Custom capacitive fuel sensors

[htx9a]

I'm working on some more refined capacitive fuel sensors that can compensate for (and likely distinguish between) different fuels and still maintain accuracy. Does anyone know if this has been done before in experimental aircraft? I suspect this is how Boeing/Airbus sense fuel levels.
<blog post>

[pilot2512]

Can't help you with the experimental side but Boeing uses capacitance probes with a densitometer to compensate for fuel densities, except for the 777 which uses ultrasonic probes and a densitometer as well.

[htx9a]

Interesting, I hadn't considered density as another variable. Maybe I'll add a temperature input on version 3 for compensation!

[DennisRhodes]

I would like to see a capacitance fuel gauge that also included a 1 to 5 vdc analog signal out so it could be easily set up on GRT EFIS as well as others. My limited exposure to that was trying to set up Vans capacitance gauges to read on a GRT EFRI . About all it found was that the capacitance gauge operated on a output frequency rather than a linear voltage amplitude that could be scaled. There were some modules that could convert except they were rather expensive time two. Settled for the same old EI gauge that has been in there since 2003 and still works well.

[Jpm757]

I have the Van's capacitive setup, and am using the Dynon capacitance to voltage converters. They work well connected to my AFS 5600's. $50 ea. from Spruce.

[N49ex]

The 4 input circuit pictured - where can you get that?

[dmn056]

You don't need to worry about density. Boeing needs it because turbine engines traditionally manage fuel quantity in terms of mass (lbm or kg), whereas we use volume (litres or USgal), which can be measured directly just using capacitance.

Also, 28-bit resolution may be much more than needed, depending on the capacitance range of your transducer. The Van's capacitance plate setup (one at each end of tank) gives a range of very roughly 150pF (empty) to 300pF (full of avgas). For my RV-8's 21 USgal tank, this gives about 7 pF/USgal. The precise numbers will vary a bit depending on details of tank construction, but they won't be too far from these. If you then use a 10-1000pF transducer, for example, 12-bit A-D conversion would resolve about 0.25pF, or better than a tenth of a USgal.

I've developed a similar system, using an Arduino with 10-bit A-D conversion, and get good results with it during initial bench tests.

[htx9a]

Originally Posted by N49ex

The 4 input circuit pictured - where can you get that?

This is a custom board I designed around a TI FDC2214 frequency-to-digital converter. I may sell a few if there's enough interest after I do a bit more testing/tweaking. The next rev will probably ditch the U.FL connectors and put BNC directly on the board.

I agree 28 bits is excessive, but it's pin-compatible with a slightly cheaper 12-bit chip. I wanted to be sure I would have plenty of resolution if my capacitive plates ended up on the small side.

Analog output would be easy to add with an opamp.

Energy per volume would vary slightly with temperature. If we can measure volume with 28-bit precision then why not be excessive and try to output a constant value independent of temperature.

[Snowflake]

If it can be retrofitted to an existing airplane with the float-type sensors without major surgery, i'd be interested.

[Nick]

I would take at least one, most likely 2.