Mosfet switch for DIY ESD gun

[pisoiu]

Hi all,
I need to build an ESD gun for in house pre-compliance tests, because Santa doesn't have a big enough bag for a proper ESD gun.
I have studied various ways to make the discharge relay and none of the classic solution doesn't seem fit for me. Vacuum relays are too expensive and reed relays does not seem to do a proper job from experience of other users here (contact wear, inconsistencies of results over time, etc)
I would like to explore a mosfet switch but I do not have experience with high voltage application of mosfets.
I attached a draft schematic, please share your thoughts.  I know 5 stages are required for 15kv voltage, considering max voltage of mosfets but I drew only 3 of them for convenience.
Idea is to use S1 to charge C1 capacitor to required discharge voltage, max 15kv, may be lower. The resistors and zenners have 2 purpose, one to charge capacitors for triggering the gate, other is to keep al mosfets voltage in balance. When optocouplers are triggered, capacitors are discharged into gates, putting mosfets in on state.
Does it seem to work on paper?


LE. I need to recalculate r2,3,4 they need to be much higher

[tszaboo]

When a MOSFET's gate sees high voltage, like more than a few dozen volts, it will start conducting and turns into a really bad resistor. MOS gates are very ESSD sensitive. I dont think they are really fit for such an application.

[NiHaoMike]

Try a triggered spark gap.

[David Hess]

A triggered avalanche series string of high voltage bipolar transistors would be more suitable.  Only the lowest voltage transistor needs to be triggered.  The pulse is coupled into the output via a high voltage capacitor which dispenses a defined amount of charge into the output.

Of course you would need to find the right transistor to use by testing them.  They are not normally specified for avalanche operation.

[T3sl4co1l]

Optos are too slow by about four orders of magnitude.  You need quite a few MOSFETs in series, and they all need to sync up as well as drive fast enough -- and this generally isn't possible I think.

Avalanche transistors are neat, but a bit too low on peak power: a full 15kV through 330R (IEC 61000-4-2) is 45A and a fraction of a megawatt; even reducing these figures by a few factors of 2 (accounting for air vs. contact method, impedance of the target, etc.), you're well beyond the rating of any avalanche transistor and need either many in parallel, or a whole mess of them synced up and combined with a power combiner.  Which will give you a very different output impedance from the electrostatic gun, so it's still not great.  (This is feasible for EFT, though -- at a level of say 4kV into 50 ohms, you only need about a hundred avalanche transistors, or a dozen MOSFETs -- or half a dozen SiC, even.  In an RF ground-plane setup, driving MOSFETs fast enough, syncing them up, and using power combiners, is perfectly feasible.)

So, the vacuum relay truly is king here.

A spark gap has some possibility I suppose, but it needs to be adjustable -- the tip voltage in general isn't known, and a triggered SG depends critically on its voltage drop to be triggered at all, while avoiding spurious flashover.

For example, a typical test case is probing a lone screw in a plastic enclosure.  It doesn't jump to anything, just sits there getting charged up.  Either it's discharged between cycles by manual intervention (or auto cycling in the gun, when available and enabled), or it just charges up in one or two cycles and that's basically that.

And an adjustable spark gap just sounds like a HV relay with extra steps.  So you might as well do that.  But keep in mind the spark gap will have different results from the vacuum switch: it will ALWAYS spark, so loses some voltage that way; and, the spark acts to sharpen the pulse, so use it carefully and account for the structure either side of it, so as to avoid introducing stub lengths or whatever that can propagate the sharpening.  Calibrate with a >500MHz scope and adjust output resistance, inductance and damping to keep the leading edge reasonable.

Tim

[PartialDischarge]

In practice ESD generators do not switch the high voltage, but a low voltage with a thyristor that is applied to a pulse transformer. You may find some pulse transformers on ebay, this one for example https://www.ebay.com/itm/363117434259/
In this sense attached is the schematic of an old Schaffner 1.2/50 pulse generator (model NGS504). It has some settings to deliver more or less energy to the primary of the transformer

This said, solid state mosfet switches are not done the way you picture. Only the lower mosfet is swiched and the rest need a carefully calculated RC divider. This paper talks about some methods:
https://www.semanticscholar.org/paper/Series-operation-of-power-MOSFETs-for-high%E2%80%90speed%2C-Baker-Johnson/a5533753a2bff95b39cf565e5b658dd7839e9b91

Also, you mention vacuum relays as expensive but there are many cheap ones on ebay.  I recently purchased a V2V-1V W2W 4kV russian model from Ukraine, and successfully tested at 15kVdc (meaning no leakage). I was about to test its switching speed by overdriving the coil at 50V for a few 10s of miliseconds and 25V after that. However I dropped it to the floor and broke it.

