Opto-isolator CTR degradation

[AlfBaz]

My understanding is that through out the lifetime of an LED it's output decreases and consequently in an opto-isolator the current transfer ratio should drop.

I'm trying to repair an offline smps. Initially it had no output but soon found a bad cap on the hot side was preventing the switcher IC from getting enough Vcc to operate.
Now I find the output "hunting" and have determined that the output voltage is ramping up but cutting out on overcurrent through the switching mosfet.

The switch control IC on the hot side gets feedback through an opto and was wondering if it's plausible that the opto's CTR has degraded enough to cause the control IC to drive too hard

I have a bunch of components off board at the moment so I can't take any measurements. There is a trim-pot and reference chip near the opto's input so maybe all I have to do is adjust that to compensate

[Bertho]

Although possible for the opto-coupler to degrade, it is not a common problem. If you drive them to the spec limit, then they will follow the "normal" degradation curve; lower drive will normally extend the degradation curve. You could check the transfer-ration off-line because the hot side may have damaged the transistor and the secondary may have over-driven the led..

Since you already found one bad cap, I think you need to check the capacity of the caps on the secondary side. It could well be that they lost a major part of their capacity.

[AlfBaz]

Thanks Bertho and welcome to the forum

As I mentioned, I have components off board, and that's what I am doing, checking the caps and active components. All of the electrolytic caps are within tolerance and have nominal ESR. I have even checked the ceramics for abnormal leakage. So far nothing found, unless there's something wrong with the transformer.

I have struggled to find any generic data showing CTR degradation over time and this is why I was asking if it was plausible. The device the smps is out of is from the early 90's but I have no way of knowing how much of that time the unit has been on or if that timespan is long enough for reasonable CTR degradation

I'm in the process of putting it all back together and will measure the output to see at what level it rises to before it cycles

[dfmischler]

Since you already found one bad cap, I think you need to check the capacity of the caps on the secondary side. It could well be that they lost a major part of their capacity.

Actually, the secondary caps ESR (equivalent series resistance) is usually much more critical than the capacitance.

[Bertho]

Actually, the secondary caps ESR (equivalent series resistance) is usually much more critical than the capacitance.

The ESR is important, but there are several issues here. The capacitance is important because it determines ripple. The transformer in a smps is a current device and will push the current into the capacitor. When you are in the balanced situation, where the load is removing the same amount of current than the secondary is pushing, then you only have the ESR to worry about (the capacitor only sees the ripple current). However, when the psu is turned on or the load changes, then the output capacitor needs to do a bit more and swallow the current pulses.

The real importance lies in stability. If the capacity falls by 50%, then your ripple rises with a factor 2. That may seem little, but can cause the loop-gain of the psu to pass into an unstable region and go into shutdown. If the ripple becomes too large, then a single current pulse may drive the feedback to signal overvoltage.

[Bertho]

Thanks Bertho and welcome to the forum

Thanks, glad to be here.

The device the smps is out of is from the early 90's but I have no way of knowing how much of that time the unit has been on or if that timespan is long enough for reasonable CTR degradation

For a 20 year old psu it can certainly be possible to have degraded components. It does depend a bit on how much it has been running.

With a PSU this old, are you testing it with a load attached? Many of the "old" ones need to have a minimum load (on all independent outputs) for them to function properly. If there is no load, then they will shutdown, often due to output overvoltage.

[AlfBaz]

With a PSU this old, are you testing it with a load attached? Many of the "old" ones need to have a minimum load (on all independent outputs) for them to function properly. If there is no load, then they will shutdown, often due to output overvoltage.

That never occurred to me. I may well have fixed it when I replaced the original bad cap!

I will indeed try it shortly, thanks

[AlfBaz]

Thank you very much Bertho, loaded up the +-5V 10A outputs with an amp each and away it went

[SeanB]

I have had optocouplers fail, but generally they will fail as no go. As they age and lose sensitivity you will find the output voltage will climb slightly as the led gets turned on harder to get the same regulation point. Generally the failure is going to be either the led open circuit or the transistor going open circuit ( same result in both cases but in the first there is no led current) and stopping it working.

[Zero999]

I agree the CTR will degrade over time but if it's designed so the normal forward current is below the maximum rating and there's enough gain, it should compensate for it as the emitter will be driven harder to compensate for any degradation.