>
>
> > I'll probably write this up in more detail sometime, but I had a recent
> > experience where about 25 VDC managed to cross
>
> a barrier rated at 2000 + or VDC.
>
> > When the primary side catches fire so that the isolated IC is incinerat=
ed
> > and an arc forms across the broken primary circuit and then finds a nic=
e
> > carbon char path into the primary side 'fun things can happen'.  And di=
d.
> >
>
>
>
>
>
>
>
>
>
>
>
>
> * This is a reason why an optocoupler by itself isn't a good way to
> isolate between different voltages in safety critical applications. If yo=
u
> design for Ex, you need to put one or two zener diodes and one fuse to
> limit voltage and current into the LED side of the OC in order to safely =
be
> able to maintin isolation between the solid material inside the OC. You
> also need to limit energy through the output transistor in order to
> guarantee meltdown. Instead of an optocoupler you can use discrete IR
> diodes and receivers with a long air distance between them. Any type of
> fault in one of the components can't break the safety provided by isolati=
on
> barrier.*


Yes.  And not just safety critical but "don't blow up your control
electronics desirable" as well.
In this case I had an ARM and an ATMega328 running with multiple
connections between them on the low voltage side and the ARM survived (I
think) and the ATMega328 still operated but would no longer walk properly.

In my case the isolator was a Hall sensor which was rated to run at up to
20A as part of its "normal operation". Isoating the energy source from the
IC is not an option in this instance. Isolating the output signal to the
required isolation specification is necessary. This was a prototype. The
ultimate solution was intended to use a digital low energy interface and
this has nicely emphasised what I already knew but failed to allow for well
enough in reality.

If the IC was in fact made to the same specification as the name brand part
and/or was the name brand part specified then it was rated to not fail at
under about 100A. What happens when it does finally fail is not part of the
specification but what it will do is clear enough. Breaking 100A DC cleanly
inside a small IC body would be an interesting challenge, which these
designers were not expected to have addressed, and didn't. The energy
source was  rated at at least 400A short term at ~=3D 25V, so ...  :-) .

More anon.


         Russell
--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.