Your Groups Are Too Small

[willfrye027]

What Harvey said. My conclusion from your data would be that 4831 is a better powder than 4350 for that bullet.

Simple load development=change powder or bullets to achieve desired group size. That much can be easily demonstrated as you have shown with some high level data.

I don’t believe that powder charge or seating depth affect meaningful differences, id love to see someone prove otherwise

 

[Longleaf]

I now load till I hit pressure and back off. Swapped scopes yesterday on my 6creed and shot a sub half minute group at 200. Ill try a 10 shot next time.

Carbon6/Bergara 108gr@2970

 

[Harvey_NW]

I have done load development and validation that improves these guns by 50-100% with a larger number of shots.

I don't think anyone responding here would argue that you're able to do that, but rather question your process of getting there. IMO, you shot 80-100 shots to get a conclusion that could have been identified in 10, and then said if that doesn't work start over with another combo. That amount of shots could have encompassed 16+ different small sample tests of combinations of components, any of which might end up shooting better than your result from the original combo.

In the video you talk about shooting a 5x5 sample, so for example something like an OCW test, and identifying a trend often referred to as a "node", and starting there because the trend showed up. Then you up your sample size because you're making smaller changes and the results are harder to identify. My issue is that you're still comparing those results to a small sample size, one of the original 5x5 groups. According to ballisticians and establishing the cone of fire, if you shot 50 of each of those combinations you would get the same group ES, or the lower charge might produce less overall dispersion. I would assume mean radius would be fairly linear as well with ES.

As for muzzle exit timing, I undoubtedly believe there is a vibrational frequency throughout the barrel with each fired shot. I think being able to "tune" it with a realistic ES of 30+ fps muzzle velocity is a complete fallacy. And everyone I've interacted with that believes those theories or methods, will absolutely refuse to shoot and provide a statistically valid sample size.

 

[BBob]

I don’t believe that powder charge or seating depth affect meaningful differences, id love to see someone prove otherwise

As a blanket statement this would be false but under the context of field or hunting rifles may be true. The measure of “meaningful“ might be greatly different depending on what discipline we’re talking about.

 

[solarshooter]

What Harvey said. My conclusion from your data would be that 4831 is a better powder than 4350 for that bullet.

Simple load development=change powder or bullets to achieve desired group size. That much can be easily demonstrated as you have shown with some high level data.

I don’t believe that powder charge or seating depth affect meaningful differences, id love to see someone prove otherwise

That I agree with. If we're questioning the validity of the powder sweep that's valid too, however I'm only looking for large deltas, which small samples are still useful for. Sometimes you see a large delta with powder, sometimes you don't. But let me ask this, WHY is one powder "better" than another? I would posit that you are making a large change to bullet time, and may get closer to a "node" of your barrel by shifting powder.

IMO, you shot 80-100 shots to get a conclusion that could have been identified in 10

But I don't follow this, I don't think I can tell a 20% difference in 10 shots. Are you saying I should still test but take fewer shots?

My issue is that you're still comparing those results to a small sample size, one of the original 5x5 groups.

Once I begin finer tuning, I am comparing higher sample sizes to each other. If those happen to match one of the first tested combos, I'll include those shots in the running tally of mean and sd radius. But for smaller changes you need more samples to tell them apart, I don't think I'm contradicting that in my method.

I think being able to "tune" it with a realistic ES of 30+ fps muzzle velocity is a complete fallacy.

The variable you are changing is bullet time, not MV. 30fps is ~1% of total MV, so the bullet time of ~1.5milliseconds may vary by the same proportion, call it 15 MICROseconds, 15e-6s. For kicks, I calculated the modes of a 30" 1.25" straight contour steel barrel:

The first mode is 2Hz, aka an oscillation every 0.5s. The fourth mode is 70Hz, aka 14 milliseconds. You can see that changes on the order of grains of powder (or type of powder) and 10s of % of MV/bullet time will have an effect on bullet time on the same order as the barrel modes, this is what I'd call tuning. MV scatter on the order of 1%, which is really +/-0.5%, in the extreme, wouldn't necessarily compromise this tune.

 

[Harvey_NW]

That I agree with. If we're questioning the validity of the powder sweep that's valid too, however I'm only looking for large deltas, which small samples are still useful for. Sometimes you see a large delta with powder, sometimes you don't.

