Capacitance Variations in Cylindrical Capacitors

[mparvin]

The capacitance measurements of “Fuel Gauging Tubes” (FGT) that are composed of carbon composite inner and outer tubes shows unacceptable level of variations and I need to reduce the variation of capacitance.

I would like to know what are the contributing factors to capacitance variations. I am aware of the effect of concentricity, straightness, and circularity of tubes. There is a clamp shape connector mounted over the outer tube that attaches a wire to it by using conductive adhesive. A wire is also glued to the inner tube by the same type of adhesive.

The probe assemblies are located in the aircraft fuel tanks and are utilized to sense the fuel height and density. The probe assemblies are linear coaxial plate cylindrical capacitors, utilizing identical plate spacing. The probes are electrically stimulated by AC signal.

Mike

 

[berkeman]

The main variables would be spacings and the permittivity of the material between the plates. Can you correlate some of the variations with variation in temperature? Are these variations with just one sensor over time, or between many sensors that you receive for production runs? If it's unit-to-unit variation, just calibrate them at the factory. If it's temperature variation, add a temperature sensor and do temperature compensation for the measurements. If it's vibration, make them more robust mechanically, or use a reference unit in a dry spot to subtract out the mechanical variations...

Do you have a web link to a typical one of these devices?

 

[mparvin]

The measurements are done on identical units at room temperature in a static situation and the dielectric between the tubes is air. Calibration is not an option. We have to meet a spec. that calls for limited range of capacitance variations.

I don't have a device uploaded yet.

 

[berkeman]

What do the curves look like for the two units for capacitance versus fluid level? Do they have the same slopes, but different starting values? Or are their slopes different too? If (a), then you can add capacitive ballast to overlay the two curves. If (b), then there is a geometry inconsistency between them. Have you been able to make careful measurements of the geometries of these two units, and compare those to calculations of what those differences would do to the capacitance vs. fluid level curves?

Also, are these devices that you are fabricating yourselves, or do you buy them from somebody? If you buy them, what do their spec. sheets say about the tolerances and variations? Are you trying to figure out how to help the manufacturer improve the product so that you can meet your overall system spec?

And why can't you use calibration to fix this? It's a very common technique in sensor circuit design, as long as you know what variables affect the accuracy of the sensor.

 

[mparvin]

It seems that I have failed to give a clear explanation about my problem I try again:

The observed variation of capacitance is not related to actual application of the device. It is observed during testing and measuring the dry capacitance. The measurements must fall within plus or minus 0.2 pf to be acceptable. We are getting a much large variation and trying to find out the source of it.

Based on the formulation of capacitance: c = 2pi * epsilon*L / log (ri/ro), having assumed that the dielectric of air is constant, the determining factors are length L and radius of tubes. In the formula, it is assumed that the two cylinders are concentric. I would like to know if eccentricity is a contributing factor. To exaggerate the situation, what would happen when the inner tube is very close to the outer tube? In this case, there will be varying distance between the two substrates. How would the capacitance of this off center capacitance compare with a concentric situation? The diameter of the out tube is about 1 inch and inner tube diameter is 0.6 inch. The length is 6 inches.

 

[Bystander]

(snip). The measurements must fall within plus or minus 0.2 pf (snip)

You are using a "jig?" Handling the devices with automated equipment? (Or, at least using gloves?)

0.2 pf change? You get that by moving test leads. Double shielding might help on the instrumentation end. You've still got the problem of locating the capacitor relative to ground in your measurement fixture.

Based on the formulation of capacitance: c = 2pi * epsilon*L / log (ri/ro), having assumed that the dielectric of air is constant,

Bad assumption

the determining factors are length L and radius of tubes. In the formula, it is assumed that the two cylinders are concentric. I would like to know if eccentricity is a contributing factor. To exaggerate the situation, what would happen when the inner tube is very close to the outer tube?

The coaxial geometry is selected for standard capacitors, construction of dielectric cells, level probes, and the like for its insensitivity to misalignment of axes.

You might want to hunt down a copy of Oliver & Cage if this is going to be a long term measurement program.