Author Topic: Ceramic capacitor is of too low voltage?...som,etimes blows up at switch on.  (Read 5635 times)

0 Members and 1 Guest are viewing this topic.

Offline ocsetTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 1516
  • Country: 00
We have an LR8 linear regulator downstream of our mains diode bridge (240vac). The capacitor at the bridge output  (at the input to the LR8) is  just a 100nf, 1812, x7r, 630vdc capacitor. It   sometimes  blows up at switch on. (one in 50 times)
It surely cannot be blowing up because of over voltage above 630v, because the sensitive LR8 would surely have blown up first.
What we suspect is that our board stuffer has accidentally placed a 250V capacitor there, and most often these are surviving in circuit, but every now and again the product is switched on at mains peak, and this 250v part blows up. What do you think?
Do you think this is likely?...ie the board stuffer has accidentally placed a 250v capacitor here, instead of the required 630vdc capacitor?
Surely it’s the only explanation?
Strangely, the capacitor , when blown up, has a hole in it and there’s tarnish  on the PCB near it, but when desoldered the capacitor actually still measures 100nf.
LR8
https://www.microchip.com/wwwproducts/en/LR8
 

Offline Giaime

  • Regular Contributor
  • *
  • Posts: 72
  • Country: it
Maybe too strong of an inrush current?
 
The following users thanked this post: ocset

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2322
  • Country: au
Maybe too strong of an inrush current?

^ This. Inductance in the wiring leading up to the capacitor can create an LC circuit; an LC circuit can (ideally) see a hard switch-on event leading to the voltage peaking to twice the steady state. Interestingly, and perhaps unintuitively, simply adding a capacitor with ESR (i.e., a bog standard electrolytic capacitor) in parallel to the ceramic cap can alleviate this a lot; the ESR damps the LC tank.
 
The following users thanked this post: ocset

Offline Delta

  • Super Contributor
  • ***
  • Posts: 1221
  • Country: gb
Don't forget that the mains voltage in the UK regularly reaches 292vAC RMS for periods of several minutes.
 
The following users thanked this post: SeanB, daveshah, wraper, tooki, ocset

Offline Giaime

  • Regular Contributor
  • *
  • Posts: 72
  • Country: it
^ This. Inductance in the wiring leading up to the capacitor can create an LC circuit; an LC circuit can (ideally) see a hard switch-on event leading to the voltage peaking to twice the steady state. Interestingly, and perhaps unintuitively, simply adding a capacitor with ESR (i.e., a bog standard electrolytic capacitor) in parallel to the ceramic cap can alleviate this a lot; the ESR damps the LC tank.

What makes you think the OP doesn't have an electrolytic in parallel? He can't be that stupid.

Yep we don't know. Maybe that's the input of a boost PFC stage? In that case, I would still expect a plastic X2 capacitor for EMI suppression  :-//
 
The following users thanked this post: ocset

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20355
  • Country: gb
  • 0999
Don't forget that the mains voltage in the UK regularly reaches 292vAC RMS for periods of several minutes.
If the mains supply regularly does that, then there's a problem with your electricity supply.
 
The following users thanked this post: ocset

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
What makes you think the OP doesn't have an electrolytic in parallel? He can't be that stupid.

Did you check which OP we're talking about here? :scared:

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: wraper, tooki, drussell, ocset

Offline Terry01

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: scotland
Maybe too strong of an inrush current?

^ This. Inductance in the wiring leading up to the capacitor can create an LC circuit; an LC circuit can (ideally) see a hard switch-on event leading to the voltage peaking to twice the steady state. Interestingly, and perhaps unintuitively, simply adding a capacitor with ESR (i.e., a bog standard electrolytic capacitor) in parallel to the ceramic cap can alleviate this a lot; the ESR damps the LC tank.

I get that you guys will see this as easy stuff but to a newby like me using the ESR in a cap like that is genius! You learn something new every day right?  8)
Sparks and Smoke means i'm nearly there!
 
The following users thanked this post: ocset

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Don't forget that the mains voltage in the UK regularly reaches 292vAC RMS for periods of several minutes.
If the mains supply regularly does that, then there's a problem with your electricity supply.
That was a joke about some insane thread in the past.
 
The following users thanked this post: Ian.M, tooki, ocset, BergRD

Offline Ice-Tea

  • Super Contributor
  • ***
  • Posts: 3229
  • Country: be
    • Freelance Hardware Engineer
Have you, oh, I don't know... Measured the voltage on the cap?
 
The following users thanked this post: ocset

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 10172
  • Country: gb
That was a joke about some insane thread in the past.

Yes, there should really have been a smiley on that!

(https://www.eevblog.com/forum/beginners/mains-in-uk-can-go-up-to-292vac-for-periods-of-several-minutes/ for context reference, but please don't resurrect it!)
« Last Edit: July 05, 2018, 10:01:00 am by Gyro »
Best Regards, Chris
 
The following users thanked this post: tooki, ocset

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Inductive kick of some sort?
 
