Group size is all that matters. You can get lost interpreting numbers.
Load Development Using Chrono Velocity Ladder
- Thread starter CharlieNC
- Start date
Gabe22BR
Gold $$ Contributor
I seen a lot of Radars on benches at the 600yd Nationals past weekend…
Yet they rendered useless by them using it… which took up time to set it up & tear it down.
5 people actually loaded at the range -during the match to stay on top of their game .
Mr. Sauter was one of them that loaded at the range.
I had mine out 2 days prior to the event-testing &tuning.. on match day, it (radar) didn’t block my view of the flags that was out..cause I didn’t set it up or load at the range.
I carried 3-4 different loads for each rifle to the bench- depending on temp
Yet they rendered useless by them using it… which took up time to set it up & tear it down.
5 people actually loaded at the range -during the match to stay on top of their game .
Mr. Sauter was one of them that loaded at the range.
I had mine out 2 days prior to the event-testing &tuning.. on match day, it (radar) didn’t block my view of the flags that was out..cause I didn’t set it up or load at the range.
I carried 3-4 different loads for each rifle to the bench- depending on temp
Jager
Gold $$ Contributor
Amen.
I use a chronograph during load development so I know what chamber pressures I'm climbing into. Beyond that, count me among the great unwashed who believe that the only numbers that matter are the kind you measure downrange with calipers.
Fresh off the press. Thanks for posting this. I hope this gets some traction.
If SD is 10 fps 99% of all shots are withing -30 fps, MV-ave, +30 fps
To be meaningful, the sample size of MVs should be large.
To know what +- 30 fps does to the vertical dispersion at your target distance, use your favorite ballistic app.
68–95–99.7 rule - Wikipedia
RegionRat
Gold $$ Contributor
To make it more graphical to understand why velocity flat spots can be slippery...
If you take Keith's example of a population of say 1000 shots, and see what happens with 3 shot samples, versus 5, versus 10, etc., and plot the high value of the averages and low value of the averages for the given number of shots in a group, and then show the SD highs and lows, you get a plot that shows you start to converge as your sample numbers go up, but you still have some risk that you have some errors.
Here is what it looks like for a set of 1000 shots with an average of 2750 and an SD of 15 and ES of 90.
On the left is the velocity Average of a given group size on the lower axis, on the right is the SD estimate of the given group sample.
You can see that for a given number of shots in a group sample, the averages, ES, and SD can be off by more than the size of the "flat spot". Even with several shots in the group, the population can show an ES of 90 FPS if the SD is actually 15. (ES ~ 6 x SD for a normal distribution)
Some folks might think that if their prep and loading is disciplined, they may escape fate. It is certainly much better to have tight loading discipline, but when it comes to predicting velocity on small samples, even an SD of 5 doesn't save us. Here is the same analysis on 1000 shots with an average of 2750 but an SD of 5, and ES of 30.
YMMV
If you take Keith's example of a population of say 1000 shots, and see what happens with 3 shot samples, versus 5, versus 10, etc., and plot the high value of the averages and low value of the averages for the given number of shots in a group, and then show the SD highs and lows, you get a plot that shows you start to converge as your sample numbers go up, but you still have some risk that you have some errors.
Here is what it looks like for a set of 1000 shots with an average of 2750 and an SD of 15 and ES of 90.
On the left is the velocity Average of a given group size on the lower axis, on the right is the SD estimate of the given group sample.
You can see that for a given number of shots in a group sample, the averages, ES, and SD can be off by more than the size of the "flat spot". Even with several shots in the group, the population can show an ES of 90 FPS if the SD is actually 15. (ES ~ 6 x SD for a normal distribution)
Some folks might think that if their prep and loading is disciplined, they may escape fate. It is certainly much better to have tight loading discipline, but when it comes to predicting velocity on small samples, even an SD of 5 doesn't save us. Here is the same analysis on 1000 shots with an average of 2750 but an SD of 5, and ES of 30.
