OK, have been wanting to run this experiment for a while, and finally found the time 2 weeks ago. Have been collecting range brass over the past 4 years and just kept it in a plastic bucket, never shot it since it is unsorted, different manufacturer, different batch and different weight. Would not be wise (or safe) to mix and match brass. My main question was: How much do you have to back off on powder charge if you switch to much thicker heavier brass?
Went through the 1000 or so i have and selected 3 to 5 of each make: Hornady, Nosler, Winchester (probably the same as Hornady?), Prime, Federal, Sig, and Norma. [Lapua was excluded because it is a small rifle primer case, and did not want to pick up speed differences due to primer changes.]
Neck turned all of them to 0.0125” (12.5 thou). Trimmed all of them to 1.908” as some were quite short and some were quite long, well beyond trim length. Chamfered and deburred, cleaned primer pockets, tumbled for 2 hours (fairly clean). Idea was that the necks must not contribute to weight differences (bullet takes up most of the neck volume before ignition starts, so reasoned neck’s impact is less inportant). Later results seem to confirm this.
Always wondered if weight differences are mainly caused by wall thickness variation vs. variations in the web thickness. Cases were not yet deprimed, so could measure case volume for a few of them. Used 91% rubbing alchohol to measure case volume on a lab scale to within 0.04 gn. Used a rubber plug to close off the neck so it will not leak or spill once on the scale, and this reduced minuscus errors, as the excess liquid would spill out of the case when the plug is slipped over the neck (on the outside), then dried the case with a cloth. Alchohol evaporates quickly, so a clean operation, just lots of fumes to deal with. Good ventilation helps. It is easy to calculate alcohol weigth (and therefore volume) from the difference in weight between an empty dry case, and the same case filled with liquid. Liquid seems to fill up the primer pocket as well due to the low viscosity of alchohol.
Found a remarkably accurate correlation between case weight (considering one make of brass at a time) and case volume: Straight line least squares fit had a R-square of 0.95 meaning the best straight line fit could ascribe 95% of the case volume variation to the case weight differences. Concluded that the weight of the liquid needed to fill a case could be adequately inferred via case weight, which is much faster to do. Differences in web thickness did not appear to cause major differences in this linear relationship (so no real issue with thick web and thin case wall combos), at least for the brands and batches that I tested. [This will not be true for poorly made brass.]
Hornady brass (from different batches) varied from 143.0 grain to 154.5 grain, a difference of 11.5 gn, which is quite significant.
Norma brass varied from 153.8 to 155.2 gn, a delta of 2.4 gn (which is far better than Hornady), but all from one batch, and the batch was small (50).
Winchester brass varied from 148.1 to 150.0 gn, a delta of 1.9 gn. Prime stamped brass was around 155 gn. with fairly low variation.
Notably, Nosler brass had a very narrow weight range (160.4 to 160.9) for this random sample of 5, but it was quite a bit heavier than other brands. Maybe that explains why it lasts 2x longer than Hornady brass in terms of max nr of reloads? Nosler claim to sort by weight before they ship, and that appears to be true.
Sig stamped brass was around 155 gn on average.
The heaviest brass was Federal at 172 gn, which is a very significant difference from Hornady at 143-155 gn. Federal brass is a good 15% heavier than Hornady. [Anybody who has experience reloading Federal brass: How long do they last?]
Next selected 2-4 pieces of brass from different manufacturers to span the widest possible weight range and get enough data points for an Excel analysis. Weight range was from 143.00 to 172.28 gn. We all know you cannot load heavy thick walled brass to the same max load as thin walled brass, so i tried to be conservative and reduced powder charge from a max of 41.6 gn (for my rifle) to 40.0 gn of H4350 and loaded 140 gn Hornady ELD Match bullets, all weight sorted to be within 0.1 gn. Scale is an AND FX120i lab scale capable of measuring to 0.02 gn. Used a Labradar to capture speed. Should be pretty accurate.
Federal GM210M large rifle match primers were used, all from the same batch. This batch gave good results before. Brass was full length resized in a Redding type S die, and shoulders were bumped back by 0.003”. Neck tension was fairly light (same bushing, same neck wall thickness due to neck turning, Sinclair mandrel die used to open up the necks and get them more concentric), as that gave the best accuracy before.
