CMOS analogues switch capacitance

[Joe G \(Home\)]

Hi All,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

The 64,000 dollar question is would you expect the input capacitances of the
CMOS switch to be constant over tempreature?

The devices I have looked at are the better versions of the CMOS 4053 etc
.... a few from NXP (formerly Philips)
http://www.standardics.nxp.com/products/switches/


While there are input capacitance specifications..... there is no spec on
the relationship between input capacitance vs Temp.


Would you expect input capacitance to be fairly constant over the operating
temp range?


Regards
Joe

 

[Jim Thompson]

The short answer is that, unless someone has taken great
pains to suppress it, everything changes with temperature.

I would use an extra section of switch to measure the
capacitance of a switch, and subtract the changes of that
that from all the other measurements.

John, _Very_good_suggestion_!

The only relatively constant term would be gate capacitance, ESD and
MOS body diode capacitances will be all over the place with
temperature.

...Jim Thompson

 

[John Popelish]

Hi All,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

The 64,000 dollar question is would you expect the input capacitances of the
CMOS switch to be constant over tempreature?

The devices I have looked at are the better versions of the CMOS 4053 etc
... a few from NXP (formerly Philips)
http://www.standardics.nxp.com/products/switches/


While there are input capacitance specifications..... there is no spec on
the relationship between input capacitance vs Temp.


Would you expect input capacitance to be fairly constant over the operating
temp range?

The short answer is that, unless someone has taken great
pains to suppress it, everything changes with temperature.

I would use an extra section of switch to measure the
capacitance of a switch, and subtract the changes of that
that from all the other measurements.

 

[mkaras]

Hi All,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

The 64,000 dollar question is would you expect the input capacitances of the
CMOS switch to be constant over tempreature?

The devices I have looked at are the better versions of the CMOS 4053 etc
... a few from NXP (formerly Philips)
http://www.standardics.nxp.com/products/switches/


While there are input capacitance specifications..... there is no spec on
the relationship between input capacitance vs Temp.

Capacitance measurements may also be influenced by the series ON
resistance of the switch. The Ron of a switch can vary depending upon
the voltage level of signal so you may also need to take this factor
into account.

- mkaras

 

[Joe G \(Home\)]

The short answer is that, unless someone has taken great pains to suppress
it, everything changes with temperature.

I would use an extra section of switch to measure the capacitance of a
switch, and subtract the changes of that that from all the other
measurements.

Yep, I like it! Like a Weatstone Bridge arrangement.....


In Telecom POTS.... to convert 2wire to 4 wire... a Weatstone bridge
arrngement is often used to cancel the local Tx audio from going in to the
Rx local.

http://en.wikipedia.org/wiki/Wheatstone_bridge

Regards
Joe

 

[Joerg]

Hello Joe,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

The 64,000 dollar question is would you expect the input capacitances of the
CMOS switch to be constant over tempreature?

The devices I have looked at are the better versions of the CMOS 4053 etc
... a few from NXP (formerly Philips)
http://www.standardics.nxp.com/products/switches/

While there are input capacitance specifications..... there is no spec on
the relationship between input capacitance vs Temp.

Would you expect input capacitance to be fairly constant over the operating
temp range?

No :-(

Consider going from coach to business class, for example to a device
like the SD5400. They have become expensive and hard to procure but for
a test station that doesn't enter mass production might be worth a look.

Heed John's advice. Clamping away drifty capacitance by monitoring one
unused section on the same chip is the way to go, just like you'd clamp
out Rdson if that were a concern.

Last but not least, if this is a test station where space and power
consumption might not be of great concern, how about putting the head
assy in a heated box? Doesn't have to be cooking, just far enough above
the highest expected temperature. Peltier regulating the chip is another
option but now it becomes esoteric. If you wanted to go that route you
might be able to scavenge parts from a laser diode arrangement.

 

[[email protected]]

Hello Joe,



No :-(

Consider going from coach to business class, for example to a device
like the SD5400. They have become expensive and hard to procure but for
a test station that doesn't enter mass production might be worth a look.

Heed John's advice. Clamping away drifty capacitance by monitoring one
unused section on the same chip is the way to go, just like you'd clamp
out Rdson if that were a concern.

Last but not least, if this is a test station where space and power
consumption might not be of great concern, how about putting the head
assy in a heated box? Doesn't have to be cooking, just far enough above
the highest expected temperature. Peltier regulating the chip is another
option but now it becomes esoteric. If you wanted to go that route you
might be able to scavenge parts from a laser diode arrangement.

There's nothing all that esoteric about Peltier junctions. Analog
Devices and Linear Technology both sell chips designed to drive them -
ADN8830 and LTC1923. They don't offer great stability (only +/-0.01C)
and they don't work all that well over a wide temperature range, but
they'd be fine in this application.

