ESD zap

On a sunny day (Tue, 29 Apr 2014 03:45:22 -0700 (PDT)) it happened snipped-for-privacy@yahoo.com wrote in :

Dunno, any data on that?

Some years ago I bought a nice overvoltage protection plug in socket for in the mains, it has thsose spark-gaps, and a bunch of VDRs, and a neon for signalling power is [still] on...

Den tirsdag den 29. april 2014 00.23.58 UTC+2 skrev John Larkin:

how about something like this at the input?

-Lasse

Interesting, the spec sheet says the terminal needed to be potted in something to obtain the max voltage...

George H.

That makes a lot of sense actually. I'd bet a good cleaning and conformal coat would greatly raise the sparkover voltage for ordinary parts too. Sharp edges, leakage paths, and all that...

Cheers, James Arthur

Those are really slick, if John can tolerate the 25v breakdown.

The USB2.0/3.0 transient protectors are golden, too.

Cheers, James Arthur

OK, this is about what I have so far:

Looks ESD hard, protected to +-50 volts overload. The 1G resistors and the lower ADC channel provide some failure detection.

The main requirement is for acquiring 100 ohm platinum RTDs, but we will make the current source programmable down to 10 uA (for cryo diodes) and 1 uA (thermistors).

My customer is also requesting +-50 volt overload protection. Which is why I'm leaning towards a low-current (ie, not shunt) protection scheme.

Not really when it only has to come on in times of bzzzzt and you just run the regular clamp diodes against it. Zeners also work well here and when at Or use two diodes to clamp rails that are weakly biased up to, say,

But that's umpteen or sometimes many dozens of parts versus just four to six. Muntz would have had a hissy fit.

I think one of these would do two lines of +-50V (SMDA24)

-Lasse

How about a home-made spark gap diverting it to earth?

A couple pointy bits on the PCB and pull back the solder mask.

Put a hole (non-plated) at the tip so it truly is an air gap and not across the surface of the PCB.

Here's an example where they didn't use a hole (or drilled slot, this

See page 22.

Relevant principle applied elsewhere:

Best regards, Spehro Pefhan

The depletion mode FET is probably just a few tens of ohms. If someone applies +30V at the E- terminal for a longer time it could go *PHUT* along with the 100ohms resistor and then afterwards the ADC.

What about tying the gate of the depletion mode FET to a low logic rail or maybe even ground? The FET and the 100ohm should be rated appropriately, like a solid watt or so.

The fet we have in stock, DN2530, will need about 100 ohms in its source, to limit around 20 mA maybe. That cartoon doesn't show all the gory details. I'll clamp the 100r to +3.3, too, with the other half of a dual diode. The 100r will need to be switched, too, for the uA current-source cases. 10K maybe.

That's a pretty good tutorial.

Plus, you don't need a permit for a girlfriend.

Wouldn't series ferrite beads (OK, surface-mount chokes) do that job as well, with less impact on signal slew rates?

The ADC presumably has low dI/dt on its inputs, but not necessarily low dV/dt.

It'll work, but if one modulates the current source, those capacitors cause transient currents. Howzabout this:

the inductors are just tiny ones, so bursts of current into the ADC don't fry the little internal clamps before the external ones conduct.

On Mon, 28 Apr 2014 15:23:58 -0700, John Larkin Gave us:

ESD potentials as low as 20 volts can cause circuit damage. No, it would not arc over (nor though) such a device as your input resistor.

You want all current clamping resistance elements to be of the bilk type. Carbon comp. There are surface mount equivalents. Metal film fails to properly abate and mutates upon each exposure. That form factor or larger should be good, but I'd get the bulk type meant for surge abatement, not simply a plain Jane COTS part. (yes, I know the parts I describe are also COTS).

ESD fields are also a consideration. This is why we wear ESD smocks.

They'll saturate at the drop of a hat. Might take off a nanosecond or two. Which might still be helpful out of IEC 61000-4-2, which is pretty snappy, but won't do much for the bulk of the energy.

Tim

You could use higher voltage transorbs (36V say) and series resistance, then let the ESD diodes of the IC take the remaining current. Or add caps there too.

I would do that, especially for low speed signals like an RTD.

I did some tests with a high voltage generator a long time ago, and it seemed the key was to put a fast clamp right at the wire where it came on to the board, to a groundplane. (In the case of board-level proection). The idea is that you don't want the end of the wire, and associated tracking, waving about at KV/ns.

This limits the spike amplitude to something reasonable, any resulting current into the circuit can be easily controlled via series R or L (since time is so short).

I started with minor disturbances crashing the micro, and ended with it surviving zaps with 2 inch arcs. We had a ~100kV generator from another project. Fun stuff.