throwing powder statistically

[taliv]

i'm using a harrell precision thrower. it's usually +/- .2 for stick powders which is super annoying, so i always have to trickle.

what i'm wondering is if i could throw more accurately by throwing multiple times. does anyone do that?

for example, if I need 40g, instead of setting it to 40 and throwing once, what if i set it at 10 and throw 4 times? would my variation tighten up?

 

[Olon]

I never thought about it but that would be a neat thing to experiment.

 

[denton]

An interesting question.... and the answer is, "It depends."

If your error is random, and not from shifts in the process, the answer is that 4 dumps of 10 grains will typically be more precise than one dump of 40 grains. The standard deviation comes down by the square root of the number of samples. The square root of 4 is 2, so your variation will come down by a factor of 2. There are industrial processes that use this trick.

However, it probably doesn't matter.

Random variation adds by the square root of the sum of the squares, and that doesn't behave like most people expect. Here's an example:

Suppose you are loading Varget with 150 grain bullets in a 308. There are a number of factors that induce variation: bullet weight, seating depth, neck tension, barrel temperature, etc. For purposes of illustration, suppose that you have a perfect powder scale, and are making powder charges down to 1/10 of a granule of powder, essentially perfect. Also suppose that you're a careful reloader, and that the standard deviation of your muzzle velocity is 20 FPS. That's better than most.

Now suppose that you switch from your 1/10 granule routine to using my Lee Perfect Powder Measure, which, with old style 4831 has a standard deviation of .1 grain. So 95% of all charges will fall between plus and minus .2 grains, pretty much as you describe.

With Varget, a grain of powder is 58 FPS. So an SD of .1 grain translates to 5.8 FPS.

Now combining your 20 FPS base SD with 5.8 FPS, we get:

square root (20^2 + 5.8^2) = 20.82 FPS.

So abandoning your 1/10 granule system, and using a powder measure about like the one you've got boosts the SD of your MV from 20 to 20.8 FPS.

So, as I said, it probably doesn't matter.

 

[troy fairweather]

Try wiping it with a used dryer sheet. The more throws the more off your charge will be.

 

[higgite]

what i'm wondering is if i could throw more accurately by throwing multiple times. does anyone do that?

I don't know anyone who does that, but you could try it yourself and not ask us to guess if it's going to work with your powder measure. Let us know what you find out.

 

[entropy]

This is why I use Lee Dippers; I choose the one next below the charge I want and scoop up another dipper full and 'shake' until the 505 says '0'. (on the line)

 

[taliv]

thanks for the math, Denton! that's just what i was trying to ask.

For purposes of illustration, suppose that you have a perfect powder scale, and are making powder charges down to 1/10 of a granule of powder, essentially perfect. Also suppose that you're a careful reloader, and that the standard deviation of your muzzle velocity is 20 FPS. That's better than most.

I have no doubt that 20 FPS is better than most, but for long range precision, much lower is required.
I try to keep SD in the low single digits. sometimes it creeps up to 10-12 fps. but an SD of 20 fps would be right out!
I do weigh to the kernel of powder and I do use varget in 6.5x47L and 6dasher. IME a kernel is about .01-.02 grains, but it's hard to tell since my acculab scale only resolves to .02 grains (i.e. it goes from .00 to .02 to .04, etc)

so the real question is, can i skip a time consuming weighing process by simply throwing 4 times. if my +/- .2 dropped to +/- .1 grain (the factor of 2), and in my dasher, for instance, .1g is about 9fps, which would be 18 fps at +/- .1g. so if i start with say, an SD of 6 fps, and add 18 to it, trying to use your math, I get an SD of 19, but I don't think that's right at all.

here's an example of two (not particularly good) 6dasher loads shot round robin with 31 and 31.2g varget. i've used excel to calc the mean, ES and SD for both groups individually. and if I also pretend it was one 31.10g group with +/- .10 grain variation, it gives me an ES of 36 and SD of 11. (though obviously, that's not how it would work as shots would be clustered more in the middle. Ie if it were one group there would be one bell curve where obviously I’m combining two curves here. So i know actual results would be better but I don’t know how to do the math)

Using those numbers, the SD of 11 gives me a .2 mil vertical at 1000 yards, which is 7.2" while the SD of 6 is a .1 mil vertical or 3.6" (for reference, F-class x-ring is 5" diameter) so I don't think I'd even want to use that to practice. Basically, I'd have to throw 16x 1.9g charges instead of dropping 4 times.

