Fortron FSP300-60ATV repair

[everell]

Back to the Power Q/Fortron power supply. After recapping, time to apply power and see what the 5vsb is doing. Nothing. Since it is a two transistor circuit and been through high temperature, I decided to simply gut the two transistor circuit and install a DM311 pwm chip. Put my LED tester on the main power supply connector and applied power. 5vsb LED came on.....success! So I grabbed my voltmeter to check the 5vsb voltage, and was surprised to find that it measured 17 volts. Yes, 17 volts.....a "regulated" 17 volts. Looks like the protection provided by the DM311 is overcurrent or overpower protection, NOT overvoltage protection. This is the first time I have ever seen a DM311 circuit go overvoltage.

The two key components to set the voltage level are the 431 regulator chip and the optoisolator chip. I decided to change the optoisolator chip first. Using a pullout optoisolator from another power supply, I installed it and applied power. Now it measures 5.2 volts. Now I am confident the 5vsb is fixed, no critical capacitor to worry about, and it will probably run for a long time. Now that the 5vsb is fixed, I can move on to troubleshoot the rest of the power supply.

[everell]

With 5vsb up and running, I turned the switch on (shorting green wire to black wire). Nothing. So I began by looking at the FSP3528 chip. Pin 1 Vcc was 5.2 volts. Pin 17 ground was zero volts. Next, pin 12 Vref (should be 3.5 volts) was only 1.2 volts. Pin 6 PS-ON (should be about 5 volts) was zero volts. No need to go any further. To turn the power supply on, PS-ON must go from 5 volts to zero volts, or nothing will happen. If PS-ON is already at zero and nothing is happening, then the chip is bad. I looked at pins 19 and 20 (pwm signal outputs) and no signals were present.

The FSP3528 is a propriety chip, so no way to order from Digikey. But I had a chip on the burned out board in my very first post in this thread. So I removed it from the burned out board and installed it on this board. Applied power, switch on, and the LED lights on my power supply tester came on. See attached picture.....the power supply is alive again, the LEDs are on, and the FSP3528 from the picture in my first post gets a second chance to operate in a power supply.

[Wester547]

So DM311 doesn't provide overvoltage protection? Fairchild's datasheet claims it does. Unless it's making reference to something else? Or is a second 5VSB optocoupler needed to pull that off with a crowbar circuit in tow (SCR)? The datasheet does show 20V for OVP, though.... (But I believe that refers to the primary side and not the output when it comes to 5VSB)

[everell]

The overvoltage protection that the data sheet is talking about is NOT the output of the 5vsb circuit. It is talking about the Vcc voltage used to operate the chip. If you increase Vcc to above 20 volts the chip will indeed shut down. The operating voltage of the 5vsb circuit is determined by a 431 voltage regulator which drives the input of the optoisolator chip. The output of the optoisolator provides the feedback which determines the pulse width output of the "pulse width modulator" function of the chip.

You will see that there is a voltage divider on the circuit to be regulated. This voltage divider drops the voltage to the 431 regulator to about 2.5 volts. At 2.5 volts, the 431 chip causes the optoisolator to provide the right amount of feedback to the DM311 such that the 5vsb is held at 5 volts. By changing the value of these two resistors, you can set the output voltage value to whatever is desired. For instance, a 431 regulator chip is used on many power supplies to regulate the 3.3 volts.

The problem in my power supply was not the voltage divider used to set the 5vsb operating point, not the 431 regulator chip which was doing its job, but the optoisolator was defective and providing the DM311 with the wrong feedback voltage. The DM311 used this incorrect feedback value to set the 5vsb output to the wrong voltage level. So my 5vsb had a well regulated 17 volt output. Replacing the bad optoisolator fixed the problem and now the 5vsb is 5.02 volts.

[Wester547]

Yes, it makes more sense that the datasheet is giving specification to the VDD/VCC voltage of the chip itself and not something that could be customized by other means. So then there really isn't overvoltage protection against a bad +5VSB optocoupler (or for that matter a bad pair of resistors that divide the voltage and a bad 431 shunt in that circuit)? Or would that need to be established on the secondary side a la what's found in the ATX-250-12Zs, ATX-300-12Zs, ATX0300D5WCs, some Hipro PSUs, etc, a SCR gate/circuit on the secondary side (what causes the PSU to latch) activated by zener diodes and the optocouplers (or does it not work that way?)? I remember you posted a thread that detailed out an account of a "flakey" ATX-250-12Z but you were able to short +5VSB to +12V and the power supply shut down immediately, indicating it has proper OVP.

