Safely testing 22 kV capacitors

[Ben Bradley]

In rec.crafts.metalworking and sci.electronics.design, On Thu, 30 Mar
2006 19:19:12 GMT, Ignoramus27088

Anyway, here is my question. I have a 9 kV DC power supply. (a
Franceformer). How can I safely test these caps before selling
them. At 22 kV, they can store about as much energy as a .22 bullet,
according to my calculations.

It would be less at 9 kV, but still, obviously, very deadly.

So. What is a safe way to charge them, verify that they hold the
charge, and then DIScharge them at 9 kV.

I've got an old Eico capacitance bridge/tester, it has a voltage
test function, you can crank it up to 500V and see the magic-eye open
as the capacitor charges (closed means it's pulling current). You turn
the voltage knob back down, and it discharges through the tester. You
can also use the bridge function to measure the capacitance (with
about 10 to 20 percent accuracy).
I agree with whoever else said it, say in your auction you tested
them to 500V or whatever you get them to, and you don't don't feel
comfortable and don't have the equipment to safely test them at higher
voltage. If anyone has any questions about this, just send 'em a
groups.google link to this thread. Well, maybe not, unless you don't
mind your buyer seeing the auction where you bought them.

Here's info on the Eico:
http://home.comcast.net/~btse1/fortrade/eico950.htm
ftp://bama.sbc.edu/downloads/eico/950/
Heathkit and perhaps others made similar devices.

 

[David Billington]

For some reason to do with silly things to do with science this reminded
me of the glass phenomenon called "prince rupert drop"
http://en.wikipedia.org/wiki/Prince_Rupert's_Drop , if you see a glass
blowing demo then ask about it maybe they will show you. If your brave
then have one go of in your hand, numbs it for a minute or so but much
like catching a baseball incorrectly.

 

[[email protected]]

You cant test them properly so why bother trying. As far as safety is
concerned just remember that any charge above 1kv on them is likely to
kill you, above 3kv and its allmost a certainty. Capacitors like these
can recharge themselves long after youve discharged them. Just get rid
of them to someone who knows how to handle them.

After the deadly explosion, as you float above your body, you think to
yourself, "Yup, it failed the test, now what...?"

 

[Rich256]

After the deadly explosion, as you float above your body, you think to
yourself, "Yup, it failed the test, now what...?"

I doubt that the shock from a 1u cap at 20KV is going to kill unless
still hooked to the power supply. However, the reaction (falling
backwards and hitting your head) sure can do damage.

I saw a fellow knocked across a room by 30KV on a large string of
capacitors in a radar pulse forming network.

 

[James Waldby]

....
Why hasn't anyone mentioned yet that if you DO get a HV probe for your
meter it will serve as a resistor to discharge those caps.

I agree with Jeff that using the HV probe is a safe way discharge the
caps. Also, you probably should buy one on ebay, since that will give
you a probe with a proper high-voltage-safe housing and high-voltage-
insulated wire. If you are feeling adventurous and have some high
voltage wire on hand, http://www.repairfaq.org/sam/hvprobe.htm has
directions for making a probe, using a 200-megohm "bleeder resistor
from a defunct video terminal". (Although in newer monitors the
bleeder may be built into the flyback assembly and harder to use.)

It will take a while, but you can observe the progress on the meter it's
hooked up to and when the voltage gets down to something that won't jump
out and grab you, an easier to obtain lower voltage rated resistor can
be used to pull them the rest of the way down quickly.

How long "a while" is depends directly on the probe's resistance. For
example, if it is fairly low, like 200 megohms, the time constant RC
is 2*10^8*10^-6, or 200 seconds. Voltage goes as V0 * e^(-t/RC).
At time t+RC, you will have 37% as much voltage as at time t. If you
start at V0 = 10000V, it will take 4.6 RC, or about 15 minutes, for the
voltage to drop below 100V. If the probe's resistance is reasonably
high, ie 1000 to 5000 megohms, the same drop would take 75 to 375
minutes - assuming the capacitor has no leakage.

Finding out whether (or how badly) each capacitor leaks is probably
the most important thing to test. If you measure the voltage, then
detach the meter, reattach exactly a minute later and remeasure, you
can figure out the capacitor's leakage resistance via the equation
above: R = t / (C*(log V0 - log V1)). (base e logs)

MAKE SURE you watch out for "dielectric adsorption" which is the
property of some capacitors to "recharge themselves" when left open
circuited after you've discharged them. Needless to say, that effect can
make for some nasty suprises. To play it safe, I'd leave a shorting link
or low value resistor connected across them after they're discharged.