YMMV
Attachments
I want to point out something relative to ES. When we do the prediction mathematically the ES will be around the 6xSD point and represent the 0.3% of the population at and beyond 6xSD (99.7%). In real life ES is a different value and will likely exceed the 6xSD. This ES is due to errors in loading charge, bad primers, wrong lot of brass, excessive brass volume variations or instrument errors. Or as we used to say, "S__T HAPPENS!". These are often considered as outliers.
The real life ES is larger than 6xSD is because SD is not meaningful for small sample sizes. As the chart above shows, when "measuring" SD (say real is 5), as the number of samples increases the measured SD would converge from 0 SD to 10 SD down to 3 SD to 7 SD
The functions used to generate the the random numbers and the normal distribution function in Excel will not generate ES numbers slightly more than 6xSD. Real life testing unfortunately includes numbers such as I described.
Lets keep in mind that we are comparing loads to other loads and not not necessarily caring about the absolute statistically significant values of accuracy and velocity themselves but how these loads statistically compare to each other.
For example lets say a good load has half the velocity SD of a bad load. The probability that a good load displays a smaller SD number than a bad load over five shots is roughly 85-90%.
Those same two loads sampled with 10 shots each changes that number to 95-97%
But solid (not necessarily great) SD numbers are not the only thing we are looking for. We also look for group size and barrel harmonics (either in terms of POI stability or positive compensation depending on application).
A good load has all three characteristics. If you use the meeting of these three characteristics as a sort of heat map to point in the right direction and use metrics that utilize every value in a sample (SD instead of ES, average distance to center vs farthest center to center, ect...) you can leverage your probabilities to winnow a good load from bad loads (or just eliminate bad loads) with less than statistically significant numbers of measurements. The closer you get to a good load the more likely these criteria are met.
The more precise and repeatable your equipment and sharper your skills the fewer shots required as variables are lessened.
The more data you can collect per shot the fewer shots required to select a load.
For example lets say a good load has half the velocity SD of a bad load. The probability that a good load displays a smaller SD number than a bad load over five shots is roughly 85-90%.
Those same two loads sampled with 10 shots each changes that number to 95-97%
But solid (not necessarily great) SD numbers are not the only thing we are looking for. We also look for group size and barrel harmonics (either in terms of POI stability or positive compensation depending on application).
A good load has all three characteristics. If you use the meeting of these three characteristics as a sort of heat map to point in the right direction and use metrics that utilize every value in a sample (SD instead of ES, average distance to center vs farthest center to center, ect...) you can leverage your probabilities to winnow a good load from bad loads (or just eliminate bad loads) with less than statistically significant numbers of measurements. The closer you get to a good load the more likely these criteria are met.
The more precise and repeatable your equipment and sharper your skills the fewer shots required as variables are lessened.
The more data you can collect per shot the fewer shots required to select a load.
What matters is what can be measured on a target, whether it's at 100 yds. or 1000 yds. The only way I feel chronograph data helpful with regard to load development is that it tells me how well my reloading process produces ammo that's consistent enough, or not. Sometimes I chrono 10 shots for what I feel produces a good reading, but usually only do 5 as my numbers for 10 aren't much different from what I get for 5. Over time, the accumulated results tell me I'm at where my goal is (getting single digit SD's with ES's in the teens). But yup, there's always room for improvement. 
Excellent Post
CharlieNC
Gold $$ Contributor
Regarding sample size, the other important factor which is not often mentioned is how small of a difference does one want/need to detect. You do not need "large" sample sizes to detect "large" differences vs the background variability. For example if I load 5 each at a different charge weight and the measured velocities are 2715, 2700, 2705, 2725, 2740 vs 2815, 2800, 2805, 2825, 2840 the es for each is 40 and the averages are statistically different with "small" sample size. The rule to determine sample size depends on the variability, degree of difference to detect, and level of statistical significance desired. There is no "right" sample size.