It makes me wonder, though, since PWM chips by themselves on the primary in substitution of a two transistor +5VSB circuit are not enough for guaranteed protection (I understand that PWM generated +5VSB circuits do have some sort of decent regulation whereas two transistor circuits rely on a critical capacitor for regulation and that can be distastrous).... how many OEM power supplies with PWM generated +5VSB circuits actually have proper protection for +5VSB? I do know some early Fortrons, IE the FSP300-60BTV, used KA1H0165R to generate +5VSB (and some Hipros). I'm starting to wonder if they have the proper OVP for +5VSB as well, or for that matter if the other DM311 equipped Fortrons do.

[everell]

Good questions. I think I have already said too much this weekend.....looks like I just passed 1000 posts on this forum this weekend! Hopefully I have said something useful and am not just blowing air.

[mockingbird]

You're certainly not blowing air... Always a pleasure to read your fascinating posts. Keep up the good work.

[everell]

Continuing with my story - repair of the Q Power/Fortron FSP300-60ATV. I was experimenting, trying to optimize the resistance in series with the diode in the optoisolator. The original circuit had 390 ohms. I tried 220 ohms, 270 ohms, and 470 ohms. With no load, it made no difference. The output was always 5.20 volts. But when using a 2.5 ohm load resistor, the 390 ohm original value gave the best (highest voltage) at 4.92 volts. I then used various load resistors between 10 ohms and 2.5 ohms. That's when the problem began. Each time I took readings, the voltage values were fluctuating. So I thought that maybe my output capacitor replacements were not adequate. I replaced them again. No improvement. So I replaced the optoisolator, then the 431 regulator chip, then the rectifier diode. Nothing seemed to help stabilize the output voltage. Finally I changed the 5vsb transformer. I found a transformer in a CasEdge Model F300-A which had an identical Fortron pc board and the transformer part number markings were the same. Voltage now stabilized.

Now to power up and check the 12 volt, 5 volt, and 3.3 volt rails. To my disappointment, the power supply was dead. More troubleshooting, and I determined that the FS3528 pwm chip was dead. Ironically, it was reading the same voltages as the original FS3528 that failed. Perhaps the bad transformer was a chip killer. As the pictures show, this power supply got very hot, especially around the main switching transistors and the 5vsb circuit. I am thinking that the heat caused the transformer wiring to begin arcing internally when under a load.

With the new transformer, the power supply came back to life. To test it dynamically, I hooked it up to an old Pentium 3 Biostar M6VLQ mother board. So far, working great.

[TELVM]

Great thread, thanks to Everell and everybody!

Modern PSUs use an integrated PWM controller/switching transistor IC for +5VSB regulation, like the Infineon ICE2QR1765 in the Seasonic G-550. Do these things have OV protection to save the mobo from disaster?

[intrax]

I would like to revive this thread as I'm trying to repair my FSP200-60ATV psu...

Power supply is starting and all rails are present, but 12 volt rail is low at 10.8 volt. Looks like pwm is working as I can see it throttling between 10.8 and 10.9 volt.

I suspected the optocoupler or pwm chip (FSP3528). The optoisolator measures ok in circuit, but the chip doesn't have the correct voltages as I see here https://www.badcaps.net/forum/showthread.php?t=22171 . Voltage on vcc pin is 5.2 io 4.99. Could the supply circuit to the chip be faulty ?

How to proceed ? Buy new FSP3528 ?

[Behemot]

have you put some small load on the rails whether it helps?

I doubt you'd find that customised chip anyways (perhaps a used one, with luck).

[intrax]

I managed to get the rail up to 12 volts by adjusting the variable resistor 201 (orange in picture) on the output rail, which I initially thought was only for the 5 volt rail.

FYI: The fsp-3528ic can be found on aliexpress no problems.

Thanks your reply!

[Behemot]

I see you still have original craps in there and also the yellow glue. Remove all those shit caps, clean the glue especially between electrodes of components (could be there are some small currents running where they are not supposed to, if it already degraded to a state with high carbon volume) and replace with quality low-ESR caps.

Do have large stock of caps suitable for these units if you wanna.

[intrax]

First up is recap of my dell gx-280 mobo that died. This is a spare power supply for testing so I'm not bothering atm. Are you using Nichicons ?

[Behemot]

Handful, couple Rubycons, Pana, Samxons, majority Chemi-Cons as I'm working with a distributor who offers reasonable conditions (MOQs etc.).

And they have great product seminar each year, definitelly taking swimwear next year for all those hot pools and stuff X-D