Bullets were intentionally loaded long (0.150”) to force all bullets to soft seat as the bolt was closed. All cases closed very easily, due to low neck tension. Net result would be a small (but unknown) amount of bullet jam.
Results: Plot of speed vs case weight formed an almost perfect straight line, with the lightest cases (143 gn) achieving 2699 fps, and the heaviest Federal cases (173 gn) getting to 2759 fps from a 30” Shilen Match Select barrel. This is about the same speed than i typically get with 41.5 gn of the same powder with Lapua cases, for the same 140 gn bullet. [Higher than expected speed from the Federal case was clearly caused by the 15% lower case volume. Jamming the bullet was also a contributing factor, likely to a lesser extent.]
Speed difference between lightest (143 gn) and heaviest case (173 gn) was 60 fps, which is significant. According to Excel, we gain 2.2 fps per grain of additional case weight, at these relatively low speeds (for a 30” barrel). Statistical fit was good: R-square was 0.97, after rejecting one outlier which was way off (possibly a user mistake).
Wondered if the soft seating was affecting my results too much, so repeated the experiment with 140 ELD-M loaded to book length (2.80” COAL), about 0.050” jump. Used the same brass as before, this time body sized then neck sized (equivalent to FL sizing), same powder charge, speed varied from 2653 to 2749 fps. R-square was less impressive this time, at 0.89, but still adequate. Excel said we now gain 3.04 fps per grain of case weight.
Repeated the experiment a third time at max load (will try to avoid mentioning powder charge or exact speeds). A small nr of half moons appeared on the case heads, and flat primers, indicating pressure was probably getting too high. Speed increment per gn of case weight increase was much less than before, at 1.2 fps per grain. This is in line with published BR experience: There comes a point where speed will not increase anymore. Degree of fit was also very poor (Rsquare of 0.37) indicating other random factors (e.g. the case not adequately gripping the chamber every time, and moving backwards to make contact with the bolt face). Also hints that shooting max loads might not be the most accurate approach (SD will open way up).
Last experiment: Tried neck sizing only (once fired since last full length resizing), 40.0 gn of the same batch of H4350, same 140 gn ELD-M bullet, loaded to book length. Speed gain was 2.12 fps/gn of case weight increase, very similar to the previous result i got with FL sizing. Speed range was 2656 fps to 2725 fps. Seems to indicate that neck sizing leads to lower speed (expected, case has more volume prior to expansion and prior to gripping the chamber), but that the speed gain vs case weight increase is almost identical between FL and neck sizing. Rsquare was 0.895, which is good. [One possible outlier was kept. Nice straight line except for the one data point.]. Around 70 fps speed gain going from a light Hornady case to a heavy Federal case. Same delta speed result as before (as long as you are well away from max loads).
What can we do with data like this? Some (possibly useful) conclusions:
1) Good quality brass is available (like Nosler and Lapua) that does not need to be weight sorted for shooting at steel targets at 600 yards. But if you want to get a further reduction in SD, you will likely gain a little bit (1-3 fps) by at least culling the very heavy and very light cases from your chosen 50. If the large batch varies by 3-4 grain (Hornady, Winchester), speed impact is 6-12 fps if we believe my Excel result, which is still OK at 400 but not OK at 1000 yards, so weight sort your (economy) brass. Takes only an hour or so, so worth it.
2) If you compete, or you are trying to shoot the smallest possible 5 shot groups (<0.35”) or trying to improve on your personal best, then weight sort your brass into groups of 5 that have very similar case weight and serialize your brass (each get a number, via a permanent marker or an engraver, stick to the web area where the case is very thick, use small letters or numbers so it does interfere with a concentricity guage). Yes this is a pain and takes a lot of time. And people look at you funny. Then sort your loaded ammo based on serial number. I don’t always do this, it depends what i am trying to accomplish (e.g. this is pointless on hunting ammo). When i shoot beyond 500 yards at very small targets, i do this and also cull the loaded ammo based on a concentricity check. In my experience, SD comes down from the 8-12 range to the 5-9 fps range. Not always a huge improvement (there are other factors too), but enough to push up your hit percentage on small steel targets, and it helps when you are shooting groups at diatance. Best 4 shot group at 500 yards = 0.97”, but that has only happened once. Typical result is 1.2-2.5”. Wind is the enemy.
