Cap dropper capacitor fail?

[Faringdon]

Hi,
Would you expect C102 to fail? (LTspice and jpeg attached)

N.B. There is a big 220uF lytic there to absorb most mains transients.

The C102 is 68nF, Film, 400VDC, 200VAC

B32521E6683K000
https://product.tdk.com/en/system/files?file=dam/doc/product/capacitor/film/mkt/data_sheet/20/20/db/fc_2009/b32520_529.pdf

As you can see, it does have 235Vrms on it (which is above its rating)....but only a few mA's (6mA) of RMS current.
It can handle up to 400VDC.

(With 6mA RMS current, the ESR would need to be >5555 Ohms to cause more than 200mW of dissipation)

[floobydust]

Yes. "fail" how?

[Zero999]

D11 to D14, R22 and C1 are irrelevant.

If C102 can handle 400VDC, then it should be fine wih 235VAC, assuming it's a sinewave and there are no nasty high voltage spikes.

EDIT:
I've just read the data sheet linked. It appears they derate a lot for AC, which is sensible given the spikes on the mains.

How long does it last for? I wouldn't expect it to fail quickly. R103 will limit the current to some extent. If it's not practical to use a higher voltage capacitor, then try increasing R103.

[T3sl4co1l]

68nF * 5.6kohm = 380.8 us, surge is 20us long, how much peak voltage do you think would show up on it?  What is the peak (short term / test) rating?

Tim

[wraper]

400VDC film caps are almost guaranteed to fail in long term in such application. You need at least 630VDC or X2 cap for more or less reliable operation.

N.B. There is a big 220uF lytic there to absorb most mains transients.

Waste of parts.

[Peabody]

Each time I've had a dropper/zener power supply fail on me, it was never the dropper capacitor that had gone bad.  It's always the output smoothing cap - C109 in this case.  I know that's completely illogical, but that's been my experience.

Edit:  Also, this circuit looks strange to me.  What does the FWBR do?  And shouldn't the rectifier come before the zener?  Well, I guess my amateur status is showing, but it just looks odd to me.

[magic]

No, it's correct.

Capacitor current has to flow both ways. Forward current goes through 1N4148 to the load and excess is shunted by the zener, backward current flows through forward biased zener from "ground" while 1N4148 blocks.

[bdunham7]

What is the purpose of D11-14, R22 and C10?  Is there another HVDC section to this power supply?

[mikerj]

The capacitor is clearly underrated for the application, so I wouldn't expect it to have a long life. 

[Zero999]

The simulation shows a maximum voltage of 335V on the capacitor, but in real life the mains will have higher voltage spikes on it, so the capacitor should have a higher voltage rating to handle them.

[jonpaul]

Just another FTTS troll post.

No reply will ever   solve the problem of Farrington, which is not electronic but psyc.

j

[floobydust]

Whaaa I'm using a part way underrated, please validate my choice to be too cheap and tell me the part will last 
Pretty much sums up OP's threads lately. I guess his product warranty period is low enough.

[Faringdon]

What is the purpose of D11-14, R22 and C10?  Is there another HVDC section to this power supply?

Yes thats the electrolytic for supplying an offline flyback...which supplies a control card etc.

[tggzzz]

Would you expect C102 to fail? (LTspice and jpeg attached)
...
As you can see, it does have 235Vrms on it (which is above its rating)....but only a few mA's (6mA) of RMS current.

No, of course it won't fail when you exceed its rating.

Satisfied now?

Sheesh

[bdunham7]

Would you expect C102 to fail? (LTspice and jpeg attached)

The C102 is 68nF, Film, 400VDC, 200VAC

B32521E6683K000
https://product.tdk.com/en/system/files?file=dam/doc/product/capacitor/film/mkt/data_sheet/20/20/db/fc_2009/b32520_529.pdf

As you can see, it does have 235Vrms on it (which is above its rating)....but only a few mA's (6mA) of RMS current.
It can handle up to 400VDC.

(With 6mA RMS current, the ESR would need to be >5555 Ohms to cause more than 200mW of dissipation)

It's actually not a bonkers question, IMO, despite the fact that you would be running it over the spec.  According to the datasheet it is OK with 200VAC out to about 1kHz, so given it's 400VDC rating why can't it handle a somewhat higher AC voltage at 50Hz?  IDK, and I suppose you would ask TDK.  They may or may not have a solid reason for speccing it that way. 

