Dell PS-5251-2DF2 missing -12V line

[everell]

Re: Dell PS-5251-2DF2 missing -12V line

Some of the glue on your supervisory board is yellow, and some almost black. Definitely one of your problems.

Kaboom.......where are you when we need you. Tell this man about glue on power supplies!!!

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Well, after rewinding the -12v coil, and measuring the toroid transformer out of the pcb, there seem to be no shorts.

One bit of information : without the -12V coil connected , voltages of 5 and 12V are in spec ( 5.12 and 12.2-12.4 ) , but the 3.3V is strange ( 2.7V ) .

With the complete toroid coils, voltages go out of spec ( 4.7V, 12.7V, 0.94 to 1.7V , -10.6V ) . I changed another cap, a 220 x 25V one, but probably will ressolder the old caps, just to be sure it is not one of my new caps that is causing it.

One part of information : it seems that while I was thinking about the toroid, with too many people in the shop offering their "advice" [ to be honest, it was the boss ] , I tried the psu without that second coil to the left, that belongs to the 3.3V circuit. Could this have damaged something ? I know it was probably foolish, but well, he is the boss.

Here are the pics of the supervisory board. Most of that black area is glue. Will clean it all and test the resistors, also changing the transistor. Could not find an A1023 here, possibly the MJE350 is a suitable one ?

[everell]

Re: Dell PS-5251-2DF2 missing -12V line

I couldn't understand the circuit either. I simply replaced the A916 transistor, the 431 regulator, and the other bad parts in drawing, and the psu started working fine. I spent much time repairing and analyzing that thing. Later, I repaired a couple more by replacing that A916 transistor on the supervisor board. I like the Bestecs much better. Easy to understand, and when fixed right, work well!

[momaka]

Re: Dell PS-5251-2DF2 missing -12V line

I ran across the notes I made at the time I was fixing my psu like yours. Looks like the A916 was not the only bad component. I don't remember looking at the -12 volt rail, but the problem did cause problems with the 3.3 volt rail. So take a look and see what you think!

Is it me, or does this PSU look stupidly over-engineered?
I was looking at the TSM111 datasheet and noticed this on the Pin Description:
ADJ (pin 4) - "Adjustment pin for 3.3V OVP. This pin is to be used for an OVP other than 3.3V (eg for uC power supply = 2.7V). When not in use, this pin should be grounded. When in use, VS33 should not be connected."
But according to everell's schematic of the supervisory board on the previous page, VS33 IS connected. So what the hell did Lite-ON do here? ADJ is obviously not grounded and if anything, seems to be related to the 12v rail and that weird circuit which contained the burned A916 transistor.
Also, from everell's supervisory board schematic, you can see that UV (pin 11) on the TSM111 IC is connected through a 100 ohm and another resistor in parallel to pin 3 on the supervisory PCB. Now if you trace pin 3 back on the underside of the PSU, you will see that it goes to that small orange capacitor C206, which in turn is connected to the AC output of the -12v rail. The description for UV (pin 11) of the TSM111 says: "Undervoltage detection, control and detect main AC voltage failure".
But, pin 3 on the supervisory board is also connected to ZD381, a resistor, 2 2222A transistors, and ADJ on the TSM111 IC.

So, I wonder if this circuit could be simplified...
Would have been helpful if the TSM111 datasheet (both SGS Thomson and ST Microelectronics) was a little better. Whoever drew the sample circuits on those datasheets didn't do a very a good job. On that sample circuit, UV (pin 11) is connected to a zener diode (which seems to be bypassed by a wire?) and then is connected both to ground and the 3.3v rail at the same time. Seriously?

Anyways, perhaps if we ground ADJ and regulated UV to 1.26v, then maybe those complex circuits can be removed completely? Any comments on this? Because I really don't see what all of those complex circuits are doing there. I know the power supply was working fine before and I don't blame the design, but if many of these power supplies start failing due to that A916 transistor, I think the supervisory board should be modified (read: simplified). Both the sample circuit and the evaluation board seem much more simple.