Know anyone with a small Van DeGraf generator you could borrow? One of
those could probably charge up those caps slowly enough for you to watch
the voltage rise on the meter, and you could turn it off when you got to
20 KV.

....

I think you should read the high-voltage probe page mentioned above, and
also http://www.angelfire.com/electronic/funwithtubes/Restore_cap.html
which has a number of comments about testing capacitors. Also, I don't
think you should start testing at 9kV all that quickly; ie, test at low
voltages first to get your test setup in order, and to weed out any of
the caps that are too bad to test at HV.

AIUI, a Franceformer is a neon-sign transformer that delivers AC rather
than DC; if so you will need a rectifier. Correct?

If you use line voltage or some other kind of transformer, such as an
oil-furnace igniter, you'll need to use a high-voltage resistor in
series with it and your rectifier to limit the inflow current when
charging a cap. A neon sign transformer probably doesn't need such
a resistor.

To avoid damaging terminals with sparks, you might want to attach some
6" wires to each cap, with hooks or loops formed on the wire ends to
let you hang a test-leads carrier (fiberglass strip at end of leads)
or a shorting bar from a distance.

-jiw

 

[Ian Malcolm]

After looking at HV resistors on ebay, I realized that I may have a
couple in my junk pile (pieces from a radar test set). Will check
tonight. There are probably 5 kV rated, but I could put 2 in series.

i

For the discharge resistor, leave your self some margin. If 5KV rated,
put three identical ones in series if you are testing at 9 to 10 KV.

What procedure are you proposing to non violently check the caps are
actually discharged? Personally I'd want to know they were fully
discharged (well past 99%) before handling them. Also does your plan
include checking the HV probe after use before relying on it to indicate
they are safe?

Insulation can be a problem, beware of surface leakage, all surfaces
need to be extremely clean and absolutely dry. I've used glass up to
about 35KV with a grounded guard ring (in case there is a flaw I haven't
spotted) with no trouble so if you have a large enough quartz glass tube
that should be suitable to hold the resistors. Some uncolored plastics
are also suitable for high voltage work. There is no way you'd get me
using wood unless it was freshly kiln dried then vacuum impregnated with
paraffin wax. Corona discharge can also be a real problem. It
shouldn't be TOO bad at 10KV but you still want to be really carefull to
get nice smooth connections with no sharp edges or points.

High voltage har a really annoying way of finding the slightest
weakness or pinhole in any insulation and many TVs have had a fine
fireworks display from the side of the LOPT on my bench. Its a bit
inconvenient if one is just trying to get the TV powered up successfully
to decide if its condition is good enough to justify LOPT replacement
but with your setup it's potentially lethal. Its also important to
confirm the integrity of the grounding wires in your test circuit. I
left off the ground lead to the aquadag on a large screen TV by mistake
ONCE, (in 12 years) now I'm rather cautious. You goof up your grouding,
yoy'd better allready have got a good deal on a pine box . . .


Wild wacky and way out there idea!! Does anyone have any idea if a
suitable length of carbon core vehicle ignition cable would make a
satisfactory bleeder resistor? Iggy would be dumping 100 joules into
it, its got distributed resistance and sufficient insulation so the main
worry is can a maneagable length take the peak power?

 

[Ian Malcolm]

Wild wacky and way out there idea!! Does anyone have any idea if a
suitable length of carbon core vehicle ignition cable would make a
satisfactory bleeder resistor? Iggy would be dumping 100 joules into
it, its got distributed resistance and sufficient insulation so the main
worry is can a maneagable length take the peak power?

AARGH 1/2*CV^2 so 50 joules for 10KV (up too late!). Hmm, carbon core
cable is even more appealing to me.

 

[J. Clarke]

Last time I did a boom with a capacitor, it broke the "broomhandle" AND
the capacitor terminal leaving a nice leak hole...
And that was ONLY 300V, but 10uF.
I think the resistor way is safer.

For some reason I'm not seeing the original post, so I'll tack this on
here--if their condition is unknown it would be a good idea to have a solid
barrier between you and the capactor when charging and discharging the
first time, and do one at a time..