But there's 40-50% variability from the average with 5 shot groups. So a small delta could be indicative of a stable load, but a large delta where the high end is above the acceptable standard will likely never be adjusted or "tuned" to consistently shoot acceptable, aside from possibly dropping the powder charge and sacrificing velocity for less dispersion. If the delta spectrum was within your acceptable standards (say a 5x5 test with groups from ranging from .3"-1"), I would just consider that random distribution.

But let me ask this, WHY is one powder "better" than another? I would posit that you are making a large change to bullet time, and may get closer to a "node" of your barrel by shifting powder.

Magic, lol. Interesting theory, and I'm excited to see what Ammolytics can accomplish with their super fancy high speed cameras on the matter.

But I don't follow this, I don't think I can tell a 20% difference in 10 shots. Are you saying I should still test but take fewer shots?

I'm not sure where you quantified a 20% improvement to a load of the same components by making changes, or identifying a node. I'm saying if it's worth it to you to shoot 80+ shots to make a 20% improvement and you can quantify it, continue mission. But the painless method will use substantially less components to arrive at, or very near, the same result.

Once I begin finer tuning, I am comparing higher sample sizes to each other. If those happen to match one of the first tested combos, I'll include those shots in the running tally of mean and sd radius. But for smaller changes you need more samples to tell them apart, I don't think I'm contradicting that in my method.

Do you have any examples of a load you developed where the sufficient sample size produced less dispersion than the biggest group in the initial small sample testing? Like one of the 5x5 shot groups was 1.2", but with the exact same components you were able to develop a load and shoot a 20 shot sample with an ES of .9" by changing the powder charge?

The variable you are changing is bullet time, not MV. 30fps is ~1% of total MV, so the bullet time of ~1.5milliseconds may vary by the same proportion, call it 15 MICROseconds, 15e-6s. For kicks, I calculated the modes of a 30" 1.25" straight contour steel barrel:
View attachment 659220
The first mode is 2Hz, aka an oscillation every 0.5s. The fourth mode is 70Hz, aka 14 milliseconds. You can see that changes on the order of grains of powder (or type of powder) and 10s of % of MV/bullet time will have an effect on bullet time on the same order as the barrel modes, this is what I'd call tuning. MV scatter on the order of 1%, which is really +/-0.5%, in the extreme, wouldn't necessarily compromise this tune.

I appreciate the data driven approach, and while I understand the concept those formulas are above my level of understanding. I would agree it might be a plausible theory for the differences in powder burn and pressure dynamics, however I still think being able to "tune" timing with a single combination of components is a fallacy. There are a lot of people looking for statistically valid proof on target that it exists.

 

[huntnful]

Using mean and SD radius is known and common only in professional munitions specification and testing, like government rifle/ammo specifications or artillery/ballistic missiles/you name it. Applying it to small arms load dev doesn't seem common at all. I haven't seen much of it at least.

Oh gotcha. I wasn't trying to undermine your new usage of that stuff in your work. I was mostly just referring to the fact that groups vary a lot, and you need to shoot a larger group to truly know what's going on. And micro tuning a load every 3 shots is basically useless lol.

Sometimes I'll just load 10 rounds of a given combo and go shoot it. I pretty much know, that there isn't much I can do those specific series of components to significantly improve the group. But normally it wouldn't be necessary anyways, because the groups are well within acceptable for a 1000 yard hunting rifle.


OR I load 7 rounds of whatever combo I want to try. 2 foulers and 5 for a group at 300 yards. Normally 2 or 3 different bullets that I'm comfortable hunting with and some known powders and charge weights. They normally all shoot good, so it doesn't matter much. But then I load 20 of whatever shot the best and verify group size and velocity with 10, and then drops at 1000 with the last of them. Very very simple and very effective.

With a quality barrel and setup, I've almost never had 10 shots go over 1 MOA with any bullet/powder combo seated near the lands.

This was a brand new test. New powder, guessed at a decent charge weight, and seated somewhere near the lands (I don't even precisely check where they are). This is pretty common with quality reloading and rifle components for sure.

