ES and SD

[Matt P]

Hi All
When using small samples (10 shot groups), what SD and ES do you deem acceptable for 1000 yrds ??
Matt P

[Quick]

Id aim for single digit on both but I dont chrony all that often. I think ES under 15 and SD in single digit would be good.

[Aubrey]

http://precisionrifleblog.com/2015/04/1 ... sd-matter/

[AlanF]

Matt,

Using a 7mm 180 VLD at 2820 fps as an example :

Theoretically, ES of 25 and SD of 5 would keep you vertically inside the 6 ring at 1000yd 99.99% of the time, BUT that assumes no other errors. In my opinion velocity spread typically accounts for about 2/3rds of vertical error, so ideally you would want ES about 18 and SD about 3.3. I've calculated that over a large number of shots, ES is about 5 times SD. As every statistician knows, you need to do at least 30 shots to get reliable statistics.

In the real world of course, positive and negative compensation can respectively make things better or worse.

All that said, if I could get SD of 6 or 7 (over at least 30 shots) then that would be very satisfying.

Alan

[Brad Y]

Honest answer is I will be happy with what ever numbers I get when the gun holds X ring elevation. Currently those numbers arent brilliant but stoked how the gun shoots at long range.

[RAVEN]

Accurate loads coinciding with a SD single digits 4 - 8fps ES 10 -20fps for 10 shots you would have a very acceptable load for long range
some recent testing with neck tension in my barrels gave me more consistent SD & ES with light neck tension from the 284 with 180VLD's.
I also tried an oddball primer recently with great results more testing to be under take before I'm curtain on its repeatability.

RB

[williada]

How many times do we get done by the teaser group which is so tight that an outlier screws a perfect shoot. We blame ourselves, curse the imaginary wind change or apparent aiming mark shift up, when in fact it went down. We come up with a litany of excuses. We still run with that load and we get them all next time. It must have been me last week. Then it lets us down again. Our performance does not stabilize.

I am here to tell you it was not your fault. Your load development was built on indicators of success, and not certainty, and the odd shot if you fired sufficient practice shots would have shown up. That’s why indicators have to be confirmed again a couple of times because one swallow does not make a summer and be cross checked with other tests to increase your confidence in the load.

How comfortable would you feel if a range officer exercised a personal view not in accordance with the rules, saying the rules are only guidelines, or someone else saying, it’s a rule not law when it suits them or in ignorance. If decision-makers are in breach of the spirit of the rules which the vast majority agree to uphold, then confidence is lost too.

How confident would you be in the drugs you swallow if the standards lapse with regards to your personal safety. All this stuff is based on statistical method. To really understand that is important.

The chronographs spit out average speed, extreme spread and standard deviation (SD) as well as individual velocities but they have to be used in a mathematical/statistical context to be accurate to be representative to predict with high probability what will happen in the future. One hundred percent is certainty. Would you take odds of 30/70 or 50/50? How confident would you be really? Ninety-nine to 1 has more chance of success.

The ten shot group is only an indicator, but we treat it as gospel because our thinking is skewed to not wearing the barrel out. However, if that ten shot group was loaded from a larger statistical sample or cross checked with other tests you would have greater confidence in the load. So the question should then be what is the standard deviation figure for my 1000 yard shoot which is a ten shot match based on a minimum sample of statistical significance or 30, preferably more? Not a small sample, as that may mislead people as to what should work unless you have cross checked and confirmed that to equate to a larger sample size.

But the figures tossed around as being great but many do not represent how groups form as Alan has suggested. The SD is a measure of dispersion from the arithmetic average or mean. It assumes the shots are evenly distributed when in fact many may not be. i.e. the group centre should be the mid-point (median) of the group or where do most of the shots fall (the mode). The median, the mode and the mean are measures of different averages in a group and all can be used to maximise a result.

On a rough day, we use the mode, when we want most of our shots in and are prepared to lose a couple by playing percentages. On a tricky day we must centre our group so we don’t lose that shot on the line. On an easy day, we can relax and run with our average, but we may get done by an outlier away from the rest of the class who has the distinction of being out the top above average. The average may not take into account the extreme spread of your group. If the average is tightly bunched we are all in the guts, if the average is spread out too far the extreme spread of outliers may put you out.

Now here is the problem with the SD figure. It assumes a normal distribution of your shots represents the shape of a bell, although it is still a handy measure. That means the mode, the mean and the median all neatly line up in the middle. If that is the case, 68% of your shots will fall within + or – 1 SD from the mean, 95% will fall + or – 2 SD’s from the mean and 3 SD’s equates to 99.7%. The more shots that are fired the more confidence you can have in the accuracy of the data representation. If there are only ten shots, then it may not represent a normal distribution of the shot data. This is what Standard Deviation on your chronograph means and by firing only a small number of shots it does not give the big picture.

There are two other problems represented by the SD analysis. The distribution of shots maybe skewed to the left or right. This is not a normal distribution to which the SD figure is applied. Groups form that way sometimes and I recognise them as teaser groups and try to tune the problem out.

Just as a variable group differs on the horizontal axis, by being skewed one way or the other, the scatter, spread or dispersion can look like a mountain or a hump. The distribution of the group can be very pointed with wide tails or humped with short tails.

To further complicate things the ballistic programs that simulate SD dispersion assume the launch angle is fixed when in reality it is far from it with positive compensation which can impact on the distribution of the shots. We go to the range believing the marketers know, and our groups may fail to live up to expectation.

Load development has to be tweaked to reflect the reality of what happens. That is the benefit of Ozfclass where we can keep raising the bar through discussion. David.

