@Sparky308
I bought the EJ-54D2 on sale from Scalesgalore ($357 shipped). Price is now over $450.
It is a load cell type scale.
It has less range than many milligram scales, 52 grams in the high range. It also has a low range of 22 grams with an extra digit displayed (counts by 0.2 milligrams, or 0.005 grains).
Since it's there I use the low range with the extra resolution.
I have a range of "Good" mass standards and have checked linearity, full scale, and zero drift with autozero OFF.
With a 20 gram calibration on the low range I have a linearity error of one or two counts (-) @ low grams values (1-5 grams) and a (+) count at 3/4 range, 15 grams. My main loads are around 30 grains. I can pick the calibration value and at 10 grams calibration it is linear from 1 to 10 to within a count.
AandD Specs:
With a good zero full scale should be very stable with the temperature coefficient of the EJ series higher than the FX series,
20ppm/C vs 2ppm/C.
Repeatability spec (Std. Dev not Extreme Spread) is 1 mg for both the FX120 and the EJ-54D2 (0.4mg on the low range)
Linearity, error at lesser values is +/- 1mg with the FX120 and the EJ-54D2 high range (0.6mg on the low range).
Repeatability for the EJ-123 is 3mg, Linearity for the EJ-123 is +/- 3mg.
A "Good" check weight near your target load is a good idea to verify you are not in a high(er) error zone due to linearity error.
I would run your scale with autozero turned off (if possible) to see what is going on behind your back.
Then turn it on after you have troubleshot any stability issues. A pan should ALWAYS return to zero with autozero ON
.