Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Parasitic Capacitance compensation

Status
Not open for further replies.

jg001

Newbie level 6
Joined
Aug 12, 2010
Messages
11
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,382
Hi,

We are trying to design a readout circuit for a capacitive sensor. The capacitive sensor itself is only a fraction of a pF and the changes in its capacitance due to some mechanical input is in the fF range. On top of these small values, two restrictions (no compromise) have been imposed on the readout circuit:

1) It has to be 10s of feet away from the sensor (connected via a multi-wire cable)
2) One end of the capacitive sensor must be grounded

With such small capacitance value to have to work with, my guess is that just the parasitic capacitance from the cable will be 100s of pF connected in parallel with the sensor itself, which will severely degrade the sensitivity of my output.

So, does anyone have any idea on how to compensate for this relatively large parasitic capacitance? I've thought of using negative impedance converters (NICs), but PSPICE simulation says it will not work.

I have attached a two-page PDF (mostly circuit diagrams) which will give the reader some background info and my NIC implementation attempt. Thanks in advance for any suggestions.

-- JG
 

Attachments

  • Parasitic Capacitance.pdf
    89.8 KB · Views: 114

In your schematic, it looks like you're only accounting for parasitic capacitance from the wires to ground (which can be dealt with by using shielding). The more troubling thing is parasitic capacitance directly between wires, which will appear in parallel with your sensor. You can also use shielding for that as well.

Negative impedance probably won't help you, since the parasitics are going to vary with the environment, and your circuit has no way of knowing that.

Also you'll want to be careful in designing your amplifier. Look at bootstrapped unity gain buffers.
 

Quote: I've thought of using negative impedance converters (NICs), but PSPICE simulation says it will not work.

Hi JG,

the NIC circuitry (your attachement) does NOT represent a negative capacitance: The input impedance is Xin=R1R2/Xc.
That means, in order to get a neg. capacitive input impedance you have to interchange R1 and C.
 

mtwieg,

I am more concerned about the inter-wire capacitance than anything else. This will be high. Unfortunately, shielding for this will probably mean a metallic foil between the wires, right? This is almost but impossible because our wires are very thin (thinner than 36-gage). Do you know anyone who will manufacture a 12-wire cable bundle where each wire is a micro-BNC? please let me know.

LvW,

I have to disagree with you. see my attached analysis of the NIC circuit. my analysis says when R1 = R2, my Zin = -Zc.

-- JG
 

Attachments

  • NIC Circuit Analysis.pdf
    46.8 KB · Views: 61

LvW,
I have to disagree with you. see my attached analysis of the NIC circuit. my analysis says when R1 = R2, my Zin = -Zc.


You are right - sorry, I was in error.
LvW
 

I would use a bridge with your sensor+cable as one element and 3 other caps for the other elements*. try and keep the value roughly comparable. Energise the bridge from AC resonable sinewave (not a squarewave) with a frequency in the high audio range so you can filter out any mains pickup. An alternative approach would be to use the cable +sensor as the capacitive part of a high frequency oscillator, very easy to measure frequency to parts/million with a digital counter. The downside is the temperature compensation of the oscillator or the beast will be measuring the temperature rather then the sensor displacement.
Frank
* if a DC return path is required carefully balance this out by using DC to set the bridge up.
 

Hi JG,

in case you try again the NIC approach, I like to repeat that you should interchange R1 and C.
As mentioned before, I agree that your component allocation results in a negative capacitive input impedance - however only under ideal conditions!
In reality the capacititive divider within the positive feedback branch creates problems because
(a) there is no dc bias current path and
(b) the parasitic ohmic resistances of the caps provide unwanted positive dc feedback.

Thus, you should use the classical NIC arrangement with a grounded capacitor.
 

Main problems of the parasitic capacitance for a measurement circuit will be capacitance variation due to temperature changes or bending of the cable and possibly noise gain. None of them can be removed by a compensation circuit, e.g. a NIC. If grounding of one sensor terminal is mandatory, only a driven shield (usually in combination with a triaxial cable) can cancel the effect of capacitance variation.
 

LvW --- I agree with you. Your suggestion would be more robust.

However, as FvM has pointed out, the NIC will not be much help for my situation.

FvM --- I'm not clear as to how a triax will help. Can you please elaborate on your suggestion. I'm assuming that the outer conductor of the triax will be grounded while the inner will be the signal path. How will the middle conductor be used? What is "driving" this conductor? Do I need to connect the middle conductor to some kind of circuit? Thanks.

--JG
 

Driven shield means, that the "hot" signal is connected to the center wire and the shield is driven by a +1 buffer to the same voltage. Because there's (ideally) zero voltage difference between center conductor and shield, the respective cable capacitance is effectively eliminated. The second shield provided by a triax cable can be used as ground return and also prevent interferences to the shield potential.
 

Hi,

Did you finally get it work?
How did you do it?

I am facing the same problem but i have bought chip like Mpr031 or AT42qt1011. The problem is that i want to connect the sensing plate to the chip through 1 metre cable. I am not sure if that will work with GND shield, driven shield...

Regards!

Hi,

We are trying to design a readout circuit for a capacitive sensor. The capacitive sensor itself is only a fraction of a pF and the changes in its capacitance due to some mechanical input is in the fF range. On top of these small values, two restrictions (no compromise) have been imposed on the readout circuit:

1) It has to be 10s of feet away from the sensor (connected via a multi-wire cable)
2) One end of the capacitive sensor must be grounded

With such small capacitance value to have to work with, my guess is that just the parasitic capacitance from the cable will be 100s of pF connected in parallel with the sensor itself, which will severely degrade the sensitivity of my output.

So, does anyone have any idea on how to compensate for this relatively large parasitic capacitance? I've thought of using negative impedance converters (NICs), but PSPICE simulation says it will not work.

I have attached a two-page PDF (mostly circuit diagrams) which will give the reader some background info and my NIC implementation attempt. Thanks in advance for any suggestions.

-- JG
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top