Resistor Extreme Pulse Overload experiments

[Gyro]

WARNING: Not the sort of experements to be carried out by the inexperienced, this is NOT the Beginners Section. The output current of the Insulation tester is very strictly limited and has a current trip. The capacitor though is capable of life/underwear threatening energy delivery!

Second warning: Long text!

A couple of weeks back, I repaired a Hunting Hivolt 10kV Insulation tester...

https://www.eevblog.com/forum/repair/hunting-hivolt-_check-it_-10kv-insulation-tester-restored-to-health/

While testing the  leakage of various transformers in my scrap box, it rapidly became clear that I needed a discharge probe! It's surprising how much energy the inter-winding capacitance of small a well insulated transformer can retain.

The problem is, how to safely and painlessly discharge these capacitances at up to 10kV? The largest high voltage capacitor in my collection is an EFCO 22nF plastic film (ok it's rated at 5kV but exhibits no leakage 10kV), it actually measures just short of 25nF (before and after). This seems like an adequate upper capacitance limit - I'm never going to test anything with more parasitic capacitance than that (at that voltage anyway) and never going to use it on connected equipment!

Ok, lets get started...

Test 1: A straight ground lead. This produced a bright blue metal removing arc and a loud crack that startled my wife two rooms away.  Well that wasn't going to be a viable long term solution! I wasn't quite expecting such drama from 22nF! I tested with a 250nF cap at 3kV, much less dramatic, but still noisy. No surprising as it would be holding a similar amount of energy (around 2.2J) as the 10kV one.

Time to try current limiting resistors (going from bottom to top in the attached photo)...

Test 2: A 500R 5W RCL/IWAKI Wirewound resistor. Theoretically this would limit the discharge current at 10kV to 20A, ignoring the drop of the arc discharge which is presumably quite low impedance. Surprisingly better, still a crack and a neat blue arc but nowhere near as dramatic. Similar results at 3kV with the other cap. The question was, how much voltage was actually being dropped across the resistor. 20A also seemed like quite a damaging amount of discharge current for the test subject too.

Time for a larger value...

Test 3: A 68k 2.5W(ish) film resistor (I believed thick film - it looks thick, but not so sure now). Not a good one! It blew the side out of the resistor and made a very loud crack and flash saying "I am not a resistor at 10kV!". Unsurprisingly it read open circuit after the test. I'd obviously jumped too high in resistance at the same time as using the wrong resistor type - Yes, I was half expecting something like this to happen - but not the expulsion of material.

Ok, Let's go for the highest value wirewound resistor that I've got, wirewounds are good for surge...

Test 4: 15k 6W Vitreous wirewound. Rather successful, no destruction, no flashover, reasonably quiet. Strangely the discharge changed from thin, brigh blue and sharply defined, to a less well defined yellowish coloured arc. I wasn't sure whether this was due to lower current density, longer discharge duration...or maybe oscillation due to resistor inductance. The other thing on my mind was the potential for flashover between adjacent turns of the element winding, possibly leading to unpredictable loud cracks a-la Tests 1 and 3 (although I'm sure the glaze wouldn't fly off). I did test it over a fair number of discharges though.

One more try for now then - I'm running out of resistor types. Carbon Composition are supposed to have good surge tolerance, aren't they. Let's try one of those...

Test 5: 22k 2W 1970s Carbon composition. First of all, it had aged up to 25k before I even started testing, however a very successful test (surprisingly), no drama, no flashover. The arc has the same yellow colour as Test 4 (not resistor inductance then). I've repeated this test 50 - 100 times now, with many of the discharges in quick succession (as fast as the tester could recharge the cap). It got very slightly warm at times but that's it. Measuring the resistance afterwards, it is still around 25k (pretty much hidden in the temperature coefficient). I've also tested repeatedly with the 250nF at 3kV too.

I knew that Carbon composition had a good reputation for surge resistance but this is pretty extreme. Ignoring the arc drop, it is passing around 400mA at 10kV, thats a 4kW pulse through a 2W resistor, repeatedly! Of course I've no way of knowing how much the resistance is dropping during the discharge, or the voltage drop across the, now cooler, arc. Even so, the fact that its measured value hasn't changed implies that any internal flashover isn't causing damage, and it is definitely limiting the discharge energy at least as well as the 15k wirewound.

I've adopted this resistor for the probe for now (top of photo). Yes, I've been dipping into the Lidl heatshrink kit again ( ).  It should at least limit any flying debris if the resistor comes apart!

I know it's not very scientific because I was using whatever I had in the parts bin . Don't throw all of those old Carbon composition resistors just yet though!  Thoughts and ideas welcome.

[Tomorokoshi]

That should make for a nice Jesus Stick.

Perhaps one upgrade would be to run a couple 10K or few 4.7K in series to spread out the chance of breakdown.

[SeanB]

I got in a pack some 56K 10W ceramic boat resistors, probably internally metal film with a spiral cut, which are pretty good for this. Just use them with the insulation tester as discharge resistors, though the insulation tester also has a feature of having a built in discharge unit, that provides a current limited discharge when the test button is released. A good addition to the tester is to add a neon lamp and some series resistors ( probably 3 1M carbon film resistors in series each side of the neon lamp) to give a slightly useful indication that there was high voltage on the capacitor being discharged. Would not trust the neon though, it really is just a hint that there was a voltage if lit, not lit can be either neon failed or voltage below 90V.

