The Designer's Guide Community Forum
Modeling >> Passive Devices >> Dielectric Absorption

Message started by senscho on Jan 6th, 2022, 11:42pm

Title: Dielectric Absorption
Post by senscho on Jan 6th, 2022, 11:42pm

Hi all,

I just finished reading the following article.

The way it explains is that the phenomenon appears when a capacitor is charged for a long time then discharged quickly.

When I look at the modeled circuit, it should also happens when a capacitor is discharged then charged quickly.

Since the article never mention the opposite scenario I wonder whether the circuit model is incomplete (like hysteresis must be modeled. Or the circuit itself is correct in all scenario and the dielectric absorption is also observed when a cap is being charged from a discharged state?

Title: Re: Dielectric Absorption
Post by Ken Kundert on Jan 7th, 2022, 10:07pm

Dielectric absorption is also observed when a cap is being charged from a discharged state.

Asymmetric differences between charge and discharge behavior would imply the capacitor is nonlinear.  Dielectric absorption is not a nonlinear phenomenon.


Title: Re: Dielectric Absorption
Post by senscho on Jan 8th, 2022, 12:05am

Thanks Ken for quick confirmation.

I have a follow up question which might not be related to the modeling but regarding what motivated me to study your article.

Is dielectric absorption observed also from photodiode or PCB (including underfill, mold or whatever material existing) ? If so, what time constant / series R (or current) should I expect as a ball park?

More details below

I fabricated a photo sensor chip and I suspect something is happening on the board (not the chip). My sensor reads very small photocurrent (via TIA / ADC). 1LSB of the sensor corresponds to 0.2nA. Small offset current (a few nA) is measured when the sampling is done within tens of microseconds from PD bias (100mV applied by TIA feedback). Confirmed by various method is that the PD voltage is well settled (also sim confirms with a large margin) within the time. This offset is temperature dependent.

I am currently thinking of doing some PCB or PD investigation. Any tip to confirm dielectric absorption?

Title: Re: Dielectric Absorption
Post by Ken Kundert on Jan 8th, 2022, 11:26pm

Yes, you could very well be experiencing dielectric absorption from the board.  This is a known phenomenon.  I believe it is possible to upgrade your board material to reduce the problem.  Or you could reduce the area of the trace that is exhibiting the dielectric absorption.  Or you can eliminate the ground plane beneath the trace.  Anything to reduce the capacitance.

As far as what time constant to expect, that is the thing about dielectric, there is a continuum of time constants from very low frequencies to very high frequencies.  If you want to test for dielectric absorption, build a capacitor by etching parallel plates on your board material.  Then charge the capacitor for a long time.  Then short capacitor with a ammeter, preferably one that allows you to plot the current over time.  The current through an ideal capacitor will be an impulse.  Or, if the the ammeter has a resistance, then there with be an RC time constant.  Any other time constants  are associated with dielectric absorption.  You will find that the remaining time constants will be hard to pin down because there is so many.

If you want to actually build a model for the absorptive capacitor, you should follow instructions in the paper.


Title: Re: Dielectric Absorption
Post by Maks on Feb 1st, 2022, 11:05pm

I would not exclude semiconductor device related effects - I observed various tricky time-dependent effects in photodetectors and photodiodes.
Especially if you have floating nets, floating wells, etc. - lifetime in semiconductors can be large, seconds or longer (depending on temperature).

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.