Also, have good ventilation in the test area--if one blows you probably
don't want to breathe what comes out of it.

 

[Ignoramus27088]

You cant test them properly so why bother trying. As far as safety is
concerned just remember that any charge above 1kv on them is likely to
kill you, above 3kv and its allmost a certainty. Capacitors like these
can recharge themselves long after youve discharged them. Just get rid
of them to someone who knows how to handle them.

Thanks, I definitely will sell all of them, maybe except one.

i

 

[Ignoramus27088]

In rec.crafts.metalworking and sci.electronics.design, On Thu, 30 Mar
2006 19:19:12 GMT, Ignoramus27088


I've got an old Eico capacitance bridge/tester, it has a voltage
test function, you can crank it up to 500V and see the magic-eye open
as the capacitor charges (closed means it's pulling current). You turn
the voltage knob back down, and it discharges through the tester. You
can also use the bridge function to measure the capacitance (with
about 10 to 20 percent accuracy).
I agree with whoever else said it, say in your auction you tested
them to 500V or whatever you get them to, and you don't don't feel
comfortable and don't have the equipment to safely test them at higher
voltage. If anyone has any questions about this, just send 'em a
groups.google link to this thread. Well, maybe not, unless you don't
mind your buyer seeing the auction where you bought them.

Thanks Ben, I appreciate the advice.

i

 

[Ignoramus27088]

Why hasn't anyone mentioned yet that if you DO get a HV probe for your
meter it will serve as a resistor to discharge those caps.

It will take a while, but you can observe the progress on the meter it's
hooked up to and when the voltage gets down to something that won't jump
out and grab you, an easier to obtain lower voltage rated resistor can
be used to pull them the rest of the way down quickly.


MAKE SURE you watch out for "dielectric adsorption" which is the
property of some capacitors to "recharge themselves" when left open
circuited after you've discharged them. Needless to say, that effect can
make for some nasty suprises. To play it safe, I'd leave a shorting link
or low value resistor connected across them after they're discharged.

Know anyone with a small Van DeGraf generator you could borrow? One of
those could probably charge up those caps slowly enough for you to watch
the voltage rise on the meter, and you could turn it off when you got to
20 KV.

Good idea. I will definitely buy a HV probe.

i

 

[Ignoramus27088]

I doubt that the shock from a 1u cap at 20KV is going to kill unless
still hooked to the power supply. However, the reaction (falling
backwards and hitting your head) sure can do damage.

I saw a fellow knocked across a room by 30KV on a large string of
capacitors in a radar pulse forming network.

Just for reference, the energy stored in that cap at 22 kV is
approximately 242 joules. About as much as small caliber rifle
bullet.

i

 

[Ignoramus27088]

For the discharge resistor, leave your self some margin. If 5KV rated,
put three identical ones in series if you are testing at 9 to 10 KV.

Ian, would you or anyone have any idea if what I have is HV resistors?


They are blue, about 3 inches long, and are marked VICTOREEN MOX-3 12
MEG.

What procedure are you proposing to non violently check the caps are
actually discharged?

I would use a HV probe. And then a real short for extra safety, and
then I would short their terminals to cases.

Personally I'd want to know they were fully discharged (well past
99%) before handling them. Also does your plan include checking the
HV probe after use before relying on it to indicate they are safe?

Insulation can be a problem, beware of surface leakage, all surfaces
need to be extremely clean and absolutely dry. I've used glass up to
about 35KV with a grounded guard ring (in case there is a flaw I haven't
spotted) with no trouble so if you have a large enough quartz glass tube
that should be suitable to hold the resistors. Some uncolored plastics
are also suitable for high voltage work. There is no way you'd get me
using wood unless it was freshly kiln dried then vacuum impregnated with
paraffin wax. Corona discharge can also be a real problem. It
shouldn't be TOO bad at 10KV but you still want to be really carefull to
get nice smooth connections with no sharp edges or points.

Thanks for the tips.

i

 

[Martin H. Eastburn]

Come on - slowly ramp up the voltage to 22kv - hold steady - have cap in a cage
box - in case it blows - and then ramp the voltage down. The supply
will draw the charge off the cap. Just remember, if they blow, the case blows...

