# Capacitive level sensor

Dear all,

I need to design a capacitive continous level sensor for a fuel tank. The sensor it self is a rod placed inside tube having empty capacitance of 47pF and with the tank filled about 150pF. The usual method is to use the capacitance in an RC oscillator and than measure the frequency than convert it to volumetric leveling. The success is unfortunately temperature dependent and should be compensated. Is there any better way to build up this sensor? I would need at least 10bit over 100pF so a definition of 0.1pF is requested. One important remark is, that one of the measuring capacitors armature is grounded to the tank (tank is made out of regular steel sheets). I have looked some of the CDC (AD7151) chips but those seems to work only for feed-in capacintances from excitation to the CDC and the output is only treshold based.

Any hint or advice is welcome!

Thank you , Endre

• Good afternoon Endre,

The only CDC device that we have that allows for capacitance to ground measurement is the AD7747. The device has a CAPDAC allowing a guaranteed minimum common mode capacitance compensation of 17pF and the full dynamic range of the device is +- 8pF (max=25pF).

It is theoretically possible to add a capacitance in series to bring your sensor into the required range of the part, but we do not have any experience using the part in such a configuration.

From your specifications below, I calculate a series capacitance of ~20pF would achieve the desired range. Using common values for capacitors, I suggest using a 22pF capacitor, which will give you a maximum reading of 19.8pF and a minimum of 14.9pF, for approximately 5pF range. Using a larger capacitor; 27pF, will give you more range, but will bring the maximum to 22.8pF which is close to the 25pF max.

Regards,

Sean

• Sean,

I want to use AD7148 to measure the position of a metal plate, like a metallic "robot finger". I was thinking I should ground this robot finger, as the human is considered to be a ground in the models I see in the AD7148 data sheet.

But if the robot finger is grounded, then the AD7148 would be measuring the "capacitance to ground", which you say is only allowed on the AD7747, not the other capacitor sensors. And if "capacitance to ground" measurement isn't allowed on this AD7148 sensor, just what model and potential is the human, if not a ground?

So I'm confused, is the human not a ground? Should I let my robot finger (a small metal plate) be floating (not grounded)?

BTW, I calculate a maximum of a couple pF for the size and closest position of the small plate relative to the sensor pad, so I'm good there.

Thanks,

-Larry

• Hello Larry,

I was being too general when I said the AD7747 was the only device we have that operates with Capacitance to Ground measurements. That statement is true of the CDC family of parts which include the AD774x and AD715x devices.

The AD7148 is part of the CAPTouch family of devices and also operates using a single electrode/capacitance to ground/self-capacitance family.

Regards,

Sean

• Sean,

Thanks for the clarification on the CDC versus CAPTouch differences in your capacitive sensors.

So to clarify further, the CAPTouch family works only if I ground the metal "robot finger" position probe, in direct contrast to the the AD774x and AD715x devices? As such, I do NOT want to float the metal plate(s) on this "robot finger"? (And grounding would help on accuracy and noise I think?)

I'm looking for a position indication using the capacitance read from the CDC_RESULT_Sx registers only. Is there any reason to use the CDC sensors instead for this position sensing application? This CAPTouch looks nice in that it has more channels that I need, and similar 16 bit accuracy with a single one-sided electrode.

Thanks

-Larry

• Hello Larry,

The CAPTouch family use single sided capacitance measurements and require the "other" terminal to be connected to ground; in this case the robot finger.

The majority of the CDC parts use two electrodes to create a sensor and rely on interference of the electromagnetic field to detect change in capacitance.

From your description, I would continue to use the CAPtouch parts, especially if you require more channels.

Regards,

Sean