Load cell project: bridge signal ampl. & arduino ADC, EM noise from brushless DC motor

[pdehairs]

Hi everyone,

I'm new to this forum and have never really been on a forum yet. I'm a student in aeronautical engineering and I've been asking myself some questions concerning the electronics implementation for an interesting project I'm working on.

I'll give a small introduction first:
The goal is to design a force transducer or load cell which will measure the couple (0-60Ncm) and traction(0-30N) of a propeller mounted brushless DC motor (840W max continuous, 15VDC).

To measure the loads I'm using 2 DC excited (Vexcitation=5VDC) fully active wheatstone bridges (operating with strain gauges) with an estimated output range of 1mV/Vex & 0.42mV/Vex for respectively the torque & thrust sensing wheatstone bridge. The respective precision required is 2.5µV/Vex & 1.4µV/Vex (this corresponds to 401& 301 values resp.)

A negative bridge output is possible, and the ADC has a range of 0-5V (10bit resolution=>5mV resolution) it thus can't measure negative voltages. Moreover, I specify that the maximum resolution I'll need on the ADC is 10mV, therefore I'll need a voltage step of 1V & 2V resp. after the amplifier (negative voltages bridge outputs will thus yield in measurements smaller than 1V & 2V resp.). The maximum output signal thus needs to be amplified to 4V& 3V resp. before the step. This corresponds resp. to an amplification factor of Av=800& Av=1450.
In short: The analog signal of interest will thus lie in the range of 1-5V & 2-5V resp. and must be measured with a 10mV precision.

Question:
-The electronics are operating near the brushless DC motor (electronically commutated motor), there will thus be EM noise that will be picked up in the electronics and the sensing circuit. However all the usefull information is DC, if I a use a low pass filter to filter out the noise and further filter with software, should this be enough or do I need physical isolation like a faraday cage maybe and coaxed cables for powering the motor? And should I put the low pass filter before or after the amplifier?
-What are other specifications I should look at for operational amplifiers except for Av range, CMRR, offset with t° and time?
-What's the simplest way to implement a voltage step in the circuit?

I think this is basic stuff for you guys, but I'm not used to working with electronics! The more I learn about it the more I'm interested !! Which is nice because I used to find it not so much of interest . I was way more passionate about mechanics and aeronautics. But to play around and experiment at home, electronics are WAY less expensive, it's amazing .

Thanks in advance for your time!!

 

[JimB]

Good grief!
What do you think you are doing, we cannot have this sort of thing here on ETO !
A fully and correctly defined problem, written in correct English with no textspeak.
Well done and welcome to ETO.

-The electronics are operating near the brushless DC motor (electronically commutated motor), there will thus be EM noise that will be picked up in the electronics and the sensing circuit.

OK

However all the usefull information is DC, if I a use a low pass filter to filter out the noise and further filter with software,

Simple low pass filtering is a very good place to start, and depending on if you are looking for fluctuations or transients in your torque/thrust measurements, you should ba able to make the cut-off frequency quite low, say 10 or 20Hz.

should this be enough or do I need physical isolation like a faraday cage maybe and coaxed cables for powering the motor?

Physical isolation is good and generally costs nothing, make use of it where possible.
Faraday cage, probably not much advantage. But bare in mind that the amplifier may be usefully mounted in a screened box if you have big noise problems.
Coax cable is probably overkill, but the wiring to your strain gauges would benefit from being twisted. ie pairs of wires, the go and return legs of a circuit are twisted together. This give good cancellation of interference from magnetic fields.
Coax cables are good for screening from electrostatic fields.

As for power to the motor, twisted cables would be good but I guess that you are looking at high current (50-60 amp at 15v) so the cables will be a bit thick. So just make sure that they are run parallel to each othe as much as possible, and keep them away from the signal cables.

And should I put the low pass filter before or after the amplifier?

Filtering before the amplifier is good, there is no point in amplifying the noise.
It may be usefull to have some filtering after the amplifier as well, but before the amplifier is best.

JimB

 

[pdehairs]

Thanks for the quick reply and the useful info !

Good grief!
What do you think you are doing, we cannot have this sort of thing here on ETO !
A fully and correctly defined problem, written in correct English with no textspeak.
Well done and welcome to ETO.

Well I don't see the point in posting an incomplete problem, as we'll get solutions which are great maybe but not applicable . So I prefer you have a clear view of the problem ! I'm a bit perfectionist, so never got used to typing in textspeak.

Simple low pass filtering is a very good place to start, and depending on if you are looking for fluctuations or transients in your torque/thrust measurements, you should ba able to make the cut-off frequency quite low, say 10 or 20Hz.

Fortunately, transients are a second order requirement; the load cell needs to give me information in a steady state regime. So I think the cut off frequency can even be lower than 10 to 20Hz, no?

Physical isolation is good and generally costs nothing, make use of it where possible.
Faraday cage, probably not much advantage. But bare in mind that the amplifier may be usefully mounted in a screened box if you have big noise problems.
Coax cable is probably overkill, but the wiring to your strain gauges would benefit from being twisted. ie pairs of wires, the go and return legs of a circuit are twisted together. This give good cancellation of interference from magnetic fields.
Coax cables are good for screening from electrostatic fields.

I'll need two amplifiers, which I'm going to mount an a breadboard. I was thinking of using an amplifier like https://www.analog.com/static/imported-files/data_sheets/AMP01.pdf
So maybe I'll put the breadboard in a screened box I think.
Yes twisting the bridge go and return cables is a good solution! It also keeps the wires at the same temperature, that will increase the temperature compensation of the circuit.

As for power to the motor, twisted cables would be good but I guess that you are looking at high current (50-60 amp at 15v) so the cables will be a bit thick. So just make sure that they are run parallel to each othe as much as possible, and keep them away from the signal cables.

The cables' diameter is about 4mm, and they're about 3cm away from the bridge go and return wires. Twisting them could be an idea but, the cables need to remain relatively souple or else their stifness will affect the measurements I think. That's because the motor is fixed to the load cell, which in its turn is fixed to a fixed support, the speed controller also is fixed to that support, thus the wires go from the speed controller to the motor and in doing so they physically bridge the load cell.

Filtering before the amplifier is good, there is no point in amplifying the noise.
It may be usefull to have some filtering after the amplifier as well, but before the amplifier is best.

Ok, seems logic! I'll have to calcutate the filter in function of the power(which will be quit small), the tension and the cut-off frequency I guess.

The amplifier that I selected has the necessary amplification, and CMR is high (which is good but I don't really know why ). On page 7, the drift and noise parameters are specified. The drift isn't such a problem as I'll tare with software before measuring. What I don't understand well are the noise specifications; the voltage noise density and voltage noise. That's the noise created by the amplifier right? So it could be necessary to put a filter behind the amplifier as well.

Thanks!

pdehairs