Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

capacitance

Status
Not open for further replies.

abawan

New Member
Hello every body, I am facing a problem in designing a circuit to measure changing capacitance, I hope that now I have reached at right place for true help.
I want to measure changing capacitance between a diaphragm and ceramic plate. My capacitance range is 10pF to 50pF. I want to measure change in capacitance less then 0.01pF. I need a circuit which takes 12-30V as input and gives 4-20mA current as out put.
 
This is interesting because there are various ways to do this. So some brainstorming is called for to start with. Because I am an RF guy, the first idea that comes to mind to get this fine resolution is to build an oscillator that uses this capacitance to determine its resonant frequency, then measure the frequency of the oscillator. This has some advantages and disadvantages. The biggest disadvantage is that the accuracy will be highly dependent on oscillator drift and general stability, and calibration will be critical. The advantage is that the fine resolution will be possible to achieve, and it isn't very difficult to build an oscillator and a frequency counter circuit.

I have a concern that the +/- .01 pF resolution when measuring 10 to 50 pF is +/- 0.02% (at 50pF). It simply is not reasonable to expect a current loop output (the 4-20mA) to support this resolution. Plus, you suffer reduction in accuracy when the device that is to measure the current at the other end also suffers its own resolution and accuracy limitations.

Hmmm, starting to think that this can't be done. I recommend that you consider using some other interface from the measuring circuit's output, preferably a digital one. Perhaps a serial data link of some sort?
 
Nigel Goodwin said:
Regardless of the current loop problem, 0.02% accuracy is probably somewhat ambitious?.

I estimate that an oscillator running at 13 MHz using the "50 pF" diaghram in parallel with a 100 pF fixed cap resonated with an inductor (1 uH I think it was) would see a shift of about 430 Hz if the capacitance shifts 0.02%. A frequency counter with a 10 ppm accuracy crystal time base would have an accuracy of about +/-130Hz at 13 MHz, so this counter could accurately count the +/-0.02% and resolve the count easily down to 1 Hz, which is approx 0.0002% or so. So resolution is easy, and accuracy is reasonably feasible.

The question of calibration could be resolved if the capacitance change to be measured is only a relative change. In this case, calibration could be done any time the diaghram is at full travel one way or the other, and could be done automatically.
 
Last edited:
Your application sounds similar to process control pressure transmitters that also use a metal diaphragm and measure the deflection of the diaphragm in response to the pressure being applied. However while early designs used capacitance detection, most changed to small resistance strain gauges for better accuracy and repeatability.

Lefty
 
RadioRon said:
I estimate that an oscillator running at 13 MHz using the "50 pF" diaghram in parallel with a 100 pF fixed cap resonated with an inductor (1 uH I think it was) would see a shift of about 430 Hz if the capacitance shifts 0.02%. A frequency counter with a 10 ppm accuracy crystal time base would have an accuracy of about +/-130Hz at 13 MHz, so this counter could accurately count the +/-0.02% and resolve the count easily down to 1 Hz, which is approx 0.0002% or so. So resolution is easy, and accuracy is reasonably feasible.

So your frequency counter is accurate to 0.0002%? - I hate to think how much it cost you, and how much you must spend per year having it calibrated? :)
 
Nigel Goodwin said:
So your frequency counter is accurate to 0.0002%?

Carefull, dont make the mistake of confusing accuracy and resolution.

He is only claiming 0.02% accuracy, but 0.0002% resolution.
Resolution is the smallest measurable change.
Accuracy is the absolute value of the measurement.

The best description I can think of at the moment (in miserable, wet, pouring with rain Oslo)

JimB
 
JimB said:
Carefull, dont make the mistake of confusing accuracy and resolution.

He is only claiming 0.02% accuracy, but 0.0002% resolution.
Resolution is the smallest measurable change.
Accuracy is the absolute value of the measurement.

The best description I can think of at the moment (in miserable, wet, pouring with rain Oslo)

JimB

Right you are. The whole subject of accuracy is both complex and fascinating to me. Over simplification and manufacturer's suspect specifications can really confuse and frustrate one, it really is a science in it's self.

Lefty
 
Thank you JimB and Lefty for chiming in on that question. My meaning was that building a counter with a 10 ppm timebase is not very difficult, and 10 ppm allows measurement of capacitor change down to about .01 % in the example . To answer your question Nigel, my own counter woud indeed be accurate to 0.0001% if I bothered to have it calibrated but as it has not been checked in three years it may be a bit worse than that.

It is not difficult to buy a timebase with thermal drift and 1yr aging combined at better than 0.1ppm (.00001%)
(ie. https://www.vectron.com/products/ocxo/co705.htm)

One of the most economical ways in the last few decades of getting truly accurate time bases is using an oven oscillator that is corrected to NBS WWV (or equivalent European broadcaster) time standard. For example:
**broken link removed**

but surprisingly, things have gotten even cheaper lately in that an appropriate choice of GPS receiver will deliver an extremely accurate 1 pulse per second time base with accuracies as good as +/- 0.0000000001 % (1 e -12). This reference explains how to make such a choice:
**broken link removed**


and for the true accuracy junkie out there, here is an interesting little history of the cutting edge:
https://www.hpmemory.org/wb_pages/wall_b_page_01.htm
 
Last edited:
RadioRon is of course correct that time is the physical property we can measure most accurately. Here is a quote from an article in Scientific American, September, 2002:

"The primary time standard for the U.S., a cesium fountain clock installed in 1999 by the National Institute of Standards and Technology (NIST) at its Boulder, Colo., laboratory, is good to one part in 1015 (usually written simply as 10-15). [edit]...space-based clocks set to fly on the International Space Station by 2005 are expected to tick with uncertainties better than 10-16. And successful prototypes of new clock designs--devices that extract time from calcium atoms or mercury ions instead of cesium--lead physicists to expect that within three years, accuracy will reach the 10-18 range, a 1,000-fold improvement in less than a decade."

Oh, how I love exponents! One part in !0E-18 is less than one second from the beginning of the Universe.

I did not download the whole article, but as I recall from when I read it in 2002, the author from NIST made a comment that they could tell on which floor of the building they were by the relativistic effects of gravity on time.

Now, there are easier and far cheaper ways to tell on which floor of a building one is standing. Which gets back to the OP's question of measuring capacitance. Despite the accuracy of time measurement that is readily available, can the other variables be controlled (with reasonable cost) to attain the accuracy needed for the capacitance measurement requested?

That is meant as a question, not a criticism, because I really don't know the answer. However, just thinking of a few of the factors that can lead to changes in capacitance, makes me think that Nigel's comment about being an ambitious is goal right on. John
 
RadioRon said:
I estimate that an oscillator running at 13 MHz using the "50 pF" diaghram in parallel with a 100 pF fixed cap resonated with an inductor (1 uH I think it was) would see a shift of about 430 Hz if the capacitance shifts 0.02%.
Remove the 100pF fixed capacitor?

It will give you a greater frequency range and the variable capacitance will dominate making it easier to measure.

[quore] A frequency counter with a 10 ppm accuracy crystal time base would have an accuracy of about +/-130Hz at 13 MHz, so this counter could accurately count the +/-0.02% and resolve the count easily down to 1 Hz, which is approx 0.0002% or so. So resolution is easy, and accuracy is reasonably feasible.
Isn't long term stability more important than initial accuracy?

For example if the inductor is 10% out you could calibrate it so it doesn't matter but if it drifts by more than 0.02% due to a slight change in temperature it makes things a lot more complicated.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top