In message 5A9C4644.5030601@yandex.com, Charles Steinmetz writes:

I fully agree, and if it were me, I would absolutely use the best
capacitor I could find at a non-insane price.

But that was exacly my point:  The only thing you (might!) get by
spending an insane amount of money on that capacitor is PTFE and
its better dielectric absorption.

But there are three good reasons not to.

First, they are HUGE, typically a couple of inches in diameter and
four or five inches long[1]

Second, there are no reputable suppliers of ~5µF PTFE capacitors
that I have been able to find, there are only audiohomoeopaths.

Nowhere have I seen anybody buy an audiohomoepathy PTFE capacitor
and publish a traceable DA measurement for it, much less information
about tolerance, lot variations etc.

I am certainly not going to shell out $785.07 for what is claimed
to be a newly produced PTFE capacitor[2]:

https://www.v-cap.com/cutf-capacitors.php

Neither am I going to shell out $34.70 for something which may or
may not be USSR army surplus and which may or may not be PTFE[3].

https://www.ebay.co.uk/itm/Lot-of-1-piece-K72-11-Teflon-Capacitors-0-22uF-4-7uF-125V-1000V-NOS-Tested/132443841947?hash=item1ed644a59b:m:mPF1aOO4F7dJXTSF0qHKcSQ

And third, the difference in DA between PTFE and Polystyrene is
barely a factor five and it is from 0.05%(PS) to 0.01%(PTFE).

That is simply not going to make any difference in an HP5065A.

The trick here is to do the math on the S/N ratio of the optical
signal:  The A9-capacitor is primarily a low-pass filter, and pretty
much any sane capacitor can do that.

Poul-Henning

[1] Almost any other type of capacitor is wound from two layers of
insulator on which a thin layer of metal has been deposited by
evaporation or sputtering.  PTFE must be wound from two PTFE films
and two metal films, which means a lot more metal, because it must
have the mechanical strength for the winding operation.

[2] Notice the claimed "dielectric coefficient of 1.45" ?  Either
he means "Dielectric Absorption" in which case the number is in %
and horrible, or he means "Dielectric Constant" in which case the
number is physically impossible.

[3] Because USSR wasn't very good at PTFE to begin with.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Poul-Henning wrote:

I agree, no PTFE (as I said in a previous message) because it is simply
impractible (or even impossible) to fit on a board that fits into the
5065A.  But as I said in the same message, that is also true of PS
(also, I'm not aware of any currently available in integer uF values).

Last I knew, there are (were?) at least 3 reputable suppliers of PTFE
capacitors, but none of my designs in current production uses them so I
don't have any recent experience.  I do see that Electrocube still lists
the 463D series on a web page with a 2018 copyright date.

I concur with everything you say about alleged PTFE caps from
"alternative" suppliers.

So, no PTFE or PS for practical reasons.  That leaves us with PP, PPS,
and perhaps PC.  These all have reasonably similar DA, so as long as one
checks the particular capacitor's datasheet to verify, DA should not be
much of a factor in choosing between them.

Best regards,

Charles

On Sun, Mar 4, 2018 at 1:04 PM, Charles Steinmetz csteinmetz@yandex.com
wrote:

FWIW, WIMA are claiming 0.05% for their MKP 4 PP capacitors.  They
performed about the same as the original integration capacitor in my 3455A
DMM - and about the same as some Polystyrene I found online somewhere - at
the time intervals that the 3455A uses.

It depends on the quantity they buy. I just looked for X7R 0.1uF in 0805 at
Digikey and the automotive grade (10%, -55 to +125) is cheaper than the non
automotive grade with worse tolerance and more limited temperature range.
Next time you buy it may be reversed...

On Feb 25, 2018 1:51 PM, "Gerhard Hoffmann" dk4xp@arcor.de wrote: