Excellent power model IT ATX 300w - shuts down after a split second

[kotel studios]

Hi,

Recently I gave a shot at recapping this Excellent power IT ATX 300 lc clone doohickey PSU. After I replaced all bulging caps and the 47uF 50v in 5vsb I turned it on. 5vsb and ps_on are spot on. So I shorted the ps_on and it dies after a split second. 5vsb still is generating, all other lines also generate voltage for a while but they all discharge so fast.

Now I am sure this isn't a problem with the 5vsb damaging the PWM IC or BJT circuit, but I have no idea where to go next. Maybe replacing all the smaller caps?

[CapLeaker]

That sounds to me like the PSU is protecting itself. Post clear, high resolution, straight shot pictures (front and back) of the PSU and post them using the attachment function. Look at the supervisor IC and see what it is.

[kotel studios]

Look at the supervisor IC and see what it is.

By that you refer to the 2003 PWM IC or did lc use really use a supervisor chip?

[kotel studios]

Here are some pics. I also noticed the PG line is completely missing. Unsure where to go next...

[stj]

first check the output diodes for shorts

[прямо]

Here's an ATX power supply schematic that shows the pinout of the 2003 PWM controller + supervisor combo IC.

[momaka]

Oh Deer!

I see a lot of heat was generated on the PCB near the output toroid - probably from when the old caps failed would be my guess. In such cases, it's a good idea to check all of the solder joints on the PCB. I looked at the pictures you provided, and can already see what appears to be at least two bad/cracked solder joints (to the bottom left of the pins of secondary side of the main transformer.)

Also, as the others stated, check the output rectifiers (TO-220 diodes on the secondary heatsink) to make sure they are not shorted. To do that, you should use your multimeter set on resistance and measure the resistance between each major rail (3.3V, 5V, 12V, -12V) and ground. BEWARE that since this is a Deer/Allied/Solytech power supply, they often used very low resistance "dummy load" resistors on the output rails as a minimum load, which might make you wrongfully conclude that there is a short on one of the output rails when there is none. In most old Deer and L&C units I ran into, it wasn't uncommon at all to see something like 6.2-10 Ohms for the 3.3V rail, 15-47 Ohms for the 5V rail, and 100 Ohms for the 12V rail - all of these being large 2-3 Watt resistors right next to the output caps. I have seen many people who tried to troubleshoot these PSUs by using the "continuity" test on their multimeter on the output rails, and they got confused into thinking that there was a short-circuit / shorted output diode on the output(s). So keep that in mind when you do the resistance testing.

You say the PSU tries to briefly turn on but then stops. In that case, try measuring the voltage on each rail (3.3V, 5V, 12V, and -12V) to see if the voltage changes at all to some slightly higher value or not (even if it's the mV range.) Maybe even post the values here too. I had one PSU before that didn't want to turn on in the same manner. Long story short, the 3.3V rail output toroid appeared to be soldered normally, but wasn't making a good connection. I was able to double-check myself by running a continuity / low-resistance test from the Cathode of each rail's rectifier to the output wires.

Lastly, just a question of how you are testing the power supply: do you put any load on the output rails (5V and 12V) when you short the PS_ON signal to ground or are you running the PSU unloaded? In general, Deer/Allied/Solytech PSUs don't need a load to turn On and I never had an issue turning one on without a load. Nonetheless, always consider or try to put some kind of a small load on the 5V and 12V rail. Some people suggest old hard drive or two. I'm personally not a big fan of that method, as the HDD(s) don't put an "instant" load on the 5V and 12V rail. So instead, I prefer to use small 12V incandescent car/auto/home light bulbs (typically 5 to 20 Watt -rated.) This gives the PSU an "instant" load so the output voltages are able to balance better.

As for PG signal missing - it's missing because likely one of the output rails (3.3V, 5V, 12V, -12V) is either overloaded (under voltage) or showing over-voltage, or is simply not there at all (missing / open-circuit.) The PG signal comes up high only when all of the voltage rails have come up to proper values and are stable.

And yes, it's normal for the 5VSB to unaffected from failure of the main PS outputs (3.3V, 5V, 12V, -12V), since it's on its own circuit. PS_ON is generated from the 5VSB rail, so that will also be unaffected.

