Safely testing 22 kV capacitors

Certainly unpleasant, in the ballpark of a defrib.

Well, you need a HV supply: neon sign xfmr + HV doubler, or a flyback/rectifier from a color TV set. Read it on a DVM with a HV probe, and then just leave the probe on there until it's mostly drained, and kill the last few hundred volts with a resistor.

No big deal.

But if it was a 22KV cap, and it behaves like a 1 uF cap at low voltages, it's very likely fine to sell.

John

Are you sure that Fermilab is the actual seller? Based on what I know of DOE labs, it's not very likely that they would be selling direct on Ebay or even to the public at large.

[ ... ]

Without one, you are SOL -- unless you get other things made for the purpose.

Take a look at eBay auction 7603726273. This is an electrostatic voltmeter. It draws *zero* current. The range is set by pulling out or pushing in the small knob on the end of that long insulator sticking out the back. I can't get this particular auction to show me the other images, so I can't see what the scales really are. Some recent eBay auctions lock up Opera with weird JavaScript.

O.K. Using Mozilla, it looks as though it will go up to 30KV full scale, if I am reading that label on the back properly.

Here is another -- 7605561926

What is it about the sellers of these which keeps them from posting the voltage scales? But these are both dirt cheap for what they are.

You can perform an interesting experiment with one of these and a variable capacitor of the sort used for tuning old AM radios. Set it up with the sections fully meshed (maximum capacitance), and connect it to the meter. Momentarily touch a source of voltage enough to get some deflection on the scale. Now rotate the shaft of the capacitor (which is grounded to the frame, and you will see the voltage swing *way* up, as the number of electrons used to charge it interact with the capacitance changing.

Especially ones which you can *trust*. Generally, the skin oils from handling will produce a leakage path significantly lower than the highest value resistors. I have seen some as high as 2.5 terra Ohms. A spiral carbon film deposited on a ceramic body, and covered with a transparent shrink-on plastic sleeve.

Well ... the discharge route is best done with a large value resistor, attached to the end of about an 18" clear PlexiGlass rod. From the end attached to the rod (drill and tap the rod for a screw), run a wire off to ground. Pick about five 22 MegOhm 2W resistors in series -- solder with short leads so it is fairly stiff, and touch this to the capacitor's terminal. This will put about 4W in each resistor at the full 21 KV. Too much for the resistor long term, but probably enough for the purposes. Make sure that the lead from the rod is *well* grounded. And observe the voltage on a meter as it discharges.

Yes.

Well ... it is possible for them to have failure modes which will only appear at high voltage, and your purchasers will probably expect them to work at their rated voltage.

That is wise.

I've been bitten by as much as 60 KV, but not at high current.

My worst one was actually at 2KV -- but *lots* of current -- from a 100 uF bank of capacitors.

Enjoy, DoN.

According to Ignoramus6399 :

Maxwell Labs caps are really nice quality pulse discharge capacitors.

Hmm ... It says "impregnant MIPB" which sounds like something different from PCBs. You might check whether they are still in business, and ask them.

And these *might* actually surivive a dead-short discharge -- but it is still kinder to them to not do that.

Enjoy, DoN.

DoN, thanks a lot. I will buy a HV probe, the advantage of which is that it can also help me discharge the caps. When I try them, I will let you know. I have a plexiglass rod of some sort (from a hospital dumpster), I will try to test it with a megohm meter and see if it is suitable.

i

Measuring the capacitance will tell you that they are still capacitors, but not that they will hold rated voltage. You could cgarge them up to the capacity of your HV supply, let them sit a while then discharge them with a well insulated shorting bar to see if they still held a charge.

BTW be careful about using the HV from a color TV, they can run well over

22KV on larger sets,

If you have a high voltage probe, the load from the probe itself may be suffcient to discharge the 1 uf caps if you are patient. Charge the cap to only a few hundred volts then put the probe on and see how fast the cap discharges from only the probes load. If this procedes fast enough, you can estimate how long it should take to discharge from a higher test voltage. This has the advantage of requiring no other equipment besides the HV probe. Make sure you read the instructions on how your HV probe is supposed to be used, especially if it requires a separate meter/scope. The meter may need to have a certain minimal input impeadence for the probe to read correctly! I still have a bank of 3 150 mfd oil filled 10KV caps that I used to discharge thru a air gap into a coil for magnetic forming of thin sheet. They weigh about

75 lbs each. They fortunatly have built > Got myself some capacitors for $10 apiece.

