How many rounds do you guys shoot to determine SD ?. I've been looking around the net and can't find much info on this. What I have seen is guys shooting 5 shot strings. Maybe you guys can tell me what you do to get a good SD reading.
Standard Deviation
- Thread starter axarob44
- Start date
In statistical process control circles, a 30 piece sample is considered the minimum to establish process capability. Standard deviation is one of the parameters used to calculate process capability.
Others will argue that 5 or 10 data samples is good enough. In many instances what passes as "accepted standard" in the gun world is not even close to correct when you look at it through a professional's eye.
Others will argue that 5 or 10 data samples is good enough. In many instances what passes as "accepted standard" in the gun world is not even close to correct when you look at it through a professional's eye.
I just got a new Labradar and my SD is not so great at 18 with 38 shots recorded. My cold bore shot was really low and I had a couple really high ones too. I was really disappointed to say the least ! Now I have to figure out how to improve loads.
Last edited:
This is true. Automotive standards allow us to use a 50 piece capability for ongoing validation, but initial SD must be established with 100 piece capability studies.
That having been said, I routinely use 10 shot samples while developing loads. Right or wrong, it gives me a good enough idea of what to expect out of a load/process.
Edit: After all, gathering data on your load is all about understanding your process control and the load's capability. One part is the components being used; barrel, chamber, brass, projectile, and powder (also primer to some degree). The other part is the repeatability of your process: brass preparation, powder measurement and loading process control.
Last edited:
I just got turned on to Ballistipedia. They go into great detail on using statistics to measure just how good that load is.
http://ballistipedia.com/index.php?title=Home
http://ballistipedia.com/index.php?title=Home
I'll chrono them all during load development for multiple reasons and keep all the data (high, low, average, S/D, accuracy). Then when I settle on a range and fine tune it Ill chrono those all as well until I decide on one.
At that point I'll have 15-20 of that specific load at least.
At that point I'll have 15-20 of that specific load at least.
I'd consider, 10 shots a start. With the Lab radar I can start shooting with the LR from the first shot of a new rifle/barrel. When shooting for fun I'm always running my LR for data and reasons why I miss a shot. I try to break in a rifle w/100 before I start figuring out a load. Then load development (50-100) rounds, high side. Every time I load up a new batch of 100 rounds I usually check the velocity(SD). If somethings up than I'll try to trouble shoot. I did have a barrel start to speed up after the 400 rd. ish point 25-40 Fps. I didn't look into it much as I sold the barrel soon after my shooting season ended. My guess was it was plateuing?? Between the LR and my Gempro 250 I average around the 10 Fps SD and haven't been able to get below on a regular basis. I'm always pushing my loads for max velocity, witch may be costing me??? Maybe an upgrade in the scale may be in order soon?
I should have clarified that acceptable sample size varies a bit within industries. Aerospace and other lower volume industries can't often afford a 100 pc sample to gather PPAP data. The fail-safe comes in the form of larger sample sizes and/or shorter sampling frequencies after PPAP and during maturation.
I have no interest in shooting a 25-50 shot string for data gathering purposes. Who cares whether the data is "statistically significant" or not? SD is not an end result, it's just a predictor of whether a load is likely to have minimal vertical dispersion at distance. So if a load is shooting well and you're not going to be tweaking it, why bother shooting all those rounds to gather data? And if you're in the load development stage, why would you want to burn 25 rounds on something just to find out if it's better than another load option? You'd shoot the barrel out just doing load development with 25 shot strings of each test load.
I shoot 5 shot strings for simplicity sake. If the SD is good enough, it's likely to be good enough over a large sample size. Who cares if my data refines with a larger sample and the SD of 3 was actually an SD of 6 or 7? All of those are "good enough". And if I shoot and get an SD of 15, I know that I'm not going to be happy with that load and I need to go back to the drawing board, don't need another 15 rounds to be sure.
Yes, a big string of shots is more accurate. But to me that seems to be treating SD as an end unto itself rather than a load development diagnostic tool.
