Garmin Xero and Primer Sorting

[Rocketvapor]

Primer weights and shot order.
If I don't get the fired cases mixed up I'll weigh the fired primers

Round Robin instead of 5 round groups is more complicated but maybe less likely for data drift.
Maybe I'll number the cases tonight.
SHOT ORDER, 4 foulers, 10 shots, 15 shots, 15 shots.


I'll shoot the four foulers, adj scope if necessary, then freeze scope adjustments.

 

[Rocketvapor]

Finalized shot order, post above, #41.
Now to get some rest before heading out for the range.

 

[Rocketvapor]

For those that think in grams, CCI450
Worst, one each max high/min low
3.775 grains, 0.2446 grams
3.610 grains, 0.2339 grams
0.165 grains, 0.0107 grams
4.6% range

A few very high/very low
3.765 grains, 0.2440 grams
3.620 grains, 0.2346 grams
0.145 grains, 0.0094 grams
4% range

Many within range (3.66gr +/- 0.04gr, or +/- 2.7mg)
3.700 grains, 0.2400 grams
3.620 grains, 0.2346 grams
0.080 grains, 0.0054 grams
2.2% range

Tight sort (3.66gr +/- 0.02gr, or +/- 1.3mg)
3.680 grains, 0.2386 grams
3.640 grains, 0.2359 grams
0.040 grains, 0.0026 grams
1.1% range
Well within the capability of a milligram scale.

 

[Rocketvapor]

Garmin velocities added to post #41.
Wife helped keep track of case numbers.
Will deprime and weigh spent primers, maybe tonight.

I'm gonna guess that velocity variation is mostly primer mix and NOT Cup/Anvil.

 

[bloc]

Wife and I hit the range in Bogalusa today for some trigger time before the next 600yd match.
She practiced with her Savage 22N bolt gun and I used my 22N AR-15.
Did a small primer sorting test with the AR using velocities from my Garmin.
Scale used for reloading was an A&D EJ54D2 on the low range.

Average primer weight for a hundred primer flat was 3.65 grains.
Loaded up 20 rounds with weight sorted twice fired/annealed brass (22 Nosler/Dogtown), weight and length sorted Berger 85.5 bullets, 31.70 grs of SB6.5, and primers from 3.645 to 3.685 grains (3.65 +/- 0.02 grains)
Found a few light ones at 3.62 to 3.63 grains, and some heavy at 3.73 to 3.75 grains.
Loaded up 5 light and 5 heavy, same brass, bullets, load.

Outlier cases were marked Red and Blue around the primer and shot round robin, one light, one heavy, one light, etc.
Shot a few foulers then Shot 10, then the the outliers, then another ten.

Garmin Xero velocities:
First 10 round middle weight primers, 2993.8 avg.
Light primer avg (5 shots) 2990.3
Heavy primer avg (5 shots) 3018.5
Last 10 round middle weight primers, 2998.7 avg.

The light/ heavy individual velocities:
Light 2989.3
*Heavy 3039.3
*Light 2975.2
Heavy 3007.7
Light 2996.5
Heavy 3018.5
Light 2993.2
Heavy 3011.4
Light 2997.4
Heavy 3015.7
Double checked post for errors. Think I got it right.

AVG for all 30 shots was 2999 including that fast heavy primer 3039.3 and the slow light primer 2975.2.
*Eliminating those two odd ones the 28 shot avg was 2998.4.

Here ya go. Ran those numbers out for you.

Light	Heavy
2989.3​	3039.9​		Anova: Single Factor
2975.2​	3007.7​
2996.5​	3018.5​		SUMMARY
2993.2​	3011.4​		Groups	Count	Sum	Average	Variance
2997.4​	3015.7​		Column 1	5​	14951.6​	2990.32​	81.567​
			Column 2	5​	15093.2​	3018.64​	158.188​
			ANOVA
			Source of Variation	SS	df	MS	F	P-value	F crit
			Between Groups	2005.056​	1​	2005.056​	16.72587​	0.003487​	5.317655​
			Within Groups	959.02​	8​	119.8775​
			Total	2964.076​	9​

A P-value of <0.05 means the two groups of primers are slightly different. In your data, of which there really isn't enough to hang you hat on yet, the P value was 0.003487 so for what little there was we would say those two groups were different statistically.

If you can muster a set of about 30 of the heavy and the light, then you would know for sure, but as it is there is a pretty good chance they will be about a 2" difference at 600 yards on average. Some folks don't care about 2" at 600, some do. YMMV

Question: you were testing 2 populations (heavy and light). Why did you use a one-way ANOVA? I seem to recall that that test is generally used for more than two populations, and a t-test was generally used for testing 2 groups.

I ran the same numbers using a 2 sample t-test, and the results are quite different and show no statistical differences:
t = 0.853779
df = 9
p-value (one-tailed) = 0.207686
p-value (two-tailed) = 0.415373

I used this: https://www.statology.org/two-sample-t-test-calculator/

Is a 2 sample t-test inappropriate for this test? Am I wrong? It's been decades since I did any statistical analysis.

