Bi-polar Aluminum Electrolytics

For those interested our SRS capacitance bridge reports the following C and R's For two back to back 100uF 50V Nichion caps. (Freq in Hz, C in uF and R in ohms)

Freq C R

I also looked at the DC voltage across each cap as I polarized the pair to 20 V. The reversed cap starts out at ~7 volts and then 'leaks down to 3.0 Volts or so.... now down to 2.4V. When you trun the bias voltage off both caps become biased in the proper direction at about

George H.

If ESR is important you can either use some of the astonishingly small low-voltage ceramic parts* or you can use back-to-back tantalums.

I've seen tantalums used back to back with a "so high it don't matter" resistor to ground to keep them biased correctly. It worked very well**.

• They make nice microphones, too, if the environment is high enough vibration.

At very low frequencies, that pair looks like 100 uF. At higher frequencies, it's 50. Has to do with the longterm behavior of the aluminum oxide under reverse bias.

I've seen a pair of back-to-back lytics with a diode across each to prevent back-biasing. That looks like C for big signals, C/2 for small ones.

Messy.

John

...

Thanks Tim, 100uF 50V tantalums are going to be expensive I bet. Did you ever look at aluminums?

Ahh a diode, that is a good idea!

I came back after lunch and the reversed cap was down to 0.6 Volts.

Since this is just low pass filtering I don't care too much what the exact C value is, As long as it's big.

Oh, and I'd rather not have it fail at some point in the future.

Thanks George H.

It was an avionics application. Aluminum 'lytics dry out at altitude, _fast_. That kind of restricts your cap selection ability.

I see no reason that the trick wouldn't work with Aluminum caps, if you could stand the ESR and leakage.

Can you use a 22 uF 1206 ceramic or something, or a few mikes of film cap and some BIG resistors?

John

These look like fairly high ESR values for a power application. You talk about a multiplier circuit, but very few of these require polarity reversal, and those that do are high-frequency circuits, where 100uF is likely out of place.

At the same time you say the application is low-pass filtering....

What kind of mess are you getting yourself into?

Sepic or Cuk coupling?

RL

My book says: never put electrolytics in series one way or another.

Just use bipolar electrolytics. These are made to handle AC.

That's what bipolars are. As long as the AC current is (much) less than the surge current it should work.

ance

that

or

E...

aks

as

Not really too much power. Say 10 Watts or so.

Yeah a weird circuit. I'm measuring Johnson noise from a resistor down the end of a probe. There is a heater down the end of the probe too, the wires go near each other and there is some capacitive coupling. So I wanted the filter the noise on the power leads. The variable supply for the heater comes from the (+) and (-) 15 volt rails. With no current returned through ground. So one lead for the heater is filtered (-) 15 volts and the other is variable from ~ -15 Volts to +15 volts. Anyway I've got it all working now. (Without any back to back bipolar's) I tied the 'ground' side of the electro on the output filter to the filtered (-) V line. This is not as good as tieing it to ground (noise wise) but it is good enough.

Thanks again everyone,

George H.