Project Details

This Amrel PPS-2322 suffered failure of both channels sometime prior to my acquisition. I purchased this because the front panel display and keypad operated without issue, in spite of the loss of both channels. On opening the unit, the 39 ohm R142 and R155 were clearly burnt, and upon removing the board to desolder these and inspect the power transistors I found that there were a  total of ten components failed symmetrically on each channel. Several power resistors were unidentifiable due to the heat changing the color of the coding bands. Fortunately, blogger Kerry Wong posted detailed photos of his teardown of a functioning unit, from which I determined the values for those resistors.

 I’ve listed the replacements in the below table.

Installing these replacement components brought the unit to a state where it was outputting a positive voltage on both channels, but the setpoint and the actual output voltage differed greatly (~9 V observed intermittently with an external DVM with a 0 V setpoint). Additionally, over current was triggering on both channels, even with no load connected.

Several unproductive paths were explored at this point (since there are no full schematics available for the unit). I checked D32 and D34 due to minor discoloration of the adjacent PCB, but these were fine (17.5 V drop across the diode, might be a Zener, have not reassessed yet), I pulled and checked all the op-amps and comparators to no avail. Frustrated and ready to drop the project straight-in-the-bin I kept checking components and voltages in the hope something would present itself.  Eventually I realized that I was seeing intermittent errors and warning beeps every time I disturbed the computer board slotted into the analog board, and concluded that there must be intermittent contact at the edge connector.

I pulled the computer board, gave its contacts an isopropyl alcohol cleaning (99% pharmacy grade) and when I noted that there was still something a bit ‘off’ with the appearance of the edge connector contacts I gave it a few light swipes with 1500 grit sandpaper then re-cleaned it. I also lightly brushed the contacts of the edge connector slot on the analog board with the 1500 grit, checked pin tension, then reassembled. No faults were observed. Though the voltage output still differed grossly from the set point, I could set different voltages and see a repeatable change in the output. Apparently, the unit had lost its calibration setpoints.

The operator's manual is available from a few sites online and copied as an attachment below. Calibration procedures from the manual will require a DVM and a 0.1 ohm power resistor (a 1 % tolerance was sufficient for my purposes). When an operator starts the calibration procedure, the operator should be prepared to carry it through to completion, though one can exit out by power cycling the unit. If one enters grossly incorrect values by mistake (connecting the DVM or load to the wrong channel at the wrong time and reading an incorrect value) then the unit enters into an error state. This is annoying as the buzzer will emit a constant tone and the operator has to keep canceling warnings while again navigating the calibration procedure. But, once calibrated, the unit should now output voltage and current close to the setpoint.

Once the unit is working consider replacing the fan (which in my case was likely to blame for the cascade of component failure). Avoid high flow-rate fans, rather look for a modest flowrate ~20 dB noise rated fan, to keep the noise reasonable.  An 80 mm “NMB Tech model 08025SA-12K-EA-D0” works well. (I mounted it between the heatsink and the back of the case to help with air exchange.) I also added a couple of compact  blowers to drive air across the power resistors and power transistors.  A couple “Same Sky Model CBM-4010B-140-178-20” work well and again keeps the noise low.

With the unit running, calibrated, and cooled I examined the outputs in detail. Ch2 was unremarkable while on Ch1 There was a substantial ‘ripple’ with a frequency around 0.6 MHZ and an amplitude that varied with output voltage, roughly 150 mV on 10 V. With a ‘scope AC-coupled and tied to the Analog ground point of the +/- 15 V & +/- 25V power connector, I traced this 0.6 MHz ripple as far back as the circuit containing the Digital-to-Analog (D-A) converter, Zener voltage reference, and Op-Amp. Pulling the computer board entirely does not eliminate the ripple, therefore I can be confident that it’s attributable to a component in the D-A circuit.

Another blogger (eblc1388) has surveyed this region of an Amrel PPS-1002, a single channel version of the power supply, and posted an inferred schematic. Reviewing his drawing, I can see that there is potential for feedback with the op-amp and the D-A converter. The D-A converter is an early generation LTC7541. The spec sheet for that vintage indicates susceptibility to minor voltage reversal on certain terminals. The present generation is the LTC7541A and with luck, installation of the modern replacement solves the ripple problem.