223 testing results

[Shadow9mm]

Finished up some testing today. Pretty interesting results I think. Thought I would share. Colt ar 15, Faxon 16in gunner profile barrel, 4-14x44 optic (for testing, not regular use) Larue MBT 2 stage 4lb trigger, on bags.

I watched the Hornady podcast about sample sizes being too small. It basically boiled down to the concept that you need to shoot 20 rounds per group to get reliable data. Whether that be zeroing your rifle, doing chronograph testing, or group size testing.

The other interesting thing that was noted was that, while you should shoot 20 rounds, it does not have to be 20 straight. Shoot how you use it. for a hunting gun 3 round. the idea being 1 round to hit the animal, 2nd round in case you had a bad shot, 3rd round in case everything went wrong. so fire 3, let the gun cool, repeat.

Heres the podcast if your interested https://www.youtube.com/watch?v=QwumAGRmz2I

So I loaded up 140rnds. 20rnds, 7 groups. each group had 0.003 variation in seating depth. went out and shot it at 100yds.

Since I'm in an AR platform, I estimated 20 rounds at once would be worst case. About 2 seconds per shot, enough time to get a good steady sight alignment and squeeze the trigger. gave the barrel a 10min cool down between groups.


1 4.09moa (mean radius 1.18moa)
2 3.98moa (mean radius 1.08moa)
3 2.64moa (mean radius 0.78moa)
4 4.04moa (mean radius 0.86moa)
5 3.36moa (mean radius 0.87moa)
6 3.59moa (Mean radius 1.06moa)
7 3.79moa (mean radius 0.93moa)


Considering this was fairly rapid fire, the barrel got quite hot, and the bullets were bulk fmj, I am ok with the groups. It was also interesting to see the groups did not walk, apparently Faxons stress relief process works. All the group were centered in about the same place on the target, just a hare low and left.

I will be going with group 3 going forward, and plan to do some slower fire testing to see how much the heat caused the group sizes to open up.

Before I forget. Data pulled from Hornady 11th.
55g Hornady fmj
CFE 223 27.0g
CCI 450 (small rifle magnum)
once fire Lake City brass.
Col, varied by load, but within the cannelure area on the bullet.

 

[44 AMP]

1 4.09moa (mean radius 1.18moa)
2 3.98moa (mean radius 1.08moa)
3 2.64moa (mean radius 0.78moa)
4 4.04moa (mean radius 0.86moa)
5 3.36moa (mean radius 0.87moa)
6 3.59moa (Mean radius 1.06moa)
7 3.79moa (mean radius 0.93moa)

Your math (or your description of it) is confusing me.

"mean" is short for mean average.

Radius is half the diameter of a circle.

So, where I'm confused is your group size is not double the radius, its 3 to 4 times the stated radius. I don't understand where that is coming from. The only thing I can think of is that you measure groups vastly differently than I do.

Could you explain further, please?

 

[Shadow9mm]

Your math (or your description of it) is confusing me.

"mean" is short for mean average.

Radius is half the diameter of a circle.

So, where I'm confused is your group size is not double the radius, its 3 to 4 times the stated radius. I don't understand where that is coming from. The only thing I can think of is that you measure groups vastly differently than I do.

Could you explain further, please?

So The app I use to measure my groups provides both moa and mean radius.

My understanding is that, you create a circle around the group to find the center of the group. Then you measure from each impact to the center of the group. then average the result. This allows each bullet to be a data point, rather than simply looking at the bullets that are furthest apart. kind of like SD vs ES.

Here is the article I read a while back. https://www.ammoman.com/blog/measuring-groups-with-extreme-spread-vs-mean-radius/

 

[stagpanther]

I believe mean radius is simply the mean of all shot's CTC impact relative to the POA. This isn't necessarily the same thing as overall MOA (dispersion) of the group. On Target will supply a POA--itself an optimized average-- if you don't actually designate one.

 

[tangolima]

Not the POA but the center of the group, or the average impact point. CEP (circular error probable). Sometimes it is called R50 method.

I use both R50 and R90. It makes more sense than extreme spread (ES).
https://en.m.wikipedia.org/wiki/Circular_error_probable

-TL

Sent from my SM-N960U using Tapatalk

 

[mehavey]

CEP.... I get it.
I do so love the smell of neutrons in the morning

 

[Unclenick]

Note that MOA is just a measure of the angle subtended by a length. Your whole group diameter or the diameter of the circle defined by your mean radius, or by CEP, or by any other method mentioned below can all be measured in MOA.

Geoffrey Kolbe points to a hard-to-find paper called "Statistical Measures of Accuracy for Riflemen and Missile Engineers" by Frank E. Grubbs, Ph.D. (November 1964, 2nd edition 1991). Even though the industry and the military have adopted various comparative standards over the last couple of centuries (CEP from artillery hit probability; the smallest circle that encloses 50% of a 200-round group (common ammo maker's measure); mean radius; etcetera), the statistically most efficient measure is radial standard deviation. To get that, you just square each hole's distance from the group center, sum those squares and, divide them by one number less than the number of shots, then take the square root of that result.

Kolbe then goes on to show that, from a probability standpoint, if you insist on using the extreme spread as your group measure, 7 shots per group is the most efficient at producing a statistical result, though you have to shoot a number of 7-shot groups to get to a certain confidence in the result. This funny shot number is because of the counter-intuitive situation that for samples of 7 or less, you actually get a more-likely-to-be-accurate estimate of population standard deviation by dividing the group extreme spread by a number called xi of n (written ξ; see below) than you do by using the modified root mean square method I just described for radial standard deviation.¹ This fact results from small samples of a random event not offering outliers enough chance of appearing to outweigh the extreme spread as an indicator. Once you get to 8 samples and up, the odds of an outlier increase to the point you need the root-mean-square (modified by using n-1) method I described in the first paragraph to tame the influence on the evaluation of that outlier to get the best estimate. But up to 7, for the ES method, more shots per group provide more statistical certainty than a smaller number does.

The Kolbe article is here and describes the 7-shot method and compares it to other shot count evaluations in some detail.

If you want better standard deviation estimates than your chronograph gives you for shot counts of 7 or less, take the velocity extreme spread and divide them by the ξ numbers for the different shot counts in the table below. just keep in mind that, like the group sizes in Kolbe's article, it will take a number of such groups to have high confidence in how representative they are.

¹Note that all methods of finding sample standard deviation are an attempt to estimate what population standard deviation will ultimately turn out to be.

 

[Shadow9mm]

Note that MOA is just a measure of the angle subtended by a length. Your whole group diameter or the diameter of the circle defined by your mean radius, or by CEP, or by any other method mentioned below can all be measured in MOA.

Geoffrey Kolbe points to a hard-to-find paper called "Statistical Measures of Accuracy for Riflemen and Missile Engineers" by Frank E. Grubbs, Ph.D. (November 1964, 2nd edition 1991). Even though the industry and the military have adopted various comparative standards over the last couple of centuries (CEP from artillery hit probability; the smallest circle that encloses 50% of a 200-round group (common ammo maker's measure); mean radius; etcetera), the statistically most efficient measure is radial standard deviation. To get that, you just square each hole's distance from the group center, sum those squares and, divide them by one number less than the number of shots, then take the square root of that result.

Kolbe then goes on to show that, from a probability standpoint, if you insist on using the extreme spread as your group measure, 7 shots per group is the most efficient at producing a statistical result, though you have to shoot a number of 7-shot groups to get to a certain confidence in the result. This funny shot number is because of the counter-intuitive situation that for samples of 7 or less, you actually get a more-likely-to-be-accurate estimate of population standard deviation by dividing the group extreme spread by a number called xi of n (written ξ; see below) than you do by using the modified root mean square method I just described for radial standard deviation.¹ This fact results from small samples of a random event not offering outliers enough chance of appearing to outweigh the extreme spread as an indicator. Once you get to 8 samples and up, the odds of an outlier increase to the point you need the root-mean-square (modified by using n-1) method I described in the first paragraph to tame the influence on the evaluation of that outlier to get the best estimate. But up to 7, for the ES method, more shots per group provide more statistical certainty than a smaller number does.

The Kolbe article is here and describes the 7-shot method and compares it to other shot count evaluations in some detail.

If you want better standard deviation estimates than your chronograph gives you for shot counts of 7 or less, take the velocity extreme spread and divide them by the ξ numbers for the different shot counts in the table below. just keep in mind that, like the group sizes in Kolbe's article, it will take a number of such groups to have high confidence in how representative they are.

¹Note that all methods of finding sample standard deviation are an attempt to estimate what population standard deviation will ultimately turn out to be.

7 groups of 7 shots is 21, awful close to Hornady's recommended 20 shots.

I also really like the concept of mean radius in that it uses every shot as a data point. Rather than the groups itself, or rather the ES being a single data point.