According to

"Use a MOT diode: typically they are 9kV 450mA and very cheap." Apparently MOT means microwave oven transformer. Note, the 9kV RMS AC of your Franceformer presumably peaks at about 13kV, so you would need a variac on its input if using a single 9 kV diode. Also see in same thread, which implies that obvious tubes like 1B3 won't handle 30 mA.

-jiw

Nah- 9kVAC is 12.7kV peak, call it 15kV peak. The rectifier has to stand off full voltage on the backswing, so you need 30kV diodes, minimum. 40 or

50kV would be even nicer. ;)

If you make a doubler (which uses diodes of the same rating, but produces twice the output voltage) you can test the capacitors at a bit over rated voltage (maybe 26kVDC, 118% of ratings). All the caps I've bought are rated for 150 or 200% of rated voltage for some time, though that doesn't mean these are.

If you bring up the voltage slowly, through a resistor or variac perhaps, you can monitor it (assuming you get a voltage probe) and stop right at 22kV or so.

Yeah, but conversely, I recall reading an article which stated that modern diodes are avalanche rated, meaning that if the voltage across one diode increases to say, rated PIV, current starts going up (it looks like a really high voltage zener diode), pulling it back into balance.

Capacitors across the diodes were also recommended, but today's diodes are more rugged to pulse and avalanche conditions (we've come a long way from "top hat" diodes!) so this isn't necessary either.

Could rip the diodes out of a few TV sets/monitors. Or use the flybacks themselves, LOL.

Tim

-- Deep Fryer: a very philosophical monk. Website:

["Followup-To:" header set to sci.electronics.design.] On 30 Mar 2006 19:43:34 GMT, Dave Hinz wrote in Msg.

It says "Impregnant: MIPB". That's monoisopropyl biphenyl which doesn't seem to contain halogenes (PCB is polychlorinated biphenyl).

robert

["Followup-To:" header set to sci.electronics.design.] On Thu, 30 Mar 2006 21:44:08 -0600, Jon Elson wrote in Msg.

A great aphorism!

robert

On Thu, 30 Mar 2006 19:19:12 GMT, Ignoramus27088 Gave us:

You bought them just to resell them?

Ok, we will use ONE cap as an example.

A: You do not need to charge the cap up very far, voltage wise, to test/verify the value of the cap.

B: If you wish to test it for leakage, you need a voltage that is closer to the max rated value than your 9kV supply. Do you have an old PC CRT monitor or a TV laying around? The anode voltage of those devices are typically pretty high, and you can look the model up to see exactly what each particular one operates at.

C: You need a high quality high voltage probe to do what you are asking.

I am not sure about your choice of comparative mechanisms, but you shouldn't use these old caps at 22kV. Even when they were new.

One does NOT operate a cap AT its rated voltage (at least, it isn't considered "proper"). They are ALWAYS placed into designs where the applied voltage will never exceed about 70% of the maximum rated voltage. This is true for ALL capacitors actually. Some should even be designed in at only 50% of the expected voltage, meaning that one should use a cap at half its rated voltage in such a case.

For testing, however, with caps that old, I would go no higher than

90% rated voltage.

This is a strange remark. I am not sure that we should be giving you further assistance after seeing this.

The test is very simple, but the procedures are what are strict, and somewhat difficult, and the parts requisite to construct a safe test circumstance may be outside your budget.

If you think that your declaration that "they have been tested" would change their value somehow, you are likely wrong. HV caps are NOT all that easy to test, and your declaration is likely to be taken with a grain of salt by anyone knowledgeable in HV settings.

The test:

Procedural rules.

1: YOU WILL NOT EVER work on or make changes to a live setup. You only get to break this rule once.

2: Making "big fun" arc discharges is a bad thing for the caps, and you supplies, and basically just not a good practice.

3: Ideally, you should have a bird cage or other large metal cage around these test setups, and you should energize the circuits from a distance of several feet.

Parts requisites:

A: A high quality HV probe. 40kV is typical for the level you are at.