[kotel studios]

Oh Deer!

I see a lot of heat was generated on the PCB near the output toroid - probably from when the old caps failed would be my guess. In such cases, it's a good idea to check all of the solder joints on the PCB. I looked at the pictures you provided, and can already see what appears to be at least two bad/cracked solder joints (to the bottom left of the pins of secondary side of the main transformer.)

Also, as the others stated, check the output rectifiers (TO-220 diodes on the secondary heatsink) to make sure they are not shorted. To do that, you should use your multimeter set on resistance and measure the resistance between each major rail (3.3V, 5V, 12V, -12V) and ground. BEWARE that since this is a Deer/Allied/Solytech power supply, they often used very low resistance "dummy load" resistors on the output rails as a minimum load, which might make you wrongfully conclude that there is a short on one of the output rails when there is none. In most old Deer and L&C units I ran into, it wasn't uncommon at all to see something like 6.2-10 Ohms for the 3.3V rail, 15-47 Ohms for the 5V rail, and 100 Ohms for the 12V rail - all of these being large 2-3 Watt resistors right next to the output caps. I have seen many people who tried to troubleshoot these PSUs by using the "continuity" test on their multimeter on the output rails, and they got confused into thinking that there was a short-circuit / shorted output diode on the output(s). So keep that in mind when you do the resistance testing.

You say the PSU tries to briefly turn on but then stops. In that case, try measuring the voltage on each rail (3.3V, 5V, 12V, and -12V) to see if the voltage changes at all to some slightly higher value or not (even if it's the mV range.) Maybe even post the values here too. I had one PSU before that didn't want to turn on in the same manner. Long story short, the 3.3V rail output toroid appeared to be soldered normally, but wasn't making a good connection. I was able to double-check myself by running a continuity / low-resistance test from the Cathode of each rail's rectifier to the output wires.

Lastly, just a question of how you are testing the power supply: do you put any load on the output rails (5V and 12V) when you short the PS_ON signal to ground or are you running the PSU unloaded? In general, Deer/Allied/Solytech PSUs don't need a load to turn On and I never had an issue turning one on without a load. Nonetheless, always consider or try to put some kind of a small load on the 5V and 12V rail. Some people suggest old hard drive or two. I'm personally not a big fan of that method, as the HDD(s) don't put an "instant" load on the 5V and 12V rail. So instead, I prefer to use small 12V incandescent car/auto/home light bulbs (typically 5 to 20 Watt -rated.) This gives the PSU an "instant" load so the output voltages are able to balance better.

As for PG signal missing - it's missing because likely one of the output rails (3.3V, 5V, 12V, -12V) is either overloaded (under voltage) or showing over-voltage, or is simply not there at all (missing / open-circuit.) The PG signal comes up high only when all of the voltage rails have come up to proper values and are stable.

And yes, it's normal for the 5VSB to unaffected from failure of the main PS outputs (3.3V, 5V, 12V, -12V), since it's on its own circuit. PS_ON is generated from the 5VSB rail, so that will also be unaffected.

Answer to question 1: (diodes from closest to furthest to main transformator on the trace side of PCB picture)
D17 resistance:
GND - both anodes show 0 ohms, kathode 14ohms
5V - both kathodes 14ohms, anode 0 ohms
-12V - all pins 450K ohms
12V - all 1.4M ohms
3.3V - anodes 6.8ohms, kathode 21 ohms

D14 resistance:
GND - 0omhs everywhere
5V - kathode 730K ohms, anodes 14ohms
-12V - anodes 450K ohms, kathode 900K ohms
12V - kathode 0ohms, anodes 1.4M ohms
3.3V - anodes 6.8ohms, kathode 0ohms

D13 resistance:
GND - 6ohms kathode, all anodes 0ohms
5V - anodes 14ohms, kathode 21ohms
-12V - all 450ohms
12V - all 2M ohms
3.3V - anodes 6.8ohms, kathode 0ohms|

Answer to question 2:
Yes, they seem to rise, albeit not fully (that is to be expected)

Answer to question 3:
I put an 7v 35w bulb on the 12v line, but even with no load at all the PSU doesn't start. It's fans just twitch slightly. However I was able to make it go a little bit further due to replacing all of the crapacitors (apart from two near the middle transformer due to there being no way to replace without desoldering the heatsink).