 

[taliv]

I don't know anyone who does that, but you could try it yourself and not ask us to guess if it's going to work with your powder measure. Let us know what you find out.

it's a math question, not a question of what works with my powder measure.

 

[denton]

thanks for the math, Denton! that's just what i was trying to ask.


I have no doubt that 20 FPS is better than most, but for long range precision, much lower is required.
I try to keep SD in the low single digits. sometimes it creeps up to 10-12 fps. but an SD of 20 fps would be right out!
I do weigh to the kernel of powder and I do use varget in 6.5x47L and 6dasher. IME a kernel is about .01-.02 grains, but it's hard to tell since my acculab scale only resolves to .02 grains (i.e. it goes from .00 to .02 to .04, etc)

so the real question is, can i skip a time consuming weighing process by simply throwing 4 times. if my +/- .2 dropped to +/- .1 grain (the factor of 2), and in my dasher, for instance, .1g is about 9fps, which would be 18 fps at +/- .1g. so if i start with say, an SD of 6 fps, and add 18 to it, trying to use your math, I get an SD of 19, but I don't think that's right at all.

here's an example of two (not particularly good) 6dasher loads shot round robin with 31 and 31.2g varget. i've used excel to calc the mean, ES and SD for both groups individually. and if I also pretend it was one 31.10g group with +/- .10 grain variation, it gives me an ES of 36 and SD of 11. (though obviously, that's not how it would work as shots would be clustered more in the middle. i think that's more of a worst case)

Using those numbers, the SD of 11 gives me a .2 mil vertical at 1000 yards, which is 7.2" while the SD of 6 is a .1 mil vertical or 3.6" (for reference, F-class x-ring is 5" diameter) so I don't think I'd even want to use that to practice. Basically, I'd have to throw 16x 1.9g charges instead of dropping 4 times.
View attachment 915816

Once again, the peculiar way that variation adds comes into play: variation in your technique, plus variation in MV.

The intuitive approach to vertical distortion of the group is exactly what you've done, and what most people assume.

It's late, but I'll fish up the article I wrote about that a few years ago tomorrow and post some numbers. IIRC, with a 1/2 MOA rifle and 1,000 yard targets, groups will be ovals that are not too badly out of round. It is unusual for all the relevant factors to align in one direction. Many small errors tend to partially cancel each other.

In some circles, a 1/4 MOA group is not exceptional. If you're in that category, then, yes, small MV variations become more important. OTOH, my little 223 routinely does 5/8" groups at 100, with an MV SD around 30 FPS. Controlling the MV SD down into single digits doesn't improve the groups detectably.

 

[taliv]

1/4moa is definitely not exceptional. But let’s focus on the math. Feel free to start another thread if you want to talk about effect of variation in velocity on group sizes.

thanks for your help. I just looked up my measure and I think it only goes down to 6 grains. So I could throw 5 times to get to 30g. Harrell also makes a Schutzen pistol thrower that goes down to 2g but I don’t know how it would do with stick powders. With 2g I could throw 15x to get to 30g. I’m not going to do that lol but In theory that would cut my spread to +/- .05g charge weight which would give me an extreme spread of around 20. Not good enough for matches but I’d take that for practice ammo.

 

[Walkalong]

Back when I started all I had was a bushing style pistol powder measure and I had to throw two or three charges to add up when doing rifle, or just weigh it all out like I had been doing. Overall it seemed to be less precise weight wise, I wasn't choronoing anything at the time either. I did continue doing it that way for .223 plinkers until I got a bigger measure. I had the little pistol measure set up to drop automatically on my old Projector.

 

[styles]

i'm using a harrell precision thrower. it's usually +/- .2 for stick powders which is super annoying, so i always have to trickle.

what i'm wondering is if i could throw more accurately by throwing multiple times. does anyone do that?

for example, if I need 40g, instead of setting it to 40 and throwing once, what if i set it at 10 and throw 4 times? would my variation tighten up?

4 isn’t large enough sample to statistically reduce the variance. I always explain to people a coin can come up heads 12 times in a row but not 1000. There would be many instances of -2,-2,-2,-2 tenths I’m afraid.
To statistically overcome the mechanical variance you have would take something like 30 Throws. To test that would be like 600.

