cyberdragon]

Two meters both say 4.8 volts.

Unfortunately, software calibration only, with a command line computer in which you have to write the calibration program manually. It has some strange connector on the back, which, even if I found the wire, would almost certainly be a parallel connector.  I'm not against soldering wires to the board to connect it. I could find some sort of parallel to serial converter, and actually write the program in a compiler.

But is there any way to do a manual calibration? Trimmer resistors or anything? Nothing like this is in the manual. (they don't want you to take it apart )

[wn1fju]

Per Appendix A of the Operating Manual: 

"Because there are no internal or external hardware adjustments, your power supply can be calibrated without removing the
covers or removing it from the cabinet if it is rack mounted. Calibration is performed by measuring actual output values and
sending them to the supply over the GP-IB. The power supply uses these values to calculate output, readback, and
overvoltage correction constants. These correction constants are stored in a non-volatile memory on the GP-IB interface
board of the supply. Thereafter, when a command such as ’’VSET1,5" is sent to set the voltage on output 1, the power
supply uses the correction constants to accurately program the output."

If there is nothing obviously wrong with the sense connections, it looks like the cal constants in memory got corrupted. 
Unfortunately, to recalibrate, you need 1) a GPIB cable, 2) a computer and 3) a GPIB device such as a USB-GPIB or Ethernet-GPIB or a PCI GPIB card (along with the proper drivers).  More unfortunately, USB or LAN GPIB devices can be a bit pricey ($300+) even on the used market.  Your best bet is to find a friend or colleague that has the proper equipment.  While the 662x series are fabulous supplies, HP wasn't exactly looking out for the hobbyist...

[Towger]

What sort of 'non-volatile' memory is used?

[wn1fju]

From what I can gather from the service manual, the non-volatile storage is a 1K EEPROM.  It has only an HP part number and HP manufacturer code.  But the layout diagram shows 8 pins, so presumably it is a serial EEPROM.  Incidentally, besides the cal constants, it also holds the power supply model number so that the thing knows how to configure the output board(s).

[free_electron]

cal memory is a 93c46 eeprom.
these supplies indeed need cal through gpib. BUT ... are the sense wires connected ?
these supplies have 1K in the sense wires...

[Cyberdragon]

Yes, the sense wires are connected.

And schematics have been acquired. I don't get how the computer could have just corrupted. Unless somebody did a bad re-calibration.

If I disconnect the sense leads the voltage on the output goes up so it's definitely the sensing. But I'm also getting nearly full voltage just across the sense connections (disconnected), with that mysterious voltage drop on them. So they are electrically pulling down the output.

Is something shorted? I thought sense connections were supposed to be purely inputs.

EDIT: Found a way to hot-wire the interface to a micro-controller on the internet. It's pretty crude but it says it should work for simple operations. I'll try that to see if there is anything wrong with the computer.