 

[Unclenick]

7 groups of 7 shots is 49 shots. 21 would be 3 groups of 7 shots. Kolbe's uses 6 groups of 7 shots in his loosest estimate, so, 42 shots.

Radial standard deviation also uses every shot if you use the standard calculation method (the modified RMS method) I described. You can adjust the modified RMS method result by an error factor for sample sizes of 7 or less that are listed in a Wikipedia article on square root bias, IIRC. Kolbe's method takes 6 groups of 7 shots to know 90% of future groups of 7 shots will be within 15% of the average size you get that way. 10-shot groups will be 10% bigger than 7-shot groups, and 20-shot groups will be 28% bigger than 7-shot groups. The reason for wanting to know outside group diameters in the future is simply that target shooters want to know what size scoring ring they can expect the gun to be able to stay within.

 

[9MMand223only]

I like how you are honest. You don't make fake groups, and post it to look cool, so SO MANY people do. AR15's, even with decent barrels, do not shoot even close to 1 MOA using FMJ ammo. And if its bulk, less premium FMJ, 3 MOA would be normal. Especially using a ball powder like that. I think of all the powders I have tested, I think the ones I like the least is....probably in this order....PP 2000-MR, CFE223. Most inconsistent powders I have tested. Wildly inconsistent.

CFE223 might be obsolete now, since they have StaBall Match, which is basically CFE223, temp stable.

 

[Shadow9mm]

I like how you are honest. You don't make fake groups, and post it to look cool, so SO MANY people do. AR15's, even with decent barrels, do not shoot even close to 1 MOA using FMJ ammo. And if its bulk, less premium FMJ, 3 MOA would be normal. Especially using a ball powder like that. I think of all the powders I have tested, I think the ones I like the least is....probably in this order....PP 2000-MR, CFE223. Most inconsistent powders I have tested. Wildly inconsistent.

CFE223 might be obsolete now, since they have StaBall Match, which is basically CFE223, temp stable.

I appreciate it. I don't have anything to prove, I just try and share my results for feedback and to hopefully help others.

I bought during the shortage and I could not get my preferred powders at the time. Now have about 10lb of CFE223 that I need to use up as that was what I could get. Trying to find a way to make it work reasonably well for a training load.

I'm excited to see the results with slower fire to keep the barrel cooler. as well as with match bullets and better powder. Hopefully I can get out and do the testing in the next few weeks.

 

[USAF Ret]

There is data. Not good or bad data, just data. It is what you do with that. As a PM, very familiar with analytics.

Looking forward to adjustments and how those affect your goal.

And information helps us all.

Much appreciated. Just lets me know I need more range time.

 

[Nathan]

Thank you for your effort here. This is a ton of effort. Good stuff!

Can you share your oal’s 1-7? Also, how was the charge weight arrived at? Is it relatively high or low compared to published data?

One of their key points was the seating depth changes don’t make a huge difference unless drastic like 0.040” changes. If you see, other than group #3, your data basically trends from 4moa @ #1 to 3.7moa @ #7. I wonder what happens if you reshoot #3 what happens. Statistics would say you will shoot between ~2.4-2.8moa….still making it an anomaly.

I also wonder what shooting 20 at 0.030” shorter than #7 yields….does it keep shrinking the group? Or does it bottom out somewhere and start getting bigger?

 

[Shadow9mm]

Thank you for your effort here. This is a ton of effort. Good stuff!

Can you share your oal’s 1-7? Also, how was the charge weight arrived at? Is it relatively high or low compared to published data?

One of their key points was the seating depth changes don’t make a huge difference unless drastic like 0.040” changes. If you see, other than group #3, your data basically trends from 4moa @ #1 to 3.7moa @ #7. I wonder what happens if you reshoot #3 what happens. Statistics would say you will shoot between ~2.4-2.8moa….still making it an anomaly.

I also wonder what shooting 20 at 0.030” shorter than #7 yields….does it keep shrinking the group? Or does it bottom out somewhere and start getting bigger?

According to Erik Cortina, the barrel harmonics act kind of like a sine wave, kind of an ebb an flow. So in theory the groups would get smaller to a point, then start getting larger again, to a point, and keep on repeating. He stated he believes a barrel has several good nodes.

Here is the video where he kinda explains. https://www.youtube.com/watch?v=oRXlCG9YZbQ

Charge weight was arrived at in previous testing. I have worked that load up in 3 other barrels and it produced around 2900fps, give or take, and no pressure signs in all barrels at that charge weight in my testing. Still saving up for a new chrono, so no velocity data on this particular barrel yet.

