TDA8380-based SMPSU can't start under load

[philpem]

I'm trying to fix a switchmode power supply from a piece of broadcast equipment (an NTL/DMV/NDS/Tandberg System 3000 satellite receiver) which won't start under load.
The PCB bears the legend "MADE IN UK PROFESSIONAL PSU RD.S.5497PCB.D MOD 4", with "5497-0401" and "SSL9601 5497 05416" on sticky labels near the output connectors.


The issue is that it fails to start when there's a load greater than 2.3A on the digital 5V rail, but works fine below that.
The motherboard in the equipment pulls about 2.5A at startup - which is why this is an issue...
If I let the power supply start up (either no-load or with a ~2A load) then connect D5V to the motherboard, everything works fine.
When the PSU is running normally I see around 6V across the slow-start capacitor; when it's stuck trying to start up I see around 1.4V.

The satellite receiver itself works fine when tested with an ATX power supply providing the 5VD rail -- or indeed providing all the rails (with a few external regulators to generate 33V and 8V)



The PSU is a TDA8380-based thing with a bunch of output rails -- +33V, 13V/18V switchable (LNB), +12V for the fans, +8V, +/-5V analog, +5V digital.
From what I can gather, it's based on the Philips 70W reference design (http://educypedia.karadimov.info/library/APPCHP4.pdf page 70) but an AVS1BC automatic voltage switch and triac have been added to allow it to run from 110V/230V without needing an input voltage switch.

Initial failure on arrival was a blown line fuse, bridge rectifier stage (all four diodes blown) and a blown SGSIF344 chopper transistor (dead short across all three terminals). I later found a dead 22uF 25V gate drive capacitor (C16 in the reference design) and a few more shorted diodes which were also replaced.

I've replaced the SGSIF344 chopper, bridge rectifier diodes (and some other dead diodes), the startup capacitor (C5) and the gate drive capacitor (C16). The TDA8380A was also replaced (with a TDA8380) to resolve an issue where the PSU went out of regulation after a few minutes operating (I guess it was damaged when the chopper failed).
I've also tested the other diodes and ESR tested the capacitors, all are good.

I had to replace two series diodes (A-K--A-K) which were across the current sense resistor; I've replaced these with 1N4001s but I'm not sure what they're intended to do.

If I try to probe the current-sense line (pin 13) with a DMM, the power supply shuts down and starts chirping (second-level slow-start mode).


Does anyone have any ideas why this thing might be failing to start under load, but work fine if it's started up on a lower load?

Cheers,
Phil.

[KhronX]

Might be worth increasing the value of the start-up cap, perhaps?

I've encountered a couple smaller flybacks that wouldn't start even unloaded with a 10-22uF cap in that position, but started up just fine with a 47uF.
You'll of course want to use a low-ESR one there.

[philpem]

Do you mean the capacitor on pin 5, with the high value resistor to the positive rectified HV?

That's a 330uF 35V already (I wish I was joking!), with 2x47k series (94k) to the HV+. It takes a second or two after AC is applied for the PSU to try and start.
I swapped in a Panasonic "FM" 105C low-ESR on the last repair round.

[xavier60]

The diodes across the CS resistor are to stop the CS resistor fusing and causing even more damage when HV appears across it.
The CS resistor might still get damaged. Its value needs to be accurately measured.
 I regularly test CS resistors by passing a constant current through them and measuring the voltage drop.
Also check other components connected to pin 13.

[philpem]

Thanks for the hint about the current sense circuit -- I found the problem!

I measured the current sense resistor at 0.5 Ohms, but it turns out there weren't two 1-ohm resistors in parallel. There were FOUR. Two chip 1206 type and two quarter-watt axial type.

Of course, the two chip resistors were blown open-circuit. Several K-ohms on the meter.

Swapped those and the power supply came up perfectly with a 3A static load. Took it up to 4A on powerup and it still came up fine (albeit with a very loud squeak), so I put it back in the receiver.

Works like a dream! I need to hook up a HackRF and set up a test signal, but the debug LEDs and status OSD are doing vaguely sane things so I'm confident!