 

[taliv]

it seems unlikely that the variance would be reduced in a single giant step at 30 throws. it seems far more likely that 4 throws cuts it in half, and 16 in half again and 64 in half again

sure the probability of -2 -2 -2 -2 is something like 2 throws per 1000
(assuming choices of -2 -1 0 +1 +2 there is a 1 in 5 chance of -2 per throw, so 20% x 20% x 20% x 20% = 0.16%)

 

[Nature Boy]

I ran the numbers on my Dillon thrower and discussed the results here

https://www.thehighroad.org/index.p...ariation-of-the-dillon-powder-thrower.849682/

 

[taliv]

cool, but i can't exactly throw 4 charges in a progressive into the same case lol

 

[Nature Boy]

cool, but i can't exactly throw 4 charges in a progressive into the same case lol

You could, but it would be tedious

 

[styles]

it seems unlikely that the variance would be reduced in a single giant step at 30 throws. it seems far more likely that 4 throws cuts it in half, and 16 in half again and 64 in half again

sure the probability of -2 -2 -2 -2 is something like 2 throws per 1000
(assuming choices of -2 -1 0 +1 +2 there is a 1 in 5 chance of -2 per throw, so 20% x 20% x 20% x 20% = 0.16%)

Agree and you are correct, but after 30 throws the margin of error S/B less than can be measured in 1/10th grain. I'd say more like 39 throws, but I was making a general observation about the impracticality of 4 throws.

 

[Olon]

After thinking on it, it seems you may be worse off making multiple throws. I could be wrong but here's my line of thinking:

Theres a specific combination of variables every time you throw a charge (as somebody else pointed out). Technique, orientation of the powder flakes, fill in the hopper etc. These variables will stack up differently each time, adding onto the inherent inaccuracy of your powder measure. So I would think that by throwing several times you'd be increasing the added uncertainty proportionally to how many throws you made.

That's only true if your powder measure is consistent in it's error throughout the charge range and if that error is proportional to the charge thrown. It might be less precise with small charges than large ones or vice versa in which case you'd have to go off of that.

I'm not a statistician or anything but it looks like the water is pretty muddy.

 

[crestoncowboy]

After thinking on it, it seems you may be worse off making multiple throws.

That's my guess as well. Especially with stick powder.

 

[taliv]

Agree and you are correct, but after 30 throws the margin of error S/B less than can be measured in 1/10th grain. I'd say more like 39 throws, but I was making a general observation about the impracticality of 4 throws.

oh i see. but still, 4 throws is twice as good as 1 throw

for me, it's mostly moot. i'm not going to do 16 throws, so i'm stuck weighing every charge. but for someone who is not going to spend time and money on 0.02g resolution scales and only has a thrower, doing 4 throws would be a big improvement over just 1 and would be noticeable at distance

 

[taliv]

i decided to try it, and started with a control group of 30 throws, target 31.00g varget. but when i went to try 7.75g, my thrower won't hold even a grain on it. it's all over the place at that weight.

 

[murf]

and it's all to the high side. seems your thrower is not linear and the variations are not cancelling out. bummer

luck,

murf

 

[taliv]

median is 31.04, mean is 31.06. means there's actually more samples below the mean than above, right?

if i turned the measure down one click, i'd prob wind up with same SD, but mean around 30.92.

 

[murf]

i was referring to your target, not the mean as to the thrower not being linear. meaning you can't just set it at one fourth the desired result and expect the result. you know that already and know how to compensate.

luck,

murf

 

[bangswitch]

i'm using a harrell precision thrower. it's usually +/- .2 for stick powders which is super annoying, so i always have to trickle.

what i'm wondering is if i could throw more accurately by throwing multiple times. does anyone do that?

for example, if I need 40g, instead of setting it to 40 and throwing once, what if i set it at 10 and throw 4 times? would my variation tighten up?

If you have a consistent probability of error of +0.2/-0.2 for a given powder throw, I'd say offhand that in 4 throws, each with the same probability, there is a probability for 4X error. In some cases, (2 throws at +0.2 and 2 throws at -0.2) they would cancel out, but if one is -0.2 and the other 3 are +0.2, you have an overall error of +0.6. And worse case, you could have +0.2 on each throw, for a combined error of +0.8. I'd estimate if you did what you are thinking of, your overall error would increase.

When I load, I choose a number for my desired load, within safe load guidelines. I adjust my powder throw until I get that number, and then I do 10-15 throws, and weigh each one to see how close each one comes to the desired number. If I get a consistent variance of no more than +0.1/-0.2, I accept that and start to load, with an actual recheck every 10th throw. This is for practice or SD ammo. If I'm loading precision loads, primarily for long distance rifle, each and every throw is measured, and then trickled to within the accuracy of the scale to the precise weight.