Starpower with bad capacitors

[everell]

Here is a Starpower Model PIV550 (peak load 450 watts), probably better known as a POS not really worth messing with. However, I accepted the challenge. The 5VSB was working, but the power supply was otherwise dead. I began by using a scope to look at the output signals on the 7500 PWM chip. No signals. Next I looked at the Vcc on the 7500 pwm chip which was a pulsing signal rather than DC voltage. I determined that the filter capacitor on the 5VSB feedback line was totally open. Replaced this 470 uF capacitor and the power supply came back to life, but no power good signal. So I decided to do a capacitor study on this power supply.

All of the output capacitors were Sus'con, and all of the small capacitors were Fuhjyyu. Strange combination! One of the output capacitors on the 3.3 volt rail was Sus'con 2200 uF/10 volts but measured 1480 uF with esr 0.09. Replaced it with a Jamicon pullout measuring 2350 uF esr 0.02. All other Sus'con capacitors on the outputs measured OK for capacitance and esr.

ALL of the small Fuhjyyu capacitors were a different story. I used measured pullouts for replacements. What I found was as follows:
1.0 uF/50 volts measured 2.16 uF esr 5.2 replaced with Jamicon measured 1.09 esr 2.9
1.0 uF/50 volts measured 1.33 uF esr 8.0 replaced with Jamicon measured 1.06 esr 2.9
10 uF/50 volts measured 11.4 uF esr 1.9 replaced with Teapo measured 10.53 esr 0.75
2.2 uF/50 volts measured 2.25 uF esr 32 replaced with Jamicon measured 2.31 esr 1.9
4.7 uF/50 volts measured 3.94 uF esr 13 replaced with Nichicon measured 4.65 esr 0.53
4.7 uF/50 volts measured 4.41 uF esr 10 replaced with Nichicon measured 4.43 esr 0.55
4.7 uF/50 volts measured 3.70 uF esr 18 replaced with Nichicon measured 4.58 esr 0.59

5vsb output capacitors Sus'con:
470 uF/10 volts measured 316 uF esr >99 replaced with Sus'con 1000 uF measured 930 esr .06
470 uF/10 volts measured 364 uF esr .42 replaced with Sus'con 1000 uF measured 930 esr .06

-12 volt output capacitor Sus'con
470 uF/16 volts measured 282 uF esr 28 replaced with Chemicon measured 474 esr 0.01

-5 volt output capacitor Sus'con
470 uF/10 volts measured 340 uF esr >99 replacement not logged

Swithcing transistor base driver capacitors Fuhjyyu
1.0 uF/50 volts measured 1.62 uF esr 11 replaced with new Nichicon 4.7 uF
1.0 uF/50 volts measured 2.05 uF esr 7.3 replaced with new Nichicon 4.7 uF

As can be seen, many of the Sus'con capacitors were in good shape, but ALL of the small Fuhjyyu capacitors were very bad. This demonstrates that replacing the bad bulging large capacitors but assuming that the small non bulging capacitors are good is a bad idea.

I also added line filter coil and capacitor. Now I don't have to call it a cheap POS.

[jiroy]

Re: Starpower with bad capacitors

You reminded me of one damaged old monitor . Last week , I decided to retrieve all its electronic components for spare parts . In the process , while de-soldering with a 80 watts Iron , at least three to four small capacitors got out with one leg only . I could see clearly the other empty rusty leg hole ..

[everell]

Re: Starpower with bad capacitors

Yes, little capacitors can be real trouble makers. What surprised me was that all of the 1 uF capacitors increased in capacitance value with aging. I would have expected their capacitance to decrease over time.

[Agent24]

Re: Starpower with bad capacitors

High leakage current fooling your capacitance meter?

[everell]

Re: Starpower with bad capacitors

Has anyone else worked on a StarPower psu lately? After a recap and adding a line filter I think this one should be good for at least 150 watts. What do you think?

[momaka]

Re: Starpower with bad capacitors

After a recap and adding a line filter I think this one should be good for at least 150 watts. What do you think?

Yes, should be okay for 150-200 Watts continuous, depending on the capacity of the input and output filter caps. The heatsinks definitely look decent enough.

The only thing I will pick on is the large "dummy load" resistors near the output capacitors. Anything dissipating more than 2/3 of a Watt I would either remove or replace with a different value resistor. Otherwise your "new" replacement output caps might get cooked soon too.

In general, I recommend no less than 15 Ohms for the 3.3V rail (22 Ohms or more preferable), 47 Ohms for the 5V rail (68-100 Ohms preferable), and 270 Ohms or more on the 12V rail (I usually do 500-1000 Ohms, or none at all on newer PSUs that have heavy 12V rails) - all of this if the resistors are rated for at least 1 Watt of power. This output resistor modding will lower the heat on the secondary side capacitors while also improving PSU efficiency,

Other than that, looks good.

I'm guessing the burned PCB is from the opened output capacitor on the 5VSB. I've seen that happen many times. And it's exactly like you said - 5VSB is unstable, which makes the secondary side auxiliary rail that supplies Vcc power to the PWM controller also unstable, and that brings the whole PSU down.