Protection against high voltage spikes/ shoot through: Help needed

[myID]

Hi-

I have a question:

How to protect the rather expensive NIM stuff and my MCA in case of a coupling capacitor failure or similar?
I will probably have 3kV on the output of my (then dead..) preamp in this case for a few ms until the PS shuts down (plus the high current pulse from the capacity of cables/ filtering caps,..).
Sure I do use high quality overrated caps but still I am afraid to blow some k$ of equipment...

Will it help to connect the BNC central wire (signal- or in failure mode 3kV) with some TVS to GND (i.e. shield of the cable)?
And to further enhance protection (MCA is the last in the chain and the most expensive..) put a TVS in front of every stage to system ground?
I expect the small capacitive load does not interfere with the signal if one for example uses 1.5KE18 or similar?

Thanks for your advise!
Roman

[David D Speck MD]

Roman,

You may gain a degree of protection by connecting NE-2 neon glow lamps between the low voltage signal traces and ground. They break over at 90 volts, and have a forward voltage of about 60 volts.

They have low capictance and nearly infinite resistance unless they are triggered. I just don't know if they would break down fast enough to adequately protect a sensitive device.

I've also seen the suggestion of putting 5 or 6 ordinary 1N4148 signal diodes stacked in series across a sensitive circuit. They would act as a Zener diode and shunt excessive voltages. Again, I would check what the maximum normal voltage would be in the signal portion of your circuit, and calculate how many 0.470 volt drops you would need to avoid affecting your valid signals. Stacking the diodes reduces their effective capacitance.

Dave

[Carl Willis]

Hi Roman,

It is common for some manufacturers of preamplifiers to use diodes (like 1N914s) as a clamp at the input to protect the preamp. However, these will only do so much. They help drain small amounts of static charge that otherwise would get routed to the gate of the FET, while adding an acceptable bit of noise to the signal. They really are not meant to handle the catastrophic breakdown of a DC blocking capacitor. There isn't much you can do about that, other than choose caps with very conservative ratings to prevent it from happening. Personally, I have never had such a cap fail.

I think your concerns are probably better directed elsewhere. I have toasted a number of preamps doing common risky things--turning off the HV too quickly, attaching a detector that I hadn't shorted before attaching, disconnecting a detector with some HV still on the connection; powering up a bad detector that had sparking problems. It is easy to toast a preamp just in routine use, especially if you are not too disciplined about protocol. They range from trivial to very annoying to repair. Some use low-noise Japanese JFETs that are hard to find in the US at least. In any case, it is not $1000's down the drain when this happens.

-Carl

[bpaddock]

http://www.supertex.com/pdf/app_notes/AN-D11.pdf
shows "A ±500V protection circuit for low voltage high impedance
measuring instruments". Mouser caries the parts.

Stacking those may be option but probably not needed.
What you need to do is limit the current during a catastrophic failure, just long enough to open some type of fuse, preferably a self-reseting PTC.

> TVS in front of every stage to system ground?

TVS are frequently combined with gas discharge tubes. While a NE2 will work for simple static build up, what you really need to meet your requirement are tubes intended for lightning protection.
The TSV gets you speed until the gas tube can carry the current, the gas tube gets you power absorbency.

[David D Speck MD]

Carl,

When you have a chance, could you put together a compendium of suggestions and best practices to avoid trashing preamps used with He3, PMT or silicon detectors.

Thanks in advance,

Dave

[Doug Coulter]

I am avoiding the expensive NIM stuff here, since if you fry some of it, you can't find parts (like the custom hybrid chips some of it uses), and I utterly reject the "cargo cult" idea/philosophy of having to trust things I don't know all about the innards of anyway.

But, having said that, I do have some, and it's pretty robust stuff -- it was designed for people who aren't hands-on electronics experts. (eg anyone with a PhD or any grad student mostly, as well as many people here). One trick is to avoid the problem in the first place -- don't use a coupling cap with a lot of stored energy! You are in general looking for some pretty skinny pulses and just don't need that low frequency response for anything much.

Here I've gone to just winding little transformers, which works better anyway and seems a lot safer. They're easy to make -- just little cup cores or toroids with about 10 turns per winding. Just about any small high mu core will work fine, and it solves some noise problems too, no ground loops, and no high impedance points, since the low impedance of the NIM stuff will reflect back through the transformer to the detector side too. You're just looking for a current pulse, this is pretty easy. I use little pieces of type 77 ferrite, small common mode chokes from TV set video chains (not the ones for 60hz, too big and slow) and just about everything I've tried has worked fine.
With the secondary of the transformer grounded at the NIM end, the other end no longer matters, and if there's a short, it won't be the NIM stuff that fries, though the power supply may.

Although otherwise out of context, you can look at the first picture in:
http://www.coultersmithing.com/AuxCP/CV.html
in the upper right hand corner for one of the tiny transformers I use. Even that is overkill.

[Richard Hull]

I have a couple of charge amps that use a 5pf @ 5kv coupling cap. Not much stored energy there at normal 1700 volt bias levels. Still, Like we all know and should be the rule, lower and raise your HV on sensitive charge amp input detector bias supplies slowly.

Others may couple in as they choose, of course. I haven't blown a charge amp front end yet with tried and true procedures.

Richard Hull