Re: RCBS Chargemaster VS Hornady LNL Auto Charge
Having torqued of some of you for my barrel 'survey' I am hoping to redeem myself with some information on digital scales and their accuracy.
My first digital scale was an RCBS Rangemaster, I thought it was the cat's a$$ so I bought a Lyman 1200DPS. Comparing the two to each other and my 10-10 I found a lot of variation. So I bought an Acculab VIC, from Sinclair, which reads to .01 grains and is lab quality. Using this as the 'master scale' I dumped the Rangemaster (which was way off) and upgraded to the DPS3. The DPS3 was not any better so I finally did the right thing and bought the RCBS ChargeMaster Combo. Finally I started to get 'similar' results between scales.
Having been a quality engineer I needed data to confirm I was feeding my boys uniform loads. I used a statistical test called an "Instantaneous GR&R" to compair the 3 scales (scales only not dispensing capability). If they can't weigh accurately they can't throw accurate loads. I do not have access to a Hornady LNL AutoCharge so I could not include it in the study. If anyone wants the excel spread sheet with the data and formulas, please PM me and I'll send it to you. You could use one matrix to check your Hornady if you wanted to.
The test consisted of ten samples weighing from 20gr to 277gr. Each sample was weighed 10 times on each scale giving me 100 data points per scale and 300 total. This test has a 95.5% confidence level.
The average was calculated for each set of 10 measurements for each sample for each scale
The StdDev was calculated for each data set of 10 readings for each sample on each scale. It was then averaged for each scale as an indicator of measurement error. 4 StdDev = 95.5% of the probable measurement error.
The Range within each group of 10 data points was also averaged for each scale. This gives a practical view of the scales variation across all of the sample sets for each scale
The 30 measurements for each sample (from all 3 scales) was averaged. This is the Grand average and is the best estimate of the true weight of each sample. The difference between each scales average for that sample was subtracted from the Grand Average and is shown as ERROR. Numbers in () indocate the Average was lighter than the Grand Average.
The RCBS was almost as accurate as the Acculab, and very consistent. The Lyman had significant variation and other problems during the test (lost zero and calibration often)
The RCBS was the furthest from the Grand Average in all cases, which means it reads slightly heavier than the actual weight by approximately .059gr.
• This may be a calibration error, I did calibrate all scales before the test though. I do not see this as a problem because it is consistent which means the loads you develop on it will always be the same.
95.5% of the time the weight displayed will be within .04gr of the actual weight which is excellent since it can only read to .1gr anyway!
• This means the RCBS unit has a very good algorithm for discriminating halves of tenths (in other words it does a great job rounding up or down to the nearest tenth).
IMPORTANT - all the scales were powered by an industrial quality power filter. You must use a high quality power filter or UPS (Uninteruptable Power Supply) for digital scales because any line noise ruins accuracy and consistancy.
>Avg Range - Average of the (Max - Min) of the ten readings for each sample on the same scale
>Avg StdDev - Average of the Standard Deviation of ten readings for each sample on the same scale
>Avg ERROR - Average of the difference between Grand Average and the Average for ten readings for each sample on the same scale (##) indicates lighter than Grand Average
>Grand Avg - Average of 30 readings for the same part from all three devices - best estimate of 'true value' (central limit theorem)
>*2 - 2 times the StdDev indicating that most (95.5%) readings will be this much plus or minus of the actual value
>*4 - 4 times the StdDev indicating the overall band of error for 95.5% of the readings
Lyman 1200DPS
-------- Average .. *2 ... *4
Range .. 0.140
StdDev.. 0.048 .. 0.096 . 0.191
ERROR ..(0.048)
Acculab VIC
-------- Average .. *2 ... *4
Range .. 0.034
StdDev.. 0.013 .. 0.025 . 0.051
ERROR .. (0.010)
RCBS Chargemaster
------- Average .. *2 ... *4
Range .. 0.050
StdDev . 0.020 .. 0.040 . 0.080
ERROR .. 0.059