B: Twenty to twenty-five feet of 30kV TV anode wire, or better quality "soft" 40kV HV wire, solid ignition wire, or hard teflon HV hook up wire at 30kV or 40kV or even 60kV insulation strength.

C: A ten-pack or two of standard medium sized alligator clips with the rubber sheaths.

D: A good Digital Multimeter with at least 4.5" digits.

E: Four 5" diameter pyrex dishes at about 1.5 to 2" tall.

F: About one quart of good, clean transformer oil.

G: Twenty-five ten Meg Ohm ten watt resistors at 5kV or 10kV voltage rating (for Capacitance Value test load and Quick Discharge Wand).

H: A good anode supply at 20kV set point (use probe to set) (for HV leak test at max voltage charge).

I: A simple, lower voltage supply (300V-500V) (for Capacitance Value test)

J: A 3 or 4 foot length of dry wooden broomstick handle (for Quick Discharge Wand)

K: A 2 foot length of solid #12 or #10 bus or house wire (unsheathed).

L: Some string (2 feet?) for tieing the resistor "wad" together.

M: The biggest cheap teflon cutting board you can find. (18x24, 20x30 whatever the biggest you can get is).

N: A couple or few four ounce sticks of modeling clay.

O: A stopwatch or timer capable watch.

The pre-prep:

A: Cut 4 24" lengths of the HV wire, and strip and tin both ends carefully at a quarter inch length. If it TV anode wire, it will likely already be fully tinned.

B: Attach (install the boots first) two alligator clips to the ends of each wire. These are your HV test jumpers.

C: Take a short piece of the #12 wire, and make a pencil diameter (inside) loop with a two inch tail on it. Sharpen a point on the tail (not needle sharp, just taper it down on the end, it can even be rounded on the point after that. A good x-acto knife will cut copper a bit (enough for this). Bend the loop over 90 degrees. This attaches under your HV probe's end point (unscrew it, then re-tighten).

D: Take two of the Ten Meg Ohm resistors, and snip the lead on one end of each at 3/8". Lay the two short cut ends together, and solder them in "lap joint" fashion. Use a shear to cut the leads )on your wire stripper, not cinch type snips (side cutters or diagonal cutters). Shear cut the remaining ends at approximately 1", and use some of the string to tie each resistor of the pair at the tip of the broomstick handle. Two strings per resistor at their end caps (just inside). Take an eight inch segment of the #12 inch wire, wrap two to four turns of it around the end of the broom stick handle, and leave a two inch length pointing off the end of it, or cut it to that length. Lay the 1 inch resistor lead from the top resistor (closest to stick end) along the coil of large copper wire, and solder it. Wrap the resistors and coil with electrical tape to keep them from shifting. Cover all four string ties, and the center node, and the end coil with a few to several wraps at each location, leaving to remaining resistor lead accessible. Attach one end of an eight foot length of the HV wire to the exposed resistor lead, solder the connection, and tape down about five inches of the HV wire to the top of the stick. You can even zig zag it to remove pull stresses from the lead attachment. Tape it down with several wraps, and include the remaining solder node you just made. This should result in a stick with the last ten or so inches taped up, and a point sticking off the end. Add an alligator clip to the remaining end of the free flying seven foot segment of th HV wire. This is your Safety/Quick Discharge Wand.

E: Place the cutting board on your test bench. Keep the pyrex dishes handy as well as the transformer oil, and clay.

The set-ups:

Capacitance Value Test:

Cutting board components will be the cap under test, the probe, the meter, one HV jumper, one pyrex dish, the medium voltage power supply and the timer. Optional components are one or more of the load resistors, requiring two additional HV jumpers, and one additional pyrex dish.

The long, slow, high statistical reliability test simply uses the probe to load the cap, and the timer to test for the discharge rate. The probe tip, and HV jumper end go into the dish, and the leads get held down to the cutting board by a clump of clay. Oil covers exposed leads in the tray, and the probe rear end can be held elevated by a empty paper towel roll tube cut to length to fit the angle that points the probe tip down into the pyrex dish. The probe should always be in place when energizing the test circuit/setup. Never use the probe on live gear in "free hand" mode.