For more detail, and some references to the literature, see my comment
in Rev.Sci. Instrum.

http://scitation.aip.org/getpdf/ser...75000003000788000001&idtype=cvips&prog=normal

The Linear Technology application ote is pretty good, but ti doesn't
include the formula for calculating the heat transferred (in joules per
amp) of the Peltier junction, which I did include in my comment. Jim
Williams claimed that every equation in an application note halves the
number of readers - I not so sure that this applies to application
notes.

 

[Joerg]

Hello Bill,

There's nothing all that esoteric about Peltier junctions. Analog
Devices and Linear Technology both sell chips designed to drive them -
ADN8830 and LTC1923. They don't offer great stability (only +/-0.01C)
and they don't work all that well over a wide temperature range, but
they'd be fine in this application.

I know but you'd have to get the peltiers which means that you might
have to butcher a laser diode assembly. Nothing wrong with that but this
tends to be expensive.

For more detail, and some references to the literature, see my comment
in Rev.Sci. Instrum.

http://scitation.aip.org/getpdf/ser...75000003000788000001&idtype=cvips&prog=normal

The Linear Technology application ote is pretty good, but ti doesn't
include the formula for calculating the heat transferred (in joules per
amp) of the Peltier junction, which I did include in my comment. Jim
Williams claimed that every equation in an application note halves the
number of readers - I not so sure that this applies to application
notes.

Nah, but the number of multiple integrals with lots of summation terms
behind them does. Because usually those articles remain too far in the
theoretical realms. Or are the usual products of the publish-or-perish
folks. Same for invoking old Maxwell too many times ;-)

 

[Fred Bartoli]

Joe G (Home) a écrit :

Hi All,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

How did you arrange your measuring setup.
It's very possible to make it insensitive to parasitics to ground.

Still, if your setup is sensitive to switch parasitics, then it is also
to your cable capacitance, which will also vary...
So your best bet is to use a setup free from parasitics dependency.

 

[John Popelish]

Yep, I like it! Like a Weatstone Bridge arrangement.....


In Telecom POTS.... to convert 2wire to 4 wire... a Weatstone bridge
arrngement is often used to cancel the local Tx audio from going in to the
Rx local.

http://en.wikipedia.org/wiki/Wheatstone_bridge

Keep in mind that PCB material also changes dielectric
constant with both temperature and humidity. So, if
possible, you might want to mimic a typical layout for the
test switch , so you get a change to compensate that, also.
To get really fancy, you could have one test switch that
has minimal trace area, and one with lots, and by test in a
temperature and humidity chamber, calculate the optimum
correction factors for each other switch as a combination of
these two.

 

[[email protected]]

Hello Bill,


I know but you'd have to get the peltiers which means that you might
have to butcher a laser diode assembly. Nothing wrong with that but this
tends to be expensive.

Farnell stocks a number of parts from the Marlow range of Peltier
junctions - 6W to 71W capacity, for from about $50 to $75 in small
numbers. You would be able to get them via Farnell.

There is at least one website that offers direct ordering

http://www.quick-cool.com/shop/formular.php

Buying a laser diode assembly for the Pelteir junction probably doesn't
represent the optimum procurement strategy.

Nah, but the number of multiple integrals with lots of summation terms
behind them does. Because usually those articles remain too far in the
theoretical realms. Or are the usual products of the publish-or-perish
folks. Same for invoking old Maxwell too many times ;-)

The formula in my comment doesn't include any integrals - it is just a
simple quadratic. It isn't all that accurate, but it worked well enough
in the project for which I had to work it out, and I think we ended up
shipping a fair number - of the order of hundreds rather than tens - of
that system.

It's written up in Measurement Science and Technology - volume 7, pages
1653-1664 in the November 1996 issue.

There isn't a lot of theory in the article - no more than I needed to
explain why we designed the system the way we did - and I published it
because I didn't have enough to do at the time. My chance of getting
work doesn't seem to be dependent on my publication record - and in
fact the Dutch organisation for space research (Stichting
Ruimte-Onderzoek Nederland, http://www.nnv.nl/NNV/CORE/00/00/0.HTML)
who regularly advertise for electronic engineers to do instrument
development, actually refused to interview me on the grounds that the
sort of hands-on engineers that they wanted didn't publish papers in
refereed journals. They can't have read the papers.

I suspect that what this really meant was that they want younger,
cheaper and more biddable engineers, but when pressed they claimed to
have a 58-year-old on their staff ...

If you want a copy of the article, e-mail me - my address is real.

 

[Robert Baer]

Hi All,

I am measuring capacitance in the pF range and I need to switch monitoring
inputs.

We can calibrate out any constant capacitance.. However

The 64,000 dollar question is would you expect the input capacitances of the
CMOS switch to be constant over tempreature?

The devices I have looked at are the better versions of the CMOS 4053 etc
... a few from NXP (formerly Philips)
http://www.standardics.nxp.com/products/switches/


While there are input capacitance specifications..... there is no spec on
the relationship between input capacitance vs Temp.