[RAVEN]

Agreed Dave
I also use SD of group calc on 5-6 5 shot groups shot at any distance other than 100yrds
and I'm a realist also
Numbers are a guide
Holes in targets is what really counts
RB

[Matt P]

Agreed Dave
I also use SD of group calc on 5-6 5 shot groups shot at any distance other than 100yrds
and I'm a realist also
Numbers are a guide
Holes in targets it what really counts
RB

Hi Richard
Couldn't agree more the paper tells the end result, reason for my question, I finally have my Labradar working and want to put the numbers to good use.
Matt

[DenisA]

I have no valuable info to offer Matt, but while we’re on the topic…….

Through my experience in short range testing I’ve found that sometimes you can produce repeatable brilliant groups with terrible ES and SD, terrible groups with brilliant ES and SD and then brilliant groups with brilliant ES and SD.

The lack of apparent relationship between grouping ability and good ES and SD has me scratching my head. Firstly, group shape and size is a product of barrel harmonics and bullet exit time. By my way of thinking, brilliant ES and SD are a result of consistent internal ballistics.
These days we have the control of that down pat with the ability to maintain consistent weight sorted brass, consistent neck hardness and seating tension, consistent bullet seating depth to the thou, consistent powder weighing to the kernel, etc.

The baffling part to me is when you load a batch of rounds to test powder charge and you know that you’ve crossed all your t’s and dotted all your i’s, everything across the batch except powder charge is as close as it can be and every round in each powder charge is as close as it can be. Depending on powder type, each powder charge might only be 3 – 5 kernels different from the last. With such a level of control, theoretically I would expect to see consistent ES and SD for each group and as the bullet exit time varies then different group size and shape for each depending on the muzzle harmonic.

I don’t believe it’s related to the condition of the bore because over 40 rounds for example testing between 2 nodes, the ES and SD can be huge. But if you test 40 rounds of a good charge then the ES and SD can be small.

So then why does ES and SD vary so much within a highly controlled testing batch?

For a silly prick like me that doesn’t shoot very well, I don’t understand and therefore have to work on a basic principal.

Use the chronny to find the velocity node I want, explore the length of the node based on group size, shape and POI, recheck ES and SD and hopefully the best are in the middle of that node or close to. If not, pick the middle group anyway and post a thread on Ozfclass and tell everyone how confused I am and hope for miraculous sudden clear understanding.

[RAVEN]

good post Denis

I take a holistic approach where others seem to think that chasing the holy grail of small numbers that you will shoot bug holes all day long
Too often I see ppl post SD & ES for 3 or 5 shot and in some obscure way determine that this is the load that everyone should be aspiring to
I have a term for these ppl that I won't share here
I have tuned many barrels for myself and other and sometimes because of circumstances beyond my control these had been done without a working chrony
and the tuned load has worked very well for them I would then adjust one thing at a time and see if the chrony number improve but if they don’t and the load holds
a consistent low SD of group I just go and shoot it easy really not rocket science just proven method.

RB

[RAVEN]

Hey Matt will be interesting to hear your opinion of the Lab Radar
once you have put it through its paces.

RB

[jasmay]

Denis,

You should try introducing a bipod into your load development and watch your head explode.

I agree with Richard and Denis, I use a chrony to get into the ball park then work on results on paper.

I also agree with David, sample sizes must be of a reasonable size to be meaningful 5 shots is useful only to get an average velocity IMO, 10 shots gives an indication of where we are headed with ES & SD, 20,30,40 shots builds confidence.

[williada]

Boys, I only use a chrony on a new barrel to know when I am over the hump. Having done so much of this stuff I read the tea leaves on paper. The final arbiter. The groups tell me what I want to know from years of testing and thousands of rounds through a machine rest. Always cross check and confirm.

Denis is right and probably the sample size is two small for what we do in competition to marry with the results. The variables, and there are so many and we can not always predict how they will interact exactly do not always harmonize to form a node where muzzle size is relatively stable. The stress waves, transverse waves and reflected waves screw with expectations of low SD's because they form deceptive nodes that are stacked at times. When the bullet is leaving the muzzle these waves speed up and create chatter. If you get the exit timing right when there is the least excitement your results will be better. If you are leaving on a true node, there is less chatter. Remember, that a true node is strictly velocity dependent, so if the day is hot velocities change and vice versa. This changes the chatter too, so your nice SD or extreme spread set on a different condition might not mix. In the fact as the barrel warms its harmonic length changes so an OCW tune or a compensation tune may give better results. So that's why I have been working with group shapes and trends for many years to minimize things we cannot control precisely.

[DaveMc]

Peter Smith and I have done a lot of testing between us - mapping velocity vs elevation at 1000 yards. Much at the old prawn farm 1000 yard "test paddock" with Magnetospeeds and PVM chrony's and more recently testing with the labradar at various ranges. A lovely set of calm and overcast conditions in Townsville recently for the FCWC squad practice with a variety of rifles.

A phrase Peter has used from time to time and I totally agree with. (You can quote him on this)

ES - Good at explaining what has happened.
SD - Will tell you what to expect.

Mapping velocity against elevation with rifles that are in tune will explain the great bulk of elevation at 1000 yards.

Generally speaking an SD of 2.5-4.5 are the hummers that hold x ring for most of the time at 1000 if you have the right load. - 5-8 will see you holding the 6 ring for the majority of shots (there are ALWAYS exceptions) But - We have not seen an exception to this rule of thumb yet!!!! (And very keen to if someone has some real data). Get a rifle in tune and a barrel and load that holds low sd and you will be accurate all the way through.

BUT There is a far bit of difference in projectile choice and speed here and Peter did a great chart of this recently which I will try and modify to post. A 230 grain pill out of a 308 is FAR more susceptible to velocity variation than a 180 gr (VLD or Hybrid or similar) out of a SAUM. - Around 1.5 times as much. So you need a tighter SD in these combinations to hold elevation than you do in others.