Dry Fire Efficiency Test

[Hilltop Customs]

thats why I liked the data in g's much more accurate data, and since the data is more accurate, calculations made using the data are much more reliable.

what type of scale did you use for this? a postal scale? I have a small digital scale but its limit is 500 g's.

This could also be used for testing out different barrel efficiencies (and testing/verifying optimal solid bore barrel length)

edit: and accessory("upgrade") effects on efficiency

[brycelarson]

What's the accuracy on the scale?

[GoatBoy]

I have a couple of run-of-the-mill digital scales at home; one is a Pelouze and the other is some off brand picked up at Harbor Freight. They both measure in terms of grams, so that's roughly the accuracy. Obviously, a single shot can't be measured, so that's why you do things in sets of, say, 100.

[brycelarson]

I have a couple of run-of-the-mill digital scales at home; one is a Pelouze and the other is some off brand picked up at Harbor Freight. They both measure in terms of grams, so that's roughly the accuracy. Obviously, a single shot can't be measured, so that's why you do things in sets of, say, 100.

sure, but using a measuring device that's +/- %10 is a tough sale when measuring something so preciese.

the other number to be concerned with in your instrumentation is repeatability.

what's the display - just grams - or does it give a decimal value?

[GoatBoy]

sure, but using a measuring device that's +/- %10 is a tough sale when measuring something so preciese.

the other number to be concerned with in your instrumentation is repeatability.

what's the display - just grams - or does it give a decimal value?

It's just grams; no decimal places. Think about it for a bit. If I put 1.1g on there, what might it read? if I put a 1.6g on there, what might it read? Why do you think I did the test in batches of 100?

[Hilltop Customs]

even if the error is +/- 1g(worst case round up from .01, or down from .99) that would mean the max the data could be off would be +/- 2 grams since your using 2 measurments to calculate the average......divide that +/- 2g by 100 and your left with +/-.02 g per shot which is only about 10% or less of the value for calculated for g/shot.

as for repeatability, check out the value for the 2 dry fire tests using 50 shots....same g value of 10g. the value for 100 shots is 20g, which fits exactly on the linear relationship of .2g/shot. Id say that shows some good repeatability, but this could be verifyed by more tests, or tests with a more accurate readout/device.

If you want to reduce the error even more, add more shots to the test, or repeat the test and average the results. figure out the g value over 1000 shots and you will still be MAX+/-2x(scale error)......so then the final error will be +-2(scale error)/1000 or +/-.002g/shot.

I just want to say both tests are great, only real difference is what they refrence each shot to(psi or grams) and since pv=nrt they are both essentially equal(ex:1/4 the number of molecules/mass will mean 1/4 the pressure). Since the goal of a paintball gun is to fire the same amt of air at the same pressure for each shot, they both should vary linearly with the number of shots.

What it comes down to is what is the +/- of reading the guage vs the +/- of the digital scale readout.....I'd go with the digital, even if the psi guage is the most accurate in the world you still have to rely on someone to interpret the reading.(look at the guage at a skewed angle and you could easily be off more than the error of a bad digital scale)

[GoatBoy]

bryce:

I don't know how the scale works internally, but intuitively, you have two choices: round or truncate. Let's assume truncation. I'm measuring a delta, and I measure something that is 2.9g on one object, and 1.0g on the other object (I believe this is worst case for truncation). The machine truncates the first measurement to 2, and the other is 1. The measured delta is 1g, when in reality it was 1.9g. I'm off by less than 1 gram when I do the difference, because the same scale should skew the measurements the same way. If I was measuring such small quantities, then my percentage error is huge.

Now, I split the "with paint" into two 50 round measurements completely by accident, because I did 50 rounds and forgot to reload, went and measured and started breaking the gun down to prepare for the CO2 version and suddenly realized I forgot to do another 50. I really intended to do 100 straight, and as my note indicates, I probably need to re-do the data, but at least my repeatability was pretty good.

However, let's assume that I had done it correctly (and I probably will re-do it), and I still yielded a difference of 20g for 100 rounds with paint. I'm off by 1g at most (again, see above), so my max error is really about 5% for that range.

In any event, you're sweating the details, and missing the big picture on repeatability. Let me reiterate.

Your test implementation is not easily repeatable... by other people. Your procedure starts like this:

Step 1: Buy a 13ci tank.
Step 2: Buy an obnoxiously large, unwieldy, and possibly expensive HPA gauge.
Step 3: Install the gauge on the high pressure side of the 13ci tank.

Those three steps right there are already a barrier.

The alternative procedure starts out like this:

Step 1: Get a tank. Any tank. Can be CO2 or HPA.
Step 2: Get a scale that can accommodate your gun+tank's weight.
Step 3: Err, there is no step three, because you don't have to install the scale on the tank.

And then in either case:

Step 4: Fire enough shots to make the average instrumentation error go down

Anyways, I've run the math, and it seems to justify my methods. My experimental result with the emptying of the tank was within 1g of calculated weight with a 67g delta. Perhaps my scale, my gauge, and my math were all wrong in exactly the correct way to be within the same estimated error? If I'm wrong, then perhaps someone would like to... rerun my test and validate my results? You see where this is headed?



Don't get me wrong, I have no problem with your results. I trust you guys are trying your best to be diligent. I'm not trying to contradict your data, because I know there are way, way too many variables involved. It's the design of the test in the grand scheme of things that could use some work.