The platform is based on four 1 kg load cells connected to instrumentation op-amps. The idea is to measure the force distribution on the four sensors according to the finger position on the plate. Please see schematics and picture below.
However I have inconsistent data coming from the four sensors. One sensor seems to be at end of its range even without pushing on the plate, another seems to be at the other end of the range, there's only one providing expected data. I wonder if it has something to do with the electronics or rather with the mechanics.
could the cable length from the load cells to the op-amps (INA 122) be an issue ? Short cables have a 20 cm length, long ones have a 40 cm length.
could the load cells be inconsistent amongst themselves ? These are load cells built from China on eBay.
is it correct to tie the top of the four load cells with a rather stiff plate (8 mm acrylic) or should I use a more flexible material ?
can I tie both top and bottom of each load cell to a common plate as shown on picture or should I let the bottom free ?
I am also looking for a solution to set an equal distance between each load cell and the bottom plate to set the top plate in the most horizontal position and therefore distribute the initial forces as well as possible. Currently it depends on the screws tying the load cells to the bottom plate only. Maybe I could use a kind of spacer with one screw between the load cell and the spacer and another screw between the spacer and the bottom plate ?
I presume you have removed the plate and found that all sensors give about the same signal output? This would put your mind at rest on wire lengths. Also, have you tried an oscilloscope on the amplifier dc outputs to see if there's any parasitic oscillation present?
amundsen:
The platform is based on four 1 kg load cells connected to instrumentation op-amps. The idea is to measure the force distribution on the four sensors according to the finger position on the plate. Please see schematics and picture below.
However I have inconsistent data coming from the four sensors. One sensor seems to be at end of its range even without pushing on the plate, another seems to be at the other end of the range, there's only one providing expected data. I wonder if it has something to do with the electronics or rather with the mechanics.
With four mounting points it would be possible to have some in-built twist due to the plate and the thing the force sensors are mounted to not being exactly the same shape. This would introduce a preload on the sensors which could easily be massively more than the finger pressure. If you only had three sensors then you could avoid the possibility of this preload. The calculation of the contact point would change slightly but not be substantially any harder than your 4-point calculation.
amundsen:
it seems I have electronic issues (several inputs reacting to the same sensor, very strange...).
If you have floating inputs (i.e. inputs not electrically connected to anything) the input state is undefined but could be affected by the state of other inputs.
Your sensors will interfere with each other via the REF pins and the common voltage divider. The higher the values of R1 and R2, the more they will interfere.
I agree that it would be better to support the plate on 3 cells rather than 4. Think of a 4-legged table placed on a floor that isn't quite flat. If you want to use 4 cells, you will need to have some sort of adjustable spacer to support one of them so that you can get it level.
dc42:
If you want to use 4 cells, you will need to have some sort of adjustable spacer to support one of them so that you can get it level.
Not such a frightening perspective. I'll test the four-sensors configuration as soon as the electronic issues are solved and shall test the three-sensors config afterwards if it does not work.
