Chrono Sample Size?

[RobT]

Hi Guys

Just starting to use a chrono (Pact MK IV). Wondering what to use for a sample size working up data on a particular load. I know many people go with 5 round strings - but that seems to small given the variability of loads I've seen and measured so far. I could get out the old statistics texts - but only as a last resort.

Any thoughts? - thanks in advance.

RobT

[Loves2Shoot]

10 rounds minimum for me

[Tman33_99]

Like L2S said, I use a minimum of 10 rounds for new load.

Before a big match or every once in a while, I will run 5 or 6 rounds as a check against my prior data with that load. If it matches pretty close I am good to go.

[MasterLefty]

I run ten (10) rounds.

The average must between the tolerance I allow, nominal pf is 168pf, and no rounds can be below 165pf.

I also record the following data:

Primers: type and lot number

Powder: type and lot number

Seating Depth

Crimp

Bullet: type and lot number (weight is based on a sampling plan to ensure 99% confidence level of the weight)

Temperature

Humidity

Barometric Pressure

Kenny

[wide45]

20 is a full mag.

[TriggerT]

20 is a full mag.

10 is a full mag for we who shoot the single stacks.

[eric nielsen]

At least ten of each load. A full [18-20] mag of whatever is "match" ammo.

The thing that creates a spread in velocities [often] is mixed brass. 2nd place for a huge M.A.D. is not seating each primer with the same "oomph".

[GvU]

I usually measured 2x6, but lately I measure 4x3, as that is the way the chrono is performed.

From a clean gunbarrel (dry patch) the first 3 are lower, the next are a bit higher.

Last batch at my chrono had a low sd. The difference at the latest match (CNC2004) was -2 PF-points with older cases. A3435 had a difference of 0 PF between measurements at home and this match.

As long as de sd is low (at least 10-12 rounds) a small sample for verification should be enough, but only from 1 "production" lot (powder, cases, primers, bullets). Next lot start with full measurement.

Determining what is enough PF follows from sd in velocity and bulletweight. Usually I subtract the sd's from both and end up a bit lower. The narrowest escape I had was were I took training ammo to a match with probably 0.1 or 0.2 grains less N320 of a new lot (lower weight to volume) and ended up with 170.4 (from 3 rounds).

Caliber / gun, 40S&W / PO

DVC, John

[Skywalker]

Determining what is enough PF follows from sd in velocity and bulletweight.  Usually I subtract the sd's from both and end up a bit lower.

John,

according to statistic laws, ±1 SD means 68% probability, while ±2 SD means 95%, and ±3 SD means 99.6%.

To explain in simpler terms:

Batch measurements:

Average: 950 fps

SD: 15 fps

This means:

You have 68% probability that all rounds from your batch will chrono between 950-15= 935 fps and 950+15=965 fps.

You have 95% probability that all rounds from your batch will chrono between 950-15*2= 920 fps and 950+15*2=980 fps.

You have 99.6% probability that all rounds from your batch will chrono between 950-15*3= 915 fps and 950+15*3=995 fps.

Of course, the accuracy of the SD depends on how large is your batch sample. I use a 10-rounds batch sample as most of the posters here have reported.

[kevin c]

Welcome to the forums, RobT. You'll learn a lot here (and make friends from all 'round de worl!).

I am a real tyro when it comes to Web wrangling, otherwise I'd link you to another thread I started a few months ago (called Making Major at the Chrono). People much smarter than yours truly offered a few pointers on statistics, one of which was that, for an approximation of the values we get from our chronos (average velocity, SD, AD) 10 rounds is in the ballpark (and is what most people use), 30 is a lot better, but only firing the whole lot of ammo will get you the numbers dead on, (though by then it's a moot point).

[MoNsTeR]

http://glocktalk.com/showthread.php?s=&threadid=253355

If you're serious about ensuring your load makes PF,

SAMPLE 30+ ROUNDS.

[GvU]

Sky,

You are about right, but I forgot my confidence "border" from 170.

First: The chance of any round fired being within 1 sd = 68%. The chance of the velocity being low = 16% (same as for to high). You can read that as 1 in 6 being below 1 sd, 4 being right on the money, and 1 above sd (this is no guarantee, as this is only probabilities / expectations). At least that is were my analisis of hundreds of rounds show what to expect. From there I got to this method, where I load and shoot only 3 and repeat that 3 times more. This also shows first shots registring low!