No reason to do anything else to this other than load up a bunch and start sending them

 

[solarshooter]

however I still think being able to "tune" timing with a single combination of components is a fallacy

This is a fair point. It's possible that the only way to achieve significant differences in bullet time are by changing components, and this is the best knob to tune a load. However, short of resolving issues with engraving or bullet imbalance, I still maintain the mechanism by which this works is by moving bullet time around relative to barrel dynamic motion.

Looking at my data only for my current combo of components, here is what I did to arrive at this load, all components and dimensions held constant.

First sweep was a powder sweep of 3 shots each in 1gr increments from 50.5 to 57.5 (24rds), groups were:

• 50.5, 0.90"
• 51.5, 0.35
• 52.5, 0.40"
• 53.5, 0.90"
• 54.5, 1.00"
• 55.5, 0.50"
• 56.5, 1.00", faint ejector marks
• 57.5, 0.70", strong ejector marks

My interpretation of this was a pattern of larger and smaller groups with nodes at or around 52.0 and 55.5. Also large differences ~100% between groups. So my next sweep was sets of 5 centered on these charge weights (30rds):

• 51.5, 0.65"
• 52.0, 0.50"
• 52.5, 0.40"
• 55.0, 0.60"
• 55.5, 0.65"
• 56.0, 0.50"

So, charges tested twice did increase in size, as one would expect going from 3 to 5 shot groups. However the accuracy was still acceptable, they didn't grow 100%, and 51.5, 52.5, and 55.5 had held up to 8 shots of testing and still showed small groups. 51.5-52.5 would likely be the most accurate, but I picked 55.5 for significantly higher MV and to stay away from pressure and poor accuracy above 56.0.

Since then I've used 55.5 and now have 51 shots banked and informing the stats I've quoted repeatedly here. Mean radius of 0.27 corresponds to a mean 3 shot group of 0.54", which I think also agrees with what this initial testing showed.

 

[Formidilosus]

This is a fair point. It's possible that the only way to achieve significant differences in bullet time are by changing components, and this is the best knob to tune a load. However, short of resolving issues with engraving or bullet imbalance, I still maintain the mechanism by which this works is by moving bullet time around relative to barrel dynamic motion.

Looking at my data only for my current combo of components, here is what I did to arrive at this load, all components and dimensions held constant.

First sweep was a powder sweep of 3 shots each in 1gr increments from 50.5 to 57.5 (24rds), groups were:

• 50.5, 0.90"
• 51.5, 0.35
• 52.5, 0.40"
• 53.5, 0.90"
• 54.5, 1.00"
• 55.5, 0.50"
• 56.5, 1.00", faint ejector marks
• 57.5, 0.70", strong ejector marks

My interpretation of this was a pattern of larger and smaller groups with nodes at or around 52.0 and 55.5. Also large differences ~100% between groups. So my next sweep was sets of 5 centered on these charge weights (30rds):

• 51.5, 0.65"
• 52.0, 0.50"
• 52.5, 0.40"
• 55.0, 0.60"
• 55.5, 0.65"
• 56.0, 0.50"

So, charges tested twice did increase in size, as one would expect going from 3 to 5 shot groups. However the accuracy was still acceptable, they didn't grow 100%, and 51.5, 52.5, and 55.5 had held up to 8 shots of testing and still showed small groups. 51.5-52.5 would likely be the most accurate, but I picked 55.5 for significantly higher MV and to stay away from pressure and poor accuracy above 56.0.

Since then I've used 55.5 and now have 51 shots banked and informing the stats I've quoted repeatedly here. Mean radius of 0.27 corresponds to a mean 3 shot group of 0.54", which I think also agrees with what this initial testing showed.

So you literally loaded until slight pressure signs, then dropped 1.0 grains and that’s your best “accuracy”? Hmmmm almsot low yo could do that without a “ladder”. And if you weren’t going to take low velocity, why start at 50gr?

 

[Harvey_NW]

Looking at my data only for my current combo of components, here is what I did to arrive at this load, all components and dimensions held constant.