[pisoiu]

Thanks all for valuable information. I tried to avoid used components from ebay because I wanted to make a design with stock components, hoping that it will be produced in limited series someday. But it seems I have no choice if I want to keep costs in reasonable limits.

[tszaboo]

Thinking about it, I think you want a SCR there, like a thyristor. Someone correct me if I'm wrong please.

[pisoiu]

I could not find any to be able to withstand 15kv. Or should I think to stack those too?

[T3sl4co1l]

SCRs have risetimes on the order of 100ns at best, and typically need to be limited to a few µs to survive.  They're quite effective for surge generators, but not EFT or ESD.

Tim

[jmelson]

Perhaps I'm out of date, but all the commercial ESD stimulators I know of used spark gaps in them.  A high-value resistor charges a small cap, when the spark gap snaps over it delivers a pulse with very fast rise time through a network that shapes the output wave.  Actually a very simple device.
Jon

[floobydust]

The problem is ESD guns are ridiculously overpriced and super expensive. It's because they are hand-tuned to the standard's waveshapes ala 61000. Well, that and the HV can hurt a bit. I've never seen a tear-down, who can afford it lol.

I've homebrewed ESD testing gear using everything from car ignition coils and CCFL inverter transformers to make the HVDC power supply.
Beating a BBQ lighter is a bit of work.

[DTJ]

Beating a BBQ lighter is a bit of work.

A BBQ lighter with a ground lead worked well for me.

[amyk]

Perhaps I'm out of date, but all the commercial ESD stimulators I know of used spark gaps in them.  A high-value resistor charges a small cap, when the spark gap snaps over it delivers a pulse with very fast rise time through a network that shapes the output wave.  Actually a very simple device.
Jon

Basically a tesla coil or spark gap transmitter?

[T3sl4co1l]

The problem is, yes the waveform, and calibration.  A random piezo igniter makes sparks, sure, but how will you get any kind of consistency in using it?  How do you even control for polarity?  Neither waveform, amplitude nor sign is consistent.

They're still useful -- but you need to be aware of these limitations.  Practice your technique with respect to waveforms and amplitude, whether by contact or air discharge, and get a feel for what the typical variance is.  Add impedance as needed to get the response closer to desired.  Probably what you'll see is, when testing a board or device, a certain arrangement is very likely (like almost always) to trip it up, suggesting a very low threshold; in others, only occasionally, suggesting a modest or high threshold -- depending on what your calibration stats showed.

Hmm, where can you get those anyway?  I don't see any...  I remember years ago, a coworker picked up a stupid chintzy, it looked like a wand lighter, but without the metal nose, the plastic body tapered to a point with a rounded electrode at the end.  Meanwhile the body had a ground lead and clip coming out of it.  All yellow, but for the metal parts of course.  Fairly soft trigger I think, and you could adjust the pressure/displacement on it to vary the voltage, or bleed off charge then release to get an inverse pulse I think, etc.  Would think they'd be all over Amazon but I don't see anything but replacement igniters offhand?  Or maybe they're too unsafe and someone thought better of selling them there...

Tim

[PartialDischarge]

The problem is ESD guns are ridiculously overpriced and super expensive. It's because they are hand-tuned to the standard's waveshapes ala 61000. Well, that and the HV can hurt a bit. I've never seen a tear-down, who can afford it lol.

Because they are not sold by the thousands. If they are that expensive then there is a market for someone to engineer a new model and market it at a lower price.
It is worth remembering that the sale price of a product is not only related to that of the cost of their parts. Engineers do often forget this.

[pisoiu]

My goal is to build something as close as possible to the test equipment which will be used at certification.
For that I need control and consistency in results, not just a big spark with unknown parameters from a cheap piezo.
I need to make precompliance test for 61400-4-2 HBM and maybe iso10605.
I am prepared to spend what is required for that purpose, but not at the level of cost involved by a tool used in certification laboratories.
I already have components for RC network, an adjustable HV power supply is on the way, the only remaining problem is the switch.

Jmelson, can you give me an example of a spark gap triggered switch available for purchase somewhere? All I found are spark gap devices with 2 terminals which look to me more like an overvoltage protection device than like a HV switch. I do not understand how you trigger them. Thank you.

[floobydust]

The problem is ESD guns are ridiculously overpriced and super expensive. It's because they are hand-tuned to the standard's waveshapes ala 61000. Well, that and the HV can hurt a bit. I've never seen a tear-down, who can afford it lol.

