ATX Power supply - startup sequence?

[gmc]

I've got a faulty ATX SMPS that is faulty and I"m trying to understand how it's supposed to startup.

Watched a few videos explaining the layout of power supplies and their block diagram but still unclear on a few things.

Amazingly I found a schematic for the power supply:

https://diodnik.com/wp-content/uploads/2017/08/iso-450pp.gif

When the power supply is off there is only standby 5V on the purple connector. This is expected and provides 5V  to the  TPS3510 (The power supply supervisor which monitors 5V/3.3V  voltage) From  my understanding (I could be wrong) when this starts up (PS-ON connection goes low when pressing the PC poweron switch)  this enabled the TL494 (PWM controller) which then drives Q3/Q4, which then in turn drives the main power transistors (Q1/Q2) which produces +-12/+-5 which is then fed back to TPS3510 to monitor.

What I can't understand is the catch 22 - on startup there is no +-12/+-5 so how does the TPS3510 turn on the TL494 to turn on the rest of the circuit.

The issue I'm seeing is there is 5V SB, and 14V to the PWM but no output from the PWM to the driver transistors.  I'm thinking it's faulty but first need to work out how it's supposed to be turned on without the voltages to begin with.

[magic]

D24, R56?
This seems to be coming from the standby transformer.

[wraper]

What I can't understand is the catch 22 - on startup there is no +-12/+-5 so how does the TPS3510 turn on the TL494 to turn on the rest of the circuit.

Read supervisor datasheet.

When PSON goes from high to low or low to high, the 38-ms debounce block is active to avoid a glitch on the
input that disables/enables the FPO output. During this period the under-voltage function is disabled for 75 ms
to prevent turnon failure. At turnoff, there is an additional delay of 2.3 ms from PSON to FPO.

[eblc1388]

The TL494 should be fully powered and is generating PWM pulses internally, even with the PS is at standby. What prevents its output pins C1 and C2 driving external transistors Q3 & Q4 is the signal coming into Pin4, the DTC(deadtime control) pin. This signal is controlled via the TPS3510. A high voltage at this pin would force the duty cycle to 0%.

If you monitoring the voltage on this DTC pin to ground, you should see a high level(more than3.3V) when the PS is at standby. When you physically push the PC start button, this voltage should drop to near zero to allow the output pins C1 & C2 to drive Q3 and Q4. If the voltage do drops but there is still no output, check the pulse transformer T2 pin8 for voltage, and components R80, D9 and R11, then Q3 & Q4 and components around their base.

[gmc]

Thanks for the explantion, this make sense. I've done a few more tests, the TL494 is always powered as you mentioned, via D24/R56

Power supply supervisor is working as well. The FPO pin is normally 4.6V and drops to 0.4V when power button is pressed.
There is still no output on C1/C2. I'll double checked all the components  connected to the TL494 but it's looking like it's dead Jim.

Might take it out and put some simply circuiry together on a breadboard to see if I can get some PWM on the outputs.

[wraper]

First thing to do would be overriding supervisor output or checking it's not blocking operation.

[gmc]

Good tip.....The plot thickens.

I've removed D15 so the super visor can't disable the PWM.

This has brought it to life albeit with low voltages. Measuring 4.5V on the 5V rail, and 11.6V on the 12V raiil. So now seems the low voltages are most likely causing the supervisor to disable the PWM. 

Read up a fe more threads and a possible cause is bad caps. I don't see any leaky or buldging caps but there might be some with high ESR.  Guess it's time to get my hands on a ESR meter.

[gmc]

Success - fixed it.

Started measuring the caps and one of the 10uF caps was reading 6uF - replaced and it's fixed the issue 

Thanks for the help all. A great learning experience.

[T3sl4co1l]

Wow, what an odd image... it's not a white background, strangely enough, but it's not even solid color, looks like error diffusion dithering... but to colors!?  And there's plenty of colors left to fill up the 256-color palette, it's not like they couldn't have left it solid... I wonder what the hell led to that?  ("Web safe" or Windows palette maybe..?)

Simple enough fix, and saving as PNG, greatly reduces file size.  Not that a meg or two matters these days, just... weird that that exists at all.

Tim

[magic]

 

Another confirmation that electronics don't fail, capacitors do.
Sometimes mechanical switches and relays too