Re: gEDA-user: ESR of 2.2u ceramic capacitor

[John Doty <jpd@xxxxxxxxxxxxx>]

On Feb 25, 2006, at 6:31 PM, Dan McMahill wrote:

> John Doty wrote:
> > > What you are talking about here is "charge soak" which is often   
> > > modelled as extra capacitance-plus-series-resistance in parallel   
> > > with the ideal capacitor you thought you were buying. I've seen   
> > > time constants varying from a few seconds to a number of  
> > > minutes,  depending on what I was doing. The charge is still in  
> > > the  capacitor, but it's coming out slowly.
> > It can take years. That's how "electrets" work. In the vacuum  
> > tube  era (yes, I'm that old, barely) I remember reading about  
> > folks using   this effect to make a rechargeable "bias cell" for  
> > grid bias!
>
> Thats interesting.  You don't have any schematics by any chance?

Long gone. I believe my collection of old electronics magazines got  
thrown out when my parents moved ~1975, when I was a grad student in  
a tiny apartment without space for all the stuff I'd left in their  
attic.

> Dielectric absorbtion and hysterysis in caps can cause unpleasant  
> issues with various types of integrating A/D converters.  I seem to  
> have a vague memory of someone (Nelson Wright maybe?) doing a  
> thesis at MIT many years back which was a study of dielectric  
> absorbtion and perhaps (can't recall now) some discussion of how to  
> measure it and how it affects some circuits.

I ran into this big time on Chandra. I'd designed a nice low power  
low noise dual slope integrating correlated double sampler for CCD  
video for the HETE-1 mission, and we decided to use it on Chandra.  
I'd used a nice C0G ceramic from Panasonic in the HETE circuits, but  
of course on a billion dollar space mission you can't use Panasonic,  
no, no, no. You have to use something the priesthood has blessed. So  
we put in supposedly equivalent parts from the QPL (100x the cost,  
months to procure) but when we tried the chains the performance was  
*awful*. The QPL capacitors were absolute junk electrically: 5%  
hysteresis! Why anyone would need such a crappy dielectric in a 200  
pf capacitor, and why they'd think this would improve reliability is  
beyond me, but that's aerospace for you. Your tax dollars at work. We  
eventually found a QPL  film capacitor that had no measurable  
hysteresis, and the Chandra chains are superb: system noise is 3  
electrons RMS, and nonlinearity is unmeasurable (<100 ppm).

The Panasonic parts flew on other missions without trouble. On HETE-2  
they've been working in space for >5 years.

> -Dan
>
> (who has in fact designed with vacuum tubes)

When the HETE ground station at Kwajalein was first put into service  
in 2000, the RF preamps had a very short life span. Apparently the  
Kwajalein Missile Range occasionally pointed a megawatt radar in our  
direction :-0. I was thinking 12AT7, but we found a commercial solid  
state preamp that could take the punishment.

My first undergraduate research job was preparing spark chamber  
systems for a particle physics experiment. The ECL drove the  
avalanche transistor which drove the lighthouse tube which drove the  
thyratrons which discharged 20kV from a large bank of capacitors  
through the spark chamber. That lab was *dangerous*.

John Doty              Noqsi Aerospace, Ltd.
jpd@xxxxxxxxxxxxx