Second: The sd is usually very low (less than 5 fps (loaded on a 650)) over a sample of 4 x 3. I look at each individual 3 rounds as well as the total sample of 12. The 12 round sd is lowest (exceptions are there). The average of three is usually within 1 sd of the average (12 round sample), the average plus 1sd (3 round sample) always is (but not important)!

Third: I want the "calculated" lower floor to be at 173. I usually end up at 175-176. This is only compensation for the measuring equipment used at any match.

Forth: If by any chance the sd is high and the first 3 are on average to low the probability is, that with the next 3 added the highest 3 out of 6 will make the PF needed. The probability is then much better! (At the chrono out of six: 2 close to average, 1 average plus 1 sd and a bit). Actually you could ask; "does the sd high or low really matter?"

Fifth: I want to know if the sd is low or high (confidence) in your ammo. Also see next point.

Sixth: I want to now if the quality of the bullets (and the average weight) is good (correct). So I take that into acount as well. In the worst case you want to know what to do with round number 7!

Seventh: One may check on temperature influences on the ammo as well, some powders may be temperature sensitive (or is it maybe the temperature at which the ammo is loaded???). This is also usefull to know, especially if extreme temperature differences exist (summer - winter).

The result is that my ammo makes major at the matches. The question was how to make sure (without loading to heavy). If you would take the confidence level (certainty of almost 100%) you would probably have to load heavier than needed.

Bottom line: Loading to a PF of 175 (where 170 is major), is where most end up. So why bother? (I Like to play with this stuff and see if it holds ground.)

Remark 1: The measurement are only valid for one gun and have to be repeated for other gun(barrel)s.

Remark 2: Low sd has nothing to do with accuracy (at least not on our pistol distances), but it wont hurt either.

DVC, John

[Loves2Shoot]

http://glocktalk.com/showthread.php?s=&threadid=253355

If you're serious about ensuring your load makes PF,

SAMPLE 30+ ROUNDS.

If your serious about making PF give yourself plenty of "wiggle" room. Especially if you are traveling to an area you have never chronoed. I will give myself 5 pf for places I know and 7-10 for places I've never been. The tiny extra recoil is way better than worrying about making PF.

[shred]

No point chronoing 30 rounds in my book. All you'll get at the chrono is seven tries. What you should do is chrono multiple different days and times with multiple chronos if possible, as well as getting a really good handle on your true bullet weight.

Chrono differences, setup and light vagaries will far exceed whatever precision you get from measuring a whole lot of rounds all at once.

[RobT]

Thanks for the welcome Kevinc and thanks to all for your input.

I suspected this topic had been discussed, but my search efforts usually result in 20 gazillion hits or none

I did mention this problem to one of the shooters at our indoor league who has a masters in math. She recommended an experimental approach - record the standard deviation after every 5 rounds and when the SD "settles down" and approaches a limit, you have reached a practical sample size. Of course, it is yet to be determined if the SD will "settle down" in what I regard as a practical number of rounds to expend in trying to measure one load.

Anyhow, based on the topic posts and my unscientific gut feel, I'll use 20-30 rounds per load at my next chrono session and may try the suggested experiment.

DVC

RobT

[Flexmoney]

If this helps...

I ran the chrono for two days at a recent Major match. And, Steve Anderson ran it the other day.

Out of 180 or so shooters, we only had a handful that we need to fire more than three rounds for.

[MoNsTeR]

Like I discussed in the GT thread, the fact that they only sample 3-7 rounds at the chrono test is exactly why you must sample 30+ rounds in development.

Reasong being, there are two main sources of variation you're confronted with: the inherent variation in the load, and variation due to random sampling. Sampling 30+ rounds virtually eliminates the effects of sampling error, so you can concentrate on controlling for the variation in the load.

68% will be within one SD, 95% within two, and 99.9%+ within 3. The question is, within X SD's of what? Of the true mean, not your sample mean! Your 10 round sample doesn't tell you the true mean, it only approximates it. A 30 or 50 or 100 round sample is still an approximation, but an order of magnitude more accurate. "Just load to 175PF" can still fail you due to sampling error if you only measure 10 rounds.

