The calibration is to turn the rig through 90 degrees, so the measuring axis is vertical, and hang a weight of known value onto it. The raw value that comes from the strain gauge electronics is multiplied by a number to make a result of the same value in grams. I use a set of plastic clamps which weigh 38gm apiece. The system seems fairly linear but as I say there's room for improvement. The strain gauge is rated at 1kg and could be more compliant and my angle brackets could be less compliant!
And I'd not get too excited about the erratic readings, I think with filtering it will show a smoother, less peaky response. I'll try and have a go at that over the next days. And test some more motors.
The motors are powerful though. The standard size Sharkey I made as a flying testbed was way overpowered. I scaled it up from 35cm span to 45cm and now its about right. It'll be interesting to see what happens with motors with the nozzles drilled out.