ESD, 2nd mail for ESD, was : RE: [sdiy] midi optocoupler isolation voltage?

[Czech Martin]

Now for ESD: (I think I know this better then Midi, because
it's part of my day job to make parts survive ESD)

You raised a good point here. A decent optocoupler can stand
1kV potential difference between input and output pins,
but only few can stand 10kV or more. So something must prevent
such a high potential difference....
Similar things have to be observed in general when pluggin in on other
devices. 

Electrostatic discharge (i.e. ESD) problems are AC, indeed very fast 
AC, rise time about 10ns down to 100ps (!), lasting 1us to 1ns of time.
Most of these events are so weak you will not notice, nevertheless
they can destroy circuits/parts very fast.



There are severall types of ESD situations:
Human body model discharge  (HBM) and charged device model discharge
(CDM). These are intended for modelling ESD during manufacturing.
I.e. what happens in your lab at home. Most parts should today
endure 1kV - 2kV ESD (HBM) precharging level today. Many manufacturers
do not spec this, because one can not measure it during production.

What does this mean? A person on a chair or walking on a carpet
will charge up to 30kV on a bad day! The upper limit of 30kV is because
a corona discharge will set in on your nose or glasses.
I recently noticed this in a hotel:
every time I pressed the elevator knob I got a painfull zap in my finger.
The spark was longer then 7mm, so I can guess that it was far above
10kV. Using your wrist strap you can assume that you will keep charging
down under 1kV (not average but peak).  But not much lower!

It did not happen often to me, but in some cases ESD stress has made
some circuits appear very strange, and it took hours until I found
out that some subtle error was introduced in the circuit.
I mean, a short is easier to detect, but a slight leakage can
sometimes be a cause for real headake!
So I would very much recommend to anyone who is building circuits:

Use an antistatic table mat with wrist strap and safe (!) ground
connector. Use soldering iron with discharge connection
to that safe ground. If your iron has no such connection
(very bad) try to connect a 1 MegOhm resistor from the iron
metal to the mat ground point. Do not open or manipulate
the mains power part of your iron , unless you know exactly
what you're doing (sorry, but better say this than death).
Also do not connect to mains earth other than using a safe ESD
ground plug. After all, you don't want that the wrist strap
will shock you in case of a earth fault nearby!
Some of the ESD gear is very bad, so you should measure if it will
realy make a connection to safe ground.
Keep away from styro foam or other materials that will easily
build ap static charge. Some CRTs will easily intruduce a large
E-field that can lead to static, too. Especially when switching
on or off. Keep away from that, too.
So this way you can cope with ESD on your lab table.

Believe me that parts get really damaged otherwise.
It depends on climate, furniture etc. but it will happen.
The trick with coping with ESD is not hard discharge, but 
"dissipative" discharge, i.e. severall MegOhms to ground.
The black foam CMOS parts come in is pretty ok in this respect.
Also most tubes. It is NOT OK if you get parts on a piece
of styro foam (whoaoaoao!!! ever wondered why styro chips
keep sticking to your clothes, this is static!!) 
covered with alu foil.
You still see such "packaging" today . Incredible!
I never got something like that from quality suppliers like
farnell, but from the Radio-Shack-alikes.


The other thing is "system level stress".
There is not much knowledge about that, but some people have measured
the discharge of a "man with screw driver" into some hf
current target. This is really hard discharge.
So this is the kind of ESD that will happen
if you actually use or calibrate your gear. Since no wrist strap is
used, the discharge will be horrible of course. Severall 10 A
and 25kV can be easily found.
We guys are lucky: since we build with discrete components and audio
frequencies,
we can user zener clamps and resistance networks to care for that.
Tube guys are lucky, too!
The IC or RF guys really have a lot of trouble with that, especially with
smaller technologies << 1.0um gate length.


So any expensive circuitry should have protected inputs and outputs.
I would never dare to connect some CMOS microcontroller frequency
counter directly to the BNC-jacks on the panel, for example.


There are some myths about ESD which will never disappear, e.g.:
-ESD does only bite other persons, never me
did you ever have some strange IC death or circuit malfunction
you never were able to explain?
-ESD has only to do with CMOS
certainly the most advanced technologies are CMOS. But
bipolar circuits will suffer from ESD, too. 
Of course, a 7805 is somewhat more solid that, than say a CMOS
micro processor. But this has more to do with device area (volume)
of the protection devices.
Look into the app-notes of an OP07. Without series resistor
the low offset can turn out as high and unknown wandering offset after
the device became zapped.
-I have everything anti-static, so no problem
Yes, but it is really hard to guarantee a 100kOhm-10MegOhm
discharge path under all circumstances, wear out, etc.
Manufacturers of "anti-static" gear can do pretty good
volume conduction, but surface conduction does still make
trouble here and there.
One day a guy showed me an IC in a typical "anti-static"
tube. He made the IC slide up and down in the tube and
turned on an old AM reciever. What did happen?
After some up and down some kind of noise could be heard
in the speaker, actually ESD was happening in the "anti-static"
tube! Parts must be fixed in those tubes, ok, but the example
showed the difficulties. The same is true for mats, carpets,
chairs, tables etc. etc.

But no need to ignore the ESD problem, because it seems so
complicated. With a little care fore equipment (table mat, wrist strap,
soldering iron, black foam, good tubes with fixed parts etc.) 
and some protection circuitry one can be pretty much safe.

The other question about relais coil kick-back is more related
to EOS (electrical over stress). In contrast to EDS these are 
mistreatments of semiconductors or other gear in a loger time
range, and often the total energy is higher. Reverse polarity supply
or regulator malfunction is a good example.
Now, some V-N-MOSFETS can stand pretty high Uds. The three leg devices
have always bulk tied to source, so the drain/bulk diode is
there and will protect the device to some extend, certainly against
reverse polarity Uds. If you fear that for some reason Uds will be
above the speced limit, use a zener diode.
(zeners will be really good only above ~8V breakdown voltage,
but that's ok for most V-MOSFETs I know).
The relais itself should be pretty much indestructable.

Hope this helps...


m.c.