[everell]

Re: Dell PS-5251-2DF2 missing -12V line

I ran across the notes I made at the time I was fixing my psu like yours. Looks like the A916 was not the only bad component. I don't remember looking at the -12 volt rail, but the problem did cause problems with the 3.3 volt rail. So take a look and see what you think!

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

[QUOTE=momaka;171442]Yes.
Check that none of the windings for each rail shows a resistance to any of the windings for the other rails. If this is not it, rewind the toroid with a few extra turns for each winding, especially the 12v rail, since it's a bit higher.
Might be a good idea to save the original wire from the toroid, too, if it's not too late already.

Also, what capacitors did you use for replacement (brand and series)? This could definitely cause a problem if the ESR of the new caps is either too low or much higher than the originals.

[QUOTE]
Thanks,

I believe they show some resistance, yes. Will get another DMM and some new batteries just to be sure. The wire for the -12V, that is the only I messed with thus far, is still here.

The original OST ones measured the capacitance ok. I substituted them with some new Hitanos we had here, more to see if it would make some diference. But the voltages kept the same, with the OST and with the Hitano. They changed only after I messed with the toroid coils. Still don´t have an ESR meter, but I`m working on getting / building one.

I´m waiting for a shipment of Panasonic FC, as I understood from the forum that they could be a good substitute for most caps in a PSU. After fixing the other more pressing problems of this psu, will change most , if not all, with them.

PCBonez,
The voltages were correct even without load ( excepting that pesky -12V ) , before I messed with the toroid . Now, they are wrong without load, and when I put a load ( an old hard disk we found in a drawer ) , the psu won´t start, and I need to disconnect it and remove the jumper in the atx connector for it to start again. Will get another, known good hard disk, so that I not be mislead for a maybe shorted "load".

Thanks.

[PCBONEZ]

Re: Dell PS-5251-2DF2 missing -12V line

Well, now I have -12V. That is, all voltages are somewhat wrong now :

12 -> 12.6V
5 -> 4.7V
3.3V -> 1.35 V
-12V -> -10.5V

As the 3.3V was reasonably good before I interfering with the toroid, I believe a good toroid, or maybe just rewinding it all, would improve things here.

What do you think, is it normal, that is, problems in the toroid could make the 3.3V regulation goes crazy like this ?

If you don't have some load on it then the regulation will be erratic.
[Some won't even start.]
An old hard drive is usually enough.
.

[momaka]

Re: Dell PS-5251-2DF2 missing -12V line

What do you think, is it normal, that is, problems in the toroid could make the 3.3V regulation goes crazy like this ?

Yes.
Check that none of the windings for each rail shows a resistance to any of the windings for the other rails. If this is not it, rewind the toroid with a few extra turns for each winding, especially the 12v rail, since it's a bit higher.
Might be a good idea to save the original wire from the toroid, too, if it's not too late already.

Also, what capacitors did you use for replacement (brand and series)? This could definitely cause a problem if the ESR of the new caps is either too low or much higher than the originals.

Well, got another diode, just in case. UF5408, 3A, 150A reverse, trr 50ns.

If the original -12v diode wasn't bad/shorted, just put it back in. The UF5408 would also work, though.

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Well, got another diode, just in case. UF5408, 3A, 150A reverse, trr 50ns.

Been reading about the tricks to hand-wind toroidal transforms, will need to practice a little before trying to do this one...

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

I will put it back, and search for a more adequate replacement later.
Actually, as I have to buy wire, will go out and search for it now...

All the other voltages were present. 5V, 12V, 3.3V, 5VSB, all normal.

Going back from 7912 pin 2, it goes to the pin of the questionable toroid.