Now if you've got a Lab Radar or free floated chrono and you happen to just gather all the data from your range trip where you would have shot those rounds anyway, that's fine and good info to have. I just wouldn't tell someone they need to go run a 25 shot string in order to do it "right".
I shoot 5 shot strings for simplicity sake. If the SD is good enough, it's likely to be good enough over a large sample size. Who cares if my data refines with a larger sample and the SD of 3 was actually an SD of 6 or 7? All of those are "good enough". And if I shoot and get an SD of 15, I know that I'm not going to be happy with that load and I need to go back to the drawing board, don't need another 15 rounds to be sure.
Yes, a big string of shots is more accurate. But to me that seems to be treating SD as an end unto itself rather than a load development diagnostic tool.
Now if you've got a Lab Radar or free floated chrono and you happen to just gather all the data from your range trip where you would have shot those rounds anyway, that's fine and good info to have. I just wouldn't tell someone they need to go run a 25 shot string in order to do it "right".
I’m with Sheldon N. I free float a Magnettospeed during development to gather data but I would never shoot 30 rounds just to get SD. If my load is shooting the way I want and has a good SD & ES through 10 rounds than I’m happy. That’s not to say that 30+ is wrong. It would certainly yield a more accurate SD, it’s just not something I feel is necessary for me.
Last edited:
I used to be in the 30 round camp , until I realised my brass prep
was shit , scales were shit etc etc . As my reloading has improved
over the years , so has my confidence in lower sample sizes to
gather data .
In addition , many other factors , particularly barrel condition ( round count )
and barrel fouling state are also very important . Even weather conditions
have a greal effect on this ; my Magnum cals foul worse in stinky humid
conditions than cold dry conditions for example . Gotta look at the whole
picture , not just one long string in one weather / barrel condition .
was shit , scales were shit etc etc . As my reloading has improved
over the years , so has my confidence in lower sample sizes to
gather data .
In addition , many other factors , particularly barrel condition ( round count )
and barrel fouling state are also very important . Even weather conditions
have a greal effect on this ; my Magnum cals foul worse in stinky humid
conditions than cold dry conditions for example . Gotta look at the whole
picture , not just one long string in one weather / barrel condition .
D
Deleted member 10043
Guest
Actually, if you generally shoot 5-10 shot relays than 5-10 shots on the Chrono is your sample frame and not a sample if you are carefully selecting cartridges
and you will be surprised that more than 10 will not materially change SD. The defining factor is the estimated relative precision. It may be 32% for 10 and 30% for 30. Statistically, both are acceptable. All of this is beyond the scope of what we need but only supports it. Nevertheless, the greater the number shot the better up to and including the whole population. So, it is just preference. Also, loader induced problems are going to create extraordinary shots that are going to skew the results. That pretty much means back to the chalk board.
and you will be surprised that more than 10 will not materially change SD. The defining factor is the estimated relative precision. It may be 32% for 10 and 30% for 30. Statistically, both are acceptable. All of this is beyond the scope of what we need but only supports it. Nevertheless, the greater the number shot the better up to and including the whole population. So, it is just preference. Also, loader induced problems are going to create extraordinary shots that are going to skew the results. That pretty much means back to the chalk board.
Last edited by a moderator:
Yeah, SD reduces in error the more samples (data) you include. Or rather becomes more accurate.
In statistics, the result is directly dependent on the amount of data analyzed. In short, you get better results.
Sometimes it's better to shoot multiple five round groups and then use that as your data.
D
Deleted member 10043
Guest
Of course, what do you think the confidence is for 10 out of 100? Like intimated by industry, 30 gets you around 90% but did the SD really materially change from 10? When we were able to say with 80% confidence did it really change when we reached 90%? To get right down to it
we are not really sampling if our population is less than 1000. More like just testing a load.
we are not really sampling if our population is less than 1000. More like just testing a load.
I was able to get accurate barrel velocity from 5 rounds when using a Magnetospeed. I verified velocity out to 1000 yds. Only shooting 5 rounds met my intent. Take your numbers and apply it to shooting at distance. Other posters are spot on. The more rounds shot under chronograph, the more accurate your muzzle velocity will be (if you can’t true your velocity at distance).
Similar threads
