UncleNick, I found that there were several keys:
First, I run at least two impressions on each sample. If they match I keep them. If not, I run two more and average the closest three.
Second, I run the impression test for the exact 30 seconds as specified (though I believe the impression size is by its very definition a a self-limiting process if that time limit is exceeded)
Third, I keep the pad of my index finger on top of the test die to feel for any deviation from the plunger being exactly flush -- and therefore provide the "design" spring pressure of exactly 60lbs on the ball. Slight movement of the fingertip pad can provide remarkably precise feel.
Fourth,
Third, I run the ram at its mechanical-max advantage -- thereby exaggerating any movement req'd to change the pressure (and therefore the plunger position).
Fifth (and critical), I've found a simple way to (a) mount, adjust, and stabilize the microscope so I can get exact hands-off readings; (b) set the sample on its own platform that allows me to easily align the sample to the microscope's grid (instead of vice versa); and (b) set up a strong light grazing from ~30° that clearly defines the edges of the indentation. It sets up/tears down quickly, ....and works remarkably well.
Sixth (and lucky) when first testing on RotoMetal's #2 alloy I found the setup provided a repeatable 14.9 BHN on two separate alloy shipments about 6 months apart -- giving me faith in both the the Tester process and RotoMetals. Since Lyman#2 is one of very few true standards ever actually tested (See both LASC and Cast Boolits references), I've baselined from that ever since.
Seventh (and last), when I actually ran the calculations (instead of just blindly following the the little chart they give you), I discovered the Brinell method/number was nothing more than literally the pounds/sq-inch it took to deform the metal. And as the indentation grew the "sq-inches" area was simple spherical geometry of the indent as its surface area grew enough to stop the ball from sinking any further.
The magic was gone...
postscript: The only way I could see this
not working was if I pressed the ball onto a digital scale to where it read 60#, and the the plunger didn't wind up flush. Then I'd have a weak spring and the test results would be off.
But if that happened, I just determine what force
did produce a flush plunger, plug
that number back into the Brinell formula, and get corrected numbers to compare with the ball impression.