Hello,

Background:
I am working on a project in which I need construct a linear position measurement system using concentric capactive plates to calculate the linear position of the inner plate relative to the outer plate. The plates are constrained such that the only degree of freedom is along the Z axis (down the center of the capacitive tubes). The framework consists of PLA plastic and the capacitive plates consist of copper foil on a PLA support structure. Range of capacitance to be measured is between 4 and 40 pF. Capacitive shielding is located along the entire length of the cylinder at a separation distance of 50 times the spacing between the sensing plate and the ground plate. The capacitive plates are oriented such that the capacitance varies from maximum to minimum in a linear fashion.

Current prototype:
My current working version uses a 555 timer in astable multivibrator mode, see attached schematic. Note: RA and RB are not included on the schematic. They were experimentally determined once the initial PCB board was completed and are used for changing the range of frequencies output for a given capacitance range. These values are not immediately available to me, but I can add them next week if they are deemed necessary.

Current status:
The current version offers a repeatable resolution of 0.01" for a 2.5" total travel distance when the measurements are taken while the inner plate is held in a static position.

Current problem:
During smooth, fast, movements (2.5" in 0.1 second), the transient response of the circuit is not as predicted. Instead of a smooth curve from the initial point to the final point, the output always takes a sharp, immediate drop (corresponding to a sharp drop in output frequency) and then quickly approaches the final position, see attached plot. This is not an issue during very slow movements. Note on the plot that the step input occurs at the start of the plot and that the delay for 25ms is due to the physical characteristics of the actuation system, not the measurement circuit.

Can anyone offer an explanation for the behavior of this circuit and/or a method to rectify the negative spike during fast movement?

Thanks,
Jeremy

 

You are using below the cap capability of the chip. Try adding a cap in parallel.

 

Current problem:
During smooth, fast, movements (2.5" in 0.1 second), the transient response of the circuit is not as predicted. Instead of a smooth curve from the initial point to the final point, the output always takes a sharp, immediate drop (corresponding to a sharp drop in output frequency) and then quickly approaches the final position, see attached plot. This is not an issue during very slow movements.

I think the behavior you're observing may be caused by the fact that the voltage across your sensing capacitor has a DC component (Vcc/2) in addition to an AC component (Vcc/3 peak-to-peak), and the presence of that DC voltage causes a transient current

I = V * dC/dT​

to flow as capacitance changes. (This is how a capacitance microphone works, by the way.) This current is probably interfering with the proper operation of the 555 oscillator, by causing a momentary drop in frequency as the cylinder is moved.

I suspect if you get rid of that DC voltage component across the sensing capacitor, the problem will disappear. You can either AC-couple pins 2 and 6 of the 555 to the sensing cylinder, for example with a 1000 pF series capacitor along with a 1 GΩ shunt connected between inner and outer cylinder, or you can return the outer cylinder not to circuit common as you do now, but rather to a DC level equal to Vcc/2 such as from a voltage divider.

And as @Colin55 pointed out, you are operating the TLC555 with a timing capacitor value much smaller than what it was designed for; expect significant unit-to-unit variations in performance if you intend to make more than one of these things, and they will very likely have to be individually calibrated.

 

I think the behavior you're observing may be caused by the fact that the voltage across your sensing capacitor has a DC component...

That was exactly my thought. The circuit is responding 'accurately' to the capacitor, but the quick motion of the capacitor is causing an electrical artifact that wasn't expected.

 

I would like to thank all of you for your advice.

Colin55,
The capacitive load is restricted to 4 - 40 pF due to physical constraints. Please correct me if I am wrong, but wouldn't adding a capacitor in series cause a reduction in sensor resolution?

