Just for example I want to verify a weighing scale 150kg capacity 0.05g accuracy. If this are my test wts calibration result (20kg each) : 19.996, 19.996, 19.993,19.994 and 19.990, If I am going to verify the scale to 40kg, so I am going to add the results of two test weights as reference 19.996 + 19.996=39.992kg? or 19.99 + 19.99=39.98kg? which of this reference were correct 39.992kg or 39.98kg ? should i include adding the last digit?
If the two masses are KNOWN and you're doing a span calibration at 40.000kg, you'll use the sum of the value. I would focus on corner load and repeatability.
This is one of the most common misunderstanding. The smallest change that can be read by a scale is called resolution. This may or may not be the same as the "e" value. Usually, it is not.
A 100kg x 0.05 resolution will be rated d=0.05, but the e value will most likely be 0.1 to 0.5. The "e" value is the smallest unit that can be used for pricing.
You shouldn't be using precision weights except for the span verification and calibration adjustment. Something like plastic coated gym dumbells or buckets of sand are adequate for linearity and repeatability. You really just need to write down the value for one of them.
Say dumbell one reads 20.755kg. Ok, well write it down. We'll call it "x".
Weigh it repeatedly ten times and take the standard deviation of all the values. Check that the STANDARD DEVIATION is within specs.
20.760, 20.750, 20.755 . . . .
Start in the middle, then measure half way to each direction and document. if the spec is +/- 10g at about 20kg, it means if the middle reads 20.755kg, left reads 20.735, right reads 20.775, it's out of specs.
Linearity is another.
no load + x
add about 20kg, zero, then add x
add about 40kg, zero then add x.
so forth until you get near 150kg
Hysteresis:
Repeat in increasing and decreasing direction and compare the difference.
Preload the scale to about 40kg. zero. add the 20.755kg dumbell. When you read 20.765 going up, that's a 10g linearity error. Zero it, then remove the dumbell. If you read minus 20.750 this time, your hysteresis in this span range is 15g.
A scale could have a passing span calibration using one precision weight right in the dead center and only in the ascending direction. This is the "accuracy check" as often mistakenly called. Scales can pass this and fail the other tests miserably. Normally failure of corner load and hysteresis is caused by cell damage.