EMC compliance: ESD discharge, how should DUT react?

[mcinque]

I've to test a commercial low cost project for esd immunity. I'm trying to pre-test it with a electronic BBQ igniter pistol (a very poor man method, I know...).

I'm very uneducated on this (and on a billion of other things) and I wonder how exactly a DUT should react to a ESD discharge to being considered as compliant:

Assuming a boxed case MCU that counts from 1 to 100, when I apply the ESD to the inputs that should be touched during use (pushbuttons), the count restart from 1.

I consider it as a failure. Should it behave as if nothing had happened, keeping on counting?

[jahonen]

That often depends on the relevant standard which applies to specific device type, which is considered as failure. A-criteria would be that the counting would not be effected by any way by the discharge and B-criteria is that counting resumes normally without any user intervention when discharges are stopped. C-criteria is that user intervention is necessary to resume normal operation. I think that B criteria is often acceptable, but not always.

So you should try to determine if your device belongs under some such standard and try to figure it out from it.

Regards,
Janne

[T3sl4co1l]

ESD events are usually B type, yes.  I think it would be somewhat up to the testing protocol you write, whether the behavior you describe counts as type B.  It doesn't sound like C though.

Tim

[Neilm]

I doubt that you are testing anything close to what that standards say. As the return currents will not be controlled, you could be getting "bounce" that is a lot more than you expect. ESD test guns are calibrated to meet a particular waveform. Different tests specify the the test voltage (4kV for commercial in my experience) and shape of the electrode. I have done tests with the wrong electrode and got massive differences. The total energy applied is quite small (although if you do zap your self it hurts).

As for the results you have, you could argue that this is a criteria B, it was affected but carried on counting with out you having to press any button. Quite often, the interpretation of the results can be a source of argument. If so -  know the standard and (assuming that B is still a pass in your case) point out that no user intervention was required.

[Mad ID]

Usually it's OK for the DUT to reset (i.e. watchdog reset) as long as the device continues operating after the testing. If it freezes it's a fail. Big part of the ESD problems can be addressed in software, at least in my personal experience with ESD testing and compliance. 

[mcinque]

Well, since the counting is started by the button press, I would consider it as C-criteria (so complete failure).

The strange thing is that I've tried all of the attached solutions with the same results: the MCU resets.

I'm surely doing something wrong because it behave almost exactly even WIHTOUT any MOV, TVS or LPF installed.

Just to describe better the setup:

the DUT is a battery operated device.
one terminal of the BBQ electronic igniter goes to battery GND of the DUT.
the other terminal goes to a sharp metal point used as ESD gun and put in contact with the possibile ESD entry point (say an exposed led) while the electronic igniter is on.

Theoretically, the ESD should discharge completely to GND because the TVS clamps anything over 6.8V, so basically the MCU should not be affected by the spark.

In practical, MCU resets even discharging the spark directly on the battery GND terminal (so without touching anything in the middle). Maybe the spark rises the GND potential and the voltage on the MCU drops to zero instantly?

I'm obviously doing something wrong or missing some details.

[T3sl4co1l]

You have an EMC susceptibility problem then.  Ground bounce (which can be 10s to 100s of volts from a fast spark), electrostatic induction (fast transients coupling into poorly bypassed or unshielded traces), even magnetic induction (ugly loops picking up the discharge current from the lead -- response varies with the ground return lead position).

I wonder what the capacitance and ESR of a BBQ ignitor piezo is.

Tim

[mcinque]

Ground bounce (which can be 10s to 100s of volts from a fast spark)

good! knowing the name of the phenomenon gives me the opportunity to begin to study it and knowing how to deal with