Now, as I´m trying to learn from this psu, and the good sense says something in the toroid should be bad ( the coil for that -12V read open, and shorted to the 5V line, some resistance between the 5 and 12V lines ) , I went and tried to rewind that coil. After removing the wire, it still showed some resistance between 5 and 12, but I made the experience anyway.

The wire I had here is a little thicker, so was not a good fit, and I will rewind better later, after seeing if it makes some difference.

Well, now I have -12V. That is, all voltages are somewhat wrong now :

12 -> 12.6V
5 -> 4.7V
3.3V -> 1.35 V
-12V -> -10.5V

As the 3.3V was reasonably good before I interfering with the toroid, I believe a good toroid, or maybe just rewinding it all, would improve things here.

What do you think, is it normal, that is, problems in the toroid could make the 3.3V regulation goes crazy like this ?

[Toasty]

Re: Dell PS-5251-2DF2 missing -12V line

Diode 1F4 is not a correct replacement.

1F4 = 1A, 25A surge, 150ns reverse recovery (trr), Fast recovery diode

UG2D = 2A, 80A surge, 25ns reverse recovery (trr), Ultra-fast recovery diode

Unless UG2D was reading open/shorted/leaky, it's likely okay.

Again, +12v present? Other voltages?

Trace 7912 pin2 (input) back and find which component has 12 to 25 volts present on other side of it to find problem. Use pin1 as your ground(+)

Toast

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Diode test ok ( UG2D 0432 ) , put a 1F4007 in its place.
Pin 1-2, 0.55V
Pin 1-3, 0.72V
Pin 2-3, 0.18V

Also changed two 2200 x 6.3 caps in the 3.3V line, a 470 x 25, a 1000 x 16 and 47 x 50. After removing all the glue, will change the transistor.

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Also, I may sound like a scratched CD, repeating myself in an annoying way, but again: check the -12v rectifier! Should be 1 or 2 axial diodes.
Based on the underside pictures you posted, the -12v rectifier should be right next to the toroid, close to the edge of the board and that small yellow ceramic cap C206.

Thanks Momaka, I will remove and test then again. Onboard, with the DMM diode test, they worked ok. One of them I have removed, but it teste ok, so I put it back. Better, I will remove and change them ( I believe there are three, will post a photo later ) .

Please-
+12v present?

1) Load PSU and test.
2) Measure from Pin1 to Pin2 on 7912, -not- from system ground. Voltage?

Circuit uses +12v as an inverted supply where +12 is ground.

Thanks Toasty, will do. Even with the toroid transformer in this strange condition? Well, I will change some caps from the 3.3V output line and ressolder the toroid & 7912 to take those measures. Lets see if this psu comes back to life .

[Toasty]

Re: Dell PS-5251-2DF2 missing -12V line

Please-
+12v present?

1) Load PSU and test.
2) Measure from Pin1 to Pin2 on 7912, -not- from system ground. Voltage?

Circuit uses +12v as an inverted supply where +12 is ground.

[momaka]

Re: Dell PS-5251-2DF2 missing -12V line

This is a transformer. It might be your standby power supply transformer, which usually provides standby +5 volts.

No.
This is the main filtering toroid/inductor for the 5v, 12v, and -12v rails. Depeding on the PSU design (and age), some manufacturers will have a winding even for the -5v rail.

Measuring continuity in the toroid coils, I got 0 ohms between pins 3, 4 and 6 ( thick yellow wire ) . Also 0 ohms between 2 and 5 (thick red wire ) .

Those are normal readings, so those windings are good.
The red wire on the toroid is for the 12v rail, and the two yellow wires are for the 5v rail.

But between 1 and 7, the pins for the -12V, I measure approximately 12 kohm.

Desoldering the thin wire from the toroid pins, one leg of it shows a short to the 2-5 coil, and the other leg seems to not be connected anywhere.

