I am using INA122 with strain gauges in bridge configuration with 120 ohm strain gauge pair, single-ended supply +5V and +2.5V reference voltage, gain 1000X. The gauges are mounted on an arm driven by a torque motor that generates a lot of MHz RF noise. I can filter the noise, but I am surprised to see a negative DC offset in the output that seems to be proportional to the RF noise. That makes their measurement of torque useless. The amplifier power supply is completely separate from the motor controller and heavily filtered and shows no signs of being dragged down. Any clues?
Solved! Go to Solution.
My first try would be a simple RC filter
4.2.1 and fig 6
The main cause can be identified in fig.4 of the datasheet. RF will be rectified by the 'input transistor' inherent diodes.
It certainly looks like rectified noise. A really simple RC filter on the output hasn't helped but Fig. 4 shows some additional capacitive filtering on the differential input that I will try when I finally get back into my shuttered lab. Given the very high frequency of the interference, I will have to be careful about capacitor inductance.
If you use selfmade bridge amps , some 'common' trategies:
-place the first amp near/at the bridge
-use shielded twisted cable ,shield connected on one side... I have used CATe cable (douple shieleded)
-ferrite shokes/pearls do a good job
and: how about fighting the RF at the source(s)?
All good suggestions but I have some limitations:
Strain gauges are on a moving arm, so electronics are at a distance and cabling to the arm must be very flexible.
I added ferrites initially and it definitely helped.
In this case, the source is a commercial torque motor and its power amplifier and controller. Based on previous experience with such products, these manufacturers have a future in the radio transmitter business. We have actually tested several of them and found them to be in violation of FCC and EU RF emissions standards, but no one seems to care these days.