OBW0549,
There may very well be a dc component, but I am unable to measure it as the probes on my scope are 1Mohm and stop the capacitor from cycling when probed. I tried setting the shielding and ground electrode plate to Vcc/2 and the only noticeable difference in the output appears to be due to the resistor located between the plates and the ground in the voltage divider circuit. I will order some 1 Gohm resistors and try ac coupling the electrode.

wayneh,
The prospect that some unexpected electrical artifact has been created during motion has definitely crossed my mind. Given that the mobile plate is sealed with a rubber o-ring, I do wonder if the movement is causing some electrostatic effect that is messing with the cycling of the capacitor.


I have not been able to recreate the spike effect on a stand alone circuit (by abruptly connecting and disconnecting a 10pF capacitor), so I feel confident in speculating that the issue is unique to the variable plate apparatus. I have also noted that the spikes occur at positive and negative peaks in relative acceleration between the plates. The apparatus has mechanical stops at either end and it seems that a rapid change in acceleration to or from the stops causes the spikes. I noted for the first time today that positive spikes in frequency do occur at the stop where the capacitance is a minimum and negative spikes occur at the stop where the capacitance is a maximum.

I will update once the Gohm resistors arrive.

Thanks,
Jeremy


On a side note, I am not restrained to using a 555 IC. If any of you can recommend a method of measuring the stated capacitance with a cycle of 10ms or less, I am definitely open to investigating it.

 

Just a few of quick thoughts:
1) The timer circuit needs to be faster than a 555 to get an accurate measurement with only 4pf of capacitance.

2) The timer circuit -- no matter what it is -- _must_ be right at the capacitor.

3) I would use a shielded 3 conductor cable to the circuit. The 3 wires would be power, ground and signal out.

edit:
4) The circuit must have very short leads to the capacitor.

 

Your plot of PosTimeplot is very nice, but you have not identified what this plot means. Where on your schematic is this Signal present?

Series capacitance or stray capacitance should not affect your readings as long as:

Anything in series is much much greater than your max variable capacitance. The net value of a small cap in series with a large cap is..... the small cap.

Parallel stray capacitance should appear as a constant term and thus not change the readings in a calibrated system. Note they need be constant for this to be true, so wires should be bundled or constructed out of say shielded wire that has zero room to wiggle about.

This from a guy who actually has a capacitive level sensor in active production today (though I no longer work for that company).

 

Update:
The 1 Gohm resistors arrived today. I AC coupled the sensing plate to pins 2/6. This had no noticeable effect on the spikes.

I disassembled the capacitor down to it's simplest components and started testing them one by one. I have verified that the spikes are not due to the sensing circuitry (the 555 circuit), the movable plate (which is grounded), the shielding (ground), leaving only the sensing plate. I connected the sensing plate (basically a 0.0625" thick wall PLA plastic tube with an exterior copper sheath) to the sensing electronics with a 6" segment of single conductor wire so that I could isolate the electronic circuit from motion.

Holding the sensing tube 8" from the nearest surface (with 3" between my hand and the conductive surface), I recorded a steady 25khz frequency in the 555. Placing my hand approximately 1" from the conductive surface, I recorded a drop in frequency of about 3khz. I tapped on the sensing tube with the handle of an insulated screwdriver and noted that light taps had no effect, but stronger taps caused the exact same spikes that I observed when the device was in operation, with the frequency dropping on the order of 15 khz for 1 - 2 read cycles (3-5 ms).

I have reached out to a materials professor in the hope that they can help me figure out what the issue is that is causing the spikes and how I can solve it.



RichardO,
The circuit is mounted as close as the capacitor shielding will allow. From the results that I have obtained, the 555 is fast enough, albeit the RA and RB that I am using are 1M and 10M +/-5% respectively, so the currents involved in the charging of the capacitor are very small.

ErnieM,
The PosTimeplot is the position verses time plot for the capacitor in response to a step input that begins at t = 5 seconds. This plot is the result after the frequency output from the 555 circuit is fed into a 4th order curve fit polynomial to relate frequency to position. I, unfortunately, did not think to record the raw frequency data during this batch of tests.

Thank you both for your advice.
Jeremy