Hi, I’m looking at the continuous allowed isolation voltage (‘endurance voltage’) for the LOC117 analog optocoupler. The data sheet just mentions 3750Vrms as Absolute Max spec.
If I look at other similar chips like the Avago HCNR200 series, there’s all kinds of restrictions in time duration and regulatory rules that sometimes restrict the ‘Working Voltage’ to only a fraction of the specified maximum.
My question is, is the 3750Vrms of the LOC117 allowed in actual, not time-restricted, operation?
Is there any additional info available on this subject for the LOC117?
Jan
Welcome to the forum.
That data sheet is terrible, I would choose a different manufacturers product if I really needed guaranteed isolation above 500V.
Alternatively if I was going to be buying 10,000 pieces for a design, and the price was too good to pass up, I’d request more information, in writing, from the manufacturer.
Hello,
Welcome to the Digikey tech forum. I put in a request to our product management group to see if we can get more information on this. I will post the information when I get it.
Thanks Paul, my requirement is for 4000Vdc continuous …
The usual suspects like the HCNR200 and IL300 all have restrictions based on Regulator code.
There’s an Avago test report saying that none of their tested units at 85C, 3000Vdc failed after a 5000hrs test, which is somewhat reassuring.
But I’d rather had some pertinent info from the LOC117 manufacturer. Then again, they may not ever have tested it. Yes, the DS is terse and not exactly confidence inspiring …
Jan
Steve, I would very much appreciate if you could get some more info on this.
I could not find any direct contact for Littelfuse support.
Jan
Then this part definitely is not for your application (their absolute maximum is less than your continuous)
But 3750Vrms is ~5250 V peak, no?
That’s less than my 4000Vdc.
Jan
1 Like
You’re right, I missed the rms units, Doh!
1 Like
Roses are nice, but in cases such as this it’s more useful to stop and smell the context.
Optocouplers are used to provide a protective barrier. When a protective barrier fails in consequence of being overwhelmed by an excess of the thing it was meant to protect against, the implication is that Bad Things Are Going To Happen. A company called Oceangate made headlines last summer by demonstrating this concept.
Dielectric breakdown of insulating materials is not a particularly easy thing to predict. It has dependencies on time, applied voltage, whether or not said voltage is monotonic or reversing, temperature, humidity, and probably also what a person ate for lunch last Tuesday. The effects can be nonlinear, cumulative, and/or synergistic. And there’s also that pesky manufacturing tolerance issue…
Any given optoisolator coming off the manufacturing line might be destined for a single-use disposable application, or it might go into a piece of equipment that’ll be in continuous operation for the next 50 years. About the only thing that can be counted on is that People Will Become Angry if the insulation barrier fails.
50-year tests are expensive and somewhat inconvenient to conduct, and since actual results are dependent on other factors beyond a manufacturer’s control anyhow, isolation ratings are more or less established by educated guesstimate through extrapolation and declared assumption, then backing off a couple of turns.
That sub that went to visit the titanic and ended up staying seems a rather profound analogy for the topic of isolator voltage withstand properties and lessons on how the matter ought to be approached. Withstanding a given stress once or for a time does not necessarily imply the capacity to do so repeatedly or indefinitely, and one is well-advised to avoid the realm of proximate failure when its precise location is unknown and the consequences of its encounter dire.
2 Likes
The point is, you cannot count on one of these devices to function reliably if one regularly applies a voltage anywhere remotely close to the peak voltage specified. Don’t expect a part rated for 5300V peak to work reliably above 700V or 800V. They aren’t meant to operate continuously above that.
1 Like
I have been running a couple of HCNR200’s at 2000Vdc for a few years now with zero issues.
I am also reminded of that Avago paper where they tested a bunch at 3000Vdc for 5000hrs with zero failures.
From the available lit the picture arises that it is the dV/dT that determines the failures.
In my application, the 4000Vdc is a constant and the AC between the input and output is just a volt or so of audio bandwidth signals.
It seems that I can’t get around to doing my own long-term testing.
Unless someone has some pertinent test or specification data we haven’t seen yet.
Jan
2 Likes
Lets see how it goes. starting with 3kV.
Jan
2 Likes
Hello jan.didden,
The 4000 VDC continuous is a very difficult target.
May I recommend you consider using fiber optic to pass the signal from the ground up to your high voltage section. I once worked on a cycloconverter that used this technique. As I recall, the SCR trigger PCBs floated with the AC line voltage at 6 to 8 kV peak above ground.
CAUTION! High voltage isolation isn’t a specification typically found on the datasheet for fiber optic cable. This would make your design an off-spec application. Perhaps the manufacturer could provide the data, but I rather doubt it as it requires extensive long-term tests to prove reliability. For example, what type of leakage can we expect in a condensing environment.
As a starting point to explore fiber optics I recommend the IF-E10 kit. However, in no way am I suggesting this is suitable for a high-voltage applications as that information is not in the spec.
On a related note, you may want to build you PCB with structures to shunt the high voltage. Some ideas include neon lamps, spark gaps, or even closely spaced “rake like pointy” structures in the PCB.
Best Wishes,
APDahlen
FB105-ND is another kit to also learn about fiber communication.
About the fiber optic solution - I am concerned about the ‘fidelity’ of the signal transmission.
This is about audio at the -120dB level of harmonic distortion.
Jan
1 Like
My informal test on the HCNR200 still runs, no issues.
BTW I noted that this optocoupler has a modified DIP08 package - the row spacing is 0.4in versus 0.3in for a regular DIP08.
Anyway. I found another solution requiring two 6kV coupling caps. No more ticking opto-timebombs required.
Thanks for the education guys!
Jan
1 Like
