Ammunition Testing Metrics???

[shutr]

Greetings,

Today I had a "randow thought while shaving". When we test ammo we normally make our selection based on average group size. Is this really the best method or is there a better method???

My theory is that you could define a way to weigh the erratic "flyers" more in the analysis part. Sort of like shooting 99-9X's or 100-5X's. The second one wins the match although X count isn't as high. I'd rather win the match than have a lot of X's.

Has anyone tested ammo and used standard deviation, standard error, or maybe even circular error of probability in the process? What were your results? I welcome your thoughts.

Anyone know how they test ammo at the ammo plant?

Dave Cloft

[Tony Nielson]

Dave,

Have you read the research on our website?

http://www.nielsonbrothersarms.com

Might answer some of your questions.

Tony Nielson
Nielson Brothers Arms, Inc.

[shutr]

Superb Job! I had never even considered bullet concentricity before. I've never put much credit into rim thickness or weight sorting, as too many other variables could affect the entire unit's weight. Very poigniant and insightful! What ammo tends to have the most consistent concentricity?

So in your oppinion, what makes one lot of ammo shoot better in a particular rifle than another? Is it that the entire lot is good? Or is it that the combination of random errors between the rifle and ammo somehow equalize each other?

[Tony Nielson]

Dave,

Concentricity doesn't solve everything, it just takes out what we believe and what are research shows is one of the biggest varriables. You still have the number one problem of harmonics and this is probably why one lot will shoot great in one rifle but not another. The obvious question becomes if concentricity varriations contribute significantly to group size, can you really pick a lot of ammo for your rifle without sorting by concentricity? In the benchrest world a similar question may be can you effectively set a tuner without sorting by concentricity?

One of the areas that is very apparent when sorting by concentricity is that different lots have different yields. For instance a lot of Eley from machine #3 might have 10% with 0.000 run-out where machine #4 has 17% with 0.000 run-out. The ammo from machine #3 might have 2% with 0.003 run-out where machine #4 has 0%. Obviously becomes very important when you are purchasing a lot of ammo and especially when you are using it. I don't think anyone would want one of those 0.003 bullets to end up in the chamber when they are going for record.

Be looking for some independent tests of our product and theories in upcoming issues of Precision Shooting, American Rifleman, Shooting Sports USA and the Small Caliber News. I will be happy to answer any additional questions you may have.

Tony Nielson
Nielson Brothers Arms, Inc.

[GOVTMODEL]

Anyone know how they test ammo at the ammo plant?

Dave Cloft

At the Eley works, after a batch of ammunition is made a sample goes to the proof range for testing. Approximately 500 rounds are fired through four test rifles, over chronographs. The data collected includes group size, initial velocity, and a number of other metrics.

After testing the reults are analyzed, and depending on the results, a label is assigned to the batch, i.e., TENEX, Club, Target, etc.

[Guest]

Er not quite! Tenex and Club are most definately different and always have been. Some Match EPS is Tenex that didn't quite make the grade; and some good Tenex is packaged as Match as that was what Eley wanted that day.

Tim S

Exeter, UK.

[Michael]

I looked at the test targets on the Nielson site and it seems to me that the groups were measured center to center - not outside diameter - is it correct?

[GOVTMODEL]

Er not quite! Tenex and Club are most definately different and always have been. Some Match EPS is Tenex that didn't quite make the grade; and some good Tenex is packaged as Match as that was what Eley wanted that day.

Tim S

Exeter, UK.

My error! TENEX is the top of the heap in the white bullet line, Club is the top of the heap in the black bullet line.

I stand corrected, and hope Bert forgives my error:-)

[Fritz]

Club is way down the list. Match is the top of the black box.
fritz

[Tony Nielson]

Michael,

Yes we had to build a special device in order to accurately measure group sizes and we built it to measure them center to center. It is just a measurement and you can express it anyway you want. The important part is that you have an accurate and repeatable method to take the measurement.

Tony Nielson

[shutr]

I'm working on an experimental design. First, I want to prove that there are statistically signifcant differences among group sizes between lots of the same top line of ammo that comes from the UK in a red box.

