I've purchased two of these, and checked each one out of the box against my reloading scale calibration weight set. Used per the instructions they tracked the calibration weights perfectly down through the 5 grain weights, any combination at or over 5 grains matched, and the counting function was flawless.
When I stacked smaller weights for 3 grains it would settle at 3 most of the time, occasionally 2. 2 grains, 1 grain, and <1 grain didn't land anywhere consistently. This is playing around at the bottom of the scale's advertised range and below the advertised resolution, so I was not surprised.
I wouldn't reload with it, but it seems like a fine bench/field scale for rocket components within its range, and I was very pleased to not have any calibration issues at purchase.
FWIW, this scale, and the last two digital scales I have purchased for office purposes all have a calibration procedure that involves a single expected weight (versus entering a value for one or more weights you might have). I imagine it makes the calibration software less complex. For any scale(s) you obtain I suggest reading the calibration procedure, and obtaining the necessary weight before you need it.
This is totally expected and normal for a scale of this accuracy and precision.
1000.0 gram = 15432 grain or 0.1 gm = 1.5 grain. With most things digital there is a +/-1 to +/- 2 count precision uncertainty. We don't know anything about the electronics in the unit, so we don't know how the strain gauge current is processed. Digitizders have a 2^n count range. You need a 14-bit (2^14) digitizer to obtain < 1 grain resolution, but we don't know what is inside the scale. There could be a 16-bit or 12-bit digitizer. With the 16-bit digitizer, you throw out the 2 LSBs and with a 12-bit unit you need to co add at least 4 digitizations to get the 2 extra bits, and ideally 16 digitizations to get an result accurate to 14-bit.
When you want to weigh an item that is close to the resolution of the scale, you should excercise the strain gauge to reduce the sticktion of the mechanism. Take a coin and put it on the balance and remove it several times. This loosens the mechanism and flexes the strain gauge and gives a more accurate reading.
Also use a piece of aluminum foil or wax paper as a weighing paper to avoid putting powders directly on the scale. Also wear gloves. The moisture and oils from you skin can be transferred to the paper and change the scale reading by one count. Air currents and static electricity will also effect the readings of a sensitive scale by one count or sol.
So to get back to what you observed, in the digital word, if you try to weigh 1 grain on a digital balance with a resolution of 1 grain, you legitimately will observe 1 +/-1 grain readings for a 1 grain weight, 2 +/-1 grain for 2 grains, 3 +/- 1 grain for 3 grains, etc.
To weight something to a precision of 10%, you need 10 counts at a minimum,so you really need to get a balance that has a resolution of 0.01 gram to accurately measure 1 grain to 10%, and 0.001 gram to measure 1 grain accruately to 0.1 grain.
https://www.dealextreme.com/p/uniqu...2-lcd-digital-scale-100g-0-01g-2xcr2032-65793 is good for measuring powder down to 1 grain to +/-10% with a range to 1500 grain.
https://www.dealextreme.com/p/high-precision-portable-digital-scale-10g-0-001g-2-aaa-52751 is good for measuring powder down to 0.1 grain to +/-10% with a range to 150.0 grain.
Bob