Data was taken from Hornady's 11th. For 223 max is 27.4, for 5.56 nato max is 27.5. So at an even 27.0g I am about 0.4 to 0.5 below max as my barrels are all rated for 223/5.56. Seeing as the bullets are all seated a little long, increasing case volume, I would expect that would reduce pressure a little bit too.

These COL are a little longer than recommended. I believe the manual calls for 2.200. However the mouth lined up inside the cannelure for all loads. I used a hornady bullet comparitor for measurements, no a direct COL, but I will convert it for you.

1, 2.237
2, 2.234
3, 2.231
4, 2.228
5, 2.225
6, 2.222
7, 2.219

 

[akinswi]

I believe too that the first thing he said was you needed to have good SD/ES then adjust seating depth

Curious what is your SD/ES is. It should be pretty good if your .4 grains under max

 

[Shadow9mm]

I believe too that the first thing he said was you needed to have good SD/ES then adjust seating depth

Curious what is your SD/ES is. It should be pretty good if your .4 grains under max

So in disclaimer, I am in the process of changing the way I collect and log my data as well as getting it all in one place. The goal is to get better data, and sore it is a way that is easier to use.

So here is the data and testing I did with my old barrel a couple years ago. I don't know how many rounds went over the chrono per group, I would wager 5 but I can't say for sure. Also I am not sure of the distance the chrono was placed at, I would guess approximately 12ft. I have also changed some of my loading equipment and loading techniques since then to collect better more reliable data.

But here's what I got then.

4-13-2021, 61F
Hornady 55g fmj
COL 2.215
LC brass
CFE 223 powder
CCI 450 Small magnum rifle primers
I am unsure whether these were weighed or thrown.

26.6, avg 2886, SD 15.44, ES 33
26.8, avg 2898, SD 20.36, ES 48
27.0, avg 2915, SD 16.54, ES 41
27.2, avg 2919, SD 32.30, ES 87
27.4, avg 2974, SD 24.69, ES 65

So nothing to write home about, but I had worse results using CCI 400 primers if memory serves me correctly. I would like to re-test these results using my new loading and testing methods once I get a new chronograph.

 

[NWPilgrim]

That is good data. FMJ is about the least accurate style of bullet so not likely to get small groups with that. Hornady 55 SP can be had for about $11/100 and has been surprisingly accurate for my tests. I just seat to mid-cannelure. My favorite load is the max charge of CFE at 27.5 gr. It shoots well in five of my rifles of different lengths. Not equally well in all, a couple are excellent and the others are not far behind.

I can’t get CFE to group well in the heavier bullet weights as I expected it would given how slow it is. But that 55 gr SP is a stand out load with it. I get better than average groups with CFE with 62-65 gr bullets, not not the best amount several powders.

 

[Unclenick]

I'm wondering if experimenting with primer seating might narrow the SD any. The manufacturers and Naval Ordnance claim about 0.003" compression is best for most primers. I've seen one YouTube video where Wolf primers were tested this way but found to be best with no final compression (just the anvil kissing the bottom of the primer pocket), leading me to wonder if the Russians preset the anvil compression when they make the primer rather than having it happen during seating.

"There is some debate about how deeply primers should be seated. I don’t pretend to have all the answers about this, but I have experimented with seating primers to different depths and seeing what happens on the chronograph and target paper, and so far I’ve obtained my best results seating them hard, pushing them in past the point where the anvil can be felt hitting the bottom of the pocket. Doing this, I can almost always get velocity standard deviations of less than 10 feet per second, even with magnum cartridges and long-bodied standards on the ’06 case, and I haven’t been able to accomplish that seating primers to lesser depths."

Dan Hackett
Precision Shooting Reloading Guide, Precision Shooting Inc., Pub. (R.I.P.), Manchester, CT, 1995, p. 271.​

 

[akinswi]

Also wonder if inconsistent primer holes would make a difference aswell. I was really surprised at his ES/SD numbers

 

[Unclenick]

Well, as a percent of the final velocity, they aren't all that unusual, except the 27.2-grain load seems high for a well-filled case (he should be at about 97-98% fill with an LC case, and that could be due to just one shot being a random outlier. Getting below 10 would be satisfying for him, though.

One other thought is the LC cases can have some pretty substantial burrs, and their flash hole drills have often proved to have walked a bit to the side. I've had the experience with spherical propellants before that irregular flash hole burrs upset the ignition more than with stick powders, so I would consider deburring the flash holes in those same cases and repeating to see if that brings SD down any.