6mmBR.com SPONSORS

A Scale is a Scale--Or Is It? By Brand Cole, Prometheus Tool Corp.

I was approached by the folks at 6mmBR.com to provide a Prometheus scale for testing. I did not at this time have a demo unit to send out, and was curious about the testing protocol being envisioned. That conversation led to this article. It will be kept basic, with concepts presented but not mathematically proven. For that you may consult a statistics text.

Manufacturing uses statistics to predict the dimensions (measurements) of a population (all units produced) based on measuring a sample (the few parts you actually measure). In this instance we are "manufacturing" weighed powder charges, and want to know how "accurate" the population of the charges will be by measuring a small sample. These charges will typically follow what is called a "normal distribution". This is the same thing as the old "bell curve" that was used in school to grade on the curve. You may have heard the term "standard of deviation" (SD). Most any scientific calculator today will calculate the SD once you input the raw data points. If you take the mean (average, or (X)) of your sample, and add to that plus and minus three (3) times the SD you will capture or describe 99.7% of the entire population. This technique is known as "Six Sigma" since the Greek letter sigma is used as a symbol for SD. Six-Sigma is a fairly standard technique in many manufacturing firms. There are some parameters that must be followed here, including sample size. This is a topic in and of itself, but the bigger the sample, the more accurate your results will predict the population. In most SPC work in manufacturing, between 5 and 10 pieces are typically sampled to look for trends in the dimension.

Now we need to discuss "measuring" something. If you want to know how accurate your dial calipers are that you measure your case lengths with, you do not test the calipers against a tape measure. You need precision blocks that you measure. A basic rule of thumb is to test at about 10 times the precision you want to hold. For example, if you want to measure to 0.1 gr, you need to "test" to 0.01 gr precision.

With these two concepts in mind, let's look at some test results for some scales and measures. For the "master" scale, I used a Denver Instruments TR-203 that reads out to 0.01 gr, and is linear to 1.0 UOR, or "unit of resolution". So even this scale does not truly weigh something accurate to 0.01 gr, but it will give good validity when weighing to 0.1 gr. The powder tested was Varget, a very popular 6BR and .308 Win long-range powder, which is fine-grained enough that it "should" work well through a powder measure. A charge weight around 40 gr was used. I used a sample size of 10 and ran the statistics for each device, including the Mean (average) (X), the standard of deviation (SD) and the extreme spread (ES) and then calculated the Six-Sigma with which we will predict the "population" of the charges such as if we were loading a thousand or so rounds.

PROMETHEUS

39.95

39.94

39.96

39.96

39.96

39.95

39.93

39.96

39.95

39.96

Mean = 39.95

SD = 0.010

ES = 0.03

Six-Sigma = 0.062 Process Accuracy: +/- 0.03 gr

HARRELL'S MEASURE

40.06

39.89

39.96

39.95

39.73

40.10

39.79

40.08

40.01

40.18

Mean = 39.97

SD = 0.142

ES = 0.45

Six-Sigma = 0.847 Process Accuracy: +/- 0.4 gr

RCBS 5-10 SCALE

40.05

40.19

40.06

40.12

40.19

40.16

40.20

40.14

40.12

40.21

Mean = 40.14

SD = 0.057

ES = 0.16

Six-Sigma = 0.341 Process Accuracy: +/- 0.15 gr

Old REDDING SCALE - hydraulic damped & custom-tuned

39.92

39.94

39.93

39.98

39.92

39.95

39.93

39.96

39.97

39.93

Mean = 39.94

SD = 0.021

ES = 0.06

Six-Sigma = 0.127 Process Accuracy: +/- 0.06 gr

There are some interesting conclusions to be drawn from this study. Notice that Six-Sigma is basically two times the ES. This Six-Sigma truly is the variation you will see over a big run. It's been proven in industry time and again. Most of you probably suspected the measure would be worse than the scale weighed charges, but even the charges weighed on the 5-10 was basically +/- "a tenth grain" that one sometimes hears that a good measure will throw to. I have only tested one electronic "reloading" scale, and it was not as good as even the RCBS 5-10 tested above. The Redding scale tested was a very old model that is no longer available, and I have really tuned it up. It is the slowest of the scales to operate though, as the damping system/beam takes perhaps 20 seconds to stabilize. Beam scale accuracy is a function of two things, design parameters and quality of construction.

Profile: Brand Cole is the Designer/Builder of Prometheus Powder Measures. The $1175 Prometheus is unquestionably the finest machine of its kind in the world. As Mr. Cole's article explains, the Prometheus can deliver accuracy nearly an order of magnitude better than the $250-$300 Digital Dispensers we tested. (Digital Dispenser precision of +/- 0.1 gr yields an ES of 0.2, and that's assuming no zero drift.) Top shooters like David Tubb use the Prometheus for a reason--because it will deliver loads consistent to a single kernel of powder, and the zero will remain rock-solid during the entire loading session.