The HV jumper lead that attaches to the HV side of the cap can remain bare, just like the cap node already is (as pictured). The Grounded side of the cap gets a lead to the MV supply ground, as well as the ground lead from the HV Probe (very important). The MV supply positive lead will also go to the cap's HV node. The banana jack probe leads go into the meter, typically set on the 10 volt range. Again, make sure the ground lead of the HV probe is grounded to the low side of the cap. The probe is typically 1V per 1kV. They are generally 1 Gig Ohm loads, or that is what you should get. You can calculate the discharge rate for a 1G Ohm load at a given voltage on a given cap value. You can then charge the cap with the MV supply, read the voltage on the meter, the remove th MV source lead, and time the discharge period to verify the cap value.

The alternative would be to fashion a set of the load resistors in series, tie the bundle together, immerse it in a second pyrex dish with transformer oil in it. Attach the HV jumper from the cap to one end, and from the other end to the grounded side of the test setup. Attach the remaining HV jumper to the HV cap output node, and to the HV Probe dish. This presents your load resistor as well as the HV probe's loading value to the cap. Charge it up, and time the new, quicker discharge rate, calculate cap value. Test(s) complete. Always ensure that meter reads zero volts before touching any setups. A good practice is to take the discharge wand, and attach the lead clip to ground, and touch the pointed barb to the immersed node in the pyrex dish for a few seconds.

High Voltage Max Charged Leak Test:

Need: One dish, probe, meter, HV supply (danger)

Test: Without cap, attach ground lead of probe to ground side of HV supply. Attach high side of supply to probe tip in oil bath. Turn on meter. Turn on HV Supply, and set output to 20kV. Turn off supply. Ensure that meter reads zero volts before touching anything, or use the discharge wand to ensure that there are no charges hanging around. Place the cap on the cutting board, and attach the ground lead from the HV supply to it. Attach the HV supply output lead to the cap HV node. Attach the HV jumper from the probe dish to the cap's HV node. Make sure that the HV probe ground lead is attached to the HV Supply ground lead. Turn on meter. Turn on HV supply (use single hand/hand in pocket techniques). Read meter. If cap is bad it may arc internally, but will discharge into the HV probe load at a much higher rate than expected. Use wand to discharge after testing, and always reattach shorting jumper that ALL of these caps should have on them (remove them of course during testing)This test isn't 100% definitive or ideal.

Testing to only 9kV will not allow you to declare that "they have been tested" with any validity if that is your goal, since as you say, you are selling them.

See... it isn't as easy as it appears.

On 30 Mar 2006 19:43:34 GMT, Dave Hinz Gave us:

You obviously didn't look at the auction page very well.

On Thu, 30 Mar 2006 19:47:06 GMT, Ignoramus27088 Gave us:

Low voltage testing is not definitive of their capacity at the voltages they will be expected to be working at, and will likely be off by quite a margin.

On Fri, 31 Mar 2006 05:16:04 +0000, snipped-for-privacy@d-and-d.com (DoN. Nichols) Gave us:

With those caps, it would be kinder to all elements involved, including and perhaps even particularly the person causing the short.

Sounds like the fine beginning of a do it yourself rail gun kit

Gunner

"The importance of morality is that people behave themselves even if nobody's watching. There are not enough cops and laws to replace personal morality as a means to produce a civilized society. Indeed, the police and criminal justice system are the last desperate line of defense for a civilized society. Unfortunately, too many of us see police, laws and the criminal justice system as society's first line of defense." --Walter Williams

On Thu, 30 Mar 2006 16:19:47 -0500, Jeff Wisnia Gave us:

If you were observant enough of the thread, you would find that even the original poster mentioned the HV probe loading.

On Thu, 30 Mar 2006 20:55:59 -0600, "Martin H. Eastburn" Gave us:

More likely, they are change outs. Maxwell caps are only rated for a specific number of charge/discharge cycles. After that, expensive, tightly controlled lab environments want new caps so they know exactly how much energy they are working with on each stroke of the bank.

On 30 Mar 2006 19:01:34 -0800, Winfield Hill Gave us:

Are you a mexican illegal immigrant, protestor too?

I can't believe that you wasted so much of your time doing a statistical analysis on a technical thread which you haven't even posted any contributory response to. You are the antipathy of what it is to be "smart".

You're an idiot. Enough of ALL of the responses provided are archived, and would actually make reading the question for a given response easier than scrolling back up through the thread chain. What the f*ck are you doing mouthing off? You COULD actually try posting on topic, contributory material.