2

Most optocouplers spend most of thier time isolating mains voltages.They are reliable doing this and they are cheap and simple.How would things go if the same garden variety opto was isolating a DC potential ? A work associate did a HVDC supply in a previous life and found opto failure despite using the opto well below its ratings .He said that this was due to electromigration,he started life as a chemical engineer .Is this true ? Does this mean that optos should be derated for DC isolation? What would be a reasonable derating factor? Should one not use optos period for HVDC ? Would isolating a negative voltage be better than isolating a positive voltage due to electrochemical reasoning ?

Autistic
  • 14,235
  • 2
  • 27
  • 65
  • I don't think that most optoisolators are for mains voltages. – Scott Seidman Jan 25 '16 at 22:41
  • 3
    After many decades believing opto-isolators to be kind-of invincible I was surprised to learn from an industrial electronics repair house that some companies would, as preventative maintenance, send servo drives to them to have electrolytic capacitors replaced _before_ they fail due to age. The repair company told me that, as a matter of course, they would replace the opto-isolators too. I'd be interested to know the failure mechanism although some of it could be the natural fall-off of LED output after many years of 24/7 use. – Transistor Jan 25 '16 at 22:44
  • 1
    @ScottSeidman: The insulation layer between the LED and the receiver is good for > 1kV typically. That's where the voltage rating comes in and not, obviously, on the LED or transistor, although opto-triac isolators (LED -> triac) are mains rated on the triac side. – Transistor Jan 25 '16 at 22:47
  • We used to use a plastic fiber optic system in my last company. the light budget did mean that units would fail due to the LED output dropping.. seems the MTBF for an LED is an output drop by 50% for 10,000 hours. so an LED on for 24/7 can fail in 10 years but still produce quite a lot of light. But depending on the situation maybe not enough. – Spoon Jan 26 '16 at 00:54
  • I've heard of an electromigration issue in opto's running with HVDC before too, so I don't think its a myth. – brhans Jan 26 '16 at 03:13
  • 1
    Been in contact with an applications engineer? This is an interesting question, and I'd love to hear what the manufacturers have to say. – Matt Young Jan 27 '16 at 20:58

2 Answers2

1

If reliability is a problem get a part with an MBTF listed in the datasheet, that means that the supplier built it with longevity in mind and tested it. Generally ic manufacturers must live up to the specs in the datasheet (read the legalese printed on most suppliers datasheets). Once you have a part that is guaranteed to work for a long time, make sure it doesn't get damaged by running it outside of spec, overvoltage, ESD and that it has the proper soldering time at the PCB assembly house. If your really interested in reliability, then derate it and make sure that the parts specs are over-preforming in every way. LED's generally have a lifetime of 100k hours, I'm not saying that this applies to optoisolators but the physics are the same. Maybe you could build the design so its running condition runs with the LED off.

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
  • Its good to get answers +1 My rule of thumb is 100K hours. Dont worry the led wont be working much.What bothers me is the fact that DC could give electromigration or other nasty corrosive effects . – Autistic Jan 26 '16 at 00:32
  • An MTBF of 1,000,000 hours doesn't mean that any significant fraction of the parts will last 1,000,000 hours, but rather that if 10,000,000 parts which are neither too new nor too old are run for an hour, the number that fail within that time is unlikely to be much greater than 10. – supercat Jan 27 '16 at 21:24
1

There aren't any problems using optoisolators at high voltage DC. Most of them (i.e. popular PC817 and its clones) have isolation rated for a few kilovolts.

I've been working with high voltage electronic devices (rated for 3kVDC) from several years and I've never seen optoisolator with broken isolation. Most of damaged optoisolators which I've seen had short in input element (LED) or output element (transistor, triac) caused by overcurrent or overvoltage between its terminals, not by high voltage between input and output.

Jakub Rakus
  • 2,225
  • 5
  • 18
  • 26
  • +1 I will google the PC817 .Its reassuring that you have not had any problems .I only need 2KV peak and 500V average. – Autistic Jan 27 '16 at 21:14
  • @Autistic - 500V average? Huh! In my work I will say that this a low voltage ;) Look at any AC/DC converter, such like phone charger - there is 115/230VAC on input and 2kV spikes are normal in such lines - converters has optocouplers in feedback loop and everything works! – Jakub Rakus Jan 27 '16 at 21:24