[eric nielsen]

Loves2Shoot nailed it.

7 over the power factor will pass you through most any chrono, anywhere.

[shred]

Reasong being, there are two main sources of variation you're confronted with: the inherent variation in the load, and variation due to random sampling.  Sampling 30+ rounds virtually eliminates the effects of sampling error, so you can concentrate on controlling for the variation in the load.

I submit that the greatest variation you'll see with the same load and even half-ass decent loading equipment/technique is chrono variability-- light, temperature, brand, setup, distance, wind, battery voltage, etc, etc.

Shooting 30 rounds at a time will tell you zero about different chrono setups. Three sets of ten on different days would be much better than 30 all at once.

Making Major at home on your chrono is only half the battle.

[GvU]

68% will be within one SD, 95% within two, and 99.9%+ within 3. The question is, within X SD's of what? Of the true mean, not your sample mean! Your 10 round sample doesn't tell you the true mean, it only approximates it. A 30 or 50 or 100 round sample is still an approximation, but an order of magnitude more accurate. "Just load to 175PF" can still fail you due to sampling error if you only measure 10 rounds.

This will always be through, as each round tested has the same chance being to low. The only thing you can say is, that the chance of being to low gets smaller with a higher PF.

The method to arive at the method I use is derived from many samples and looking at the overall pricture as well. Reject of accept is part of this when it is time to use a new batch of any components. The sample is then not included in the overall. Actually a new situation is created and you should start al over or use experience.

Making Major at home on your chrono is only half the battle.

So through. This is where experience with "other" chrono's becomes valuable. This also sets a trend (comparison) to your chrono. This is were you compensate x many PF-points. If that is 3 then you want something like 173 (IPSC) as your minimal PF (I mean: average - 1 sd).

DVC, John

[rtr]

I'm just curious, when loading rounds to chrono how do you do it?

Meaning do you measure each round to ensure it is the right amount of powder?

I load on a 550, if I set it to throw say 7.5 grains it will generally throw anywhere from 7.4-7.6 grains. So far I have been measuring each round to ensure that the rounds I'm chronoing are actually 7.5 grains, so I know what kind of chrono info I'll get from rounds that are 7.5 grains.

So how do you do it?

[kevin c]

You're right, of course, that you really don't know that the rounds you are chrono'ing have the charge that you are trying to test, given the variation in each "throw" of powder. In your case, it could be a tenth off in either direction, compared to what you might actually read if you actually measured each powder charge. That measurement will also vary from the "true" weight by up to a tenth, because of the vagaries of the measuring process, even using a Dillon scale.

But it doesn't matter. You aren't going to weigh each individual charge as you load your match ammo anyway, are you? What you want to know is how the ammo performs, on average, given how you set up your press and how you yourself run it. That is as much as you can expect given the nature of the loading and measuring process. IMHO, you can get that following the recommendations here in this thread, and, for our game, that is good enough.

[Flexmoney]

Yeah, what kevin c said...

Measuring each round to ensure it has 7.5 will tell you what numbers you will get with 7.5.

But, if you want to measure the ammo that your press actual pumps out, then you just need to measure the ammo that it pumps out.

[.40AET]

IMO, you need to chrono what your press is going to crank out on a regular basis. As long as you chrono 10 rounds and the slowest round makes it over the minimum PF, then you will be good to go.

Pay attention to the extreem spread of the sample. If it's 26 and you high PF is 170, then the low PF will be 167.

To determine what my powder measure is throwing I take the average of 5 charges. Watching the scale go up and down by a tenth makes me nuts.

Once I decide on a load that I'm going to use for a while, I'll chrono it every time I change any component. Also, 2 weeks before a big match, I'll check it again to try to eliminate any stupid sh*t that may have happened over time.

You aren't going to weigh each individual charge as you load your match ammo anyway, are you?

Nope...The brink of insanity is sometimes only a step away. That would throw me way over the edge.

[revchuck]

If you use mixed brass, you need to have an idea of at what velocity each brand of brass chronographs with your chosen load. I mainly worry about .38 Special, because that's the hardest to make the PF with using current published data. IME, brass headstamped "WCC" gives the highest velocities, and "R-P" gives the lowest. I load so that the R-P makes the PF, and then use the WCC at major matches to give me a bigger fudge factor.