As for the 5vsb rail I meant when the crapacitors go to being resistors and make the 5vsb kill PWM IC due to over-voltage (same thing that happened to my LC 300w psu IIRC)

[harp]

Check with test motherboard as load. I have same behaveour and befre to search problem, eliminate posibility... See end of video... http://www.youtube.com/watch?v=d7jXHcYf9wk

[kotel studios]

Check with test motherboard as load. I have same behaveour and befre to search problem, eliminate posibility... See end of video... http://www.youtube.com/watch?v=d7jXHcYf9wk

Nope, didn't work.... I have no idea at all what's the issue here, besides the PSU protecting itself.

[momaka]

OK, I can sort of make sense of the readings you did, but the way you did them and wrote them is causing some confusion on my end.
Luckily, I did buy a scrap PC last week that has this same exact PSU (except, it's the cheap L&C version, so there may be some small variances in the output caps and load resistors.)

In particular, for all of the "GND" readings on the diodes: the reason you get 0 Ohms from the Anodes to ground is because the main transformer windings are very short pieces of wire and thus measure like short-circuit in DC mode. In fact, if you measure resistance between any two output pins on the main transformer, you will see that they all measure 0 Ohms to each other, which is normal. So this reading is not really worth doing or showing.

Before interpreting the results in detail for you, note the appearance of these low-resistance readings in the results: 6.8 Ohms, 14 Ohms, and 21 Ohms. Also worth noting is that 6.8+14 is approximately 21 too.

D17 resistance:
GND - both anodes show 0 ohms, kathode 14ohms
5V - both kathodes 14ohms, anode 0 ohms
-12V - all pins 450K ohms
12V - all 1.4M ohms
3.3V - anodes 6.8ohms, kathode 21 ohms

OK, let's break these down:

GND: the reason you get 0 Ohms from the Anodes to ground is because of the low resistance between the main transformer windings, as noted above.On the plus side, at least this confirms the Anodes on D17 make proper connections to the 5V output pins of the main transformer.

5V: not sure what you mean by "both kathodes". There's only one Cathode on D17... so I guess you meant both Anodes? And for the Anode you meant Cathode shows 0 Ohms to ground? If so, that would make a lot more sense. The Cathode on D17 is connected to the 5V rail output wires, so that's why you get 0 Ohms. Meanwhile, the 14 Ohms from both Anodes of D17 is because its Anodes are connected to the 5V output pins of the main transformer, which (as was noted above) have 0 Ohms DC resistance to GND through the main transformer... and there's a low-value load resistor (probably a 15-Ohm) on the output of the 5V rail. So that's why you get this reading. Feel free to double-check this: you should see a large resistor near the 5V rail red output wires with color code brown, green, black, gold.

-12V, 12V: high-resistance readings, because the 5V rail does not have any connection to these.

3.3V: The reason the Anodes of D17 show 6.8 Ohms to the 3.3V rail orange wires is the same as why you saw them show 14 Ohms to the 5V rail red wires. Most likely, there's a 6.8-Ohm load resistor between the 3.3V rail and ground. Since Anodes of D17 = GND as far as your multimeter is concerned, you're essentially measuring between GND and 3.3V rail... and picking up that 6.8-Ohm load resistor. But again, feel free to double-check me on that.
As for the 21-Ohm reading between D17 Cathode and 3.3V rail orange wires - that's a little more tricky, but still normal. Now you are essentially reading across both the 5V rail and the 3.3V rail's output resistors as a series circuit. And for resistors in series, the resistance adds - i.e. 14 + 6.8 = approximately 21 Ohms.

With all of that said, I think we can give rectifier D17 a pass for now.

