In the data-sheet of polypropylene DC-Link capacitor (Vishay MKP1848 or Epcos B32778), I can read: "The peak-to-peak ripple voltage (Upp) shall not be greater than 0.2xUndc". Undc is max DC voltage.

Up to now I have assume that AC voltage could be 350Vrms for a 500V MKP capacitor if we assume a very clear and very safe sine voltage. In the reality, it is not the case, and a margin is necessary, like 900VDC for household line use (other email list: with 240VAC in switzerland we have Upp=71%*Undc).

However the data-sheet of DC-Link MKP capacitor, they are more restrictive: the capacitor is design for DC voltage, and Upp (peak-2-peak) should be < 20%*Undc (VDC-max). So for 500V, we should not apply more than 35Vrms and always with the same voltage sign?

Polypropylene is non-polarized capacitor, so why such restriction for DC-Link? Is there a risk to not strictly follow the data-sheet? Where is the problem?

  • \$\begingroup\$ Some non polarised capacitors are designed not to handle AC. It's generally a compramise of capacitance vs ESR vs physical volume when making it. In this instance, I suspect that the ESR is too high for the capacitor to support the current required to fully charge and discharge it continuously. The clue is in the name - DC link capacitor. \$\endgroup\$ – Matthew Sep 24 '16 at 20:34
  • \$\begingroup\$ ESR? maybe no. Check this one: \$\endgroup\$ – ABU Sep 24 '16 at 21:04
  • \$\begingroup\$ ch.farnell.com/fr-CH/vishay/mkp1848c71050jy5/… \$\endgroup\$ – ABU Sep 24 '16 at 21:04
  • \$\begingroup\$ ESR is 0.003 ohm, max current is 19A! \$\endgroup\$ – ABU Sep 24 '16 at 21:06

DC-Link caps are designed to integrate the massive current pulses from voltage ripple and low Dissipation Factor, DF or it's equivalent ESR*Ipk^2*d.f. converted to RMS or using duty factor , d.f. of a pulse.

Historically capacitors have improved and presently, the figure of merit, that I use is

  • T=ESR*C

    • to "ballpark" choose an appropriate cap.
  • 100us~3ms Alum. electrolyte, general purpose , small to very large

  • 0.5~10us Alum. electrolyte, ultra low low ESR
  • 50~150ns Plastic, self healing , high voltage , lower values with higher V

This range will improve over time with R&D and nanotube processed surface on foil conductors.

Thus when you know Ic=C*dV/dt , the Vpp ripple has a direct relation to power dissipation

  • Pd=I²*ESR

This ripple current without chokes or external resistance becomes inversely related to the ripple voltage but duty cycle also reduces so RMS Pd is constant.

  • But if ripple voltage is too large then rms power self heating will damage the product and fail well before rated hours at elevated temperatures. ( highly stressed Arrhenius Effects or highly accelerated stress screen or HASS)

When high rms ripple current is expected consider 20% margin below rated spec. of the component.

Bottom line is one uses large uF Alum caps for low frequency 100Hz and smaller value C plastic caps for SMPS off grid switcher at >=10kHz or step up using PFC when required for>100W in IEC std.

  • \$\begingroup\$ I should say that the current could be a limitation for capacitance value above about 100uF. But if we take this one: \$\endgroup\$ – ABU Sep 24 '16 at 21:15
  • \$\begingroup\$ ch.farnell.com/fr-CH/vishay/mkp1848-510-094k2/… \$\endgroup\$ – ABU Sep 24 '16 at 21:16
  • \$\begingroup\$ The value is 1uF. If I make the calculation, 240VAC/50Hz should give 73mA. Maximum RMS current is 2A. So with 240VAC, we use 3.6% of the max current! \$\endgroup\$ – ABU Sep 24 '16 at 21:19
  • \$\begingroup\$ what part of my explanation did you not understand? what f, Vpp Vdc ripple do you want? \$\endgroup\$ – Tony Stewart EE75 Sep 24 '16 at 21:20
  • \$\begingroup\$ I continue my calculation: ESR is 0.063 ohm, so the dissipated power is 0.3mW. No risk of self-heating! \$\endgroup\$ – ABU Sep 24 '16 at 21:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.