Hello everyone,
I have mercury-in-rubber (I think it's actually InGa not Mercury) strain gauges I need to use for blood pulse detection. Link here: Strain Gauges | Hokanson. I figured I would use a 24-bit ADC that I already had experience with, so I went with the HX711 Sparkfun module (with the bogde Arduino library). However I have not been able to get it to work.
I have another load cell (cheap luggage scale that I took apart) which works fine. The resistances between any two out of the four pins on that load cell range between 670 and 930 ohms. This itself doesn't really make much sense to me as I thought the 4 resistors in a Wheatstone bridge were supposed to be similar without strain, so I would expect to be close to 1.4k between at least one set of pins. Either way, it works. Maybe the bridge doesn't need to be balanced to detect changes.
The InGa-in-rubber, though, is a different story. The resistances between any two leads there measure between 6 and 8.5 ohms (that might even be dominated by contact/wire resistance from the DMM). I tried looking on the HX711 datasheet for a range of acceptable resistances but couldn't find anything. The color scheme on these strain gauges is also confusing: brown, red, orange, yellow. I assumed brown and red were excitation -ve and +ve respectively, and orange and yellow were the outputs (didn't matter which way at the moment). I've tried swapping them around and trying different combinations, nothing worked.
I also tried putting 680 ohm resistors in series with all 4 wires running into the HX711 board, then only with the two output wires, didn't work. I tried the same with 100 ohm and 27 ohm resistors, still nothing. The output (using the bogde HX711 library calibration sketch) reads a constant number which jumps around slightly due to noise or drift but doesn't change in response to strain, even when the calibration constant (sets sensitivity/scale) is very low (high sensitivity).
I tried this procedure much less carefully yesterday, and ended up letting the magic smoke out of the IC, so I was more careful today (I have two HX711 boards). I attached a DMM (measuring current) in series with the ground path of the HX711 board. I know low-side current sensing is frowned upon (IC ground potential varies with current), but it still worked fine with the luggage scale. I used the current reading to tell me if I was shorting anything. The luggage scale would cause the HX711 to draw around 70 uA while operating normally, whereas the InGa strain gauges would draw 20 uA in some configurations and 4 mA in others. I assumed this was the on-board PNP overcurrent protection kicking in (unlike yesterday for some reason).
Nothing worked. Just random jitter due to noise (and the high sensitivity of the IC). I kept using the luggage scale load cell just to make sure I didn't mess anything up; it worked fine every time I tried it all the way up until. I'm out of ideas at this point.
Anybody with previous experience with these? Do I need to make any modifications to the board to get it to work?
I've gotten these strain gauges to work with a Phidget 1046 board (Wheatstone bridge amplifier + interface board/microcontroller), but I need something small that I can embed into a wearable product down the line, so I can't use the Phidget.
Any help is appreciated, thanks in advance.