In message CAE6XXrhydntKXjzq6W8ZA46Fs048CD-Wvrcbotz7kciGd57j5Q@mail.gmail.com
, Will writes:

Actually the major trouble with Peltier is controlling them, because
they are asymetric with respect to transport direction.

--
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.

I am measuring small capacitances, just for sport but that is more difficult as I thought.
I need something that gives a known capacitance around 1 pF.
I cut several pieces FR4 in different sizes and measured them several ways, but the problem is the dielectric constant if I use K = 4.5 and distance 1.33 ( it is 1.35 thick form outside copper to outside copper so the 1.33 is also a guess) i get calculated results that are in line with my measurements. Not the same values but the same ratios.

I used a digital VNA in shunt mode, a TF1717 bridge from Marconi, the frequency shift methode as described by F.E. terman in RF measurements, a modern LCR meter ( not an expansive very good one) a function generator at 50 KHz and thn measuring the current through the capacitor, an O-opamp plugin with a setup i made to measure small currents ( delta V and delta t are constants so i measure delta i) and the last something call a Capacitance-Frequentie converter i designed, a constant DC current, a constant delta voltage integrator and comprator) and as a result a changing delta T, so frequence is related to capacitance.
All measurements are close but not enough ;-)

so I need a sturdy standard capacitance. Any suggestions, something using air will be best I think but two metal plates should be straight and mounted solid opposite. I used aluminium but forgot the dielectric constant of the oxide so it is not just air. Only if I use K=1.41 I am close. Two seperate peases of pcb ? And then there is the edge effect ect. Most formulas I find are aproximations.

I got some standard caps in the range 100-1000 pF but i want to be able to measure it in fF.

Fred PA4TIM

Fred PA4TIM

Op 13 jan. 2012 om 15:04 heeft "Poul-Henning Kamp" phk@phk.freebsd.dk het volgende geschreven:

In message E7EAFDAE-17F8-48DC-AC7B-F8E42CB367D0@gmail.com, Fred Schneider wri
tes:

I think a major part of your uncertainty is stray fields:  parallel
plates have them all over the place.

I read somewhere that "real metrologists" use a particular design
for which the theoretical value has an analytical solution.

I can't remember the geometry which was described, but using circular
symmetry is a pretty good bet, and coaxial would make even more
sense, because the electrical field is mostly contained.

Found it, where else but at BIPM:
http://www.bipm.org/en/scientific/elec/calc_capacitor/

And here's a good slide-set:
http://www.metas.ch/LesHouches/downloads/talks/24_Bachmair.pdf

--
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.

How accurate does it need to be? Does it need to be "exactly" 1 pF,
or do you want it near there but accurately known? Stable mica caps
in the 1 pF range are fairly common. It's hard to get close initial
tolerance in that range, but if you measure one on an accurate bridge
or LC meter, you can then have a known reference as good as the
measurement, as long as you can replicate the physical conditions.
The C variation in the connections can easily exceed the device
value, so the measurement system should be a 4-wire type. If it needs
to be an "exact" value, an air-variable cap in series with a mica
could be set where you want, but the physical variation issue would
be even more critical.

Ed

At 08:23 AM 1/13/2012, Fred PA4TIM wrote:

Might want to take a look a the manual for the GR 1621A and 1622A
bridges, and the 10 pF air cap standard.

-pete

On Fri, Jan 13, 2012 at 8:48 AM, Poul-Henning Kamp phk@phk.freebsd.dk wrote:

IMO, the best caopacitance bridge is the GenRad (GR) 1615, along with it's
generator and detector. It easily can measure to 10E-15.

-John

===============

Hi Fred:

Measuring components is very different from measuring things with connectors.
In order to get good answers you need to account for stray impedance as well as compensate for things like cable length.

One way to eliminate stray capacitance is to use guarding, for example see:
http://www.prc68.com/I/HP4274_4275_LCR.shtml#G

For the frequency range where they are available the four terminal pair RLC bridges are by far the most accurate method
of measuring impedance.
For more on that and a treasure trove of info on imedance measurements get a copy of the free HP Impedance Measurement
Handbook and the associated Accessories Selection Guide for Impedance Measurements.  Links to those at:
http://www.prc68.com/I/Z.shtml

For really accurate impedance measurements it's best to use an "analyzer" (sweep frequency test with graphic display) as
opposed to a "meter" that has a numeric display. The better impedance analyzers have the ability to fit a number of
models to the measured complex impedance vs frequency plot and give you the circuit values.

For semiconductors their is the HP C-V meter that uses a current source to charge a capacitiance and measures the
voltage.  This data plotted over time gives the semiconductor doping profile.

Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.end2partygovernment.com/Brooke4Congress.html

Fred Schneider wrote:

What about this?

http://www.eevblog.com/2010/07/25/eevblog-101-hacking-your-own-peltier-lab-thermal-chamber/

With best regards Andreas

----- Original Message -----
From: "Will" willvolts@gmail.com
To: volt-nuts@febo.com
Sent: Friday, January 13, 2012 2:53 PM
Subject: [volt-nuts] DIY Air bath

Hi Fred,
I'd go for a coaxial (or Triaxial with a outer screen) air dielectric design. Use largish diameter alloy tubes with a big gap. Adjust the length to suit the tube diameters. The large gap will make the effect of the oxide layer and supports (a thin plastic disc at each end). Connection strays are another issue.

Robert G8RPI.

From: Fred Schneider pa4tim@gmail.com
To: Discussion of precise voltage measurement volt-nuts@febo.com
Sent: Friday, 13 January 2012, 16:23
Subject: [volt-nuts] Small capacitance

I am measuring small capacitances, just for sport but that is more difficult as I thought.
I need something that gives a known capacitance around 1 pF.
I cut several pieces FR4 in different sizes and measured them several ways, but the problem is the dielectric constant if I use K = 4.5 and distance 1.33 ( it is 1.35 thick form outside copper to outside copper so the 1.33 is also a guess) i get calculated results that are in line with my measurements. Not the same values but the same ratios.

I used a digital VNA in shunt mode, a TF1717 bridge from Marconi, the frequency shift methode as described by F.E. terman in RF measurements, a modern LCR meter ( not an expansive very good one) a function generator at 50 KHz and thn measuring the current through the capacitor, an O-opamp plugin with a setup i made to measure small currents ( delta V and delta t are constants so i measure delta i) and the last something call a Capacitance-Frequentie converter i designed, a constant DC current, a constant delta voltage integrator and comprator) and as a result a changing delta T, so frequence is related to capacitance.
All measurements are close but not enough ;-)

so I need a sturdy standard capacitance. Any suggestions, something using air will be best I think but two metal plates should be straight and mounted solid opposite. I used aluminium but forgot the dielectric constant of the oxide so it is not just air. Only if I use K=1.41 I am close. Two seperate peases of pcb ? And then there is the edge effect ect. Most formulas I find are aproximations.

I got some standard caps in the range 100-1000 pF but i want to be able to measure it in fF.

Fred PA4TIM

Fred PA4TIM

Op 13 jan. 2012 om 15:04 heeft "Poul-Henning Kamp" phk@phk.freebsd.dk het volgende geschreven:

volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.