Perhaps this, with a sufficiently sensitive method of detecting the beam's movement, a reloading scale (presumably tuned) is sensitive enough to register the weight of a single kernel of smokeless powder. I had already determined that using a web cam for magnification, with the added advantage of being able to see the gradual and proportionate deflection as additional kernels were added. I was using Varget, and after adding four kernels, one at a time, and seeing the increase in deflection that was caused as each was added, I set the scale to .1 gr. and the pointer came back to zero position.
I was just curious what it would take to get one kernel accuracy. The promeythous 2 powder gets that accuracy but costs $3000 or so to lease it for 25 years and that is a little out of my budget for something I don't need.
I was thinking a micro version of this bowl feeder would work as a single kernel trickler with at least two switches on the beam so it doesn't take forever, a high/low speed setup.
You show one kernel sensitivity for your scale. And that is good. But it does not show one kernel weight. Or show accuracy. To show accuracy we must show repeatability.
The photo electric switch is tripped when the scale beam swings through the light. This shows that the beam has moved, it does not show HOW FAR the beam has moved. Or if the beam has moved the same amount as the last time you dropped one kernel in the pan. If you drop another kernel in the pan, what does that kernel weigh. With a very sensitive and readable scale you will be able to see that not all kernels weigh the same.
With your set up you will achieve very good results though if you throw a charge and trickle up until your light comes on. Have you ever seen the Omega powder trickler. http://www.omegapowdertrickler.com/Home_Page.php
If you want to see the difference between kernel weight you might try a web cam set up in front of your pointer like Boyd suggest. A very expensive lab scale. Or a mirror centered over the pivot point of the scale with a laser pointing at the mirror. The reflected laser light going a distance of 6 to 8 feet will in affect make the beam of your scale 6 to 8 feet long. A small amount of movement will be much easier to see if the pointer on the beam is 8ft from the beam pivot point. Something like this.... http://youtu.be/mvnUXxYjX08
I like the bowl feeder idea. And how about some type of sensor to count the kernels as they drop into the pan. Keep working on your idea.