Next:

D14 resistance:
GND - 0omhs everywhere
5V - kathode 730K ohms, anodes 14ohms
-12V - anodes 450K ohms, kathode 900K ohms
12V - kathode 0ohms, anodes 1.4M ohms
3.3V - anodes 6.8ohms, kathode 0ohms

OK, I think you made an error either when you wrote down your readings or when you typed them back.
First, you can't have both the Anodes and the Cathode of D14 show 0 Ohms to ground, yet have them show different resistances for the other readings (like for example, how you got 14 Ohms from the Cathodes to the 5V rail red rails and 730 kOhms from the Anode.) The 0-Ohm resistance would imply that the Anodes and Cathode of D14 are shorted together. But your other results show they are not.
Also, how did you get both 0 Ohms from the Cathode to the 12V yellows wires and also 0 Ohms from the same Cathode to the 3.3V rail orange wires? That tells me you made an error with recording these results.
For better clarity, please try to stick to the same format throughout your results - i.e. put the Anode readings first, then the Cathode second... or the other way around, but just be consistent.

Since there are multiple errors, I can't tell if there's anything wrong or not with D14 rectifier. Probably not, based on the bulk of the readings, but please re-do for a sanity check. What I can tell you from my PSU is that D14 is the 12V rail's rectifier, so its Cathode should show 0-Ohms resistance to the 12V yellow wires and nothing else.

D13 resistance:
GND - 6ohms kathode, all anodes 0ohms
5V - anodes 14ohms, kathode 21ohms
-12V - all 450ohms
12V - all 2M ohms
3.3V - anodes 6.8ohms, kathode 0ohms|

OK, here it's the same interpretation as for D17 (the 5V rail rectifier).

GND: 6.8 Ohms between Cathode and GND, because you are reading the load-resistor on the 3.3V rail. The Anodes show 0 Ohms to ground, because of the very low resistance of the windings in the main transformer. So all is normal here.

5V: to your multimeter, Anodes = GND, so the 14-Ohm reading is the 5V rail load resistor (15 Ohms). The Cathode reads ~21 Ohms because now you have the 3.3v rail resistor (6.8 Ohms) read in series with the 5V rail resistor (15 Ohms)... i.e. the two add up. So all is normal here.

-12V: I'm guessing you meant 450 KOhms, as that's what you previously had for the 5V rail? So typing error??

3.3V: Anodes = GND, so you're reading the resistance of the orange wires to GND, which is the 3.3V rail load resistor (6.8 Ohms). And for the Cathode, it shows 0 Ohms to the 3.3V orange wires, because this is the rectifier for the 3.3V rail. So both readings are normal here.

Answer to question 2:
Yes, they seem to rise, albeit not fully (that is to be expected)

I was hoping you'd post the values you got. Yes, I understand they may be hard to catch and vary quite a bit between trials, but that's OK. It can be of help to know if the voltage on a certain rail is rising only a few mV vs. a few Volts.

Also, did you check & address the cracked solder joints I mentioned in my previous post? Those could well be the cause of the issue.
Again, it looked like the output toroid overheated quite a bit when the caps failed, so I would suggest re-soldering everything in that area around it.

As for the 5vsb rail I meant when the crapacitors go to being resistors and make the 5vsb kill PWM IC due to over-voltage (same thing that happened to my LC 300w psu IIRC)

Probably not the case here, since your PSU appears to try to turn On.

.
.
.

So just to recap here:
1) re-do the measurement for D14 and report back.
2) check the solder joints I mentioned... or better yet, re-solder as many things as you think look suspicious on the secondary side, then see if the PSU still refuses to turn On
3) remove the load resistors from the 3.3V, 5V, and 12V rails and do these tests:
--- a) red MM (multimeter) probe on 3.3V orange wire, black probe on ground. Report resistance value.
--- b) red MM probe on 5 red wire, black probe on ground. Report resistance value.
--- c) red MM probe on 12V yellow wire, black probe on ground. Report resistance value.
--- d) red MM probe on -12 blue wire, black probe on ground. Report resistance value.
--- e) red MM probe on -5V white wire, black probe on ground. Report resistance value.
^ Basically, a-e are a more direct way of testing if there are any shorted rectifiers. I'm almost certain from your readings that D17 and D13 are probably fine. But with the load resistors removed, we should be able to tell more clearly, as the low resistance of those load resistors sometimes can obscure the reading of the output rectifiers being checked properly.