Then, I'm trying to find the right methodology to find the optimum ammunition for a specifc rifle/ammo combination.

I guess my first step would be to look at what research has already been done so I don't reinvent the wheel. Anyone know of any good sources?

Thanks!

[GOVTMODEL]

I'm working on an experimental design. First, I want to prove that there are statistically signifcant differences among group sizes between lots of the same top line of ammo that comes from the UK in a red box.

Then, I'm trying to find the right methodology to find the optimum ammunition for a specifc rifle/ammo combination.

I guess my first step would be to look at what research has already been done so I don't reinvent the wheel. Anyone know of any good sources?

Thanks!

IMHO, you're chasing the wind.

1. It's no secret that the batches of red box ammo from the UK that perform best in the makers in-house Quality testing program are reserved for the Customer Range. That makes sense to me- the folks who make the effort to do batch testing get the best stuff.

2. The methodology used there is to mount the rifle in a test fixture and fire groups. From the batches that look promising, additional groups are fired. When I went I fired about a brick of their ammunition; there were seventeen batches in stock and basically used up one box from each batch.

SUMMARY-
There is enough variation from batch to batch of even the best ammunition and from gun to gun of the best guns, that batch testing ammunition is the optimum way to select ammunition. The gun and ammunition makers are making products that are as consistent as they know how. SK Jagd opened a customer range last year. I don't know anyone who has tried it.

Richard

[Tony C.]

Regarding "finding the methodology to find the optimum ammo for a specific rifle/ ammo combo". I'll second GOVTMODEL's opinion that you are chasing the wind. fact is the only reliable way to determine how well a specific rifle and ammo will shoot is to shoot them and observe the results. Generally if lot 123 shoot well in rifle A, it will also shoot well in rifle B. If someone can identifiy what is the determining factor causes one lot of ammo will shoot well in one rifle but not another, lots of people, including folks from the ammo company. will love to hear it.

[Tony Nielson]

shutr, Richard, Tony C.

First to address the question of "experimental design to statistically prove". This was an area of big concern amoung several of the engineers who have had input on our concentricity project over the years. After much research and debate it was concluded that the minimum number of rounds that could be fired to have "statistically significant results" was 20 rounds. That is why you see the 20 round groups in our machine rest testing in 2004.
Shooting "batches" of ammo to determine which lot you should use in your rifle has been the norm in the past. However, what our research shows is that if you do not sort the ammo by run-out, you will not get any statistically significant results, which is what you are after in the first place. As far as differences in lot to lot of the same ammo, go look at the targets that we shot in 2004 that were Laupua ammo.

http://www.nielsonbrothersarms.com/HS%20Lapua.htm

If you look closely at the heading for that page it says those 20 round groups were fired with mixed lot numbers. We think that is a pretty significant result. Think about it.

Some good references you might turn to for experimental design of shooting tests include Rifle Accuracy Facts by Harold Vaughn or Understanding Firearms Ballistics. Hope that helps.

Tony Nielson

[Tony C.]

[Shooting "batches" of ammo to determine which lot you should use in your rifle has been the norm in the past. However, what our research shows is that if you do not sort the ammo by run-out, you will not get any statistically significant results, which is what you are after in the first place]

Perhaps I miss the point here, how do I know that case of ammo sitting on the dealer's shelf will shoot in my rifle without testing it ? You can hand sort each and every round, without test shooting, it still a guess. While on the same topic, why is that from time to time we came across lower grade ammo shoot far better than higher grade one? Why certain lot of ammo shot well in one rifle but not another? Why some rifle will shoot almost any ammo well? Why some rifle shot one or two lot of ammo extremly well but any others very bad?
If there are 10 case lot of ammo on my dealer's shelf, I would llike to know which case shot best in my rifle before I pay for it. The only way to be 100% sure is to test shoot. After I got them home, then I can hand sort each and every round.

[Tony Nielson]

Tony C.

What several people are doing is rather than buy a bunch of ammo in a few different lots, they are buying a couple of boxes of ammo from a ton of different lots. They then can check the distribution of concentricities for each of those lots.

