Helpe me select a capacitance bridge

[Githin]

I am trying to make a capacitance bridge using two capacitors an inductor. I need a precision ratio transformer. Does anyone have a clue where I can get it from? Can I substitute this with something else? I live in USA.

I have attached a page which describes this sort of sensor.

Many Thanks
Githin F Alapatt

 

[FvM]

manual for ad7745 evaluation board

The frequency range and intended accuracy and resolution haven't been said. If feasible at all, it's an expensive custom made component. You should consider a more recent capacitance measurement method.

 

[Githin]

capacitance bridge noise

Thank you for your reply. I don't care about frequency -as long as it is above 1Hz!

Resolution and stability are the most important things. I want to measure pressure. I am trying to move a plate of the parallel plate capacitor to cause a change in capacitance to measure pressure. I need to see a 1Pa change while applying a 1MPa pressure. (Resolution 10^-6). I can use a plate of max area 100 sq.cm.

What new capacitance measurement techniques can measure this?

 

[FvM]

lock-in capacitance bridge

Are you required to use a classical capacitance bridge? An Analog Devices Capacitance-to-Digital converter as AD7745 would be a today's solution, that basically achieves similar a resolution. (Don't know, if other conditions are contradicting).

I don't believe, that capacitance bridge operation at 1 Hz achieves a sufficient resolution. There are also parameters as current noise to consider. At medium frequency, e.g. a few kHz, the said ratio transformer should be feasible as a fairly ideal component. But unfortunately, I don't know a supplier. Thus I would opt for other more common capacitance bridge designs.

 

[Githin]

ad7746 noise problem

Thank you once again. I already have that capacitance measuring chip's evaluation board sitting on my desk. I have tried with it, there is a lot of noise. Do you know a good way to connect the capacitor to the board to reduce the noise level? I am currently using a mini coax soldered on to the capacitor. I thought the precision ratio transformer might help to reduce the noise - I am not sure. It seems getting this part is hard, so... I don't have an option to try it.

These are the results of the Analysis done on the evaluation board of the AD7745.

Average: 0.725575 pF
RMS Noise: 0.000370 pF
p-p Noise: 0.002137 pF

A 1Pa pressure will cause only a small change in capacitance, so resolution is very important for me. I want that noise as small as possible. I am in a University lab - I have all fancy equipments. But the noise is the problem!

Thanks for all the help.

 

[FvM]

capacitance measurement chip and germany

I'm not clear about the source of the noise observed in your measurement. I assume, that you already tried a filtering of the output up to the maximum suitable time constant?

However, I didn't use the chip in noise limited application yet and can't give detailed suggestions regarding it's operation.

The noise level is several order of magnitudes above the data sheet specification, either it's a matter of interferences or the additional ground capacitance of the setup (coax cables) is degrading the performance.

A state-of-the-art capacitance bridge with a low noise amplifier and synchronous (lock-in) detection is promising a better noise performance than a CMOS mixed signal chip, even if the latter performs optimal signal processing.

I think, the ratio transformer can be replaced by a different circuit in left bridge halve, either variable capacitors or even sufficient low ohmic resistors (not to add too much noise).

 

[Githin]

ad7745/6

Thank you for the reply.

From the data sheet of AD7745/AD7746 - page 11.

The noise is only 4 - 40 aF (WITH NO CAP connected) The data sheet mentions that no caps were connected while doing the noise analysis.

I get 7aF with no connections made to the board. With a capacitor connected, the noise goes up 1000 times.

1. I will do the following - shield the entire setup using aluminum foil and connect that to the power ground using a bleed resistor.

2. Change the conversion time and see if that makes a difference.

I have currently soldered the Coax cables directly to the capacitor. Is there a better way of connecting it?

I didn't understand what you meant by filtering the output. Do you mean a software filter? I have not tried that. I believe a hardware filter will only introduce more noise in the system.

It will be great if I can get this chip to work for my application because it is real cheap and easy to do when compared with the bridge method. I have attached the data sheet for quick reference.

Thanks!

 

[FvM]

ad7746 filter

It may be the case, that the ground capacitance at the C measurement inputs already increases the noise.

I understand, that the specified 19 bit rms Resolution is related to nominal +/- 4pF range, so it should be also achievable with a several pF base capacitance.

Filtering the output means increasing the conversion time and additional averaging. You should get doubled resolution with fourfold measurement time.