First sweep was a powder sweep of 3 shots each in 1gr increments from 50.5 to 57.5 (24rds), groups were:

• 50.5, 0.90"
• 51.5, 0.35
• 52.5, 0.40"
• 53.5, 0.90"
• 54.5, 1.00"
• 55.5, 0.50"
• 56.5, 1.00", faint ejector marks
• 57.5, 0.70", strong ejector marks

My interpretation of this was a pattern of larger and smaller groups with nodes at or around 52.0 and 55.5. Also large differences ~100% between groups. So my next sweep was sets of 5 centered on these charge weights (30rds):

• 51.5, 0.65"
• 52.0, 0.50"
• 52.5, 0.40"
• 55.0, 0.60"
• 55.5, 0.65"
• 56.0, 0.50"

So, charges tested twice did increase in size, as one would expect going from 3 to 5 shot groups. However the accuracy was still acceptable, they didn't grow 100%, and 51.5, 52.5, and 55.5 had held up to 8 shots of testing and still showed small groups. 51.5-52.5 would likely be the most accurate, but I picked 55.5 for significantly higher MV and to stay away from pressure and poor accuracy above 56.0.

Since then I've used 55.5 and now have 51 shots banked and informing the stats I've quoted repeatedly here. Mean radius of 0.27 corresponds to a mean 3 shot group of 0.54", which I think also agrees with what this initial testing showed.

Does this not exemplify the statistics in the podcast, and how the painless load development would get you to the same result with potentially 13 shots though? Both sets of incremental tests produced group sizes well within the statistical variability of a single charge, the second set the variability shrunk, as expected.

So I'll use your data and results to show how I do the painless method with a new rifle and components (very similar example posted here https://rokslide.com/forums/threads/painless-load-development-mine.238400/post-3283387). Assuming book max is in the 56-57gr range, I would load a 3 shot pressure ladder from 55-57gr in 1gr increments. Slight or no ejector mark at 56, ejector mark at 57. Drop .5gr from first indication to be safe, load and shoot 10 to confirm. We both end up at 55.5gr.

 

[solarshooter]

So you literally loaded until slight pressure signs, then dropped 1.0 grains and that’s your best “accuracy”? Hmmmm almsot low yo could do that without a “ladder”. And if you weren’t going to take low velocity, why start at 50gr?

But if I had gone 2gr below max it wouldn't have been good. Seems like luck more than a rule, though it would have worked in this case. I think a ladder/pressure test is a more foolproof way to get the "right" answer. For instance, I would surmise based on my barrel mode theory that for a different length barrel, the optimal charge would be different. I started low because I wanted to bracket the possible load range and see if there was a pattern. For the sake of science, I might go repeat this ladder and see what happens...

 

[Formidilosus]

But if I had gone 2gr below max it wouldn't have been good. Seems like luck more than a rule, though it would have worked in this case. I think a ladder/pressure test is a more foolproof way to get the "right" answer.

Bet you $100 that if you shot a 30 round group of at 54.5, 55.0, 55.5gr- there is functional difference, and almost certainly they will be within statistical variability- you have “engineered” yourself into believing that you can control small variables to effect a real outcome.

For instance, I would surmise based on my barrel mode theory that for a different length barrel, the optimal charge would be different.

Dude…. All of this to end up at a 1.2 MOA system?

 

[Harvey_NW]

But if I had gone 2gr below max it wouldn't have been good.

You can't make that claim without shooting the shots to prove it. Hence, "your groups are too small".

 

[willfrye027]

Oh gotcha. I wasn't trying to undermine your new usage of that stuff in your work. I was mostly just referring to the fact that groups vary a lot, and you need to shoot a larger group to truly know what's going on. And micro tuning a load every 3 shots is basically useless lol.

Sometimes I'll just load 10 rounds of a given combo and go shoot it. I pretty much know, that there isn't much I can do those specific series of components to significantly improve the group. But normally it wouldn't be necessary anyways, because the groups are well within acceptable for a 1000 yard hunting rifle.


OR I load 7 rounds of whatever combo I want to try. 2 foulers and 5 for a group at 300 yards. Normally 2 or 3 different bullets that I'm comfortable hunting with and some known powders and charge weights. They normally all shoot good, so it doesn't matter much. But then I load 20 of whatever shot the best and verify group size and velocity with 10, and then drops at 1000 with the last of them. Very very simple and very effective.

With a quality barrel and setup, I've almost never had 10 shots go over 1 MOA with any bullet/powder combo seated near the lands.

This was a brand new test. New powder, guessed at a decent charge weight, and seated somewhere near the lands (I don't even precisely check where they are). This is pretty common with quality reloading and rifle components for sure.

