ceramic capacitors explode - need urgent help

[DimaA]

Hi all,
attached a schematics for an input EMI filter for DC-DC PSU
the input voltage is 48V from external power supply
output is 5V 30A

DC_IN_POS is the +48V input
DC_IN_NEG is the 48V return

the chassis ground (the one between RFI_CONTACT1 and RFI_CONTACT2) is connected to DC_IN_NEG in the external power supply (its "-" is connected to chassis), so virtually D3,C6,C7,C8,C9 and R2 are short-circuited to DC_IN_NEG. All capacitors are rated 250V.
D1, D2, D3 are gas-arrestors (surge protectors) that start working on 100V

The problem is that sometimes some of the C1,C2,C3,C4 capacitors are explode. It happens sometimes, but we can't reproduce this. First, I thought it was because of transient voltages caused by input cable insertion, but I tried to plug and unlplug the power cable quickly about 1000 times and nothing happened.

I'm frustrated...

 

[dick_freebird]

Maybe some sort of stress cracking starts it off, and voltage
finishes the job. Have you done any sort of curve tracer
testing, looking for first-fail and all-fail voltage on a bare cap.
virgin board, life-tested board?

 

[karesz]

Hi,
I can not understand your note over shorting in external pwrs!
if the chassie in your external equipment is only to the negative line connected, but is not to the earth wire contacted_they makes not a short for the said components...
Otherwise is a relative dumb situation in my opinion.

I think too, that they are transients in play, but teir are coming from your environment, i.e. is a lift started or some bigger electromotor or an industrial switcher/RF Power source_ some welder, Thyratrons/Thyristors or even a tream is driving outside etc...?
Otherwise you must think on too high ripple current from your DC/DC converter too.
Do you have clean earthing situation?
Maybe you have at the critical time some voltage, a higher as zero on the GND/earth wire, because is a bigger industrial consumer (motor or generator) in short to earth & your aerth system is eventually not optimal_has higher resistance as normal?

I think; you need a longer time monitoring of your incoming voltages & earth line minimal with a DSO.

As another experiment is possible; the said capacitors to select from some other vendor(s), or other charge!?
K.

 

[E-design]

Hi all,


The problem is that sometimes some of the C1,C2,C3,C4 capacitors are explode. It happens sometimes, but we can't reproduce this. First, I thought it was because of transient voltages caused by input cable insertion, but I tried to plug and unlplug the power cable quickly about 1000 times and nothing happened.

I'm frustrated...

Ever had C5 blowing up?
If the answer is no, there is some issue with your shorting out connection RFI1/2 to DC_IN_NEG. If this was a perfect good short then it will place C5 in the same position as C1-4 and exposed to the same transients and stresses, with equal chance of blowing up.

 

[karesz]

Hase it eventually some resonant effects _with some second eqiupment/cabling_in your system?....
These calls for a logging/monitoring over time too...
K.

 

[FvM]

Can you tell about the 1uF/250 caps manufacturer and type? The capacitance value seems unsually high. I also wonder, if they can stand high current transients.

 

[DimaA]

attached a picture of the power supply
I don't know the reason why the "-" connected to chassis, it is off-the-shelf power supply unit, and it has passed all safety tests and has UL certification, so I don't worry about it.

Ever had C5 blowing up?
If the answer is no, there is some issue with your shorting out connection RFI1/2 to DC_IN_NEG. If this was a perfect good short then it will place C5 in the same position as C1-4 and exposed to the same transients and stresses, with equal chance of blowing up.

The C5 capacitor never blow up, but if it has to do with transients, the gas-arrestors would take the transient voltages, they start working at 100V, so if there is a transient voltage over 100V, it will clamp it to 100V. I checked them, they work fine.

Can you tell about the 1uF/250 caps manufacturer and type? The capacitance value seems unsually high. I also wonder, if they can stand high current transients.

These are 1812 size, X7R capacitors, by CAPAX, attached a datasheet
the 100nF are 1206 capacitors, same series.

 

[E-design]

Depending on the shielded cable characteristics and length you may have stray inductance when combined with capacitance causes dangerous fast transients. GDT arresters often need to be used with other protection due to speed issues. You should get a high speed storage scope and search for any abnormally's. The fact that C5 never blows up should invite further investigation.

https://en.wikipedia.org/wiki/Surge_protector

Gas discharge tube (GDT) – These rely on a gas trapped between two electrodes that is ionized by the high voltage to conduct electrical current. GDTs can conduct more current for their size than other components. Like MOVs, GDTs have a finite life expectancy, and can take a few very large transients or a greater number of smaller transients. GDTs also take time to trigger permitting a higher voltage spike to exist before the GDT conducts significant current. It is not uncommon for a GDT to let through pulses of 500V or more of 100ns in duration. In some cases additional protection is necessary to prevent damage due to this effect. GDT create a short circuit when triggered, so that if any electric power (spike, signal, or power) is present, the GDT will short this, and will continue conducting until all electric current sufficiently diminishes. Unlike other protector devices, a GDT will conduct at a voltage less than the high voltage that ionized the gas. Gas arrestors are often used in telecommunication equipment. Due to an exceptionally low capacitance, GDTs are commonly used on high frequency lines.

 

[karesz]

Hi,
Even such "resonant like" effect did I mentioned..
Otherwise; I did never see on switcher/EMC filter GDTs!
Their are in my opinion too incompatible technique with semiconductor environments...
I would change all 3 for MOVs/Varistors_they have no problem with small number of active cycles toK, their are eventually bigger, but GOOD for general transient supressing!.
K.
@DimaA,
100nF as 1210 would have 300V spec...
By Capax datasheets is me really unclear if its a working voltage or testvoltage!? :-(
For me is very interesting,makes thoughtfully, that as one of the best vendors; dont have 200V X7R in 1812...up tu 2225_only maximum 100V _or 330nF for 200V,from 1812 upwards!!
**broken link removed**

 

[FvM]

I see that other manufacturers as AVX have respective C*V values in 1825 or 2225 size. They are e.g. suggested as analog telephone "tip & ring" capacitor. 250 V should be understood as rated DC voltage, meaningful working voltages are typically 50% of it. On the other hand, the capacitors should endure an even higher voltage for short times.

One possible explanation for the capacitor failure would be an incorrect specification by capax. The fact, that apparently no major manufacturer offers a similar form factor at least arouses suspicion. But you should also consider that large MLCCs are in danger to get cracks during reflow solder. Such preceeding damages can perfectly explain "exploding" devices.

The transient overload theory is more unlikely in my opinion.

 

[DimaA]

Thanx all for your replies
I will replace the capacitors to higher voltage ratings, and will add TVS diode in parallel with GDTs. Hope this will help.

 

[karesz]

Hi Dima,

Because the possible failure source is in contact with the negative resistanses of Gas dechargings; I would change the GDTs to some other components, i.e. as TVS or Varistor...
A parallel modus is for me similarly unoptimal as if you have a dry electrolytic (high ESR) and you does parallel such other Cs for better successes.
Good progress!
K.