As a linear sensor 2 points should be perfect, for the best accuracy if this points are near the maximum and the minimum of the sensor should be great, but if you can make a calibration at 0 and the second one at the max weight you have is ok.

More than 2 points are needed if the sensor is not linear

Never heard about that, my old dyno was calibrated with one calibrated weight, ive seen Mustand Dynos are calibrated using same way. 

Posted by: Walter

Never heard about that, my old dyno was calibrated with one calibrated weight, ive seen Mustand Dynos are calibrated using same way. 

Sure, but first you also have to set zero value, then the weight, that are 2 values.

I was always think about "zero value" on the dyno is not ok as the arm that holds the load cell have a weight.. so is not real zero.

Also the weight, if is not very high and the it have a little margin of error then at high brake torque this error gets multiplied.

But all this is to be as perfect as possible, sure all this is not needed for real life 

Hi Walter, load cells are not 100% linear, and also the torque arm and chassis frame work can influence the load cell output. For systems I have worked on in the past there has been calculation in the PLC software to compensate the load cell raw value to a corrected value (calibrated). Similar to what we have in the yourdyno software, however there is a correction factor applied based on mulit-point test weights being applied (alfa/beta values to correct the curve). I am used to working with larger load cells in industrial equipment, so it maybe irrelevant with these smaller load cells on the engine dyno. Once I have my dyno up and running I'll let you know if I find anything interesting when calibrating. 

regards, D-E

Example attached of results for 2000 kgs load cell calibration using several test weights. The max working load is 1600 kgs for the load cell, so we calibrated up to that. You can see the slight variance in raw output verses the calibrated load. 

OK, so I am quite impressed with this little (and none expensive) load cell. I have 16 kgs of calibrated weights. The load cell is a 50 kgs size.

I tried various combinations of calibration, tarring the load cell then using 1, 2, 4, 6, 8, 10 and 16 kgs and the "set" weight. I then tested each weight individually and the values were all consistent. In his case, I agree that a single point calibration is accurate enough.

\thanks D-E  

Very interesting, weights were calibrated? is not very far from linear, other thing that I read is that the load cell was also affected by the temperature

Not, load cells arent linear but his error are very small, depend of quality of load cell of course.

That's how dynos normally works, a 500kg load cell is used for powerfull dynos, normally are 150/200/300Kg. never worked with +1000Kg load cells..

About temperature, yes but affects about 0.010%-0.030%. as i said, depend of quality of load cells..

Hi, yes you are very right on calibrating over a wide range of the capacity of the load cell using different weight sets instead of a single weight. This helps to tackle the issues concerning the systematic error in the system and it ensures better accuracy and of course, linearity. You can learn more about this topic in the article I wrote here on this webpage: https://tacunasystems.com/knowledge-base/load-cell-tips/calibrating-the-measuring-chain/