SMPS troubleshooting

[theozaurus]

Hi,

I'm trying to repair a SMPS. The output voltage on the +8V, -8V is double what it should be. The +12v output and -12v output are correct, but only because there are linear regulators on the input, the input voltage to the linear regulator is 32v! This looks double what it should be to me.

The circuit is attached. I've taken meassurements on the secondaries of the transformer windings. The attachments have the pin number of the transformer in the filename. I've also taken meassurements after the diodes. I was expecting to see this oscillate around to provide continuous feedback via U906, however it looks totally steady (I'm sure I've got completely the wrong end of the stick). The input to the IC902 fed by R928 is a steady 2.72V.

I've meassured R928, R929, R930 and they all seem fine.

The datasheet for the TOPY244 can be found here: http://datasheet.octopart.com/TOP244YN-Power-Integrations-datasheet-337.pdf

What on earth is going on?

[wraper]

Measure voltages across TL431 and optocoupler. Most likely it's either of them faulty. It could also be C925, C923 or D910.

[theozaurus]

Thanks wraper,

So I've meassured the anode and cathode of the U906 (optocoupler) with respect to ground and used the math function to get the voltage across the U906. See attached.

It looks like the voltage across the TL431 is 2v, and the voltage across the optocoupler is 1.25v. I'm not sure what the expected behaviour of this area is. What do you think?

[theozaurus]

Just pulled:

 - C923 meassures as 45uF with ESR of 0.66?
 - C925 meassures as 5.2uF with ESR of 2?

So they both look good. D910 meassures 0.488v with the multimeter. These all look good to me.

[dave_k]

The optocoupler might be struggling to supply enough current to the C pin (but still enough to prevent it going into shutdown).

What is the voltage on pin C?

[wraper]

IMO optocoupler is dead.

[wraper]

Just in case, check if R925 resistance has not increased.

[theozaurus]

@dave_k: The voltage on the control pin is 5.8v
@wraper: The resistance on R925 is 199 Ohms (meassured in circuit)

I have a spare optocoupler so shall try swapping that in later today.

Thanks for all the suggestions. Keep them coming!

[dave_k]

Hmm .. 5.8 is about correct according to the datasheet.

[xavier60]

The Control pin clamps at 5.8v. Current into the Control pin controls the switching duty cycle.

[SeanB]

I would still change C923, 925, as they are cheap enough, and new ones will perform better. Optocoupler though is low transfer ratio ( really common as they age) so needs replacement. Another thing that might be faulty is the TL431, which is also a cheap part easy to replace.

[theozaurus]

@seanb. Replaced the optocoupler and TL431, the 8v output is now correct. Thanks very much. There's another fault on the PSU I haven't described yet that I'll put in another thread.

Thank you everyone for your help!

[David Hess]

Optocoupler failure is common enough that I have seen off-line switching power supply designs which included backup primary side voltage regulation to prevent the output voltage from rising out of bounds and damaging something.

[xavier60]

Sharp says that the LED emission degrades with time,

"?Circuit design
(1) The LED used in the Photocoupler generally decreases the light emission power by operation.
In case of long operation time, please design the circuit in consideration of the degradation
of the light emission power of the LED. (50%/5years)
"
Optocouplers can be damaged by soldering over temperature.

[David Hess]

Sharp says that the LED emission degrades with time,

They do but this is beyond that.  They just seem to exhibit complete failure more quickly than other parts such that they limit operating life.  I assume this has something to do with their packaging because I have yet to find a failed optocoupler built from a separately packaged emitter and detector.

[SeanB]

They do but this is beyond that.  They just seem to exhibit complete failure more quickly than other parts such that they limit operating life.  I assume this has something to do with their packaging because I have yet to find a failed optocoupler built from a separately packaged emitter and detector.

You never had the misfortune to own or repair older HP laser printers, where the Sharp made optical switches are rather notorious for going low emission from the LED with time, leading to intermittent false jam indications and stopping of the print cycle. The cure, aside from a new switch, was in many cases to run the LED at a higher current, which got the coupling up enough for a long time. A lot of them use a red LED so that the light is visible for servicing,  though there are a lot with IR leds as well. I have fixed some that were either out of stock or which were simply unobtainable by using another optoswitch and swapping the LED in the package, helped by them often being a bespoke design and using the LED leads as part of the connector as well. others there is a small PCB which is thankfully a common part number amongst various switches of otherwise massively different style on the mechanical side.

[xavier60]

Delta must have ignored the design advice about the degradation of LED emission which resulted in a bunch of short lived TV power supplies.
 Ironically, these power suppliers powered some Sharp LCD TV models, sending a lot of TVs to an early grave when stocks of the power supplies ran out prematurely.

[theozaurus]

I never knew optocouplers where so sensitive. I've replaced the remaining ones on the board to (hopefully) give it a bit more life! I'll try and get some plots from the transformer windings of it working properly incase anyone is interested as well.