X2 CAP loss of capacity - powerline communication

[Gordon Shumway]

First post!

I have a problem with a X2 capacitor that will loose capacity over time.
It is used i a coupling circuit for a ST7540 powerline modem.
See page 46

https://www.st.com/internet/com/TEC...AL_LITERATURE/APPLICATION_NOTE/CD00143379.pdf

I'm transmitting at 110 Khz and L5 is 33uH and C23 is 68nF (305Vac)

Some have run for years with out problems and some fails after a few weeks. Some caps have only a few pF's left when i meassure them (PEAK LCR40). I have tried to protect them with a VDR (275V) from N to P but they still fail.

I have built the circuit (L5 - C23) and applied 3,7 Vrms at 100Khz for 24 hours and the cap worked fine after and did not fail (the circuit did get quite hot).

I have read this thread https://www.edaboard.com/threads/199723/ but dV/dt for my caps are nearly 500V/us

I have searched other powerline modem manufactors* and they say it isn't a critical component as long as it's a X2 and 10% tolerance - in fact - some say "any X2" !!

It seems like a well known problem but is there any suggestions of what i can do?

Regards,

* Arianne, Yitran, Avago and Echelon

 

[FvM]

Did you measure the AC voltage in the series resonant circuit generated by he 3.7 V modulation? It may possibly exceed the capacitor's rating. You may need to refer to a high dV/dt FKP (polypropylene) pulse capacitor.

 

[Gordon Shumway]

Did you measure the AC voltage in the series resonant circuit generated by he 3.7 V modulation? It may possibly exceed the capacitor's rating. You may need to refer to a high dV/dt FKP (polypropylene) pulse capacitor.

Thanks for your reply!

I did measure that. It was about 25 Vpp - if you refer to the point between the cap and the coil.
But the 3.7 V did NOT destroy the cap. And in the final design there is a transil (D9) to GND.

I have a test running now with a relay applying 230 Vac mains at random times (3Hz). Nice fast spikes and my Picoscope gives me Peaks going from -100V to 960V in 500nS It has been runing for a hour and the cap meassures fine.

 

[FvM]

25 Vpp won't be a problem. In this case, I can't answer about the failure cause, expect for doubting bad capacitor quality. I don't know, why your cicruits got hot in the test, but I guess, it's the amplifier.

In any case, using better quality, particularly pulse proof capacitors can be expected advantageous.

 

[Raza]

Hi gordon,
As you said

Some have run for years with out problems and some fails after a few weeks. Some caps have only a few pF's left when i meassure them

show clearly the poor quality of these capacitors. Cheaply made caps simply do not have good contact to the film/foil and reduce C over time under modest operating conditions.

I have built the circuit (L5 - C23) and applied 3,7 Vrms at 100Khz for 24 hours and the cap worked fine after and did not fail (the circuit did get quite hot).

Also shows the leaky capacitor, though it did not fail but caused overloading due to which your circuit got HOT.

 

[Gordon Shumway]

Hi gordon,
As you said

show clearly the poor quality of these capacitors. Cheaply made caps simply do not have good contact to the film/foil and reduce C over time under modest operating conditions.

Also shows the leaky capacitor, though it did not fail but caused overloading due to which your circuit got HOT.

I'm not 100% sure the heat came from the cap, could be the coil (0.14 ohm RDC). They where soldered close together.
They are not particular cheap and they are from a wellknown manufactor.

 

[nancy0116]

I'm not 100% sure the heat came from the cap, could be the coil (0.14 ohm RDC). They where soldered close together.
They are not particular cheap and they are from a wellknown manufactor.

Isn't it rather cheap now? espacially china.