Rippe current rating of aluminium electrolytic capacitor is very low.

Hi,
The Nichicon SMR series 150uF , 400V electrolytic capacitor says it has a tan delta of 0.15.
Also, its ripple current rating is stated as 1.3 * 1.41 ~ 1.83A at 33kHz.

Howcome its only rated for 1.83A of ripple current.?

The ESR = tan delta * Xc = 0.15 * 1/(2.pi.33000.150E-6) = 4.83milliOhms.
With 1.83A of ripple current, the dissipation is only 0.016W. That’s very very low. Surely it can manage more than 1.83A of ripple current?





Nichicon SMR series 150uF,400V

**broken link removed**

Your ESR calculation is unrealistic. Actual ESR numbers of similar capacitors are in a several 100 mOhm range.

The attempt to figure an ESR from ESL and
frequency is inappropriate here because ESL
is not dissipative, but true ESR is. True ESR times
RMS ripple current squared is probably closer
to internal power dissipation and temp rise;
temp rise is the killer (evaporating electrolyte
against whatever seal integrity there may be,
which for cheap electrolytics is not much).

Thanks, so tan delta doesnt actually equal ESR/Xc, but rather the XL vector comes into it too?

The reactive components (Xc, Xl) may -act-
to increase impedance @ frequency, but they
are not expressing what's internally dissipated.

The "ESR" calculation is more of an "effective
impedance" quick number, but even there it
lacks any R value, which there always must
be in addition to C and L.

The capacitor impedance vs frequency curve
should have three regions. Low freq, defined
by Xc. High freq, defined by XL. In the middle,
a trough or flat bottom which is R, the dissipative
portion.

Too bad your datasheet has nothing like that.

Now, that loss tangent is spec'd at 120Hz
probably means XL is negligible (in the power
line filter application) for this purpose, and
maybe a tangent, f and C can give you the
R. But I'd be suspicious of assumptions and
approximations in the formula.

And I would not be so quick to discount L,
when you're trying to chop at 100kHz or more
with edges in the 10-100ns timescale. That is
way off the application paradigm.

Maybe a RLC meter would deliver you the
three components reasonably accurate?
Of course I ain't got one, and I have a much
better than average electronics bench. Just
some DMMs with C capability.

ESR = tan-delta / 2.pi.f.C

however this equation likely does not extend to 33kHz, other losses factor in at 33kHz ( as resistance of bond wires and dielectric ) , also many similar sized electro's are near resonance from 30 - 100kHz, ideally the SRF should be quoted in the data sheet - or seen on a simple sig gen test ....

you can see in the data sheet - the ripple current multiplier for 50kHz is x1.43 giving max ripple at rated temp of 1.86 amps ( 50kHz )

the actual dissipation is Vac x Iac ( in the cap ) x tan-delta, so for 1Vrms & 2A rms = 300mW

at this dissipation the temp will be nowhere near 85 deg C for a 35 deg C ambient say

Hence you can run the cap harder than 2A as long as the core temp does not exceed 85 - 90 C - which is dependent on the VA in the cap and the max ambient temp ....

more explicitly - if the ambient is 0 deg C - then you can run more ripple current ( and hence ripple voltage ) in the cap - than if it is sitting at 70 deg C ambient ...

the actual dissipation is Vac x Iac ( in the cap ) x tan-delta, so for 1Vrms & 2A rms = 300mW

Click to expand...

This of course is correct...though doesnt it only apply if the AC voltage and current waveforms are both pretty well sinusoidal.
(for example, if there was a large ESR or ESL spike in the VAC waveform (due to switching currents) , and the current ripple was sinusoidal, then it wouldnt apply?

Vrms x Irms x Tan-delta always applies as the rms is calculated ....

zenerbjt said:

Hi,
The Nichicon SMR series 150uF , 400V electrolytic capacitor says it has a tan delta of 0.15.
Also, its ripple current rating is stated as 1.3 * 1.41 ~ 1.83A at 33kHz.

Howcome its only rated for 1.83A of ripple current.?

The ESR = tan delta * Xc = 0.15 * 1/(2.pi.33000.150E-6) = 4.83milliOhms.
With 1.83A of ripple current, the dissipation is only 0.016W. That’s very very low. Surely it can manage more than 1.83A of ripple current?





