high voltage self-powered capacitor charged light 1

[RyreInc]

I have a 50-400uF, 3500V capacitor bank. It would be nice to know if the caps are charged to a level any higher than 50V or so, without having to rely on an external power source, i.e. powered from the caps themselves.

It appears that a non-switching solution is not very practical--drawing 10mA at 3000V still results in 30W, and furthermore draws down the capacitor voltage a bit too quickly (somewhat more slowly and it'd be a feature, not a side-effect). If not for that a current limiting circuit would do the trick.

So now I'm pondering some sort of buck converter. But how can I generate a switching signal from such a high voltage, and over such a large range of voltages? Keeping the circuit as simple as possible is virtuous as well.

Are there other solutions out there?

Any solution should be pretty robust since it is intended for safety purposes. With that in mind this is probably a bad idea, and there is probably no substitute for established safety procedures such as manually attaching a bleed resistor. But I'm curious as to the feasibility of such an idea, and having an at-a-glance indicator of voltage present would be useful.

 

[VEBill]

Starting at 50 volts, and up to 3500 volts input?

 

[MiketheEngineer]

Pffffft???

 

[RyreInc]

Indeed, 3500V. Very pie-in-the-sky, I'm thinking, at least practically... what about theoretically?

 

[Sparweb]

Forgive the ignorant questions, but is there a ~10 Meg resistance across the cap terminals?
Has the system been designed to slowly discharge them, or do they stay charged when it's off?


STF

 

[VEBill]

How about using the capacitive detection concept embedded in one of those non-contact voltage sensing pen-like tools?

http://www.aliexpress.com/price/non-contact-voltage-detector-price.html

Most seem to be AC-only, and none reach the lofty heights of 3500 volts.


The fundamental problem is that this safety feature would eventually fail, thereby providing negative contribution to safety.

 

[ScottyUK]

Gold leaf electroscope with optical detection of the leaf using the Mk. 1 eyeball?

 

[itsmoked]

Why are you doing this? Everyone else has bleed resistors and KNOW when their device is powered on and when it's powered off. If it's off in a few minutes the bank is zeroed out.

Are you trying to use the magical capacitor power?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[RyreInc]

SparWeb, there is normally a 10k resistor paralleled with the caps, but it is disengaged when charging/discharging via a normally-closed relay. The caps are "supposed" to be discharged when not in use, but of course the relay could fail.

VE1BLL, I'll look into that solution, but I believe you're correct that they only work with AC. I agree with your other statement about the negative contribution to safety--at this point I'm more curious as to whether this is possible at all!

ScottyUK, that is a pretty novel idea, very interesting, and impractical!

itsmoked, your attitude is not appreciated. Waiting a few minutes is not practical in many instances. Our current method is to manually attach a resistor to bleed off the caps within a few seconds.

 

[itsmoked]

Sorry about the 'magic' crack. After reading your post probably 5 times I couldn't be sure you weren't a crack-pot. We get them here fairly regularly.

Anyway, if this is for life safety then I don't see how you can rely on any electronics. Can you make the shorting resistor application trivial? Like having it permanently hooked up to a knife switch that lets one visually see the contacts?

For a light, what about a neon bulb in series with enough resistance that it will not burn-out at 3kV but will still light up above 50V.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

I've seen circuits where a relay (powered by a normal upstream supply, e.g. 120 VAC) holds open the bleeder resistor circuit. When the power is turned off, the relay closes and the big scary bleeder resistor drains the capacitor in a few seconds. The advantage is that the bleeder resistor can be aggressively low resistance (quick) without wasting power and being hot.

You'll need an appropriate high voltage relay and an appropriate high voltage resistor (typically several inches long).

 

[VEBill]

The other human safety design approach is a cover with an interlock. The time constant of one-half of the dual redundant bleeders is design to be quicker than the time it takes to remove the cover (from opening the interlock to touching the Hi-V). Done 'correctly', the design docs include a little white paper explaining how it's impossible to be shocked, even if they try.

 

[LiteYear]

I think a switching circuit will be very hard to design with such a wide range, but as itsmoked suggests, a neon bulb could be a goer. They have extraordinarily flat V-I characteristics.

The challenge might actually be getting it to light at all at 50V, since the typical striking voltage is at least 90V. You may benefit from the fact that the cap voltage in question is decaying, so as long as the light it already struck, it'll maintain glow down a bit lower than 90V.

My quick calculations give an enormous 1GOhm as a good starting point for the series resistance. That'll give you about 1 nanoamps at the lowest voltage and 3 microamps at the highest voltage. Both are plenty to make a neon bulb glow and both are low enough to represent very low stand-by power draw. Of course, you'll need to adjust to suit your particular bulb, but it seems the numbers are there to make it work.

 

[VEBill]

Neon bulbs aren't the most reliable of devices. Then again, perhaps the 3+kv supply voltage would help to ensure that it strikes (lights up) reliably.

 

[RyreInc]

VE1BLL, we do have both systems you mention, although our bleed resistor will not discharge the caps so quickly that opening the door is "safe" as you describe--that could be a good improvement for us to make.

Looks like a neon bulb could be a solution--these are much more tolerant than I was aware, and the current is lower than I expected as well. I will look into this more. Any good resources?

itsmoked, I've had a similar idea, but manual solutions are not what I'm after--we have plenty of those.

The idea here is to have instant feedback, both for charging and discharging, of whether the caps are at a dangerous voltage. I agree with all that this would be a redundant and possibly counterproductive safety measure, but I think it's an intriguing idea nonetheless.

Thanks everyone for your responses!

 

[iop995]

A small switching converter (or voltage-frequency converter) with a good isolated xfmr witch drive an AC sensible meter may be a variant to study. Powered by caps bank itself.

 

[ScottyUK]

Operating over the range 50V to 3500V? That's quite an input variation to accommodate.

 

[iop995]

Sure, it's difficult to connect direct; need a stage to reduce/scale down input voltage; may be by some power resistors / resistors divider with relay shorting resistors to modify divider ratio accordingly input voltage.

 

[VEBill]

"...may be by some power resistors / resistors divider with relay shorting resistors to modify divider ratio accordingly input voltage."

That approach would be as reliable as a chocolate teapot. An unreliable solution would contribute to negative safety.

How about a voltage meter instead of an indicator light? Even the obvious limitation of resolving 50 volts on a 4kV scale is almost a non-issue; if the voltage drops upon power off, it's not likely to suddenly stop discharging at (for example) 100 volts.

 

[iop995]

Safety is as sure as sure is xfmr insulation; resistors and relays are in primary side and don't influence safety grade, maybe will melt a too close chocolate only.