10V Surge Protection Circuit

[Panikas]

Hello all,

I would like to ask you if you could please help me out with the design of a 10V surge protection box.

This circuit is meant to maintain voltage at 10V during a flashover episode. The voltage is supplied by a high voltage transformer and a main circuit breaker interrupts current at 9A.

As I am measuring current on an insulating surface and drive the signal to an oscilloscope and a measurent PCI card with +/- 10V input voltage, it is essential I maintain this voltage at an event of a flashover (high current arc).

The circuit we designed is based on a 2 level protection consisting of back to back ziener and shottky diodes (Please check the attached).

Four back to back Ziener diodes, 20W, 10V (A) and another four back to back ziener (C) and Schottky diodes (d) 9.1 V, are clamping voltage as soon voltage across the measurement resistor 1.5kOhms (B) exceeds 9.1 V or 10V if the schottkys fail. I tried it and it has worked fine. But at the 3rd flashover, the 22 Ohms,10W resistor (E) got ''fried'' as the fault current went throught it during flashover.

Could you please suggest another component that can withstand such a thermal stress during flashover? I know that there might be other suggestion on the principle of desing but I am not able to radically redesign the circuit.

Your help will be very appreciated! Cheers guys!

PS: all resistors are 10W rated and the currents I measure before flashover dont pose a problem as they are a few milliamps ( peaks of 8mA worst case)

 

[Resqueline]

Big lasers have big safety discharge resistors, and heart starter testers have quite big load resistors.
Both of these devices may operates at as much as 7kV, and their resistors can fail with sufficient use.
If you need to stick with a 10W size, then consider the different tecnologies available (film or wire-wound), and also reduce the resistance value.
Consider using varistors and/or proper transient protection diodes (instead of, or in addition to zeners - which I personally doubt can withstand high transient currents).

 

[Panikas]

^ Thanks for your reply!

I am also thinking of adding an MOV and/or a GDT in parallel to those Zieners. I was just thinking that maybe that 22Ohms resistor is not necessary after all since it cannot significantly limit the flashover current. What do you think?

 

[Resqueline]

Yes, that's my thoughts too. After all, what's 10 or 22 ohms to several kV's?
The only question I ask myself is if there's any point in starting with high-voltage limiters at the input and then have lower voltages, ending with 10V units.
I'm not sure what approach would be most durable.

 

[(*steve*)]

I think I would be considering some form of LC filter (or a series thereof), presuming that the high voltage transients have very high dv/dt. A problem could be the voltage rating of the inductor though.

Follow that with transient protection devices. Be aware though, that these devices often limit the voltage to something quite a bit higher than their knee voltage. You could follow this with a resistor and zener, but I doubt you'd be able to keep the spikes below 1 volt.

To achieve really good elimination of those transients, you would probably want to start with a slightly higher nominal voltage (or end up with a lower one) and insert a linear regulator. This way you would only need to keep the transients down to about 35V and the regulator would all but eliminate them.

 

[Resqueline]

Can't use a regulator since this is a measuring input, but it could be an idea to start out with a high voltage MOV, with series resistors to the next lower voltage protectors.
Be aware that with the fast, high-current spikes encountered here the physical layout becomes very important to avoid setting up large voltages along ground paths etc..
Note that your circuit above has a shorted input & output (but we get the point).

 

[(*steve*)]

Perhaps you need to look at some gas discharge style lightning protection devices as the first stage, then MOVs, then finally some Zener diodes.

How much resistance can you put between the input and the output?

The resistors before the first suppression device will have the hardest life. Remember that resistors have voltage ratings. If you want 100k with 1000V across it, you will probably need 5 or more resistors in series, and even that may be pushing it. Perhaps think 50 to 100V per resistor and figure out how many you'll need...

The more series resistance you can have, the less energy will be getting into your suppression devices and the happier they'll be.

You say you're measuring current, is the current expected to flow through this circuit, or are you measuring a voltage to calculate current?

 

[Panikas]

First of all thank you for your replies.

I am measuring current by measuring the voltage across the measuring resistor (1.5kOhms) and then perform the calculation ( I=V/R) with my LabVIEW data acquisition programme. I am not interested in the transient current, only interested in the steady state currents.

My measuring equipment is protected as I am driving my signal through an optic fiber that will saturate in the case of a flashover. I am practically protecting the transmitter of the optic fiber.

Unfortunately I do not know the characteristic of the main circuit breaker that open the circuit in case of breakdown. If the response time is around 1ms, even though for steady state power dissipation would be 9A*9A*22Ohms=1782W, for the transient the energy would be only around 2 Joules. So I dont think I should look for very high power rated resistors.

As for adding an inductance in series, I am afraid that this would effect my measurements so i would like to avoid that.

I understand that there are a lot of approaches to do this as you have already kindly suggested. But, if I get this right, the most swift way of doing this would be adding at the input and before the 22Ohms a MOV and GDT in shunt. In this case, don't you think that the 22Ohm resistors are useless?

Cheers guys, this is very helpful!

 

[Resqueline]

If you have 22kV at flashover then a 22 ohm resistor sees 1000A, although for only 0.1us (correct me if I'm wrong). The 22 MegaWatts might vaporize a thin & narrow film.
One additional hard part for it is the voltage gradient it sees, maybe 2kV/mm, creating internal flashovers, eroding the resistance film.
You'd need high-power resisors not for the power but just for the sake of dispersing the current density and the voltage gradient across greater widths/lengths.

It sounds like a good idea to start with a (900V) GDT at the input, then a (22 ohm)resistor to a (160V) MOV, then a (22 ohm) resistor to a 10V zener/transient diode.

A GDT, like a spark, has a very high conductivity (once ionized) and a very low voltage drop (10V) despite peak currents of 1000's of Amps.

 

[Panikas]

Alright, I am gonna have a go by adding the GDT and MOV and fingers crossed it will do the job!

Cheers guys, thanks for the help, really appreciate it!