Not good. You should have continuity just like for the 5v and 12v windings.
My guess would be that when the small transistor shorted, it overheated the -12v winding on the toroid and caused it to melt.
Since one leg of it does show a short to pins 2 and 5 (the 12v winding), then you should rewind the toroid's 12v and -12v windings.
Better yet, if you have any other junk PSUs, find a toroid of the same size and rewind it so that it has the same number of turns for each rail as this one.

Also, I may sound like a scratched CD, repeating myself in an annoying way, but again: check the -12v rectifier! Should be 1 or 2 axial diodes.
Based on the underside pictures you posted, the -12v rectifier should be right next to the toroid, close to the edge of the board and that small yellow ceramic cap C206.

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Three main windings, one with thick golden wire, one with thick red wire, one with thin wire. ( the thin wire is connected to the -12V line ) .

I will recap the output of it, just for getting rid of those OST and Ltec caps.

[Longbow]

Re: Dell PS-5251-2DF2 missing -12V line

Well, it would be interesting to see a general schematic showing where this part appears in the circuit. This is a transformer. It might be your standby power supply transformer, which usually provides standby +5 volts. But it is obvious that you have 2 main windings, one with heavy wire and one with smaller wire. The only thing you can do is to check the resistance between the larger wire and the smaller wire - it should be open (infinite). Anything less than that would indicate a primary/secondary short. If it appears fine, it probably is. I don't see any burn marks. Put it back in.

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

Hi,

Here are some photos.

Measuring continuity in the toroid coils, I got 0 ohms between pins 3, 4 and 6 ( thick yellow wire ) . Also 0 ohms between 2 and 5 (thick red wire ) . But between 1 and 7, the pins for the -12V, I measure approximately 12 kohm.

Desoldering the thin wire from the toroid pins, one leg of it shows a short to the 2-5 coil, and the other leg seems to not be connected anywhere.

I still have to find the a916 transistor or an equivalent, but will solder the toroid back ( with the -12v coil disconnected ) to measure if there is voltage in the input side.

The red line in the foto is the path of the -12V, with the 7912 in the upper side. Input comes from the toroid, output goes to the supervisory board.

[rogfanther]

Re: Dell PS-5251-2DF2 missing -12V line

As far as looking at the resistance of the switching transformer windings, this method will not yield useful information. The d.c. resistance should be low, by design. Hint: the most often replaced part in a SMPS is the transformer. The part that fails least often is the transformer. (A possible exception is a CFL inverter, where a faulty switching device can fry the transformer, which is usually very tiny.)

Hi,

I will post some photos, so that I can explai myself better, but the problem ( I believe ) is not in the switching transformer. All the other voltages are ok ( less than 5% off their nominal values ) . The only that is wrong is the -12V, measuring 0.72 V. The transformer I believe is involved in the problem is the toroidal one, that couples all the voltages before they go out of the psu.

I´m still reading and studying to learn what it does and how it works, but in a bunch of trashed boards I have, I could always find at least two terminals for each voltage, and in this problematic one, the -12V doesn´t appear to connect to other pins, so probably I will remove it to measure correctly. The wire is different gauge and color from the others, so I believe it should not be so hard to identify.

Thanks.

[Longbow]

Re: Dell PS-5251-2DF2 missing -12V line

Interesting problem and very helpful advice given. It is always good to find a problem that affects hundreds or thousands of units so they don't have to be tossed. Transistor data sheets - this particular part seems like a general purpose PNP, but the votage rating is 120 Volts, which is much higher than the typical. It looks to me like the circuit deals with + and - 12 volts, so the selection of this part seems confusing. Because of the failure rate of the part, I would be inclined to use a different transistor here. Sometimes during startup, various parts may be subjected to much higher voltages than their normal operating voltage, and this may be an explanation.

As far as looking at the resistance of the switching transformer windings, this method will not yield useful information. The d.c. resistance should be low, by design. Hint: the most often replaced part in a SMPS is the transformer. The part that fails least often is the transformer. (A possible exception is a CFL inverter, where a faulty switching device can fry the transformer, which is usually very tiny.)