EMI/RFI issue in basic circuit with homemade FSR

Hello,

I have a very basic circuit with an homemade FSR (made from IC insulating foam) and a LED.

The LED is supposed to emit light with an intensity controlled by the pressure applied on the FSR. The circuit behaves as expected except that the LED is also lit when no pressure is applied but an object is close to the circuit. So there is obviously some EMI/RFI problem here. The data from the sensor also shows there is some cyclic noise in the data.

You can see the circuit below. There is a video to show the problem here:

So what should I do to solve this ? I have some Linqstat film. Should I wrap the circuit with it ? Put the circuit in a metallic box ?

Would it help if I’d connect the metallic plate which is under the breadboard to the Arduino’s ground ? Or is it automatically connected yet ?

Should I use an opto-electric isolation between the Arduino’s 5V pin and the circuit ?

Any help would be highly appreciated.

Are those resistor values right? If so they are way too big. Then need to be down at 10K or so. You have what is known as a floating input.

The answer to all those questions is no.

Yes these are two 10M resistors in series. But if I insert lower values my signal doesn’t fit the ADC’s range. Or should I rather use the AREF and lower the ADC’s range ?

I have read your tutorial about decoupling and therefore tried to insert a 100 nf capacitor as shown below. It seems to work but is it a bad idea ?

Yes these are two 10M resistors in series. But if I insert lower values my signal doesn't fit the ADC's range.

Yes that is the problem 10M is way too high to be doing anything. I know it affects the range of readings but that is what you get with a single ended resistance sensor like this. Basically this is not a good way of doing the measurement of this transducer. It is the simplest but not the best. A better way would involve op-amps and a bridge circuit.

It seems to work but is it a bad idea ?

No if it works it is good. It might not be repeatable and might depend heavily on the actual make of the capacitor because lets face it 10M is doing very little for you. I was told many years ago that 10M was about the resistance of a finger print.

Grumpy_Mike:

Yes these are two 10M resistors in series. But if I insert lower values my signal doesn't fit the ADC's range.

Yes that is the problem 10M is way too high to be doing anything. I know it affects the range of readings but that is what you get with a single ended resistance sensor like this. Basically this is not a good way of doing the measurement of this transducer. It is the simplest but not the best. A better way would involve op-amps and a bridge circuit.

It seems to work but is it a bad idea ?

No if it works it is good. It might not be repeatable and might depend heavily on the actual make of the capacitor because lets face it 10M is doing very little for you. I was told many years ago that 10M was about the resistance of a finger print.

Repeatability will be an issue because I want to use at least four such sensors.

Using AREF with 1.1 V as reference improves the situation a bit but doesn't solve the problem completely. Can you give me an example of bridge circuit ? Also, which op-amp would you use ?

I have an ULN2803 (Darlington driver). Could I use it for this purpose ?

Try a 100nF cap between the analog input and ground - should reduce all the pickup due to the high-impedance of the circuit (brings the impedance at mains frequencies down to more like 10k than 10M...)

I have an ULN2803 (Darlington driver). Could I use it for this purpose ?

No totally wrong thing to use.

Using AREF with 1.1 V as reference improves the situation a bit

No it doesn't you are just moving the problem.

Can you give me an example of bridge circuit ?

http://www.allaboutcircuits.com/worksheets/bridge.html

I undersand a bridge is useful to measure an unknown resistance but I don't understand the point: what should I measure ? The sensor's resistance range ?

Also, you've given me the advice me to use an opamp. Which one should I use ? There are so many models...

Your sensor is an unknown resistor, you want to know its value. One way to do this is with a bridge. It is the same no matter what the sensor actually is. This schematic was designed for a pressure sensor but there is no reason why you can't use for your sensor. http://www.flickr.com/photos/33177304@N03/4666641155/in/photostream/lightbox/

Basically your problem is that you need a large range but you can't do that with an impedance low enough for a simple circuit like you have. So you have to amplify it.

Grumpy_Mike:
Your sensor is an unknown resistor, you want to know its value. One way to do this is with a bridge. It is the same no matter what the sensor actually is. This schematic was designed for a pressure sensor but there is no reason why you can’t use for your sensor.
http://www.flickr.com/photos/33177304@N03/4666641155/in/photostream/lightbox/

Basically your problem is that you need a large range but you can’t do that with an impedance low enough for a simple circuit like you have. So you have to amplify it.

Couldn’t I simply use something like a high-impedance JFET transistor to amplify rather than such a complex circuit ?

Yes if you want, and if you can design one with a linear transfer characteristic. But that was a proper differential amplifier. It depends on how accurate you want things.

Looking at your schematics is it right to say that the two op-amps located between the sensor and the differential amp are there to provide a high impedance on both inputs ?

These op-amps are not identified. Would basic models like LM358 work ?

Also, I don't understand what the function of the part located to the right of the LM324.

Another point: my sensor has only two connections (it's just IC insulating foam) instead of four. So I suppose I would connect these to the op-amps and drop the connections to the 5V and ground.

Thanks a lot for your help.

is it right to say that the two op-amps located between the sensor and the differential amp are there to provide a high impedance on both inputs ?

Yes they provide a balanced input impedance, that is to say they are both the same. This allows the differential amplifier to have maximum common mode rejection.

These op-amps are not identified

Yes they are. The op amp LM324 has 4 op amps in it, this circuit uses one chip.

my sensor has only two connections

You replace one of the resistors in a bridge with your sensor.

OK. I'll try to build a similar circuit. What about the rightmost part of the schematics. What is it for ?

Can't remember, the text of the article is here, this explains it:- http://www.doc.mmu.ac.uk/STAFF/A.Wiseman/Acorn/BodyBuild/BB_86/BBC37.txt

Thanks !

By the way I have replaced the foam with several layers of Linqstat and the sensor has a much bigger range of resistance now. I have acceptable values with a pulloff resistor smaller than 1 Mohm. But I'll try the differential + bridge circuit. I just need to order the components...

Those who work with OP AMPS eventually end up figuring out that when working really high value resistors (and the resulting low level signals), that yes, you do end up with lower current used by the circuit but you also invite noise and that noise becomes what you amplify more than your actual desired signals. This is why lowering your resistance values can have a positive benefit.