Delta DPS-800GB Power Supply No Power

[mockingbird]

Maybe Delta underspec'd the parts required in that location... This is the first time I've heard about KZH and KZE not lasting...

[Blerog]

I will have a look and see what the voltage across those caps is tomorrow. you could be right.

[261Jack]

Originally Posted by Toasty

Bad startup cap(s). First clue was NO standby voltage.

Everything is Ruby's or Chemi-con's except the bloody one that fails which is an 100µF Ltec. Reads open, as in pF instead of µF...

Nice job shootin' yerselves in the foot there Delta. jeeze...

Fans need to be removed first. Then the power plug needs to be separated from the chassis. It is caulked in place -underneath- the socket. Joy. Xacto knife time. It then needs to be pulled out so its' wires can pass through the slot in the chassis while the main board is removed.

It's a real bastard of a cap to get to because of the cramped conditions, and some super-fine wiring going to the daughter board. Trust me, you don't want to screw with them. (hair fine in Teflon® insulation)

This is a double-sided board, so lots of heat (air) plus the hot vacuum desolder tool.

The heat sink that needs to come out has a finned spreader that spans several components and must be removed first. Then the heatsink complete with the bridge, thermistor, and switchers on it. Once you get the thing loose and out a bit, loosen the bridge screw a bit and and swing it aside to expose the screw holding the thermistor. Remove the screw & nut and detach the thermistor so it does not get damaged.

Replace both caps that are in the heat-shrink tubing, next to the daughterboard. (You're there, do you -really- wanna go back later for 50¢ worth of caps???)

Reverse the above to reassemble.

T

T - My DPS-800GB turns on SOMETIMES - Pin 31 is connected to pin 34 - The fuse is good - I replaced the two [large] 270 mF caps - There is a click [I assume a relay] when the PS does start - I see only one cap [c232] next to the daughter board

Your commits please, Jack

[261Jack]

Originally Posted by Toasty

Bad startup cap(s). First clue was NO standby voltage.

Everything is Ruby's or Chemi-con's except the bloody one that fails which is an 100µF Ltec. Reads open, as in pF instead of µF...

Nice job shootin' yerselves in the foot there Delta. jeeze...

Fans need to be removed first. Then the power plug needs to be separated from the chassis. It is caulked in place -underneath- the socket. Joy. Xacto knife time. It then needs to be pulled out so its' wires can pass through the slot in the chassis while the main board is removed.

It's a real bastard of a cap to get to because of the cramped conditions, and some super-fine wiring going to the daughter board. Trust me, you don't want to screw with them. (hair fine in Teflon® insulation)

This is a double-sided board, so lots of heat (air) plus the hot vacuum desolder tool.

The heat sink that needs to come out has a finned spreader that spans several components and must be removed first. Then the heatsink complete with the bridge, thermistor, and switchers on it. Once you get the thing loose and out a bit, loosen the bridge screw a bit and and swing it aside to expose the screw holding the thermistor. Remove the screw & nut and detach the thermistor so it does not get damaged.

Replace both caps that are in the heat-shrink tubing, next to the daughterboard. (You're there, do you -really- wanna go back later for 50¢ worth of caps???)

Reverse the above to reassemble.

T

T - My DPS-800GB turns on SOMETIMES - Pin 31 is connected to pin 34 - The fuse and fans are good - I replaced the two [large] 270 mF caps [nothing changed] - There is a click [I assume a relay] when the PS 'does' start - I see only one cap [c232] next to the daughter board.

Your commits please, Jack

[261Jack]

[QUOTE=261Jack;619662]T - My DPS-800GB turns on SOMETIMES - Pin 31 is connected to pin 34 - The fuse and fans are good - I replaced the two [large] 270 mF caps [nothing changed] - There is a click [I assume a relay] when the PS 'does' start - I see only one cap [c232] next to the daughter board.

More DPS-800GB prodding - - -
I have the PS completely out of its case and sitting on my work bench - Every time I apply power I hear a faint click - If it does start, there is a much louder click [I think from a near-by relay] - Applying power from 80 to to 140 VAC makes no difference; neither does the line/neutral polarity - The PS starts an average of about 1 out of 4 attempts -

All help is appreciated, Jack

[Behemot]

Post some pics, there are none of this unit here.

Do you have stable +5 V SB rail?