3) At least for the brands of brass that i tested, case weight after decapping, trimming and and tumbling is a good proxy for case volume. Consider buying cases in bulk (200 plus) to get the same batch, then trim to length (neck turn if you want, or skip), weigh them all, and then sort your brass into a batch of 50 with the narrowest weight range you can get from that batch. There seems to be little benefit in measuring actual case volume, which is slow and messy if you use water. Water dripping through the electronics inside your scale is probably a bad idea.
4) Avoid max loads, as speed variations (SD) becomes excessive, and the increased SD will cause large vertical spread at distance. Probably best to reload for the second highest ‘node’ and avoid the top node. Watch out for thick walled (heavy) range brass, and calculate how far you will have to back off on powder charge, start low and increase in small increments.
5) The sensitivity factor that relates speed to case weight is somewhere between 2.1 and 3 for my rifle (30” barrel), and it seems to depend on the speed range. The slower the speed, the higher the factor. Your results will likely be different.
If i want SD below 10 fps, then assuming half the variation comes from case volume and the remainder from other sources (bullet weight, bearing surface, barrel fouling, barrel temperature, ambient temperature), then i can only afford 5 fps variation due to case volume/case weight differences. That means between 5/2.1 and 5/3 = 1.7 to 2.5 grain. To average that out, i’m assuming i cannot vary case weight by more than 2 grains in the same box of ammo. I usually aim for 1.0, sorted into groups if 5 that are 0.1 gn different. Probably overkill, but it helps my confidence! The placeabo effect is very real! ?
You will have to run your own experiment since this is a 30” barrel on a target / bench rifle, which is probably not typical for this community. Would be very interested to see similar results for short barrel 6.5 Creedmoor rifles, as i intend to build an 18” rifle in that caliber.
Hope this helps.
[Edit: How much to back off on powder charge if you switch to heavy cases requires another experiment: Determine effect of powder charge on speed, for your bullet jump/jam, weight, primer and powder combo).]
As always, YMMV. Hope this helps!
NH
Went through the 1000 or so i have and selected 3 to 5 of each make: Hornady, Nosler, Winchester (probably the same as Hornady?), Prime, Federal, Sig, and Norma. [Lapua was excluded because it is a small rifle primer case, and did not want to pick up speed differences due to primer changes.]
Neck turned all of them to 0.0125” (12.5 thou). Trimmed all of them to 1.908” as some were quite short and some were quite long, well beyond trim length. Chamfered and deburred, cleaned primer pockets, tumbled for 2 hours (fairly clean). Idea was that the necks must not contribute to weight differences (bullet takes up most of the neck volume before ignition starts, so reasoned neck’s impact is less inportant). Later results seem to confirm this.
Always wondered if weight differences are mainly caused by wall thickness variation vs. variations in the web thickness. Cases were not yet deprimed, so could measure case volume for a few of them. Used 91% rubbing alchohol to measure case volume on a lab scale to within 0.04 gn. Used a rubber plug to close off the neck so it will not leak or spill once on the scale, and this reduced minuscus errors, as the excess liquid would spill out of the case when the plug is slipped over the neck (on the outside), then dried the case with a cloth. Alchohol evaporates quickly, so a clean operation, just lots of fumes to deal with. Good ventilation helps. It is easy to calculate alcohol weigth (and therefore volume) from the difference in weight between an empty dry case, and the same case filled with liquid. Liquid seems to fill up the primer pocket as well due to the low viscosity of alchohol.
Found a remarkably accurate correlation between case weight (considering one make of brass at a time) and case volume: Straight line least squares fit had a R-square of 0.95 meaning the best straight line fit could ascribe 95% of the case volume variation to the case weight differences. Concluded that the weight of the liquid needed to fill a case could be adequately inferred via case weight, which is much faster to do. Differences in web thickness did not appear to cause major differences in this linear relationship (so no real issue with thick web and thin case wall combos), at least for the brands and batches that I tested. [This will not be true for poorly made brass.]