I see that certain parameters, including maximum dV/dt, are highly dependent on the lead spacing and presumably the resultant overall physical construction.  The 5-mm 630VDC version has more than 4 times the pulse withstanding capability of this model.  That might be better, but an actual X2 rated model might be the best idea.  As for whether it would fail or not it use, I'd guess that a good percentage of them would have reduced life.  You'd be running at ~330V peak in normal operation, any small surge even from motors turning off or whatever might exceed the 400V rating and start to activate the 'self-healing' process, which is cumulatively destructive.  No MOV or other device is going to have room to clamp spikes in the narrow window between normal peaks and 400V.

[tggzzz]

Would you expect C102 to fail? (LTspice and jpeg attached)

The C102 is 68nF, Film, 400VDC, 200VAC

B32521E6683K000
https://product.tdk.com/en/system/files?file=dam/doc/product/capacitor/film/mkt/data_sheet/20/20/db/fc_2009/b32520_529.pdf

As you can see, it does have 235Vrms on it (which is above its rating)....but only a few mA's (6mA) of RMS current.
It can handle up to 400VDC.

(With 6mA RMS current, the ESR would need to be >5555 Ohms to cause more than 200mW of dissipation)

It's actually not a bonkers question, IMO, despite the fact that you would be running it over the spec.  According to the datasheet it is OK with 200VAC out to about 1kHz, so given it's 400VDC rating why can't it handle a somewhat higher AC voltage at 50Hz?  IDK, and I suppose you would ask TDK.  They may or may not have a solid reason for speccing it that way. 

So it fails (more or less often), BadThings^(TM) happen, court cases ensue, and the prosecution presents the case that you deliberately used it beyond its rated values.

Good luck trying to argue your way out of that one!

[floobydust]

We've already discussed film capacitor failures in dropper applications: EEVblog 1486 - The REAL Truth Why Film Capacitors FAIL!

OP despite the 5k6 series resistor (fusible, non-flammable right?) available fault current appears low, but the charged capacitor has enough energy for a damaging internal discharge. The metallization is angstroms thick.

"The applicant must guarantee that the maximum continuous mains voltage is lower than the rated AC voltage and that maximum temporary overvoltages (< 2 s) are lower than 1.6 times the rated DC voltage or 4.3 times the AC rated voltage. Instructions can be found in the application notes and limiting conditions in the specifications."
https://www.vishay.com/docs/28153/acfilmconnectionmains.pdf

Epcos TDK products
https://www.tdk-electronics.tdk.com/download/187726/b0a343f87535444c97d4cfb5eade04ff/pdf-emiapplicationnote-x1-x2.pdf

"The loss of capacitance was due to the corona. Corona is a small but locally intense electrical discharge that injects charge into the insulating film adjacent to edges of foil/metallization or a location where air or moisture is trapped between foil/metallization and the film. The discharge is caused by a voltage gradient large enough to ionize molecules in either the film or small air pockets. Each discharge does some small but cumulative damage to the film. Corona is an important consideration for AC and/or pulse applications where the cumulative damage can rapidly accrue and cause dielectric failure. For capacitors employing metalized film, the “clearing” around the dielectric failure sites results in progressive capacitance loss."
https://gideonlabs.com/posts/ic-emi-mkp-33uf-film-capacitors-failures/

[bdunham7]

So it fails (more or less often), BadThings^(TM) happen, court cases ensue, and the prosecution presents the case that you deliberately used it beyond its rated values.

Good luck trying to argue your way out of that one!

I'm not advocating using an underrated part, although I would point out that 1) there should be a fusible component somewhere as you shouldn't count on the capacitor alone for fire protection and 2) even if underrated, a capacitor in this application will almost always fail low or open anyway.  So the worst case scenario should be that the device stops working, not flames or death.

What I was getting at was that the technical question was not 'bonkers' and hasn't really been satisfactorily answered as far as I can tell.  In a non-mains no-surge-protection-needed environment, why would this capacitor only be rated for 200VAC @ 50Hz?  There may be a good answer, but I don't know what it is.

[Berni]

Like others have said just use a properly rated capacitor for the job.

You want something that is designed to handle being plugged straight across the mains 24/7. This does not only include 240V with +10% so it could be 264V AC (The grid is perfectly allowed to supply that legally) but also an occasional nasty surge that might happen from some nasty devices or lightning.

