Thread: Testing a conventional approach to load development

Just chose it because it’s the longest length of the lot. I don’t think it’s going to matter anyway.
Go with what you’ve suggested 1.852 as it’ll hopefully prove a point to some.

My worst fears, Gimp is going to shoot out what is possibly a "once in a lifetime" barrel just testing shit . . .

Originally Posted by Tentman

My worst fears, Gimp is going to shoot out what is possibly a "once in a lifetime" barrel just testing shit . . .

The sacrifices I make, for the benefit of the community...

you should all chip in 10 bucks for my next barrel

Originally Posted by McNotty

Just chose it because it’s the longest length of the lot. I don’t think it’s going to matter anyway.
Go with what you’ve suggested 1.852 as it’ll hopefully prove a point to some.

I am interested in doing the best I can to test the truth of the matter.


at present the average 3-shot group with 24.5gr 2206H (across different seating depths) is 0.6MOA
the smallest is 60% smaller than the average
the largest is 45% larger than the average

my current understanding is that with any particular load, you would expect + or - 60-70% variation in 3-shot group sizes. So what I'm seeing with the "tuning" of different seating depths is within the normal variation expected of any load. But I am very interested to do my best to test the process as so many people believe in it.

The current 21-shot group has a mean radius of 0.088 MRAD (0.88cm) and a predicted O95 of 0.37MRAD. The current observed (overlaid) 21-shot group size is 1.09 MOA (0.32MRAD).

If tuning works - a 20 shot group of the "best" load will have a smaller mean radius and a smaller observed group size over 20rd (overlaid) than the "worst" load.

If tuning doesn't work - both loads will be similar (within say + or - 20%) of each other for mean radius (more powerful) first and observed group size second (more variable), and also similar to the current overlaid 21rd group (although they should both be better than a group with 0.021" variation in seating depths - that's not consistency...)

All current shots fall within the same general "scatter"

does it get tighter with the selection of a specific seating depth that produced a better group in testing? We will find out.

Originally Posted by gimp

my current understanding is that with any particular load, you would expect + or - 60-70% variation in 3-shot group sizes. So what I'm seeing with the "tuning" of different seating depths is within the normal variation expected of any load.

This figure comes from hornady -



And it is consistent with my observed results - for example https://www.nzhuntingandshooting.co....3/#post1656779


With the factory ammo that shoots around 3.5MOA out of my rifle, 10x 3-shot group sizes showed a variation of + or - 60-70% from the average group size.

Originally Posted by gimp

So - what now?

Speaking from experience, I would have thought the correct answer to be - use up the remaining components, while making minor yet arbitrary changes to different variables.

Then take a photo of the best three shot group, and post with a title to the effect of "load development nailed, off hunting!".

Cool to see some results, although I would have picked the charge with the lower es and ok on both sides, 25.1g�� As for the seating depth nice to see the changes in such small variations��

To be clear the "changes" visible in 3 shot groups are within the variation that happens within 3 shot groups of a single load. At present there is no evidence at all of any difference between the seating depths. SD is meaningless for a sample of less than 10 at minimum, more like 20-30. I've loaded 20 of each and I'll shoot tomorrow or Saturday and post results.

Okay so I went and shot these this morning. Conditions were good - essentially nil wind, my usual setup shooting at 100m with Atlas bipod & rear bag prone.

First, I did the process @McNotty suggested - 2x 3rd groups, and a 4rd group. Here are the results -




1.861 ("best" load identified) - produced groups of
3rd 1: 0.57 MOA
3rd 2: 0.79 MOA
4rd: 0.55 MOA

Mean radius from these 10 shots overlaid was 0.101MRAD and a 10rd aggregate group size of 0.79MOA. Looked pretty good.

1.852 ("worst" load identified - produced groups of
3rd 1: 0.85 MOA
3rd 2: 0.54 MOA
4rd: 0.89MOA

Mean radius from these 10 shots overlaid was 0.099MRAD and a 10rd aggregate group size of 1.21MOA. Looked ... actually basically identical to the 1.861 load if you look at MR rather than group size. The individual groups weren't obviously different either.


So then a 10rd group of each. Looked like this. This looks like a really bad group with the "better" load - the 1.861 CBTO. However bear with me through the results.



Best load 10rd group size - 1.37 MOA
worst load 10rd group size - 0.92 MOA


So what do the 20rd from each load look like when aggregated to one big 20rd group ?

1.861:


Mean radius: 0.116MRAD group size 1.37MOA (4cm). The predicted O95 from the 20rd mean radius is 4.8CM so the group size measured falls within expectations. This is a worse group than I "expected" but it is within the calculated expectations of all the data. Did I fuck it up and pull a couple of shots - change cheek pressure - did the wind gust a little bit L-R ? Who knows but it is what it is.

1.852:


Mean radius: 0.09MRAD, group size 1.21 MOA.

It does actually appear if you look at the MR, that either -

1. There is no functional difference between the loads and there is just a bit of natural variation due to measurement error etc showing up in the results

OR

2: I cannot shoot well enough to resolve any difference between these different loads, and the difference in the MR values is the "noise" from my shooter error

OR

3. The load identified as the "worst" through 3-shot group testing is actually slightly more consistent over large samples than the load identified as the "Best"

I do feel that I may have introduced a little error on the 10rd group of the 1.861 load and, had I not, it would have likely produced a more or less identical result to the 1.852. The variance in the mean radius is higher than you would expect from natural variation but there is likely a bit of measurement error here - the group analysis in the app isn't perfect. Regardless the results are functionally identical and fall within the cone of fire of the system even if a couple right at the edge were maybe contributed by a little extra shooter error.


Here's all 61 shots with this powder charge, and all seating depths - a 1.37MOA group with a mean radius of 0.096MRAD





So conclusion - with this rifle, bullet, brass, powder and shooter combination it is not possible to identify any difference in precision between loads with different seating depths by testing with 3 shot groups, at .003 inch increments of seating depth (as recommended by Erik Cortina, if he isn't the person to listen to for this then who is ?). Or 20rd groups of same.

All results fall within the cone of fire predicted by the 10rd mean radius for any of the 10 rd groups for any load.

Writeup is a bit messy and rushed because I'm off hunting for a week

Some further reflections -

1. I would not load this bullet in this rifle with this brass/powder. I now know this load does not meet my hit-probability requirements for precision (0.07MRAD mean radius or better, so that precision is not the limiting factor on 0.3MRAD targets).

2. I did not know that this load did not meet my requirements after 36 rounds of the conventional process. Often it is stated that "3 shot groups can't tell you how good something is, but they can tell you how bad it is and then you can rule it out". None of the 3-shot groups come close to representing the true size of the cone of fire and I could not rule the load out after 12x 3 shot groups. Clearly it was at least 0.895 MOA, as that was the largest 3-shot group - but that would meet requirements.

3. Nor did I know the true velocity consistency as measured by SD. 3-shot SDs/ES are useless to represent the true velocity consistency of a load.

I did not know enough to make a decision. A single 15-shot group of a single arbitrary load selected to be most likely to be functional would have given me much more useful information.

I am so glad fishing isn't this mind boggling. My brain would melt if I had to collect and interpret this kind of data for my lure making. Not that I am making any at the moment. Which reminds me, @gimp your lure is about an hours worth of work from being finished, been like that for about 12 months