The following users thanked this post: ocset

Online mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 14117
  • Country: gb
    • Mike's Electric Stuff
The fact it goes at switch on suggests to me the issue could be current surge. Regardless of voltage rating, there will always be some peak current value that will make it unhappy.
What limits the peak current into this cap? Is there any series resistance before it?

 
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 
The following users thanked this post: ocset

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 10172
  • Country: gb
Of course you don't actually know that it didn't fail short at switch-off, only to manifest itself at the next switch-on.
Best Regards, Chris
 
The following users thanked this post: tooki, ocset

Offline BrianHG

  • Super Contributor
  • ***
  • Posts: 8275
  • Country: ca
    • LinkedIn
Your all crazy, he just needs a 6300vdc cap.  There, problem solved...  :-DD

Next thing to blow up would be the LR8 linear regulator...

As for inductive kick, not much may be needed at all for such a small cap value while just the AC snap on with capacitance in the diode bridge as well, you can hit a massive combo double peak source voltage, + a super fast V/us slew rate into that poor 1812 cap it you switch on the power at just the right time in the AC phase.

If you are going to stick with this cap, the absolute cheapest thing I would do is add an appropriate sized series resistor either just before the rectifier or just after.  However without knowing the rest of the schematic, do not follow me here.
 
The following users thanked this post: tooki, ocset

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Keep in mind that the value is significantly less than rated.  This exaggerates the peak voltage during transient startup or surge conditions.  X7R are bad already, and worse at high voltages.

Note that C0G have more value (at voltage) for the same package volume or cost.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: ocset

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4004
  • Country: nl
My guess is that the inrush current into the capacitor is so high that Ohmic resitance heats a small part of the capacitor too much and it vaporizes some bonding wire or similar.

Did a few experiments a long time ago with a capacitive dropper to light a LED (with anti- parrallel diode) directly from 230Vac. Quite some leds exploded on turn on untill I learned to put in an extra series resistor to limit the inrush current.
A wall power socket can easily deliver several hundreds of Amps without blowing a fuse for a short time such as a switching event.

The LR8 has a current limit between 10mA and 30mA:
http://ww1.microchip.com/downloads/en/DeviceDoc/20005399B.pdf
You can probably use a series resistor between a few hundred Ohm to a few kOhm between 230Vac and your capacitor to limit the inrush current. Make use of this to turn it into a usefull RC filter.

If you have doubts about accidentally wrongly placed caps, then take one of those caps and subject it to a high DC voltage untill it breaks. Compare that voltage to a known good cap.
Beware that you capacitors may explode or release some nasty magic smoke during this experiment, but it is one of the experiments where destroying these components is actually usefull.
 
The following users thanked this post: ocset

Offline chris_leyson

  • Super Contributor
  • ***
  • Posts: 1549
  • Country: wales
No room for a film X2 cap ?
 
The following users thanked this post: ocset

Offline Terry01

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: scotland
My guess is that the inrush current into the capacitor is so high that Ohmic resitance heats a small part of the capacitor too much and it vaporizes some bonding wire or similar.

Did a few experiments a long time ago with a capacitive dropper to light a LED (with anti- parrallel diode) directly from 230Vac. Quite some leds exploded on turn on untill I learned to put in an extra series resistor to limit the inrush current.
A wall power socket can easily deliver several hundreds of Amps without blowing a fuse for a short time such as a switching event.

The LR8 has a current limit between 10mA and 30mA:
http://ww1.microchip.com/downloads/en/DeviceDoc/20005399B.pdf
You can probably use a series resistor between a few hundred Ohm to a few kOhm between 230Vac and your capacitor to limit the inrush current. Make use of this to turn it into a usefull RC filter.

If you have doubts about accidentally wrongly placed caps, then take one of those caps and subject it to a high DC voltage untill it breaks. Compare that voltage to a known good cap.
Beware that you capacitors may explode or release some nasty magic smoke during this experiment, but it is one of the experiments where destroying these components is actually usefull.

I thought a wall socket was only good for 32 Amps? I don't know for sure myself, i'm still learning. I read it on the forum somewhere.
Sparks and Smoke means i'm nearly there!
 
The following users thanked this post: ocset

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
I thought a wall socket was only good for 32 Amps? I don't know for sure myself, i'm still learning. I read it on the forum somewhere.

Some sockets? I guess??