[kotel studios]

Thank you for the reply!
Yes, I will redo the measurements (I get confused often when there's a ton of stuff happening at the same time, not sure why tho...).
I probably screwed some stuff up but I also noticed some wierd readings with my multimeter on D14, can't recall what rail it was but at first it showed 0 everywhere and after I probed again it was above 1k IIRC. It was somehwere between 5V and 3.3V rails. It'll probably show up once I get to re-measuring.

[stj]

you have to hold the meter in place for a while or capacitors charging can give you bogus readings

[kotel studios]

I have redone the measurements and GND on d14 is 0ohm anodes and 100ohms cathode.
I also redone the soldering on secondary side and replaced the original -5v filter cap (which turns out to be okay) but to no avail. Still dies after a second.
I also did the rail measurements as you asked:
5v - around 1v (MM jumps from mV to V then back to mV (expected))
3.3v - around 500mV
12v - around 5-8v

[kotel studios]

Okay I managed to remove the dummy load resistors and here are the results:
3.3v - 1k6Ω and rising
5v - 18kΩ
12v - 3m3Ω
-5v - 500Ω
-12v - 500Ω
There was one resistor missing either for -5v or -12v tho...
P.S. the soldering on the resistors cracked easly so if you can reflow them

[momaka]

I have redone the measurements and GND on d14 is 0ohm anodes and 100ohms cathode.

OK, that sounds a lot more like it.
On my L&C, there is also a 100-Ohm load resistor on the 12V rail, so that's what I expected to see on yours too.

I also redone the soldering on secondary side and replaced the original -5v filter cap (which turns out to be okay) but to no avail. Still dies after a second.

Good. Post a clear picture of the solder side, just so we can double-check it again. I find that no matter how many times I go over something, sometimes it's just better to have another person to check me to see if I missed something.

Okay I managed to remove the dummy load resistors and here are the results:
3.3v - 1k6Ω and rising
5v - 18kΩ
12v - 3m3Ω
-5v - 500Ω
-12v - 500Ω
There was one resistor missing either for -5v or -12v tho...
P.S. the soldering on the resistors cracked easly so if you can reflow them

OK, those resistance readings look good. The rectifiers are probably OK. But we will see.

I also did the rail measurements as you asked:
5v - around 1v (MM jumps from mV to V then back to mV (expected))
3.3v - around 500mV
12v - around 5-8v

Hmm... looks like the 3.3V rail might be having trouble, as it doesn't seem to not rise much in voltage. It uses what's called a "mag-amp" (short for magnetic amplifier) circuit that regulates power from the taps of the 5V rail on the main transformer. This mag-amp circuit is usually comprised of a PNP BJT transistor, a few small diodes, a TL431 current shunt regulator, and a saturation toroid (the smallest toroid near the main transformer.) The saturation toroid should be OK, but any of the other components mentioned could have gone bad and thus cause the 3.3V rail to not want to regulate properly. From what I can track from my PSU (so that you don't have to), there's no TL431 shunt regulator (it appears that it may be built into the "2003" PWM chip), but the other components are on the board, and these are: D25 (small diode, right next to rectifier D13 that you checked), transistor Q7 (an "8550" PNP BJT, close to the 5VSB wires), diode D26 (close to Q7), and resistor R54 (small, 100 Ohms 1/8W.) See if these are OK first.

If OK, let's do another test: back feeding voltage with a power supply. For this, you will need another working ATX power supply... or if not, then at least a 5V adapter and a 12V adapter (both capable of 1 Amp.)
But first, re-install all of the load resistors back into the PSU.