Not all capacitors fail gracefully at these voltages either. The X and Y rated capacitors are specifically designed to always fail open safely, but just any random film cap might not. I seen improperly rated capacitors on mains in cheep chinese products pop open and blow smoke upon failure.

I built a fair few capacitor dropper supplies before and they are very reliable really. Been running continuously for well over a decade. One of them even got wet from rain due to a enclosure seal failing. It made a mighty bang and blew the fuse, but the only thing that actually needed to be replaced for it to work again is the diodes and limiting resistor, the MCU it was powering survived.

[Peabody]

Zero999 suggested early on that R103 in series with the cap would limit the current.  But wouldn't it reduce the voltage the capacitor sees as well.  As the AC cycles, any voltage that appears across the resistor reduces what the capacitor sees.  Would that permit using a lower rated part?  Well, the resistor would have no voltage drop at zero load, so the cap would see the full AC, but there would be no current flow.  So maybe you still need the full voltage rating.  But unless the zener fails, you're always going to have some load.

Aside from that, this circuit still looks odd to me.  I guess what I'm used to seeing is the dropper capacitor and resistor feeding into a FWBR.  Here you have the bridge rectifier in the circuit, but you don't use it.  Instead, you have essentially a half-wave power supply.  I don't see why you would do that.

[Berni]

Aside from that, this circuit still looks odd to me.  I guess what I'm used to seeing is the dropper capacitor and resistor feeding into a FWBR.  Here you have the bridge rectifier in the circuit, but you don't use it.  Instead, you have essentially a half-wave power supply.  I don't see why you would do that.

The benefit of using halfwave rectification is that you can connect mains N (or L) and circuit GND together.

This is useful if you need to interface to the outside world in some ways. Like switching mains live using a triac but not needing a optotriac to isolate you from it. If you wish to measure mains voltage then you can just use a resistive divider. If you want to measure the power consumption then all you need is a shunt resistor connected to a ADC pin. If you need to sense the state of switches you can have wires going out of it that are near neutral potential...etc

This is the reason i used half way rectification most of the time in capacitor dropper supplies. The larger filtering cap (that is not even very large since these supplies can only provide a few miliamps) is worth it for the other circuit simplification that are possible with it.

[magic]

I also discovered it's possible to use cap dropper to power low voltage circuitry at the negative of full-wave rectified mains, with some fascinating side effects

[Zero999]

Zero999 suggested early on that R103 in series with the cap would limit the current.  But wouldn't it reduce the voltage the capacitor sees as well.  As the AC cycles, any voltage that appears across the resistor reduces what the capacitor sees.  Would that permit using a lower rated part?  Well, the resistor would have no voltage drop at zero load, so the cap would see the full AC, but there would be no current flow.  So maybe you still need the full voltage rating.  But unless the zener fails, you're always going to have some load.

Yes it would, but it would have to be much higher to make a significant difference.

The impedance of the capacitor is much higher than the resistor.
X_C = 1/(2π*FC) = 1/(2π*50*68*10^-9) = 39k

Increasing the series resistor will increase the power dissipation, up to a point until if falls again, but by then the current will be too low for the circuit to work properly.

Aside from that, this circuit still looks odd to me.  I guess what I'm used to seeing is the dropper capacitor and resistor feeding into a FWBR.  Here you have the bridge rectifier in the circuit, but you don't use it.  Instead, you have essentially a half-wave power supply.  I don't see why you would do that.

I think the bridge rectifier is irrelevant to the part of the circuit in question.

[T3sl4co1l]

I also discovered it's possible to use cap dropper to power low voltage circuitry at the negative of full-wave rectified mains, with some fascinating side effects

Well, yeah. Impedance from line to DC makes it look like a doubler, at light enough load where that impedance is small in comparison.

Same reason I can't have EUT probed on the 'scope and plugged into the LISN (or the LISN earth not lifted(!)), at least not unless EUT is active above some power level.

(LISN has a good ~10uF common mode, so a DC ground-return path can get pretty spicy if you were expecting just 120V.)

Tim

[tridac]

Peak half cycle voltage is 235 x 1.4=329 volts, within its rating. I would have used a 500 or 600 volt rating, but unlikely to fail if it's a decent quality. Something like that should have an input fuse at say, 250mA for safety, but that sort of  design is common in cheap led lamps, for example. Also, what is the bridge recifier and 220uF cap doing ?. Its output is not connected to anything. Take all that out and just use a single 1n4008 in series with the cap to the mains. Why spice, when ten minutes building that on the bench would tell you all you need to know ?...