Hornady brass (from different batches) varied from 143.0 grain to 154.5 grain, a difference of 11.5 gn, which is quite significant.
Norma brass varied from 153.8 to 155.2 gn, a delta of 2.4 gn (which is far better than Hornady), but all from one batch, and the batch was small (50).
Winchester brass varied from 148.1 to 150.0 gn, a delta of 1.9 gn. Prime stamped brass was around 155 gn. with fairly low variation.
Notably, Nosler brass had a very narrow weight range (160.4 to 160.9) for this random sample of 5, but it was quite a bit heavier than other brands. Maybe that explains why it lasts 2x longer than Hornady brass in terms of max nr of reloads? Nosler claim to sort by weight before they ship, and that appears to be true.
Sig stamped brass was around 155 gn on average.
The heaviest brass was Federal at 172 gn, which is a very significant difference from Hornady at 143-155 gn. Federal brass is a good 15% heavier than Hornady. [Anybody who has experience reloading Federal brass: How long do they last?]
Next selected 2-4 pieces of brass from different manufacturers to span the widest possible weight range and get enough data points for an Excel analysis. Weight range was from 143.00 to 172.28 gn. We all know you cannot load heavy thick walled brass to the same max load as thin walled brass, so i tried to be conservative and reduced powder charge from a max of 41.6 gn (for my rifle) to 40.0 gn of H4350 and loaded 140 gn Hornady ELD Match bullets, all weight sorted to be within 0.1 gn. Scale is an AND FX120i lab scale capable of measuring to 0.02 gn. Used a Labradar to capture speed. Should be pretty accurate.
Federal GM210M large rifle match primers were used, all from the same batch. This batch gave good results before. Brass was full length resized in a Redding type S die, and shoulders were bumped back by 0.003”. Neck tension was fairly light (same bushing, same neck wall thickness due to neck turning, Sinclair mandrel die used to open up the necks and get them more concentric), as that gave the best accuracy before.
Bullets were intentionally loaded long (0.150”) to force all bullets to soft seat as the bolt was closed. All cases closed very easily, due to low neck tension. Net result would be a small (but unknown) amount of bullet jam.
Results: Plot of speed vs case weight formed an almost perfect straight line, with the lightest cases (143 gn) achieving 2699 fps, and the heaviest Federal cases (173 gn) getting to 2759 fps from a 30” Shilen Match Select barrel. This is about the same speed than i typically get with 41.5 gn of the same powder with Lapua cases, for the same 140 gn bullet. [Higher than expected speed from the Federal case was clearly caused by the 15% lower case volume. Jamming the bullet was also a contributing factor, likely to a lesser extent.]
Speed difference between lightest (143 gn) and heaviest case (173 gn) was 60 fps, which is significant. According to Excel, we gain 2.2 fps per grain of additional case weight, at these relatively low speeds (for a 30” barrel). Statistical fit was good: R-square was 0.97, after rejecting one outlier which was way off (possibly a user mistake).
Wondered if the soft seating was affecting my results too much, so repeated the experiment with 140 ELD-M loaded to book length (2.80” COAL), about 0.050” jump. Used the same brass as before, this time body sized then neck sized (equivalent to FL sizing), same powder charge, speed varied from 2653 to 2749 fps. R-square was less impressive this time, at 0.89, but still adequate. Excel said we now gain 3.04 fps per grain of case weight.
Repeated the experiment a third time at max load (will try to avoid mentioning powder charge or exact speeds). A small nr of half moons appeared on the case heads, and flat primers, indicating pressure was probably getting too high. Speed increment per gn of case weight increase was much less than before, at 1.2 fps per grain. This is in line with published BR experience: There comes a point where speed will not increase anymore. Degree of fit was also very poor (Rsquare of 0.37) indicating other random factors (e.g. the case not adequately gripping the chamber every time, and moving backwards to make contact with the bolt face). Also hints that shooting max loads might not be the most accurate approach (SD will open way up).