Would you expect input capacitance to be fairly constant over the operating
temp range?


Regards
Joe

Absolutely *not*.
Use the system to measure its own capacitance (at a given
temperature) before using it to measure attached unknowns, and use
software to calculate the difference.
That capacitance should be reasonably repeatable (ie little
hysteresis) as long as the ICs in the instrument are not changed.

 

[Robert Baer]

John, _Very_good_suggestion_!

The only relatively constant term would be gate capacitance, ESD and
MOS body diode capacitances will be all over the place with
temperature.

...Jim Thompson

DEfine "gate capacitance"; i think it is subject to changes over
temperature.

 

[Robert Baer]

Capacitance measurements may also be influenced by the series ON
resistance of the switch. The Ron of a switch can vary depending upon
the voltage level of signal so you may also need to take this factor
into account.

- mkaras

Not a significant factor; what is worse in using a FET as a switch
for commutating an input to a mesuring system, is the fact that the
applied gate voltage will show up (attenuated by capacitive division) in
the source and drain.

 

[Jim Thompson]

DEfine "gate capacitance"; i think it is subject to changes over
temperature.

Duh! The gate oxide is GLASS... so its changes are quite small.

...Jim Thompson

 

[Robert Baer]

Duh! The gate oxide is GLASS... so its changes are quite small.

...Jim Thompson

Are you trying to tell me that glass does not expand and contract
over temperature???

 

[Tony Williams]

I am measuring capacitance in the pF range and I need to switch
monitoring inputs.

We can calibrate out any constant capacitance.. However

I did a 128-channel capacitor measurement test set, not
in the pF range but it doesn't matter.

What proved most useful was dedicating two channels to
reading back Cstray and (Cref+Cstray), (where Cref was
about 50% of full scale). This allowed an autocalibration
immediately before each scan of the C-unknowns.

We have walked the same 100-odd channel UUT around the
factory, across the 5 test sets. I always have a little
gloat at the consistency between test sets.

If the design is ever revisited I might look at changing the
two reference channels to (Cref+Cstray) and (2*Cref+Cstray),
where 2*Cref would be about 60% of full scale.

 

[Jim Thompson]

Are you trying to tell me that glass does not expand and contract
over temperature???

No, but the gate capacitance varies trivially compared to the
surrounding junctions.

...Jim Thompson

 

[Joerg]

Hello Bill,

Farnell stocks a number of parts from the Marlow range of Peltier
junctions - 6W to 71W capacity, for from about $50 to $75 in small
numbers. You would be able to get them via Farnell.

There is at least one website that offers direct ordering

http://www.quick-cool.com/shop/formular.php

Pretty steep pricing. When having to stabilize oscillators or phase
shifters (usually also C-drift issues) I mostly opted for the
heating-only method. An insulated housing, heat sensing and a big
resistor would keep things rock stable for a Dollar or two.

Buying a laser diode assembly for the Pelteir junction probably doesn't
represent the optimum procurement strategy.

Only for experiments. Not for production, of course.

The formula in my comment doesn't include any integrals - it is just a
simple quadratic. It isn't all that accurate, but it worked well enough
in the project for which I had to work it out, and I think we ended up
shipping a fair number - of the order of hundreds rather than tens - of
that system.

It's written up in Measurement Science and Technology - volume 7, pages
1653-1664 in the November 1996 issue.

There isn't a lot of theory in the article - no more than I needed to
explain why we designed the system the way we did - and I published it
because I didn't have enough to do at the time. My chance of getting
work doesn't seem to be dependent on my publication record - and in
fact the Dutch organisation for space research (Stichting
Ruimte-Onderzoek Nederland, http://www.nnv.nl/NNV/CORE/00/00/0.HTML)
who regularly advertise for electronic engineers to do instrument
development, actually refused to interview me on the grounds that the
sort of hands-on engineers that they wanted didn't publish papers in
refereed journals. They can't have read the papers.

I suspect that what this really meant was that they want younger,
cheaper and more biddable engineers, but when pressed they claimed to
have a 58-year-old on their staff ...

If you want a copy of the article, e-mail me - my address is real.

Sometimes when looking for a job people over here leave achievements off
their resume. It's sad but it might be the only way to avoid being
weeded out as "over-qualified". In general age discimination is
happening and can't be legislated away, doesn't work. However, some
companies have not fared well retaining only the young ones. They just
don't have the experience that is often needed.

 

[Joerg]

Hello Chris,

One other thing (that you would know well) is that the gate capacitance of
the MOS devices will depend strongly on the bias (VGS) so it is important
to measure the calibration channel and the measurement channel with the
same DC bias and about the same signal amplitude on the switch pins.

Another thing to watch is the substrate, for chips where it is brought
out to a pin. During many of my consulting trips I found that much of
the grief was rooted there. The substrate was connected to "somewhere"
and in one case the pin was left open...