Now if you want to measure the capacitance at some high voltage - t=RC
time (seconds) = resistance times capacitance. 1t = 66.6% voltage.

Odds are they are just surplus.

Martin

Martin Eastburn
@ home at Lions' Lair with our computer lionslair at consolidated dot net
NRA LOH & Endowment Member
NRA Second Amendment Task Force Charter Founder
IHMSA and NRA Metallic Silhouette maker & member

 

[Martin H. Eastburn]

Discharging caps that way can ruin the cap. Just turn off the supply or
apply a resistor load. The internal plates violently move - ripping themselves up.

You get away with it - smile - sell it - it blows up when used. Not a happy day.

Martin

Martin Eastburn
@ home at Lions' Lair with our computer lionslair at consolidated dot net
NRA LOH & Endowment Member
NRA Second Amendment Task Force Charter Founder
IHMSA and NRA Metallic Silhouette maker & member

 

[Winfield Hill]

Ignoramus27088 wrote...

Got myself some capacitors for $10 apiece.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7603382621
The specs are 22 kV, 1 uF, discharge capacitor. See
http://igor.chudov.com/tmp/cap.jpg

The seller was wrong in describing them as 22 VOLT capacitors.
They are 22 KILO volt capacitors. (the seller is Fermilab).
They are similar to this one:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7598631638
but have twice more capacitance.

Anyway, here is my question. I have a 9 kV DC power supply. (a
Franceformer). How can I safely test these caps before selling
them. At 22 kV, they can store about as much energy as a .22 bullet,
according to my calculations.

It would be less at 9 kV, but still, obviously, very deadly.

So. What is a safe way to charge them, verify that they hold the
charge, and then DIScharge them at 9 kV.

Ignoramus, I strongly protest your post. Here is an interesting
conversation that you started, with 32 posts so far, according
to the Google usenet archive, of which you have made 14, or 44%
of the posts. Yet you marked all your posts "do not archive" -
which means they will be removed from Google Groups in 6 days,
thereby severely damaging the stored thread. With this kind of
callous action on your part, one wonders why others should even
take part in the already-broken, soon-to-be-bogus conversation.
What the h____ are you doing here?

 

[Jon Elson]

OK, I think that we are up to something here.

The test setup that I may try is as follows:

- 9kV franceformer feeding the cap, with remote disconnect leading to
the HV pole

A FranceFormer is a neon sign transformer, and definitely AC.

- Wires from the cap connected to two junk steel flats, all set safely
apart
- A shorting stick (also made of junk steel) with a wooden broomhandle

Hmmm, better be sure that stick is dry, clean and free of cracks in the
wood.

The procedure would be

- connect everything
- put on hearing protection
- turn on the franceformer
- disconnect franceformer with remote disconnect.
- turn face away, short the steel flats using broomhandle
- hear BOOM
- describe testing in the ebay ad

The boom might be quite impressive, as these are probably Marx generator
pulse discharge caps, and may have a current capacity of tens
of thousands of amps.

Jon

 

[Ignoramus6399]

Ignoramus27088 wrote...

Ignoramus, I strongly protest your post. Here is an interesting
conversation that you started, with 32 posts so far, according
to the Google usenet archive, of which you have made 14, or 44%
of the posts. Yet you marked all your posts "do not archive" -
which means they will be removed from Google Groups in 6 days,
thereby severely damaging the stored thread. With this kind of
callous action on your part, one wonders why others should even
take part in the already-broken, soon-to-be-bogus conversation.
What the h____ are you doing here?

Winfield, I am sorry, I changed my settings to remove the XNA header.

i

 

[Ignoramus6399]

The boom might be quite impressive, as these are probably Marx generator
pulse discharge caps, and may have a current capacity of tens
of thousands of amps.

Yes, they say "DISCHARGE CAPACITOR" on the dataplate. They probably
were using them to power high power lasers. One of my acquaintainces
worked there on lasers at some point.

i

 

[Jon Elson]

ok... what would be a good resistor... I figure, 1 megohm or so should
be just about right, RC of these 1 uF caps would be about 1
second. Discharging most of about 40 joules in a couple of seconds.

i

Well, let's see. P = 22 kV squared / 1E6 = 484 Watts.
Well, that isn't so amazing, but I don't know where you will
find a 1 Meg Ohm resistor with a 500 W rating and a 22 KV
voltage rating.

Jon