Last experiment: Tried neck sizing only (once fired since last full length resizing), 40.0 gn of the same batch of H4350, same 140 gn ELD-M bullet, loaded to book length. Speed gain was 2.12 fps/gn of case weight increase, very similar to the previous result i got with FL sizing. Speed range was 2656 fps to 2725 fps. Seems to indicate that neck sizing leads to lower speed (expected, case has more volume prior to expansion and prior to gripping the chamber), but that the speed gain vs case weight increase is almost identical between FL and neck sizing. Rsquare was 0.895, which is good. [One possible outlier was kept. Nice straight line except for the one data point.]. Around 70 fps speed gain going from a light Hornady case to a heavy Federal case. Same delta speed result as before (as long as you are well away from max loads).
What can we do with data like this? Some (possibly useful) conclusions:
1) Good quality brass is available (like Nosler and Lapua) that does not need to be weight sorted for shooting at steel targets at 600 yards. But if you want to get a further reduction in SD, you will likely gain a little bit (1-3 fps) by at least culling the very heavy and very light cases from your chosen 50. If the large batch varies by 3-4 grain (Hornady, Winchester), speed impact is 6-12 fps if we believe my Excel result, which is still OK at 400 but not OK at 1000 yards, so weight sort your (economy) brass. Takes only an hour or so, so worth it.
2) If you compete, or you are trying to shoot the smallest possible 5 shot groups (<0.35”) or trying to improve on your personal best, then weight sort your brass into groups of 5 that have very similar case weight and serialize your brass (each get a number, via a permanent marker or an engraver, stick to the web area where the case is very thick, use small letters or numbers so it does interfere with a concentricity guage). Yes this is a pain and takes a lot of time. And people look at you funny. Then sort your loaded ammo based on serial number. I don’t always do this, it depends what i am trying to accomplish (e.g. this is pointless on hunting ammo). When i shoot beyond 500 yards at very small targets, i do this and also cull the loaded ammo based on a concentricity check. In my experience, SD comes down from the 8-12 range to the 5-9 fps range. Not always a huge improvement (there are other factors too), but enough to push up your hit percentage on small steel targets, and it helps when you are shooting groups at diatance. Best 4 shot group at 500 yards = 0.97”, but that has only happened once. Typical result is 1.2-2.5”. Wind is the enemy.
3) At least for the brands of brass that i tested, case weight after decapping, trimming and and tumbling is a good proxy for case volume. Consider buying cases in bulk (200 plus) to get the same batch, then trim to length (neck turn if you want, or skip), weigh them all, and then sort your brass into a batch of 50 with the narrowest weight range you can get from that batch. There seems to be little benefit in measuring actual case volume, which is slow and messy if you use water. Water dripping through the electronics inside your scale is probably a bad idea.
4) Avoid max loads, as speed variations (SD) becomes excessive, and the increased SD will cause large vertical spread at distance. Probably best to reload for the second highest ‘node’ and avoid the top node. Watch out for thick walled (heavy) range brass, and calculate how far you will have to back off on powder charge, start low and increase in small increments.
5) The sensitivity factor that relates speed to case weight is somewhere between 2.1 and 3 for my rifle (30” barrel), and it seems to depend on the speed range. The slower the speed, the higher the factor. Your results will likely be different.
If i want SD below 10 fps, then assuming half the variation comes from case volume and the remainder from other sources (bullet weight, bearing surface, barrel fouling, barrel temperature, ambient temperature), then i can only afford 5 fps variation due to case volume/case weight differences. That means between 5/2.1 and 5/3 = 1.7 to 2.5 grain. To average that out, i’m assuming i cannot vary case weight by more than 2 grains in the same box of ammo. I usually aim for 1.0, sorted into groups if 5 that are 0.1 gn different. Probably overkill, but it helps my confidence! The placeabo effect is very real! ?
You will have to run your own experiment since this is a 30” barrel on a target / bench rifle, which is probably not typical for this community. Would be very interested to see similar results for short barrel 6.5 Creedmoor rifles, as i intend to build an 18” rifle in that caliber.
Hope this helps.
[Edit: How much to back off on powder charge if you switch to heavy cases requires another experiment: Determine effect of powder charge on speed, for your bullet jump/jam, weight, primer and powder combo).]
As always, YMMV. Hope this helps!
NH
Last edited:
