So let me confirm that I have my parts list correct here. To use the strain gauge setup, you need 2 strain gauges A single wheatstone bridge A low noise filter An arduino or similar processor Is this correct?
-
Yes, you could use 2 strain gauges and make 2 quarter bridge wheatstone bridges. However, i used a full bridge configuration since it is more accurate and also is less influenced by external factors such as humidity or temperature.
-
The wheatstone bridge was just made using some wire, but a pcb is also possible
-
The low pass filter was implemented in software since you will probably have to tweak the settings a little. This can be exponential smoothing or some Finite Impulse Response filter, I used a FIR, but didn’t really put alot of thought into this decision to be honest.
-
you will also need an external ADC since the ADC built into the arduino has very poor resolution with respect to the voltage deflection the wheatstone bridge will put out.
Will the INA125P voltage amplifier compensate for the adc or do I need something else?
I used a HX711, which is a voltage amplfier followed by a high resolution ADC. But you could also use just a voltage amplfier in such a way that you use the complete range of the (lower resolution) arduino ADC. I don’t think you need the external ADC since the extra resolution is probably overkill.
Did you modify your HX711 to run at 80sps or left it at 10sps? I haven’t tried 80, but I saw a document mentioning that there is a lot more noise at the higher sampling rate. The only problem I have with the HX711 is that it does not explicitly indicate strain gauge failure.
@taroko So do you use the hx711 or something similar? Also, have you figured out how to bulletproof the HX711 from a strain gauge failure?
I use HX711. I don’t have a good solution for strain gauge failure yet. I just detect the following: 1) If the weight reading jumps to an abnormal reading (like a high negative value) 2) if the weight show very little variation.
These detection methods are not bulletproof, but I’ll live with it for now. For my next version, I may consider using a volt amp instead. I’m just hoping that I don’t rip my strain gauges off while I’m riding.
What I do like about HX711 is that I can convert the sensor signal to digital near the sensor.
@taroko Ok thanks for the help. I have a year long school project for which we are building one of these. I will dedicate a really good portion of that time to the programming in which I will try to bulletproof it. I will report back with any questions or triumphs.
If you get the sparkfun version, you can cut a connection on the board. See this thread: https://forum.arduino.cc/index.php?topic=479175.0
The generic breakout boards are not as convenient.
I left it at 10 sps, the higher sampling rate doesn’t really improve responsiveness if you low pass filter it like we did anyways.
If one of the strain gauges gets damaged, they basically become an open circuit and you will get readings that are abnormal, like a factor of 1000 bigger than usual. You did it the same way we did it. But we always used a remote for extra safety during testing.
Alright. I know that I am being obnoxious but I have one more question. If I am using 2 total strain gauges, one in the front and one in the back, how would I wire the Wheatstone bridge. The strain gauges are 350 ohm ones…
Are you using just 1 strain gauge for each truck? If so, I would use quater-bridge type I (see link below). With quarter-bridge your signal to noise ratio will be lower. Let us know if quarter works ok.
I’m using full bridge at this time which seems to work well, but this might be over kill. I"ll test half bridge in the future.
When you want to sense weight over the rear of the board, but do not require a precise measurement (ie. in a comparator circuit), would you still use a strain gauge, or another form of pressure sensor?
EDIT: Disregard the switch in the example circuit below.
Well done, it looks you got how to make it, it’s really unbelievable you make such great progress in just 4 mouths. but in my experience, you just got how to start and you are still far away to the complete and perfect it. From the quick demos, i can see the acceleration and braking is not smooth enough, actually, i should say the acceleration and braking both are not sensitive enough, it means you should test and modify the value of current control which is based on the weight distribution, it should be more elaborate, i’m not sure you can get a good result after a lot of test, but if you need a better idea, i would like to advise you use best weight sensor and the change the % to accurate weight, then make the current control based on the weight changes, you will get a much better performance. anyway your this project is on hold already, so you may can’t make it now, it’s a little pity that i read your post so late. Otherwise we can have a interesting communication. good luck!