140pF to 200pF with 0.1pF accuracy - small design

Hi Shyam,

My main calibration capacitance bridge will measure very accurately form 10 uF dowm to 0.000001 pF or 10 attofarads

It works on the good old ratio transformer principle and is accurate to 0.01% at least... if you're comparing capacitors then it good for less than 1 part per million accuracy!!!

The voltage used to drive the test capacitor is easily adjusted....

John.

Dear John

0.01% measurement is very good one. However, I can not use null method as this is to be done automatically in small device. Perhaps difference voltage, amplification and then digitization will be ideal. Right now using that relaxation oscillator which gives capacitor charging rate at about 10kHz.

I have measured capacitance of devices for control with a colpitts oscillator. The frequency of the oscillator will change based on the capacitance. This is usually very accurate but it depends on how high of a Q you can make. All you have to do is measure the frequency of the oscillator and know the capacitance. This method is much better for control because you don't have to know the absolute value of capacitance but I just though I would throw out the idea.

I am bit worried about the sensor capacitance change with temperature. Humidity itself is temperature dependent parameter. Perhaps one arm of the bridge I need to put at same temperature which sensor is experiencing. However, I need to record the sensor response with temperature perhaps for one humidity parameter or perhaps humidity sense information for one temperature parameter.

Temperature is expected to change from 1C to 60C range for humidity from 5%RH to 95%RH. I do not know which way to start working on this sensor.

Dear Electroman:

Which bridge you use? Can I have more details? Who is the manufactures this? Is it your custom made?

The design given here uses a capacitor bridge. The three other capacitors will have their own temperature dependencies, so if they are good quality and matched capacitors then their contribution to error is minimised. See schematic. Using op-amp with gain of 10 enables output into a precision rectifier (not shown), to convert capacitance to dc amplitude. Simulations show amplitudes at 140pF and 200pF. Excitation signal is 10V amplitude 10kHz. The circuit is nice and linear through the range.

Dear Neil

Thanks for giving detailed circuit scheme wand waveforms. Can you also push me this image on my email.

<sst (at) sensorstechnology.com> and if you have very stable small 10kHz sine wave oscillator circuit then you can give that image also. I can use an OP-amp full wave rectifier to get the DC output. Good information.

Hello Shyam,

To keep it simple:

• use a 555 timer connected as a free running 10kHz oscillator
• Then feed the TRIANGULAR wave from the oscillator into a low pass filter with some gain to get up to 10V sine (or whatever voltage you would like to operate at)
• Don't worry too much about the stability of voltage (or frequency) out of the oscillator. Rather feed the reference sine wave (which is used to excite the capacitance bridge) into a second precision rectifier and then measure the RATIO OF VOLTAGES between capacitance bridge circuit and excitation circuit
• If you are using a microprocessor the two DC signals to be measured could be fed into two A-D inputs with the excitation reference signal being fed into Vref analog input, and the uP will automatically measure correctly.

You can right click on the pictures and save each picture as a separate GIF file. If you would like finer resolution, then I will gladly email them to you.

NeilJ

Sounds OK. I will then require two precision full wave rectifiers and filters for 10kHz. That should not be difficult. Many ADCs have the ratiometric measurement with external reference so I can try that.

As a 'consultant' you are getting paid to implement this capacitance-measuring circuit for your client, yes?

No. I have capacitive Humidity Sensors and I will like to give these for students and wants to see what kind of circuit they can use.

Perhaps I can test few simple designs and let people use these for trial runs if any circuit is good enough for accuracy.

Humidity sensor HS1101 is often used with TLC555 type oscillators but other options are most welcome.

I can't say if others will not use it for commercial gain if anyone finds it useful. Anything protected for commercial use need not be made public here.

Parallax has a number of application notes on their website, including designs based on (cmos) 555s, BASIC Stamp, Propeller and Arduino processors. Parallax also sells inexpensive kits.

Yes I know.

HS1101 sensor work very well in frequency range 5kHz to 10kHz range. It does not work well in lower operating frequencies due to leakage current. Ideas with slow charging and discharge are non=linear.

I think the basic impedance method suggested by NeilJ is good one and operating it at 5kHz / 10kHz fixed frequency likely to yield good results.

With increasing Humidity HS1101 may have larger leakage current and some problem with high humidity is expected.

Forgive my curiosity, but why must this circuit be so precise if you are giving them to your students? Possibly it would be far more instructive if your students themselves came up with ways to improve the basic design which you supply. These are challenges real-world designers face every day, and your students would benefit greatly were you to gently introduce representative challenges early-on. They would certainly emerge from under your tutelage way ahead of the pack.

Many fresh engineering graduates whom I've known over the years were shocked to discover that components have tolerances within which the value could be anything! I've known MIT grad students who did not know how a colour CRT worked, nor why rare-earth magnets slid down an aluminium panel more slowly than an unmagnetised metal slug of comparable size. No real-world experience. You can give your students that if you wish. One day they'll thank you for it, for making them work harder than their peers in other schools. My teachers worked my arse off, and I hated every minute of it, but today I can't thank them enough.

Most of the Engineering students try their own circuits and we do give ten what others have tried and they can verify those also and can compare results. Engineers need to be fit for industry and everything they learn need to be useful sometime or the other. Reference designs give some idea for where to start and what not to repeat. Purpose of education is very complex and only few have special brain to reject all other designs and come up with their own new ideas.

I have published dozen designs for this sensor in Electronics Design Magazine and I do like looking at what other may think can be tried out. You can search some using my name in search on http://electronicdesign.com/

It will be a great idea if students can share and try what others are also doing but without putting lots of funds. Perhaps a club helping them is a good idea.

How do you propose engineering students can be helped? Most o them hardly find support in their institutions. Knowledge of basic industrial sensor electronics and DAQ is very essential these days.