[Toasty]

Did you replace the cap that I wrote about in what you quoted...?

T

[261Jack]

Originally Posted by Toasty

Bad startup cap(s). First clue was NO standby voltage.

Everything is Ruby's or Chemi-con's except the bloody one that fails which is an 100µF Ltec. Reads open, as in pF instead of µF...

Nice job shootin' yerselves in the foot there Delta. jeeze...

Fans need to be removed first. Then the power plug needs to be separated from the chassis. It is caulked in place -underneath- the socket. Joy. Xacto knife time. It then needs to be pulled out so its' wires can pass through the slot in the chassis while the main board is removed.

It's a real bastard of a cap to get to because of the cramped conditions, and some super-fine wiring going to the daughter board. Trust me, you don't want to screw with them. (hair fine in Teflon® insulation)

This is a double-sided board, so lots of heat (air) plus the hot vacuum desolder tool.

The heat sink that needs to come out has a finned spreader that spans several components and must be removed first. Then the heatsink complete with the bridge, thermistor, and switchers on it. Once you get the thing loose and out a bit, loosen the bridge screw a bit and and swing it aside to expose the screw holding the thermistor. Remove the screw & nut and detach the thermistor so it does not get damaged.

Replace both caps that are in the heat-shrink tubing, next to the daughterboard. (You're there, do you -really- wanna go back later for 50¢ worth of caps???)

Reverse the above to reassemble.

T

Okay, I'm back on this project after a LONG recess. I have done every thing you mentioned down to removing the finned spreader. Neither one of the heatsinks will budge. Is there needed component desoldering that you did not mention?

[Toasty]

IIRC - (It was 4 years ago...) The heatsinks have attachment points/legs that are soldered to/through the board. Did you unsolder those legs?

FWIW - $19 incl shipping on ebay with warranty. Nice to fix but... :P

T

[261Jack]

Originally Posted by Toasty

IIRC - (It was 4 years ago...) The heatsinks have attachment points/legs that are soldered to/through the board. Did you unsolder those legs?

FWIW - $19 incl shipping on ebay with warranty. Nice to fix but... :P

T

Thanks, I'm looking for them now. How do I find you on ebay?
Jack

[Toasty]

???

Search: DPS-800GB

T

[Rainbow]

I have this PSU but with a different problem. Standby works but main power does not start. With the PSU installed in a server (2nd PSU removed), power button lights orange. When pressed, power supply fault LED illuminates.

The two offending caps seem to have good ESR in circuit. Could they cause this problem?

[Rainbow]

The answer is no. Removed the heatsink (soldered on both ends, also Q804 and Q808 must be desoldered as they're screwed on) only to find that the two hidden caps are not crap. C820 is Nichicon, C904 is Rubycon.

[Rainbow]

I've also changed KN4401S (ZUA) transistor on control board as seen here:
https://www.rom.by/forum/Problema_s_...-800GB?page=16
(use Google Translate)

It tested bad (one PN joint open) before replacing. But replacing did not help. Server powered up for a moment, then shut down. Looking with cover open, saw that one fan does not spin!

Removed fans, connected to 12V - both worked fine. Connected back, tried again - server runs. The fans GND and 12V pins are connected in parallel on the control board - so it must be the fan, randomly not working.

[PeteS in CA]

If the PSU is controlled by a microcontroller, it might be sensing a fan failure and shutting down the PSU. Is the harness to the fan 4-wire (+12V, Return, PWM, and tachometer)? It may be that the fan is not receiving the PWM signal or is not providing a tachometer signal.

[Rainbow]

Yes, most server PSUs have fan failure detection and will shutdown instead of overheating and catching fire. The fans are connected using 3 wires: 12V, GND and RPM sense (likely).

[PeteS in CA]

Originally Posted by Rainbow

Yes, most server PSUs have fan failure detection and will shutdown instead of overheating and catching fire. The fans are connected using 3 wires: 12V, GND and RPM sense (likely).

You probably could search that (Delta?) fan model and learn the function of that third wire.

Modern server PSUs are basically microcontroller (uC) controlled. At least some have a uC on the primary side controlling the PFC and reporting status data to the uC on the secondary side (which controls regulation and monitors for fault conditions). Because these uCs are programmed by the PSU manufacturer, replacing a uC cannot fix this sort of PSU unless one has a programmed part.