Repair of HP 0957-2304 power pack for HP Photosmart 7510 All-in-one

[GraemeG]

After years of mostly trouble free operation, I tried to turn on my HP Photosmart 7510 printer/scanner/copier a few days ago and found it totally dead. I soon tracked the problem to the 0957-2304 power pack, the green indicator light was off and a quick check with the multimeter showed 0 volts output.
I have found and ordered one on ebay but decided to have a look inside and see whether it was worth repairing.

This is a slightly unusual power supply in that it has 2 modes, 32 volts at up to 1094mA output and a "sleep mode" of 12 volts up to 250mA. The mode is controlled by a signal fed back from the printer on a third wire. BTW I suspect that the 0957-2305 is a similar, possibly identical, unit. There are other units 0957-2259 0957-2479 0957-2271, these seem to be able to supply 32V at 1500mA but they don't have the 12V sleep mode so may be totally different internally and I don't know if they would work correctly with my Photosmart 7510.

Firstly a warning The box is quite easy to open, just 2 Torx screws hidden under the corners of the label, but a check with the meter showed over 300 volts still on the main capacitor. With the low voltage stage dead there is nothing to discharge the cap so BE CAREFUL!

I used a 100K ohm resistor and some leads to discharge the cap and then had a closer look. The board looked flawless, no signs of heat or failed components, but testing found that the main diode D5 (VSB3200) on the secondary side is short circuit and the main MOSFET Q3 (6N62K3) is open circuit except for the drain-source reverse diode. As far as I can tell all the other components are OK. All the electrolytics have been tested for capacitance and ESR, most of the diodes and several of the SMD transistors seem to check out OK too. The opto isolator is fine also. I removed it from the board and used a couple of multimeters in diode test mode to verify it. Likely the D5 diode failed and took out Q3.

I have replaced D5 with a second hand MBR5200 (5amp Schottky instead of the original 3amp unit) that I had lying around and I have some 6N62K3 MOSFETs on order on eBay, 10 units for less than USD$15 delivered. I'll update this post when they arrive.

The pictures I have attached show the board with the heatsink, Q3 and the thermistor TH3 removed. Once they were removed from the board I was able to test Q3 to confirm it had failed. Q3 and TH3 were glued together and to the aluminium heatsink with a hard green "epoxy" glue. I was able to gently lever the side of Q3 until they both cracked away from the heatsink but then had to carefully cut the thermistor free from the dead FET with a hacksaw blade! I made sure to cut into the FET body so that the thermistor was undamaged. I'm not sure yet how I will reassemble this, possibly with heatsink compound and/or epoxy. The thermistor markings are invisible under the green epoxy but as best as I can tell it is possibly a 200Kohm NTC device. Mine is still working but if you damage it I would try a 200Kohm NTC thermistor or you could try a resistor of about 150K-220K but you wouldn't have any thermal protection so it's at your own risk!

[Swake]

Are you certain the mosfet is dead?
Measure it again with a Ohms meter (not in diode check mode). Check youtube for a tutorial if you're unsure how to check a mosfet. It does not behave the same as a normal transistor showing 2 diodes.

[GraemeG]

Yes I'm certain
I'm aware of the differences between bipolar and FETs, I use the diode mode on my multimeter because it has a higher test voltage than the resister test mode. I have found that the ohm mode doesn't supply a high enough voltage to turn on some FETs. I use the following test (for N channel FET):

1. Momentarily short Gate to Source
2. Check Drain to Source is open circuit
3. Momentarily touch meter positive lead to Gate, negative lead still on Source
4. Test Drain to Source now conducts

 In fact I have a cheap Digitech QM1549 meter that only supplies 1.5V in diode test mode and that is not enough to turn on the gate of many FET's. My trusty old Fluke 75 supplies 2.5V and that works nicely.

[GraemeG]

I owe you an apology Swake, the MOSFET is not dead.

I have learned that some power MOSFETs, particularly the high voltage types like the 6N62K3 and NDF04N60ZG, need even more than 2.5V on the gate to turn them on. I found that if I add a 1.5V AA battery in series, giving about 4V total, when I touch the meter to the Gate in step 3 then the D-S conducts correctly. The gate capacitance is charged by the momentary touch to the Gate and the D-S conducts for a few seconds until the Gate charge dissipates.

So in the end all I needed to replace was the VSB3200 diode. The thermistor was reattached beside the MOSFET on the heatsink with a little heatsink compound underneath and between the thermistor and the FET and then covered with some 5 minute epoxy. I could have save myself time and effort removing the thermistor and MOSFET if only I had tested it properly in the first place!

With the Sense pin not connected it floats at about 2.4V and with no load the output is 32.6V. I used a 1K ohm resistor to pull the Sense pin to ground and the output drops to the standby level of 12V. The printer is now up and going again 

[Swake]

It works, that is the perfect outcome I would say
And we learned something about that minimum gate voltage that I totally forgot about myself.