Because they are not sold by the thousands. If they are that expensive then there is a market for someone to engineer a new model and market it at a lower price.
It is worth remembering that the sale price of a product is not only related to that of the cost of their parts. Engineers do often forget this.

$900/month rental of 30kV ESD simulator gun kit like ESD3000 or NSG438.
$200/month rental of extra Discharge Network
$9,000 16kV ESD simulator gun NSG435.
$1,175 Discharge Network
$3,450 2ohm current-sense probe
$4,750 ESD test bench and coupling plane and grounduing cables
$200 470kohm ESD ground cable bleed resistor

Naaa it's just a ripoff under the guise of certification.

[floobydust]

I did research and for the switch they are using high voltage vacuum or gas-filled relays with the tungsten contacts. Counter-intuitive but:
"If a load is switched, an arc will form. At the point where the contacts are getting very close and the current density is getting higher and higher, a breakdown occurs. This arc will have a very low voltage of 18–23 V and be quite stable in comparison to an arc in air, which is another advantage of vacuum. The vacuum relay constant arc voltage acts as a current limiter and, when considered in combination with the inherent short arc time, offers a technology that generally wears less than other types of relays, providing stable performance over the life of the relay."
"In theory, the contacts of the vacuum relay don’t arc as the contacts approach each other until the contacts are within microns of each other when field emission starts occurring."
"In a vacuum relay without a gas to ionize between discharge electrodes, current flow only occurs when metallic contact is made..."

Above 15kV they make x-rays and look a lot more Frankenstein-like. A bit more: Greg Senko, "ESD Simulator Verification"
https://www.oldfriend.url.tw/Designer/ansys_en_ESD.html

Alternatively, a three-electrode spark gap or Trigatron is what people are mentioning. I have seen UV LED's also tried to trigger them. But they are in air and a bit weird, humidity affects them and the arc is initiated by streamers.

[T3sl4co1l]

"Constant arc voltage" "current limiter" riiiight...

Vacuum contacts indeed aren't zero-wear, as sputtering and erosion occurs around that spark.  Field emission becomes so intense that atoms are ejected from the surface, or just some random gas molecules get caught in the crossfire, and -- pretty damn quickly a whole bunch of atomic shrapnel is produced and makes the breakdown very conductive.

Reclosers, used for switching power lines at some 10s kV and 100s A (handling fault currents in the kAs), evaporate substantial amounts of metal when acting -- this limits lifetime, and slowly causes contamination inside the vacuum "bottle".  A shield condenses most of it, so it doesn't form a conductive path between electrodes.

They also make 'em with oil filling, which has the advantage of high dielectric strength, so again, small gap distance when breakdown finally occurs; and also higher thermal conductivity, which is nice for power line applications (and not especially relevant here, heheh).  I wonder if that would actually be pretty feasible to do as a home build... or maybe you can get small oil-filled HV relays too, dunno.  Heh, mind that pretty much anything proper and HV-rated isn't going to be cheap itself, so, you've always got that sort of problem going here...

Tim

[NiHaoMike]

I doubt the contact erosion is going to be significant at such low energy levels. Keep in mind automotive spark plugs last well into several hundred million sparks before needing to be replaced.

[pisoiu]

Is this one ok? 15kv, 50 amps.
https://mgs4u.com/product/vc-2/
There are a lot of them on ebay and price is acceptable.

[T3sl4co1l]

Would think so.  Rather overkill, but the problem is doing it at all, and if that's available, sure.

I doubt the contact erosion is going to be significant at such low energy levels. Keep in mind automotive spark plugs last well into several hundred million sparks before needing to be replaced.

Exactly.  Hm, given that spark plugs only do a few amps peak (AFAIK), but also do some 100s µs versus < 100 ns per pulse.  So they're probably even worse.

Tim

[jmelson]

Jmelson, can you give me an example of a spark gap triggered switch available for purchase somewhere? All I found are spark gap devices with 2 terminals which look to me more like an overvoltage protection device than like a HV switch. I do not understand how you trigger them. Thank you.

Not triggered.
The units I've seen have a spark gap that snaps over when the supply voltage ramps up, so it just goes snap-snap-snap while you move it around the unit under test.  If it has a trigger button, that just turns on the HV power supply.
Some other HBM testers just charge a terminal, and you bring the terminal up to the unit under test until a spark occurs.
Jon

[Terry Bites]

I think 100V avalanche transistors (ZTX415*100) for this would cost about $1600. Ouch!
A UV LED triggered spark gap would cost next to nothing.
What about a fast sychronised driver scheme for your mosfets. Building the pulse transformer is not as bad as you might think- say 50uH per section on a well insulated ring core.