That 39.6 of comp under a 175 smk looks like a fun load.What matters is what can be measured on a target, whether it's at 100 yds. or 1000 yds. The only way I feel chronograph data helpful with regard to load development is that it tells me how well my reloading process produces ammo that's consistent enough, or not. Sometimes I chrono 10 shots for what I feel produces a good reading, but usually only do 5 as my numbers for 10 aren't much different from what I get for 5. Over time, the accumulated results tell me I'm at where my goal is (getting single digit SD's with ES's in the teens). But yup, there's always room for improvement.
View attachment 1370797
Yes, it was/is. That 24" barrel was swapped out after 5,500 rounds for a Krieger 26" heavy and I haven't shot that load in it as yet; busy with 169's and Comp now and getting to know my 6.5 PRC.
RegionRat
Gold $$ Contributor
@Straightshooter1 , I was very impressed with your data. Not only with your discipline to collect and maintain your notes in such an organized fashion, but also with the results. The performance is exceptional.What matters is what can be measured on a target, whether it's at 100 yds. or 1000 yds. The only way I feel chronograph data helpful with regard to load development is that it tells me how well my reloading process produces ammo that's consistent enough, or not. Sometimes I chrono 10 shots for what I feel produces a good reading, but usually only do 5 as my numbers for 10 aren't much different from what I get for 5. Over time, the accumulated results tell me I'm at where my goal is (getting single digit SD's with ES's in the teens). But yup, there's always room for improvement.
View attachment 1370797
I hope you don't mind me using a sample of your data to illustrate a point on the theme of prioritizing the target over the chrono. I plotted the SD versus your group size. What is shows is a R^2= 0.0592, and for those who don't speak statistics that is like saying no correlation at all.
Not to say don't use a chrono or pay attention. These data clearly show a meticulous disciplined loader at work, and a chrono is a form of Quality Control. It augments the target work, but doesn't trump it.
Here is another example of a great performing load, giving nearly a factor of three dispersion in SD values, while keeping under 0.5 MOA.
Well done Straighshooter1.
How about several single shot ladders. It's easy enough to rule out the extremes and bad areas and use fewer rounds with each string. Take your time, reload and shoot on different days and the sweet spot will be obvious.
Attachments
First question is that at what distance are those groups?
Note: Group sizes could be broken in 2 components: Vertical spread and Horizontal spread.
The MV SD/ES would affect almost only the Vertical spread, and it would be very critical at very long range.
Example, at distance of 300 yards SD mean variance is inconsequential to the group size.
Even at 1,000 yards, a 1% MV spread, say + from 3000fps +- 15 fps. Those +-fps would ammount to +- 0.1mrad or so depending on Bullet G7 and MV.
Long Benchrest shooters prefers scope with 0.128 MOA turrets, so they can center the POI as much as possible. A 1/4 MOA or 0.1mil scope would not be able to fullfill their needs.
I will share the spreadsheet for the MV at 1000 yards a bit later today
RegionRat
Gold $$ Contributor
The theme of the thread is simple.... but not if we don't show the targets....
The concept is that "velocity flat spots" may not show in single or even low sample tests, just because the natural dispersion of velocities is perfectly normal.
The issue is that velocities that are out of tune in terms of group precision, may look as good as ones that are in tune.
So not only is the idea of tuning while only looking at single shot velocity tests risky in terms of how flat spots might disappear with more samples... it is also that the optimal tune velocity may or may not correlate to the velocity plots at all.
Upgrades & Donations
This Forum's expenses are primarily paid by member contributions. You can upgrade your Forum membership in seconds. Gold and Silver members get unlimited FREE classifieds for one year. Gold members can upload custom avatars.
Click Upgrade Membership Button ABOVE to get Gold or Silver Status.
You can also donate any amount, large or small, with the button below. Include your Forum Name in the PayPal Notes field.
To DONATE by CHECK, or make a recurring donation, CLICK HERE to learn how.
Click Upgrade Membership Button ABOVE to get Gold or Silver Status.
You can also donate any amount, large or small, with the button below. Include your Forum Name in the PayPal Notes field.
To DONATE by CHECK, or make a recurring donation, CLICK HERE to learn how.