For instance take 100 rounds from lots A through E. You want as many bullets as you can get that have 0.000 run-out. After checking these 500 bullets you find that they have the following yields of bullets with 0.000 run-out.

A = 12%
B = 15%
C = 8%
D = 22%
E = 17%

This example is oversimplified because you would obviously look at the yield of 0.001, 0.002, etc. But you get the point, lot D has more zeros in it and is a better value for "buying zeros". However you then can take the zeros from each lot and determine which one of these lots is best fitted to the harmonics of your rifle system. For instance lot D might have the most zeros but because of the velocity of that lot, it doesn't shoot well in you rifle. Lot E or B might be better. You get the idea.

Hope that helps

Tony

[Bart B.]

USA military arsenals have been testing 22 and 30 caliber centerfire rifle ammo using the mean radius method. They'll shoot a couple hundred shots per test group then measure each shot hole's position to get the data. Mean radius is calculated and it's typically about one-third the extreme spread. I've seen some of the 30 caliber 250+ shot test groups fired at 600 yards that ranged from about 6 to 12 inches.

Mean radius would be a good way to test rimfire ammo, but only meaningful if the rifle was clamped in a decent (three-point based) machine rest. If someone holds the rifle while testing, the results will be the sum of rifle + ammo + shooter. And we all know which one of these three has the greatest variable.

My own method to test ammo (rimfire or centerfire) is to sling up in a proper prone position with a bag under my front hand and another under the toe of the stock. I can get my holding (wobble?) area on target down to under 1/4 moa. First shoot enough fouling shots to get the barrel ready. Then start shooting for group. If the group quickly gets pretty bad in a few shots, stop shooting; it's a waste of time to test ammo that's not gonna be any good in the first place. If the group starts out good, shoot at least a 20-shot group but stop when I think I'm getting tired.

Lones Wigger tells me he doesn't think there's any rimfire ammo on this planet any more that'll stay under half an inch at 100 yards. He thinks the best we're gonna do is about six to seven tenths of an inch at that range. When one compares the X count of the winners scores 30 years ago to what's being shot today, the numbers are lower. Which also means it's gonna be harder to realistically test ammo.

I think one good method is to first determine the limit for missing your call. The 'call' is what point on the target your sights were aligned to when the rifle fires and the bullet exits; rarely is it dead center in the middle of the target. If you don't want your shot to strike off call by more than 1/2 moa, then you gotta use ammo that you can shoot that well regardless of range. Mark each shot's call on a plot sheet, then plot the bullet's shot hole relative to that call. Finally measure all the distances from call to shot hole. If the greatest distance is larger than 1/2 moa, then that ammo may not meet your 1/2 moa miss your call limit.

The good thing about this method is it doesn't matter how well you shoot. The bad thing is if you can't accurately call you shots, then it won't work. If you can't call your shots very good, then refine your position, get help or whatever until you do. Remember the first rule of evaluating rifle/ammo accuracy: if you can't call your shots well, then get help until you do.

[Ross Mason]

Might like to check out this site. Jeroen has been doing some interesting maths with air rifle shooting, group sizes, expected scores etc etc.

http://home-2.worldonline.nl/~jhogema/ballist.htm

The stats part may be of interest, Especially the one about shots per group.

I suspect that electronic targets give the best way of calculating the standard deviation as you can get the x and y coordinates of each shot.

One of the key graphs on Jeroens page is the one showing the frequencies of 10.9s, 10.8s, 10.7s etc. The peak is the SD. So an easy way it to see what decimal is shot most in the test group and it is a pretty good indication of the SD.

Check out the scores in finals. 10.9s are rare. better ammos will give more 10.7s and 10.8s but still don't expect too many 10.9s.

Ross mason

[Jack Leslie]

There are two articles that I wrote that you might find interesting. One looks at several statistics and the other examines the number of shots per group. They are available in the features section of Florida Highpower: http://www.floridahighpower.com/Features.htm

The short version is that mean radius and radial standard deviation work best and the information content tops out with about 15-20 shots.

Jack