Now here's how to proceed:
Since the 3.3V rail seems to be the one that's not regulating (at least so far), take the working ATX PSU and connect its ground (any black wire) to ground of the non-working PSU.
Next, connect the 3.3V rail (orange wire) of the working ATX PSU to the 3.3V rail of your non-working ATX PSU via your multimeter set to high current mode (typically 10 Amps for most meters) - that is, connect the red probe to the working ATX PSU 3.3V rail and the black MM probe to the 3.3V rail of the *non-working* ATX PSU.
After this, turn on the working ATX PSU (by connecting PS_ON/green to ground) and monitor the current on the meter. It should be much less than 500 mA (0.5A) in less than a second (and probably in the few hundred mA range at most.) IF IT'S NOT AND MEASURES ONE 1 AMP OR MORE: STOP! (turn off the working ATX PSU) and report back that you encountered this. If this doesn't happen, continue below.
Now with the working ATX PSU turned On and and running (and supplying 3.3V to the non-working ATX PSU), try turning On the non-working ATX PSU as well by shorting its PS-ON/green wire to ground. See if this makes the non-working ATX PSU able to turn On and stay running. If you have a 2nd multimeter, it would be a good idea to check the voltages on the non-working ATX PSU if it's able to run.
-- If the non-working ATX PSU is able to run, then you probably found the problem... or at least that the 3.3V rail was the issue. If the non-working ATX PSU still does not run with voltage being fed back to its 3.3V rail, repeat the same setup above, but for the 5V rail - i.e. connect 5V rail / red wires on working ATX PSU to the 5V rail / red wires of the non-working ATX PSU via multimeter set to high current mode. Then turn on the working ATX PSU and monitor current to make sure nothing tries to draw too much power (again, you should see maybe 100-200 mA... or 0.1 to 0.2 Amps, whichever way your MM displays it). If all seems OK, turn on the non-working ATX PSU and see if it comes up. If not, also repeat this test for the 12V, -12V, and -5V rails. Just NOTE that when you do this test for the 12V rail, you will see the non-working ATX PSU's fan turn On, since it's powered from the 12V rail. Don't let that fool you that the non-working ATX PSU may be working, when it may not be.

Now, if you don't have a working ATX PSU to do the tests above, you can still do these tests with a 5V and a 12V adapter. The wiring is all the same. Only the 3.3V rail is more tricky - for that one, use two large(r) regular or fast-rectifier type diodes (but *not* Schotky type) rated for 3-5 Amps. Basically, put the two diodes in series, with the first diode's Anode connected to the 5V adapter positive (+) and its Cathode connected to the 2nd diodes Anode. Then connect the 2nd diode's Cathode to the 3.3V rail of the non-working ATX PSU. This will essentially give two diode drops (about 1.4V to 2V) from the 5V of the 5V adapter to the 3.3V rail of the non-working ATX PSU. So the 3.3V rail of the non-working ATX PSU will see about 3 to 3.6V - hopefully enough to make it happy. But again, for this to work, make sure that the load-resistors are re-installed in your PSU.

After doing the above tests, report back if you were able to make the PSU turn On or if it still dies the same way it did before.
Also report back your findings on the 3.3V rail's mag-amp circuit component measurements, if possible.
Hopefully all that will reveal where the issue might be.

you have to hold the meter in place for a while or capacitors charging can give you bogus readings

Excellent tip!
I should have mentioned this to O/P from the start.

[kotel studios]

Here are the results of 3.3v line components you wrote:
8550 PNP - hFE184, Uf 785mV (componenet tester I forgot the model of)
D25 (d11 on mine) - 220mV both ways (in circuit), 500mV out of circuit
Ill post the back feed results later. I can't really do good solder side pics cause the light gets reflected off the pads and whole quality is meh.

[kotel studios]

I have no idea why but I get no amps on the MM. Using an adjustable PSU but that shouldn't matter. Black probe to patient atx and red to adjustable psu. Black lead is connected to ground of patient.

[kotel studios]

D26 and an 100Ω R55 resistor close to 3.3v cable from 2003 pwm are fine.

[momaka]

[

D25 (d11 on mine) - 220mV both ways (in circuit), 500mV out of circuit

Just verifying: you get 500 mV on D25 only in one direction, correct?

I have no idea why but I get no amps on the MM. Using an adjustable PSU but that shouldn't matter. Black probe to patient atx and red to adjustable psu. Black lead is connected to ground of patient.

Did you add back the load resistors for all of the rails? The 3.3V rail had what appears to be a 6.8 Ohm resistor... so with that, you should see approximately 3.3V / 6.8 Ohms = 0.48 Amps of current going to the Excellent PSU. If that's not the case, double-check your multimeter connections. Most multimeters have a separate jack for the high-current setting, so you need to move the red probe into that jack (typically labeled "10 Amps"... or however many Amps your meter is rated for.) Black probe always stays in COM jack, though. So double-check your connection and try again.