High Voltage Spark Discharge

[Haf]

I occasionally have the requirement to perform electro-static discharge testing on the components I design. The tester we use has a 600pF capacitor that is discharged at 25kV to simulate a severe static discharge from the human body. It is connected to desired positions on the test article using low inductance cables.

The components I test have two possible breakdown paths. My question is this: will the spark energy discharge exclusively through one path only (the path of least resistance) or will it somehow be split between the two paths?

For example: I have rough estimates of the breakdown voltage for each path. These are based on approximations of the dielectric strength for the materials in each path. Let's say one path breaks down at 5kV, and the other at 10kV. When the component sees the 25kV ESD, where does the energy go?

To me it seems that it should discharge exclusively through the 5kV discharge path, but I know things can get tricky with high voltage, high frequency events.

 

[logbook]

A spark does not follow the path of least resistance. A spark has a wavefront of sub-nanosecond risetime. It will follow a low inductance path. If there is an outside path rather than an inside path it will go on the outside. (This applies particularly to metal boxed equipment). It won't want to flow into the box but rather just on the outside. If you inject it inside it will find the fastest way out.

Breakdown is an avalanche effect. The dielectric may not have time to breakdown as the cascade avalanche takes time to build up.

600pF is rather high as a human body model. 150pF is more typical. Likewise 25kV is tough. 15kV is a typical maximum.

 

[Haf]

logbook,

Thanks for your response. It sounds like you have quite a bit of experience in this area.

I should've said path of least impedance instead of resistance. Please pardon an electronically-challenged ME.

You're right that the 600pF and 25kV values are high; however, they are the standard that my corporation uses, so I'm stuck with them.

The testing I perform is typically done so that the high voltage side of the tester is applied to one of the component's electrical leads, with the low side (ground, or return path) connected to the case of the component. There are two potential paths for the spark, as I outlined above.

After talking with a couple of the EE's I work with, they indicated that the energy from the spark will tend to travel through the path with the lower breakdown voltage. The energy will not be split to other paths, neglecting corona leakage.

Does that sound right to you?

Thanks,

Haf

 

[itsmoked]

Sounds correct to me.

Since once a spark has been established there is now a very low impedance path on the order of 'ohms'. Simultaneously there is a drop in the source voltage (due to the arc) and hence reduced 'tension' with regard to any other potential breakdown path.

May I ask what kind of products you are testing? Pacemakers?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[logbook]

> they indicated that the energy from the spark will tend to travel through the path with the lower breakdown voltage. The energy will not be split to other paths, neglecting corona leakage.

>Does that sound right to you?

Well yes and no. I think the field will travel so fast that it will apply potential to both parts equally. Both will "think about it" for a while and then one will disharge the energy, stopping the other from being stressed. I wouldn't like to make the assumption that the lower breakdown path would conduct first. That may be due to a timeconstant within the material. I bet you could find materials that withstand a field for a longer period than others. Maybe your guys have more experience and have established their ideas as experimentally true. I would worry about it since I haven't tried it and can certainly think of reasons why it might not be true.

 

[OperaHouse]

Try to find the book GASEOUS CONDUCTORS by James Dillon Cobine. This is the bible on things that go arc in the night. Even though the last copyright is 1958, I have a paperback version and believe it is still in print.

 

[Haf]

logbook,

Thanks for the heads up. The EE's I work with are good, but not necessarily experts on ESD-type scenarios. I posted here to hopefully get an expert's opinion.

I had not thought too much about a potential "time constant" factor. I guess what you are saying is that the lower breakdown path may or may not be the lowest impedance/inductance path in the frequency domain of interest.

The reason I'm asking the question is that I would like to make the spark "choose" a certain path. I have little to no control over the design of one path, and fairly high control over the geometry of the other path. My planned approach is to design the path length such that the spark will preferentially discharge along that path instead of the other. Thinking simplistically, I thought I could just ensure that my designed path had a much lower breakdown voltage than the other path. You have pointed out that there may be other considerations. Unfortunately, I do not have much control over materials, so I'm stuck with manipulating geometry only.

I'm have plans for experiments that will give me a clearer answer to these questions, and I'll consider your input as I move forward.


itsmoked,

I'd rather not divulge the product, but I will tell you it's not pacemakers.

 

[Comcokid]

The key here is the risetime of the transient. I have designed boards where there was two paths - the one I wanted the ESD to travel and the one to the circuit I was trying to protect.

I had a open-drain 60-V MOSFET to a connector I wanted to protect, but no budget or room to put in a transzorb or any other device. I meandered the trace to the MOSFET so it was about 2" in length and had a few nH of inductance. At the MOSFET drain I put a small capacitor. Right at the connector pin, I put a 0603 resistor of moderate value (about 1K) directy to ground with mimimun trace length to the pin and ground-plane via. Testing with a ESD gun proved the MOSFET remained OK with this approach.

 

[itsmoked]

No problem Haf.

So Comcokid you were using the 0603 as the arc/tracking path?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Haf]

Just to clear things up - the two paths in my component are not part of any electrical circuit and do not contain any electrical components (nor do I wish to insert any). Also, the breakdown paths are not through air or any other gas; both paths are through solid material, or at a mechanical interface between two solid materials (this would involve some air, I suppose).

 

[logbook]

Understand that this would be a one off deal then. On circuit boards if you are between tracks you carbonise the board and it breaks down easier next time. If you actually break down a material then you will potentially leave a carobinised path there too.

For consistent breakdown you need an air discharge; nothing to carbonise.

Solid insulators are very flakey. If you double the thickness you do not double the breakdown strength. The increase can be sqrt(2), ie non-linear.

 

[analogkid2digitalman]

Air discharge?

http://video.google.com/videoplay?docid=4468957986746104671&q=500kv

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[itsmoked]

Yes.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com