Electronics > Projects, Designs, and Technical Stuff

"Proving" safe voltage (detecting a very small % of a high voltage)

(1/7) > >>

Hey everyone,

I'm working on an experimental setup at work that involves a quite high voltage power supply. (The experiment requires the feedstock chemicals, in gaseous form, to flow through an electric field to catalyze a reaction somehow. We designed the electrodes as two cylinders separated by a gap. The lower electrode has the outflow in the center, so that the gas mixture cannot exit the reactor without passing between the electrodes.) This project was started before I was hired, and a high voltage power supply had already been ordered with a maximum voltage likely far higher than needed: 40kV DC. So even though they'll likely never run the experiment at that voltage, the entire setup has to be designed to protect human safety even if someone were to crank up the voltage.

The power supply is current-limited to 1mA in terms of regulation, but the peak current as its caps discharge can reach up to 500 times that, according to the manual.

Additionally, the 3 meter long HV cable, being the typical coaxial construction, has about 140pF of capacitance of its own, so on the advice of my old boss (who has experience with HV stuff) I'll be adding a discharge circuit (with analog moving-coil meter) and a series resistor to the electrode to limit current.  Between my discharge circuit, and the discharge circuit built into the power supply, any voltage should be discharged within 1 second, which is faster than one could physically open the barriers between the user and the electrodes.

There will also be interlocks, interfacing with the fume hood door, as well as the door to the metal cabinet that will house the reactor itself. (And inside that, everything HV is insulated.)

So the actual question:
One of the things that the users would like is a light (like the typical industrial beacons) that indicates whether it's safe to touch or not. A theoretically simple one would be a light that simply signals when the HV output of the power supply is turned on (regardless of what voltage it's set to), but unfortunately the power supply does not have an output for that, only a 0-10V proportional analog output, so a safe-to-touch output voltage of 50V DC, for example, would mean an analog output of just 20mV -- and I can't be sure if that output would even operate if the power supply's mains input is turned off.* The even better option would be something that actually indicates a voltage of 50V or less at the electrodes, but I am not sure how I would go about measuring that in any halfway reliable way.

So how would one go about detecting that the voltage is 0-50V when the maximum is 40kV?

Or is there some entirely different way of sensing this, especially something entirely passive?

I'll be using a safety relay (probably something from Phoenix Contact) to monitor the interlocks and cut off power to the HV supply, and to control the beacon, if used.

What we don't want to do is a light that doesn't actually indicate anything useful, or that could even be misleading. So I don't want the beacon to simply illuminate when the interlocks are closed (=ready to operate), but the HV supply isn't on, since that's signaling "danger" even when it's not, which could lead to complacency.

I'm currently waiting on a reply from the power supply manufacturer to see if they can offer a solution. The power supply's own HV output on/off button illuminates exactly as I want, but I'm not opening and modifying an expensive, dangerous HV power supply without their say-so.

Thanks for any ideas!

*I did lash together a quick comparator circuit as a test, but all I had laying around was an LM339, whose input offset is too big for comparing 20mV to 0V. Even with hysteresis it flutters around.

You probably want to go as passive as possible for safety, and maybe double up too. My first thought is a simple resistor chain and indicator, either a neon or very high efficiency LED. You could then rely on this to discharge the capacitances that you mentioned - for a neon you would need a high value shunt resistor across it to ensure that you get below 50V. The problem is that you have a relatively small amount of current to play with in terms of illumination. With the loading of the indicator, you could probably add a label saying something like 'Wait 10 seconds after the indicator is extinguished before opening'.

As an alternative, you could use a MOSFET with gate clamp and high value shunt resistor in place of the LED / Neon to drive a larger indicator - as long as you can ensure that the supply to it is failsafe.

The comparator method is reliant on an independent supply and other factors that might make it more prone to failure, not to mention the difficulty in handling such a wide input range without possible noise problems at the low end.

For that kind of voltage span, 50V to 40kV a passive method is not possible, because resistors will create a large voltage drop or the power dissipation will be too high.
Maybe they can help https://saker-mv.com/voltage-detector-mvd30-mvd15/

What's going to discharge the output capacitances down to <50V? The PSU? - one for tooki to check. I'm not saying a 40kV resistive divider is easy, but it is doable with proper insulation and layout. For a few 10s of uA dissipation isn't a problem, eg. 800mW @20uA. If it turns out that a discharge load is needed anyway to bring the voltage down in a sensible time, then you may as well make use of it.

HV resistors are not that expensive. RS Pro 500Mohm rated for 20KV:

Use 4 of them to make 2Gohm 80KV + a final 200K sensing resistor. (1KV=100mV).
2Gohm load (20uA @ 40KV, 800mW), giving a 10000:1 ratio or 4V ouput for 40KV input. Adjust the final ratio as required.

Feed this to a  instrumentation op-amp/comparator, these usually have GigaOhm input impedances.
It doesn't need to come out for anything, just put it in a earthed metal enclosure. Use a small secondary psu for running the detection circuitry and turning the led on/off.

Make the circuit redundand just in case, but the interlock should be the true safety mechanism, disconnecting the HV supply and applying a bleeding resistor unless the HV PSU automatically does this when off.

As long as everything is referenced to ground (Protective earth), it shouldn't be dangerous.
Dip everything in HV silicone, perform degassing if required, should perform great.

--- Quote from: tooki on February 24, 2024, 10:55:37 am ---The power supply's own HV output on/off button illuminates exactly as I want, but I'm not opening and modifying an expensive, dangerous HV power supply without their say-so.

--- End quote ---
Or just a phototransistor reading this!
But this probably isn't a real HV indicator, but simply turned on by the driving circuitry, assuming there will be HV output unless something is wrong.


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod