I find this interesting, and have always wondered what kind of tolerance is actually meaningful in our reloading. It seems that reloaders always want to hold the smallest unit they can measure to. With the old digital scales it was +/- 0.1 gr. Now everyone has the A&D's and everyone seems to want +/- 0.02 gr. tolerance on everything. I haven't ever run in to anyone who did any testing with meaningful sample sizes to determine if the AVG, SD, and ES are actually any better when holding these super tight tolerances. The Hornady guys tease in some podcasts that they have, but I havent seen it yet (maybe I missed that podcast). Sounds like you have done the testing, and I would love to hear your results if you have them available.
Along these same lines; I have always wondered how consistent the energy distribution within the powder is. For example, does 50.00gr of powder X always yield energy of Y, or is there a margin that it fluctuates within and how big is it. If we knew this, then maybe we could mathematically calculate what our charge tolerances really needed to be held to?
Oh...and I hope Paul releases the IP system soon!
Improved precision of measurement is often a good thing but we can end up pursuing inconsequential increments. I'd posted here in the past about scale precision with strain gauge and magnetic force restoration devices (search "Adventures in Metrology"). The A&D unit is excellent and when I test with calibration weights it lives up to the 0.02gr promise. That lets one do some comparative analysis. With my pet Redding LR-1000 scale throwing a 40-45 grain charge, if I test fifty throws, the spread is about 0.5-0.6 grains. (A Harrell scale was worse). My Ohaus (and most strain gauge scales) claim 0.1 grain resolution. If you check a series of those charges on the A&D, the actual spread will be about 0.2 grains or so, and that is if you watch closely for drift.
The newer TRX-925 is a strain-gauge scale but claims improved resolution, and is the best-performing strain gauge scale I've tested (with calibration weights), but is not as accurate as the A&D. Being a strain-gauge scale, it is still subject to drift and arrives at the weight significantly slower than the A&D scale.
I was curious about the utility and have checked this on two occasions. The first was using a pet Stuteville 6.5 Creedmoor rifle and handloads with H4350 powder, CCI 450 primers, Lapua SRP brass (once-fired and annealed with an AMP annealer), and Redding bushing dies on a Forster Co-Ax press. Velocities checked with a Labradar. I did 20 or 17-shot strings. Thirty would be better, but I get bored easily and if a difference isn't apparent in twenty rounds, I'm sure I cannot shoot the difference. (It's analogous to eating oatmeal daily because it will drop your cholesterol by two points. It's an improvement but not worthwhile unless you enjoy oatmeal).
I took the velocity readings and used Excel to derive the mean and standard deviation. I then used Excel to run an F-test to look at the differences. If you compared the Redding thrown (not weighed) shots to either the strain-gauge or A&D rounds, there was a difference at the 95% probability level. I'm pretty cocky that would persist for a more rigorous 50-shot comparison. But comparing the strain-gauge to A&D weighed rounds showed no difference.
I happened to check this again last week using a different 6.5 Creedmoor rifle, but just comparing Ohaus to A&D weighed charges. When I initially looked at the results, I thought that I'd found a real improvement with precision weighing. The Ohaus rounds had a mean velocity of 2805 fps with a SD of 25.7. The A&D precision weighed rounds had a mean velocity of 2808 fps and a SD of 15.1. I was about to run an F-test but looked at the data more carefully. (The group sizes were essentially identical)
The first Ohaus round fired had a velocity of 2713 fps, well under all the others. It was about 40 degrees and this was the first round out of a cold rifle. Although H4350 isn't particularly temperature sensitive, the other rounds spent a little time in a hot chamber as I aimed. Likewise, the rifle sat and cooled for a few minutes while I fired a ten-shot group with the .308. The first "precision weighed" 6.5 CM round was 2768 fps, also the lowest velocity in that 17-shot string.
I re-analyzed the data, discarding the first "cold" round and just used the 16 rounds that followed. The Ohaus rounds then showed a mean velocity of 2811 fps with a SD of 9.8. The A&D rounds showed a mean velocity of 2811 fps with a SD of 11.5.
Ooops. I didn't run an F-test, but I'm confident that there was no significant difference after excluding that first "cold" round. I'll still keep using the A&D scale, however. First, unlike the Ohaus strain-gauge scale, the A&D never exhibits any "drift" whereas I have to watch the Ohaus and periodically tare it back to "zero" (which is a class characteristic of strain gauge scales). Also, the response and result on the A&D scale is much faster. I don't enjoy weighing individual charges but it is worthwhile for precision rifle ammunition. The speed gain with the A&D/Autotrickler/Ingenuity Precision system is substantial.
It may seem counter-intuitive that precision weighing does not reduce velocity SD. With cartridges in the .308/6.5CM class, I generally find that one grain of powder increase yields about 60-70 fps increase in muzzle velocity. When I check the Ohaus weighed "0.1gr" charges on the A&D scale, I'm decreasing my powder charge variation by about 0.10-0.15 grains with precision weighing, so you would think that the velocity SD would drop by about 6-7 fps. Nope. The trouble is the muzzle velocity depends on other factors also, like the powder temperature, case capacity, exact bullet diameter, primer ignition, and more. The total "error" in velocity is the square root of the SUM of the individual "error" squared for each of these factors. So even if one drives the error in powder weight down to zero, you don't gain measurable benefit from eliminating a powder weight variation of only 0.1 grains. (The mathematical proof is gnarly, but remember that all the errors work in both directions)
Cal Zant's "Precision Rifle Blog" has some excellent essays on these considerations. Improving the accuracy or velocity SD of our pet load can be helpful, but only within limits. One does reach a point of "painting the lily" where further improvement just does not help get rounds on target. At some point, you have to declare victory, push away from the reloading bench, and learn to call that damn wind!