[Gyro]

Thanks guys, yes both excellent ideas. I know I was really (having fun) pushing to the limits in these tests, I probably wouldn't have posted if it wasn't for the way I found the Carbon Comp resistors can handle such abuse.

I really want to try to keep the resistance value as low as possible (56k wouldn't be out of the way though), firstly because I want to get decent discharge speed at sub kV settings, but also because I want to be able to see visible discharge - as you say Sean, a neon tester won't reliably indicate all the way down (probably a fair way above 90V in fact because the high series resistance will drop the neon current to below visible levels before that). Seeing it definitely spark at a distance, and then again as you bring the probe into direct contact does give a good level of comfort. The DMM is staying well out of the way!

It's one of those itchy problems where you won't find 10kV rated resistors at values low enough to achieve quick discharge times (not unless they have a very high power rating anyway - I mentioned the 4kW pulse disipation). Thankfully in this case it isn't a worry, energy levels are relatively low and there's no problem of ongoing prospective fault current. These tests were very much 'worst case' in terms of stored energy.

I'd be interested in what the element voltage spec is for those 10W resistors is - if we're looking at the same ones it looks like about 550V. Presumably it's the voltage rating that sets the spiral track spacing and they'll reduce that as much as possible to get maximum element coverage. I don't know if you've got 10kV capability?

My main motivation actually is to keep things as compact as possible. The tester itself is very compact and I don't want something too unwieldy when probing small smps transformers.


EDIT: If I'm reading them right, the Pulse performance graphs in this datasheet seem to imply that I might not be pushing my luck that far ....

http://www.ttelectronics.com/sites/default/files/resistors-datasheets/CC.pdf

[daqq]

It's one of those itchy problems where you won't find 10kV rated resistors at values low enough to achieve quick discharge times (not unless they have a very high power rating anyway - I mentioned the 4kW pulse disipation).

You should look at the pulse rating/pulse duration rating rather than the power rating for this kind of application. You can get away with far more using far smaller parts. There are special pulse resistors that can take massive repeated pulsed powers.

For instance, we use these fellows: http://www.hvrint.com/rtrlseries.htm , which can take some serious abuse
Or these guys: https://www.vishay.com/docs/20043/crcwhpe3.pdf (see page 5 for the relevant charts)

[Gyro]

Thanks daqq,

The HVR resistors in particular look very rugged. It's good to see that there are viable pulse handling alternatives. Filed for future reference.

I also found this attempt by Vishay to provide cross references for C-Comp parts in pulse applications. I notice that their best suggested replacements top out at a blanket 3kV due to dielectric strength limitations...

https://www.vishay.com/docs/31049/carbcompcrossref.pdf

Those poor old derided Carbon Composition resistors did have a USP after all, it never occured to me that they were that rugged in pulse applications.

[David Hess]

Ceramic composition resistors are suppose to be replacements for carbon composition parts in high peak voltage applications.  25 kilovolts for a 2 watt part is typical.

My only complaint about them is that they are only available down to 1/2 watt and values are limited compared to carbon composition parts.

[duak]

Years ago I used carbon/graphite ignition wire to discharge small value HV caps.  I remember something like 5 - 10 K ohms/ft.  At higher energies, the folded back and crimped connections could be damaged so it's a good idea to confirm its resistance before using.

I also recall making my own HV resistors by mixing clay and graphite then putting it in a tube with electrodes on each end.  It sort of worked but it needed better packaging as I recall it had to be under a bit of pressure.

Cheers and Happy Turkey Day to those celebrating.

[Gyro]

Ceramic composition resistors are suppose to be replacements for carbon composition parts in high peak voltage applications.  25 kilovolts for a 2 watt part is typical.

My only complaint about them is that they are only available down to 1/2 watt and values are limited compared to carbon composition parts.

Yes, you've hit it on the head. Looking at Ohmite and Tyco datasheets for their Ceramic Composition parts, their 2W parts are pulse rated to 20kV and is still up at 10kV for the 0.5W parts. I hadn't even heard of them before. Farnell/Element14 stock the Ohmite OY series ones for not too silly money. They're still 10% tolerance and temperature coefficient of  -1300 ±300ppm/°C  much like their esteemed predecessors in that respect!

Nobody seems to say what the conductive component in the ceramic matrix is, I bet it's Carbon! (although it could be some sort of metal oxide I suppose).

Looking at the Joule rating for their 2W Carbon-Composition parts I realize that I haven't even been pushing the envelope! 

Happy Independence Day to you folks.

[daqq]

My only complaint about them is that they are only available down to 1/2 watt and values are limited compared to carbon composition parts.

If it's down you are looking for (smaller average power) take a look at https://www.vishay.com/docs/20043/crcwhpe3.pdf . At 0402 size they can survive (according to the datasheet) a ~30W pulse. Or at 1206 it's about 300W. The average power stays the same for things of similar package.

Nobody seems to say what the conductive component in the ceramic matrix is, I bet it's Carbon! (although it could be some sort of metal oxide I suppose).

Could be carbon. I opened one up and the ceramic bit inside was black.

The HVR resistors in particular look very rugged. It's good to see that there are viable pulse handling alternatives. Filed for future reference.

Others make them as well, such as these guys:
http://www.hvproducts.de/en/hv-passive-components/bulk-ceramic-non-inductive-resistors/axial-leaded-resistors.html
Then there's the nuclear option - http://www.hvproducts.de/en/hv-passive-components/bulk-ceramic-non-inductive-resistors/disk-and-washer-resistors.html