No reason to do anything else to this other than load up a bunch and start sending them
View attachment 659330

This is pretty much all I do now. I think what the engineer is saying about mean radius is that it will allow you to come to a conclusion about a difference between two combos, in fewer shots, compared to extreme spread of group size.

 

[solarshooter]

Bet you $100 that if you shot a 30 round group of at 54.5, 55.0, 55.5gr- there is functional difference

You can't make that claim without shooting the shots to prove it. Hence, "your groups are too small".

I'm assuming you meant NO functional difference. True. I guess I would need to do that to prove that the selection method is valid or not. Like I said earlier, I have already proven this for a different powder combo, but it's not certain for this powder. I could pick the worst charge from the initial ladder and shoot 20-30 to see if it remains worse than 55.5. Disprove the null hypothesis as it were. I will try and do this in the next week or two and report back. But I will be asking for that 100$ if I'm right...

 

[huntnful]

This is pretty much all I do now. I think what the engineer is saying about mean radius is that it will allow you to come to a conclusion about a difference between two combos, in fewer shots, compared to extreme spread of group size.

I agree for sure

 

[Formidilosus]

This is pretty much all I do now. I think what the engineer is saying about mean radius is that it will allow you to come to a conclusion about a difference between two combos, in fewer shots, compared to extreme spread of group size.

Sort of- what MR won’t do is tell you what size target the system will hit. Or, it won’t do it without charts, and Excel sheets, math, and general time wasters. Using mean radius (as well as large shot group ES) makes sense from a large scale testing and contract process. It does not make sense in the “I need to see if this combo will hit what I need to hit, and zero correctly” process.

Hitting living things isn’t about “average distance from center”, it’s about “worst” shot from center.

 

[solarshooter]

Dude…. All of this to end up at a 1.2 MOA system?

To be clear, all the shooting I did to inform the video was with a 22, not a big deal. Most of the rest of what I'm talking about is in interpreting the data. As far as what I've "wasted" on the 284, I think the main point of contention is the 25rds or so for the powder sweep? We agree that testing different components is worthwhile. Your method consumes 20rds to do this, if I skip the powder sweep and just find pressure and back off a grain (10rds) then proceed to validation (as much as you want, the more the better), I think my method essentially collapses to yours. Remember, the first sweep of testing I described was all different component combos.

Regarding 1.2MOA, lets be clear about what metric that relates to again - 20 shots. I think that's actually pretty good, and I would love to see other people post their 20 shot groups so I could know where I stand. Also, it's 1.5MOA informed by 51 shots. Do you have a 51 shot group for any of your guns? You know what your 20-30 shot group ES is (roughly 2 sigma level), but without insight on the distribution of shots inside that group, how do you know what your true worst case over the life of the rifle will be?

 

[solarshooter]

Sort of- what MR won’t do is tell you what size target the system will hit. Or, it won’t do it without charts, and Excel sheets, math, and general time wasters. Using mean radius (as well as large shot group ES) makes sense from a large scale testing and contract process. It does not make sense in the “I need to see if this combo will hit what I need to hit, and zero correctly” process.

Hitting living things isn’t about “average distance from center”, it’s about “worst” shot from center.

What inputs for rifle precision does WEZ use?

Also for the worst shot from center, I would take mean + 3 sigma (or 4 or 5) radius. It's not a crazy amount of math, just some basic excel. The way I see it, I spend a couple hours loading, a couple hours driving and shooting, and I can spare 10-15 mins to input data on a spreadsheet when I'm done.

 

[Harvey_NW]

True. I guess I would need to do that to prove that the selection method is valid or not. Like I said earlier, I have already proven this for a different powder combo, but it's not certain for this powder. I could pick the worst charge from the initial ladder and shoot 20-30 to see if it remains worse than 55.5. Disprove the null hypothesis as it were. I will try and do this in the next week or two and report back.

That would be awesome, I look forward to seeing your results.

Do you have a 51 shot group for any of your guns? You know what your 20-30 shot group ES is (roughly 2 sigma level), but without insight on the distribution of shots inside that group, how do you know what your true worst case over the life of the rifle will be?

FYI you're opening a can of worms, I see you're a new member and it might benefit you to search past posts on the forum.