Safety in-circuit discharging HV main caps

[sasa]

I have very simple bleeding resistor for discharging capacitors before in-circuit ESR testing for eventually replacement. It is 100 Ohm, 11W ceramic resistor with leads and it is suitable for low voltage caps.

Even I never measured more than 25V for large main caps, for instance 400V, 47uF, I have to ask what is the safe way to discharge it in-circuit if indeed contain near 400V ?

Performing simple power calculation, it will be 1.6KW through that 100 Ohm resistor, which is far too much and perhaps large metal 22K 25W resistors with montage on heatsink, or even 22K 11W ceramic one could be sufficient, however that would require long time to discharge on safe voltage level.

Personally I would leave that device alone...

[T3sl4co1l]

RTFDS -- and, if that particular resistor series doesn't have surge ratings, then you can find another similar part that will.

Very likely, 1.6kW is fine for a few hundred microseconds.

47uF isn't a lot of energy, and even 1000uF at 450V isn't a lot of energy for a modest value, larger size ceramic resistor.  (100-300W resistors are still pretty cheap.)

And if it's not, so what?  Resistor goes open, but you're double-checking the cap afterwards using the multimeter, right?

(That's the safe, approved, and oh-so-often omitted process for dealing with capacitors; but it sounds like you've probably done this, at least a few times, so my smirk may not be 100% deserved.  Good on 'ya. )

Tim

[sasa]

Resistor goes open, but you're double-checking the cap afterwards using the multimeter, right?

Even my knowledge in electronic is modest, I'm hardly an idiot.
I hope I was quite precise about.

Joke aside, even I never measured more than 25V from such device main caps as already mentioned, shock from fully charged 400V, 47uf is probably enough to disrupt or stop heart muscle.

[Gyro]

I use a 470R 10W vertical ceramic wirewound resistor. It's high enough value to limit the discharge current through a typical smps capacitor while still being low enough to be wirewound rather than more fragile film type. I've never had a failure due to surge dissipation.

A DC jolt is very unlikely to stop the heart, it's AC that causes fibrilation. On the other hand, it is very unpleasant and not worth putting to the test.

[Zero999]

What type of resistor is it?

Wire wound or carbon composition should be fine but film resistors are less tolerant to power surges and can be quickly destroyed by even short surges of over ten times the continuous rating.

[jaycee]

I've used 220 5W taped to the end of a plastic chopstick to discharge the typical line-side capacitors found in SMPS supplies. Never had a problem. Wire wound power resistors are normally rated for many times their nominal rating for brief periods... long enough for most capacitors in such supplies to have discharged.

[sasa]

After closely examin parts from local el. store, the 100R resistor is 11W, max. 750V, axial wirewound rectangular type (MIL type LRW), same as this datasheet:

http://www.megastar.com/Content/pdfs/CPW-SQ-SPS.pdf

They are dirty cheap and I doubt that any lower value than 22K will be able to handle more than specified 11W.

And as well I have access to:
https://www.digikey.com/product-detail/en/te-connectivity-passive-product/HSA2522KJ/HSA2522KJ-ND/2366270

If this formula is correct:
V_t = V₀ * e ^-t/(RC)

Then:
t = ln (V₀/V_t) * RC

For 400V, 2x47uF (usually they are in parallel) and resistor of 22K, 11 or 25W, discharge to 25V around 6 sec (or 13 sec to 1V), then negligible to full discharge with 100R, is not too bad after all.

In any case, extreme caution during connecting appropriate bleeding resistor and appropriate grade HV protective gloves and safety glasses are what I would looking additionally, especially with bigger capacitance caps.

Thank you all for comments.

[Ian.M]

Megastar Datasheet:"Short-time Overload      5 times of rated wattage for 5 seconds"
An 11W resistor that can handle a 55W overload for 5 seconds, can handle 110 joules of energy.

To not exceed the peak wattage, assuming a capacitor charged to 400V, from P=V²/R,  R>=160000/55, which is 2.9K, and to stay within the 110 joule energy limit, from E=C*V²/2, C<110/80000, or 1375uF.

Assuming 100uF of reservoir capacitance, charged to 400V, its time constant with 2.7K will be 0.27 seconds, and after six time constants (1.62 seconds), it will be under a volt, so counting to ten will give it plenty of discharge time.

At the more usual 340V, there'll be a 38% safety margin on the power ratings, so there is little chance of the resistor failing.

The peak current will be slightly under 150mA (from 400V), so there is no pressing need for safety glasses, though, assuming you have a pair with optical quality lenses that wont increase your risk by distorting your vision, they should still be worn, in case you discharge to a point that isn't directly across the capacitor and maybe explode some other component.

I've discharged plenty of such capacitors with a suitable rectangular ceramic body radial leaded resistor held directly in my bare hand - if you hold it by the sides, even if the infill cement is cracked, the risk is minimal, but it might be more prudent to tape it to a tongue depressor or other short heat reistant insulating handle.

[T3sl4co1l]

Note that the energy rating of a resistor drops as the time frame shrinks.  This is because it takes time for heat to dissipate into the bulk of the device.  The same is true for transistors, of course.

Most resistors go as t^(1/3) to t^(1/2), so the peak power goes up gradually and the total energy drops as the pulse length is reduced.  You can use this to estimate the peak rating of a resistor.  (For design purposes, please do perform qualification testing, or obtain the manufacturer's approval, before applying this in practice.)

Wirewound and bulk resistors (not carbon comp: a cermet composite is used for this) are the best for pulse handling.  Wirewound resistors have an optimal resistance range for maximum pulse handling, because small values use thick wire but not very much of it, and large values use lots of wire but it's very thin.  The optimal value uses the thickest wire, wound at the steepest pitch without short circuiting.  This is usually in the 10 ohm to 1kohm range.

Also choose a wirewound with welded terminations.  These survive thermal stress better.

FWIW, the body material and construction does not make much difference.  Cemented box, cylindrical, tubular and vitreous enamel types all have comparable thermal conductivity and heat capacity in the supporting materials, so the time required for heat to diffuse away from the element is comparable.

Film resistors are usually quite good on this, however, because they use alumina, AlN or similar substrates, which are many times more conductive than other ceramics.  They're still easily destroyed with short pulses, but this gives you an edge where low inductance is needed.

Avoid aluminum body resistors, as they use epoxy resin filler which has a low temperature rating and will decompose with repeated surge operation.

Tim

[David Hess]

For additional safety, besides the discharge resistor, I usually wire in a neon bulb and dropping resistor to indicate that the high voltage is charged.  After an unfortunate incident with a 1000 volt DC power supply, I also made it a habit to use as little capacitance as feasible.

[BradC]

I have to ask what is the safe way to discharge it in-circuit if indeed contain near 400V ?

I did just that on Saturday. I used my thumb. It hurt. A lot. Normally I use a 2k2 5W resistor that I keep handy. In this instance I simply forgot to engage my brain before grabbing the PCB.

I like the idea of the Neon.