Delta DPS-300AB-47B........dead

[everell]

Here is a Delta DPS-300AB-47B 300 watt power supply with 80 Plus Bronze printed on the label. That is.....80 Plus Bronze when working! This one is dead. No 12 volts, no 5 volts, no 3.3 volts, and no 5vsb. Took it apart and found that the fuse was open. This is serious! So I measured the big transistors and found that one of the switching transistors was shorted. Transistor is a FET, ST P20NM60FP. This power supply uses a pair of them. They are the black transistors that do not require heat sink pad, but can be directly attached to the heat sink. And not much of a heat sink!

Very interesting power supply. It has four SATA plugs and no Molex plugs. The 5vsb circuit uses TNY277. The PWM/PFC chip is a CM6802, first one I have seen. Up to this point everything I have worked with had the CM6800. The CM6802 has many new and improved features over the CM6800 (according to the data sheet). Looking at the bottom of the pc board there are lots of SMD surface mount devices including two chips, two large transistors, and four optoisolators. The main capacitor is 180 uF at 450 volts made by ishi. Could not determine brand of output capacitors.

I do have a pair of 18N50 transistors which might work for replacement switchers. But I think the breakdown voltage is a tad lacking. Not too sure if I want to fix it or not. What do you think?

[Th3_uN1Qu3]

Re: Delta DPS-300AB-47B........dead

In the PDF it says 550 volts... I'd say it's worth a shot. The difference is small, and if the devices do hit breakdown voltage it's likely to be noticeable before they blow, and you can pull them and save them. I ran some tests last year and unless you do something really stupid (like 200v rated transistors where 400v ones should have been), they won't blow right away.

Stick a temperature probe on the primary heatsink when testing the PSU, and see how hot it gets. If it runs over 70C, you need 600v ones.

[c_hegge]

Re: Delta DPS-300AB-47B........dead

I'd do something about those heatsinks too. Maybe use some thermal epoxy to stick a few old chipset heatsinks to them. I had an FSP 300W with heat sinks like that. They got rediculously hot, even with the fan running like a jet engine.

Other than that, though, I think decent 300 watters are worth hanging on to.

[everell]

Re: Delta DPS-300AB-47B........dead

I replaced the fuse and the two switching transistors with the 18N50. Line power plugged in but power supply off, 5VSB comes on OK. Turn PS on, fuse blew and same transistor location shorted. Back to the drawing board!

[Toasty]

Re: Delta DPS-300AB-47B........dead

Shades of Apple's Samsung G4 problems...

Check that mains cap.

...and use a light bulb in series...

[Toasty]

Re: Delta DPS-300AB-47B........dead

Verrrry quiet.....?

Oh, FYI ( I'm sure you know how ):

[everell]

Re: Delta DPS-300AB-47B........dead

This evening I was doing some very basic tests. With switching transistors and pfc transistor/diode out (short across the pfc diode), will the DC voltage come up to about 170 volts?.....yes Does the 5vsb still function correctly?......yes....turning on pfc/pwm chip what is Vcc?..........14 volts........what is Vref?........7.65 volts.........so removing the short across the pfc diode and reinstalling the pfc transistor and diode, will the pfc circuit work?..........yes, 454 volts (no load).

Is pwm driver transformer shorted?......had to remove it from circuit to check it. no winding to winding shorts, and all three windings have zero ohms. So pulled out my ring tester........all three windings light up all of the LEDs. So no apparent problem with the transformer.

I want to take the psu to work tomorrow (too late to do much else....I'm getting sleepy!). I want to look at the actual pfc signals, then see if the pwm signal turns on when at full pfc voltage......need to see signal coming out of the 6802, then signal coming out of the two transistor driver output (primary of driver transformer), and signal at each output winding of the driver transformer. In short I want to be sure the signals to the switching transistors is clean before looking at the switching circuit. So progress is slow, but for now its bedtime! Good night...................

[budm]

Re: Delta DPS-300AB-47B........dead

You may want to check the resistance of the output pins of the IC that drives the MOSFET's also, check between ground pin and output pins, output pins and VCC pin.

[everell]

Re: Delta DPS-300AB-47B........dead

I took a look at the voltages today, and readings looked much better.....don't know what was wrong with that voltmeter last night! (or the voltmeter operator!!!) Voltage at the bridge was 163 volts and voltage after pfc was 388 volts. Vcc to the chip was 13.55 volts and Vref was 7.61 volts. Here is what the PFC signal looked like on the base of the FET transistor. Looks pretty good to me.

[everell]

Re: Delta DPS-300AB-47B........dead

And here is what the signals look like to the two switching transistors (transistors out of the circuit).

Incoming signals look OK to me.

[everell]

Re: Delta DPS-300AB-47B........dead

After sitting on the shelf a while, I am back to fixing this one. I decided that perhaps one of the original FETs simply died. I have a couple of unknown 20N60 FETs, so put them in for a try. Power up, blew fuse and FET in same location. I don't think this is a problem with the FETs.

Measuring some other components, found that the snubber diode was measuring short. Since it is across the switching transformer primary, I removed the diode because I might be measuring the transformer winding instead of the diode resistance. Well.....the diode still measures shorted. The diode is a BYV26C. According to data sheets, this is a 600 volt 1 amp ultra high speed diode. Looking about, I have a leftover package from another project of some MUR2100EG ultra fast diodes rated at 1000 volts 2 amps. So I installed a MUR2100EG and the two surviving unmatched 20N60 FETs. Using a light load of one HDD and one CDROM, powered up, and......................................it works!

I ordered a new pair of 20N60 FETS to achieve better matching. Installed them and everything still working fine. All voltages looking good. I see no reason to change the snubber diode.

Now for a big question. Why just a diode for a snubber? Usually there is a resistor and capacitor paralleled, in series with a diode. Is using just a diode a "cheap" solution, or is this OK for a circuit using APFC and two stacked FETs as main switching transistors?