Optoisolator - choosing the right one

[everell]

I was working on a 5vsb circuit in a Bestec power supply, and thought that perhaps the optoisolator was bad. I had some pull outs in my junk box......but which would be the best to use? I built a 5vsb circuit on perf-board with an eight pin socket for plugging in the optoisolator being tested. All had the same output with no loading, but as additional loading was added, the differences began to show.

Most of the optoisolators tested were either 817 or 817C. I found a data sheet for the Fairchild series and found that the differences were due to the Current Transfer Ratio (CTR). I would have intuitively thought that the higher the CTR, the better the part. Just the opposite turned out to be true. I had one 817A, and it was my star performer.

I purchased from Digikey some Fairchild optoisolators to test, ten each of the 817, 817A, 817B, 817C, and 817D. Since the CTR range for the 817 was so broad, I figured it was probably floorsweepings and leftovers. So I tested the A, B, C, and D parts.

Test results can be seen in .pdf attachment. I concluded that the A was the best part when using a pwm chip in the 5vsb circuit because the feedback current needed is down in the microamps. The C part was probably the best for the two transistor circuits because the controlling transistor would need feedback in the milliamps.

[everell]

Re: Optoisolator - choosing the right one

Here is a picture of the test jig I used for testing the optoisolators. The load resistors are on the right. The 5vsb circuit is on the left. The VIPer22A circuit board is just barely showing on the far left. Comparing the optoisolators without resistor loading does not show any differences.

[LDSisHere]

Re: Optoisolator - choosing the right one

I was surprised by your discovery that the different revisions of the 817 optoisolator would affect the performance of your circuit. I had always figured there would be a difference between optoisolators that had completely different markings from the 817 and wondered if they could be easily interchanged, but I never have given any thought to replacing a 817 with whatever other 817 I had available at that time. (I have a attached a couple of pictures of optoisolators that are not marked 817.)

If they have the manufacturer marked on them then finding a datasheet may not be difficult which would allow you to directly compare characteristics. The ones that do not have an easily identifiable manufacturer could prove problematic to find a datasheet which would seem to make figuring out what to use as a replacement a crap shoot.

I suppose it is a good thing these devices do not go bad very often.

[Khron]

Re: Optoisolator - choosing the right one

Well, it may or may not be difficult. Fortunately, on Alldatasheet you can search for part numbers with wildcard beginnings and ends

The black ones are NEC PS2561's, and the white ones seem to be Vishay TCET1109's.

[everell]

Re: Optoisolator - choosing the right one

Khron666 thanks for the helpful information on Vishay. I have made a cross reference for the main three numbering systems on the optoisolators commonly found on computer power supplies. There are many manufacturers not mentioned, but i think most will be using one of these three numbering systems. This chart shows the differences in CTR (Current Transfer Ratio). Most of the optoisolators I have seen fall into the 200-400% catagory.

[budm]

Re: Optoisolator - choosing the right one

I ran into the same problem with the same model of Opto bewtween Fairchild and Liteon, the Fairchild out performs the speed and rise time compare to Liteon. The prototype and fist production run works great with Fairchild, then we switched to Liteon then the problem starts showing up.

[Khron]

Re: Optoisolator - choosing the right one

I actually wasn't aware LiteOn were in the semiconductor business PSU's and optical drives yes, but never knew they were into such "small scale" stuff...

Was the price difference really "worth it"? As dad says, "the shortest route is the one you know", so in some cases, it might be better to stick with what you know