 

[RegionRat]

Is a 2 sample t-test inappropriate for this test? Am I wrong? It's been decades since I did any statistical analysis.

No, it isn't inappropriate and when used in this context it wouldn't change the conclusion.
The common T-test assumes the variances are equal. If they are not, we use the Welch's.

Not sure what you punched into the windows, but it should have looked like this:

If that is what you punch into the web tool you linked, you get the values below.

The delta between the averages is more than 2 sigma apart, so the hunch is the means will be different on visual inspection no matter which method we apply, but since we only have 5 of samples of each...
The variance of the heavy group was nearly double the lighter ones, but let's just play it out and see if it makes any difference with the calculators you showed.
The two samples have averages as follows: 2990.32 AVG SD 9, versus 3018.64 AVG SD 12.6

If we run your T-test we would get:
t = -4.089728
df = 8
p-value (one-tailed) = 0.001743
p-value (two-tailed) = 0.003487

And if we run Welch's we would get
t = -4.089728
df = 7
p-value (one-tailed) = 0.002316
p-value (two-tailed) = 0.004633

Either way, when we only have 5 samples the conclusions are weak, but they still point to a difference between the two primer weight averages. All of these tests do better with more samples.

In both of these tools, the p-value is less than the Alpha of 0.05, so the two means are not the same and there is a likelihood the light primers are in fact slower than the heavy primers.

So, something went wrong with your input window, but it should be obvious on visual inspection that the two groups are not the same even without the details when you see the averages are spaced by more than two sigma. Your DF should have been N1+N2-2=8 so I think your input window may have an error?

 

[Rocketvapor]

Entering the heavy first and rounding I get
t = 4.147956
df = 8
p-value (one-tailed) = 0.001609
p-value (two-tailed) = 0.003218

When the data doesn't LOOK related, and calculation does,
the analysis might be the wrong type.
I tried making a graph of today's run, but EXCEL didn't want to cooperate.
Run the first ten of today's results.
Extreme high and low primer weights.
Probably a velocity difference between 5 light and 5 heavy.
Heavy Mean 3044.6, SD10.7
Light Mean 2987.2 SD 5.26
I agree with:
"there is a likelihood the light primers are in fact slower than the heavy primers"

 

[Rocketvapor]

Using the third group from today (not quite the extremes)
Heavy 3055.1, 3053.5, 3063.0, 3032.3, 3044.6
Mean = 3049.7 SD 10.48
Light 3010.1, 3010.3, 3013.2, 3032.3, 3003.1, 2999.9
Mean = 3011.5 SD 10.4

If you notice an increase in velocity from the first test to today's,
the difference is the first charge was 31.70gr SB6.5, today was 31.80gr of SB6.5.
The medium weight primers (3.66) were left overs from first test.
Same 85.5 bullets.
Sorry for the curve ball in loads.

The 10 Medium, 3.66 primers:
Mean 3003.4, SD 11.72
Doesn't look like I'll be getting SDs in the low single digits anytime soon with this cartridge and powder.

 

[RegionRat]

Doesn't look like I'll be getting SDs in the low single digits anytime soon with this cartridge and powder.

It is hard to get ball powders like SB6.5 to run with extruded powders in this respect.

Will you have to try something else or will you get away with it for 600 yards?

 

[Rocketvapor]

After the wife shot her first group and I shot mine we got this "Sensor Disconnected message on the shotmarker. Groups went right by about an MOA and opened up. ShotMarker error???
Anyway her first ten (31.8gr load)
with WIN 41 across the Garmin was:
Min 3031..3 Avg 3061.0 Max 3082.1 ES 50.8 SD 16.7
Next ten with CCI 450
Min 3002.3 Avg 3018.2 Max 3033.1 ES 30.9 SD 11.2
30.8 is obviously TOO FAST. I want close to or a little less than 3000 with the Bergers.

Might have to go back to IMR4350 we used in the past.
2850 to 2900.
But, wife doesn't like science experiments and is closing in on Master.
Changing right now will not get me brownie points

 

[Rocketvapor]

First ten shots, Extreme Heavy/Light.
Odd # shots were Heavy.
Don't know if I can trust the SM data for this group.
Got Sensor message during the group.

Sort of looks like group in post #5.

 

[bloc]

No, it isn't inappropriate and when used in this context it wouldn't change the conclusion.
The common T-test assumes the variances are equal. If they are not, we use the Welch's.

Not sure what you punched into the windows, but it should have looked like this:
View attachment 1538421
If that is what you punch into the web tool you linked, you get the values below.

The delta between the averages is more than 2 sigma apart, so the hunch is the means will be different on visual inspection no matter which method we apply, but since we only have 5 of samples of each...
The variance of the heavy group was nearly double the lighter ones, but let's just play it out and see if it makes any difference with the calculators you showed.
The two samples have averages as follows: 2990.32 AVG SD 9, versus 3018.64 AVG SD 12.6

If we run your T-test we would get:
t = -4.089728
df = 8
p-value (one-tailed) = 0.001743
p-value (two-tailed) = 0.003487

And if we run Welch's we would get
t = -4.089728
df = 7
p-value (one-tailed) = 0.002316
p-value (two-tailed) = 0.004633

Either way, when we only have 5 samples the conclusions are weak, but they still point to a difference between the two primer weight averages. All of these tests do better with more samples.

In both of these tools, the p-value is less than the Alpha of 0.05, so the two means are not the same and there is a likelihood the light primers are in fact slower than the heavy primers.

So, something went wrong with your input window, but it should be obvious on visual inspection that the two groups are not the same even without the details when you see the averages are spaced by more than two sigma. Your DF should have been N1+N2-2=8 so I think your input window may have an error?

Thanks, I don't know what I did wrong, but wrong it was. I recalculated and I was nowhere near accurate in the first iteration.

I appreciate the correction and the time you took to do it.

 

[JFrank]

First ten shots, Extreme Heavy/Light.
Odd # shots were Heavy.
Don't know if I can trust the SM data for this group.
Got Sensor message during the group.
View attachment 1538495
Sort of looks like group in post #5.

Isn’t this opposite of your earlier results on post #5 ?

 

[Rocketvapor]

Odd # shots in post #5 were LIGHT and LOW, First shot LIGHT.
Odd # shots in post #51 were Heavy and HIGH, First shot Heavy.


Suspect errors with post 51 ShotMarker data, position and velocity.

Gonna be the last test for a while. I'm all out of really heavy and really light CCI 450 primers

 

[Ruger15151]

Shot round robin, this is all the data you need. The velocities are what they are, and we see higher ES shoot smaller than lower ES even at 1,000 beyond half the time. Good job.

Tom

Tom

Do you even pay attention to ES or do you just pay attention to what the target tells you?

 

[Rocketvapor]

Any one of the outliers in a match would cost me a point.
Just looking for a cause of big groups (elevation).
Currently @ 600.
Pretty sure primer weight drives me outside an MOA.

Well, me, the rifle, and primer weight.

 

[Someone online]

After reading this thread, I just weighed all 1000 primers in a box of CCI 250. Weighed from .329 to .344 grams. Majority were within .334 to .339. anything weighing beyond have about 200 weighing .340 and up, and about 100 weighing .333 and down.
I kept them all in separate trays of 100. One tray of .333 minus, trays of .334-.335, .336-.337, .338 to .339.
The next time I load these up, I will load some light ones, mediums, and heavies, and see if they all average different velocities, at least 30 in each weight.
Will also need to sort the cases by volume in each set.
Then measure the powder charge down to the same hundredth grain.
This rabbit hole is going to make this loading session take all weekend.

 

[Rocketvapor]

@Someone online
0.329g to 0.344g that's a range of 4.56%
I got a range 3.775gr to 3.610gr of 4.57%


A little tidbit to keep you statisticians busy.
The first ten (extreme heavy/light) as fired primers. They came out all intact.
Didn't clean.
Reference post #41, weights below in grains.
Case #, primer weight, as fired, weight lost, velocity
1 -- 3.765, 3.400, 0.365, 3036.8
30 - 3.625, 3.375, 0.250, 2989.2
2 -- 3.760, 3.390, 0.370, 3042.7
29 - 3.625, 3.360, 0.265, 2989.0
3 -- 3.765, 3.385, 0.380, 3054.4
28 - 3.620, 3.360, 0.260, 2987.4
4 -- 3.775, 3.405, 0.370, 3058.4
27 - 3.615, 3.370, 0.245, 2993.1
5 -- 3.755, 3.395, 0.360, 3050.2
26 - 3.610, 3.360, 0.250, 2977.4

Weight range of primers: 0.165 grains
Weight range as fired: 0.045 grains
Weight range of weight loss: 0.135 grains
Velocity range, ES 81fps SD 31.3

FULL LIST IN POST #64

 

[JFrank]

Odd # shots in post #5 were LIGHT and LOW, First shot LIGHT.
Odd # shots in post #51 were Heavy and HIGH, First shot Heavy.
View attachment 1538516

Suspect errors with post 51 ShotMarker data, position and velocity.

Gonna be the last test for a while. I'm all out of really heavy and really light CCI 450 primers

Would you say that #51 the heavy shot a little better and #5 the light shot a little better ?
I don’t speak for Tom but in our Benchrest world the target is king, ES simply keeps the reloading practices in view.

 

[tom]

Tom

Do you even pay attention to ES or do you just pay attention to what the target tells you?

Oh I document the speeds every time, even if for just knowledge of about where I'm at when someone asks. But since I don't even shoot my target rifles anywhere but 1,000, I always trust the targets. I was just as competitive not using a crony, that never held me back, but they're easy to set up now and fit in a bag so I use them.

The OP posted another target (sort of a target lol) that pretty much sums up all he needed to know. No need for p max/t tax/climax or whatever these words mean. When point of impact can literally be seen as 2 different locations, and you're doing round Robin which eliminates conditions with exception of first and last shots....you don't need these giant sample sizes to know what is real.

Tom