The stated precision of the CZL602XK-50kg is .02 (on load cell) and .02% on certification.
The manufacturer claims the sets of four are matched.
The spec allows a zero balance of +/- 1.5% of FS.
I have measured a zero (unloaded) differences of 4.83% between parts just out of the box.
I'm wondering if the slopes are matched not the intercept (zero)?
It would be interesting to record what a 50 kg static load test would produce in terms of error %.
In early testing I have seen substantial differences at the high end but my test apparatus was not ideal.
Maybe we could get clarification form the manufacturer
Adrian
Load Cells Accuracy
Mon, 04/11/2016 - 03:53
#1
Load Cells Accuracy
I don't think you are reading the spec sheet correctly and understanding what a "matched" set means. I will try to dig into the specs and explain what zero balance and precision means when I have time. Good luck communicating with the manufacturer ......
This one's a bit strange. I ran "sudo apt-get update && sudo apt-get upgrade" a few weeks ago, which ended up installing a new kernel. This broke a driver or two, which I was able to work around by switching to the old drivers. The one exception was the Lux sensor, which I could not get working again. So yesterday i broke down and decided to reinstall 0.7.4 in order to take the system back to the earlier kernel. However, after setting all the configuration settings to what they'd been previously, the output from the HX711 has roughly doubled. I'm using the same slope and intercept as on my previous install. Anyone got a clue?
I've tried pretty much everything I can think of. Still no idea what's causing this. Not only is the weight reading much higher, it's also very erratic. The numbers are bouncing between 460 and 215.
Couple of questions:
1. Do you have all 4 load cells plugged in? If so, try unplugging 3 and testing each load cell individually. See if you have the same noise and the same doubling in weight with each cell. (There may be a bad crimp on one of the RJ11 connectors.)
2. "The numbers are bouncing between 460 and 215" - are these the raw number from the hx711 or the scaled value? If raw, 200 counts of noise is nothing.
3. What is your slope? They typically run in the 10^-4 range (.0001).
m=.0001035 b=-5
Those numbers are weight, not output (which is currently 2903946). I'll try fiddling with the cables tomorrow, since it's raining right now and I don't want to get the R-Pi wet. Thnx for the suggestions.
I unplugged all the leads and plugged them back in and now I'm getting a rational reading once again. Must have been a loose connection somewhere.