Nichicon SMR series 150uF,400V

**broken link removed**

Click to expand...

The spec sheet indicates that tan delta is given at 120 Hz so you should have:

The ESR = tan delta * Xc = 0.15 * 1/(2.pi.120.150E-6) = 1.33 Ohms.

The Electrician said:

The spec sheet indicates that tan delta is given at 120 Hz so you should have:

The ESR = tan delta * Xc = 0.15 * 1/(2.pi.120.150E-6) = 1.33 Ohms.

Click to expand...

To expand on this, the spec sheet gives the maximum D (tan delta) at a frequency of 120 Hz. The value of D is not the same at 100 kHz as it is at 120 Hz, so you can't calculate the dissipation at 100 kHz from the value of D at 120 Hz.

High voltage electrolytics have a substantially higher ESR than low voltage ones. I don't have the exact capacitor you referred to, but I do have a 100 uF, 400 V capacitor. Here I have used an impedance analyzer to measure ESR and D over a frequency range from 100 Hz to 5 MHz. The image shows the ESR as a yellow curve and D as a green curve. The display is logarithmic in both axes. The vertical axis goes from .001 at the bottom to 1000 at the top. There are two markers, A & B, and the values of ESR and D are shown in the upper right at the frequencies of the markers, 100 Hz for marker A and 101.7 kHz (to approximate 100 kHz) for marker B:



We can see how D increases with frequency. The measured value of D at 100 Hz is .045, but at 100 kHz it's 6.55. The measured ESR at 100 Hz is .843 ohms, and .510 ohms at 100 kHz. These are the values that should used to calculate the dissipation in this particular capacitor.

Thanks, thats interesting, because the shown values of ESR dont tally up with the equation ESR = D * XC.
I suspect its due to ESL.

The equation, with the shown values of D at f (and for 100uF) give 0.716R and 0.102R respectively, for ESR

treez said:

Thanks, thats interesting, because the shown values of ESR dont tally up with the equation ESR = D * XC.
I suspect its due to ESL.

The equation, with the shown values of D at f (and for 100uF) give 0.716R and 0.102R respectively, for ESR

Click to expand...

No, it's not due to ESL. Your calculation used the nominal value for the capacitance of 100 uF. The actual value of capacitance is quite different, especially at the higher frequency. Here are a couple of sweeps showing the measured capacitance; the capacitance is the green curve in both sweeps. The analyzer can only show two parameters at a time, so I have to do two sweeps and change one of the parameters in between. The leads on the capacitor are very short and I have to hold the capacitor in the test fixture. When I hold the capacitor like that, the warmth of my fingers heats the capacitor enough to change the capacitance and ESR for the second sweep. The ESR measured today is larger than it was yesterday because I'm making the measurement earlier in the day and it's colder now. Notice how much the capacitance changes with frequency; you can see the self resonance at about 3 MHz.






If you'll use the measured values of D, C, and ESR, the values tally up quite well.

I wonder about the electrolyte ion-drift time
and how it manifests - ESR, ESL or something
altogether different. I know semiconductor
mobilities vary a lot with temperature and
expect electrolytes do as well, maybe even
more so. Mobility and distance produce a
conductance value. But in the capacitor it
would be a dynamic event, not continuous-
time.

Is this a thing, if so in which of the various
modeling terms might it reside?

dick_freebird said:

I wonder about the electrolyte ion-drift time
and how it manifests - ESR, ESL or something
altogether different. I know semiconductor
mobilities vary a lot with temperature and
expect electrolytes do as well, maybe even
more so. Mobility and distance produce a
conductance value. But in the capacitor it
would be a dynamic event, not continuous-
time.

Is this a thing, if so in which of the various
modeling terms might it reside?

Click to expand...

In the last two images I posted you can see that the capacitance begins to decrease rapidly past about 10 kHz. This is quite pronounced in electrolytic capacitors where the manufacturer has used etched foil for the electrodes. Etching the foil provides a surface with many small channels. This increases the effective surface area of the foil, but the small channels are places where the ion movements are impeded, so at higher frequencies the effective area is reduced and from that, effective capacitance is reduced. Some manufacturer's literature discusses this effect, and shows how to model it.