You can easily draw over 2000 amperes from a typical outlet... it's just a question of, for how long.  At that rate, the breaker will trip within a cycle (< 10ms).  Which, mind you, is an electrical eternity -- say a transistor was dropping that voltage (even while drawing far less current -- transistors can't draw fault current, but a fault condition like SMPS shoot-through, or driving a shorted load, will still quickly be fatal), it will die in about 20us.  So the breaker opens after about 500 transistor deaths, by which time the one transistor has thoroughly turned into an expanding plasma ball. :)

Over longer time scales, you want to keep the RMS current below the fuse/breaker rating (whatever that happens to be -- and for UK circuits, I believe that includes a fused plug and cord, versus the US system that assumes any outlet may be asked to deliver full rated current).  RMS is an averaging process, so the peak current can still be much more than rated -- again, it's just a question of, for how long.  :)

In the present case, blowing a ceramic cap will have an equivalent RLC circuit, which will peak at around 1.5A in 3.5us, then 680V at 7us (worst case, switching at the instant of maximum line voltage, and assuming mains inductance ~500uH and capacitance 0.01uF constant).  In practice, the ceramic's capacitance will drop off rapidly as voltage rises, so the peak current will be lower, and the peak voltage will be far higher, probably over 1kV.  At that point, the LR8 is guaranteed to be unhappy; it may well happen that it can bear modest avalanche currents (<1A in this case), but it may also be gradually damaged in the process.  Of course if the capacitor fails internally, you don't get a chance to see if that happened... :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: exe, ocset

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4004
  • Country: nl
In this case it is not the (stable state) of an RLC circuit which has to be considered, but the inrush current.
Inductance of your main cables is hard to guess / measure and varies upon location.

Go blow a few leds like I did a long time ago to get some respect what 230Vac will do with unprotected electronics.
BigClive (youtube) has a very nice switch box for working relatively savely with Mains voltage & electronics.
 
The following users thanked this post: ocset

Offline Terry01

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: scotland
I thought a wall socket was only good for 32 Amps? I don't know for sure myself, i'm still learning. I read it on the forum somewhere.

Some sockets? I guess??

You can easily draw over 2000 amperes from a typical outlet... it's just a question of, for how long.  At that rate, the breaker will trip within a cycle (< 10ms).  Which, mind you, is an electrical eternity -- say a transistor was dropping that voltage (even while drawing far less current -- transistors can't draw fault current, but a fault condition like SMPS shoot-through, or driving a shorted load, will still quickly be fatal), it will die in about 20us.  So the breaker opens after about 500 transistor deaths, by which time the one transistor has thoroughly turned into an expanding plasma ball. :)

Over longer time scales, you want to keep the RMS current below the fuse/breaker rating (whatever that happens to be -- and for UK circuits, I believe that includes a fused plug and cord, versus the US system that assumes any outlet may be asked to deliver full rated current).  RMS is an averaging process, so the peak current can still be much more than rated -- again, it's just a question of, for how long.  :)

In the present case, blowing a ceramic cap will have an equivalent RLC circuit, which will peak at around 1.5A in 3.5us, then 680V at 7us (worst case, switching at the instant of maximum line voltage, and assuming mains inductance ~500uH and capacitance 0.01uF constant).  In practice, the ceramic's capacitance will drop off rapidly as voltage rises, so the peak current will be lower, and the peak voltage will be far higher, probably over 1kV.  At that point, the LR8 is guaranteed to be unhappy; it may well happen that it can bear modest avalanche currents (<1A in this case), but it may also be gradually damaged in the process.  Of course if the capacitor fails internally, you don't get a chance to see if that happened... :)

Tim

I've read this 4 times now and still don't fully understand it but I will...even if I have to read it another 4 or 5 times! It's very very interesting reading.
Thanks for the info..
Sparks and Smoke means i'm nearly there!
 
The following users thanked this post: ocset

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
You can easily draw over 2000 amperes from a typical outlet... it's just a question of, for how long.  At that rate, the breaker will trip within a cycle (< 10ms).  Which, mind you, is an electrical eternity -- say a transistor was dropping that voltage (even while drawing far less current -- transistors can't draw fault current, but a fault condition like SMPS shoot-through, or driving a shorted load, will still quickly be fatal), it will die in about 20us.  So the breaker opens after about 500 transistor deaths, by which time the one transistor has thoroughly turned into an expanding plasma ball. :)


2,000A seems a bit high for a real world situation. Looking up wire resistance I see 12AWG is 1.588 Ohms per 1,000'. If you figure a 50' run between the panel and receptacle then that's 100' round trip so 0.1588 Ohms. If we assume 170V peak and ignore the source impedance of the distribution feeding the panel, resistance of the breaker, receptacle, power cord, etc then I get 1,070A as the absolute peak you could get. Now that's still a whole lot of current, but it's not 2,000A.  Apologies for a pedantic moment.
 
The following users thanked this post: ocset

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
No worries about pedantism, 2kA was merely a rough ballpark figure.  YMMV, depending on many things, of course. :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: ocset

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
If we assume 170V peak and ignore the source impedance of the distribution feeding the panel, resistance of the breaker, receptacle, power cord, etc then I get 1,070A as the absolute peak you could get. Now that's still a whole lot of current, but it's not 2,000A.  Apologies for a pedantic moment.

There are some counties with 230VAC RMS :P
 
The following users thanked this post: ocset


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf