STUGGED UP AGAIN

[Tarquelne]

Originally posted by Bastables:

It's pretty simple Jason, you state that the front turret is proof versus 5cm KwK.

???

Could someone point out where JC states that? "1/3" and "not routinely", with specific ranges given, is what JC is pushing, IIRC. Is that "very simple" better characterized as "A twisted misintrepretation of your statements as a warmup to calling your argument factious."? Right now I'm in exasperation with B. However, I'm not opposed to redirecting the if my memory is faulty.

[Bastables]

Originally posted by Tarqulene:

</font><blockquote>quote:</font><hr />Originally posted by Bastables:

It's pretty simple Jason, you state that the front turret is proof versus 5cm KwK.

???

Could someone point out where JC states that? "1/3" and "not routinely", with specific ranges given, is what JC is pushing, IIRC. Is that "very simple" better characterized as "A twisted misintrepretation of your statements as a warmup to calling your argument factious."? Right now I'm in exasperation with B. However, I'm not opposed to redirecting the if my memory is faulty. </font>

[Tarquelne]

How about we try backing up a bit? B, how close do you think a "real life" 50L42 would need to be to "routinely" score front turret penetrations on a '41 T34. (Ideally, answer both in % of total hits and % of shots taken.) Now, what's your value for "routinely". Finally, same basic questions, but with regard to how it is in CMBB.

Jason: Same questions.

[Ex Bellator]

Post on this subject by A.E.B.:

Sorry in advance.

I know that this issue has been brought up on a number of occasions, but newbies like me don't necessarily have the time to check through 240+ pages of threads.

It is obvious that the question "how come I can KO T34s in 1941 so easily....I thought that they were invulnerable..." will just keep occurring every time a suitable thread is started. I have just finished looking through all my books on the Eastern Front with the aim of designing a couple of senarios. This is my observations on this matter.

The T34 was not invulnerable because:

* Due to a lack of spare parts and fuel, few were battle ready at the start of Barbarossa.

* The crews were very poorly trained.

* The early T34s had design faults: a shot trap around the mantle, a vulnerable plate covering the exhaust system and the cast turret tended to deflect rounds downwards into the hull deck.

* Early T34s were badly made. Even later T34s had gaps in the welding, plates that weren't correctly fitted, poorly rolled and hardened armour, etc.

* Many of the early T34s had a short (L28?) 76mm gun. These T34s were the HQ tanks of companies of BTs or T26s. This short gun was ineffective at tank killing.

* Early T34s had a two man turret that overworked the commander. This, coupled with a large turret hatch, greatly reduced the situational awareness of the commander.

* Most T34s in 1941 lacked radios and couldn't coordinate their actions with other tanks or infantry easily.

* Russian tanks in July 1941 were scattered in penny packets as infantry support, and were usually outnumbered by the Germans at the point of attack.

* The T34s superior maneuverability in mud and snow wasn't as noticable in the summer of 1941.

* Due to rapid German advances and the interdiction by aircraft that cut Russian supply lines, many T34s were lost due to lack of fuel and ammunition.

For the above reasons the Germans were initially slow to notice the presence of T34s, KV Is and KV IIs except for a few well reported incidents. Many Germans probably couldn't tell the difference between a T28, a T34, a T35 or a KV in June/July 1941 anyway.

How the Germans killed T34s with the L60 37mm or the L40 50mm or L24 75mm gun is interesting. I have a number of pictures of KOed T34s taken in 1941. None of these T34s showed the catastrophic damage (turrets blown off for instant) that is seen in T34s killed by later German high velocity guns. Most of the KOed T34s in the pictures had their turrets facing towards the rear of the tank. This suggests to me that the T34s were out maneuvered by the Germans on the battlefield. Another picture shows a T34 disabled by a large bomb crater, and another shows two T34s that have driven at speed into a swamp. Again the turrets face rearward, and it appears that the tanks bogged and were then abandoned.

Given the lack of burning or other catastrophic damage shown in the photos, I theorise that most T34s were abandoned by their crews after sustaining lesser damage.

The Germans began to notice the T34 once the weather, supply problems and attrition robbed them of the initative. It is then that the T34 running across the snow towards immobilised Germans is frequently reported. This was coupled with the Siberian units arriving on the central front with (I believe) a good number of T34s and KVs. Then the Germans took notice!

Ultimately it was the quantity of T34s produced, not the quality of each tank that determined the armour war on the eastern front. The T34 was rough and ready, which was perfect for the war in the East.

This post isn't meant to be a definitive answer on this issue. War is too random to summarise down to a few bullet points. Some T34s died hard in 1941, but the great majority were simply out maneuvered by better trained and often more numerous Germans, or they were simply abandoned by their crews.

A.E.B

[ January 30, 2003, 10:29 AM: Message edited by: A.E.B ]

--------------------

"Drag them by the gonades and their hearts and minds will follow" - Pacification for Dummies.

"The strange thing about war is that all to often there are no good guys" - A.E.B.

[JasonC]

The KV comparison was in the specific context of illustrating German "hail fire" tactics. I showed that a KV-1 that the Germans could not penetrate could still be rapidly disabled by superior numbers of Panzers, at range, via multiple hits resulting in immobilization, gun damage, and abandonment by a rattled crew. It was to show the effectiveness of such tactics even when the tank cannot be penetrated that I used a KV-1 for this. As I noted in my test, the KV did not even get off a shot. I did not say or mean that the T-34 turret front should have the effectiveness of 75-100mm of armor.

My actual proposal is that "curved" armor be treated as a high variance version of "45 degrees". Not 60, so the turret is still the vunerable point on the front of course, but also not 30 as it is now.

I gave a breakdown of "rolls" to determine the angle applied, and suggested modifications to that "roll" for crew quality and range, to simulate the ability to aim for weaker parts of the tank. I specifically intend for high quality crews at shorter ranges to be able to kill T-34s with front turret hits, with a high probability per hit. While ordinary crews at longer ranges would find it difficult, but not impossible.

My testing showed that the current behavior of 50L42 vs. the M40 model turret front is that out to 750 yards, the turret is penetrated routinely. I saw only one deflection of a turret area hit, and that was due to side angle. I saw a couple of gun hits, and a couple of partial penetrations which were still effective. Over half were complete penetrations.

Where would turret penetration become routine for a regular crew on my proposals? A clean one would be possible at 500 yards, certainly, and at that range "partials" would be common. But not every time.

The chance would rise as penetration ability does, and would be higher for 1941 (45 with 95% quality), than 1940 (100 quality), than cast turret 41s (52 with 95 quality). Against the weaker turrets (meaning not cast 52mm) with 50L42, I think my proposals would have the penetrations start becoming routine (up not from impossible, but "sometimes" at the 500m range) around 200-300m. At 100m, it would be routine even for the better cast M41 turret.

The exact probabilities at the various ranges, I haven't worked it all out. It would involve interpolating the angle effect, the range and angle value for the 50L42 there, for the dispersion of possible angles "rolled" (modified by range and crew bonuses), and for the ordinary +/- "can vary" stuff already present in every CM armor penetration outcome.

But the overall "tactical tendency" of my proposal is clear enough without working through all of the probabilities. Instead of a 50L42 hit on the turret front practically always being a kill, some would deflect, some would be partial penetrations, and some would go in. The portion going in would rise as the range falls, from 500 yards down to 100, instead of already being practically topped out at "almost always" at 500 yards or more. Therefore, closing would be strongly encouraged, and dueling hull down at medium range would be significantly riskier than it is now.

[JasonC]

I've made a bit map diagram based on the picture, with colors filled in for high deflection, middling, and good places to hit. I don't have a place to post it and I'm not very good at such things, but I can send to anyone who does and is, to post here. If anyone could do that for me quickly, please email me at -

jasoncawley1@ameritech.net

and I will send the file.

I counted the pixels, more or less (multiplying to get areas, etc). I have 25% as the portion of the front aspect taken up by the whole turret area. As portions of the whole front of the tank, the relevant areas come out to -

Gun itself (sky blue) - 1%

Turret ring and shot trap (royal blue) - 3%

Gun Mantlet (navy blue) - 7%

Middling angles (yellow) - 8%

High angles, turret "edges" (red) - 6%

"Go" portion of the turret for penetrations comes to about 40%, or 10% of the overall front aspect. Since people don't have my diagram yet, it looks a bit like a "bell curve" centered on the Mantlet (with squared off upper "corners" at the top of the bell). The "long tails" on either side are turret ring. The "big center hump" is the mantlet. The "slope" portions between the two are the shot trap areas, where low centered turret hits would be deflected downward into the upper hull.

If you estimate the size of the vunerable mantlet and shot traps as generously as possible, you might make a case for pushing it to half of the turret front. There is no way you can make it into *all* of the turret front.

CM seems to use 30-33% for the turret area of all tanks. A 30% chance of a low "angle" rolled for "curved" would give 9-10% of the overall front aspect, about the right figure.

A 40% chance (which is what I currently propose gives to vets under 500 yards - see below) of a low angle gives a generous 12-13% of the front aspect, which may cover minor disputes over the size of the vunerable mantlet or the size of the shot trap areas.

When the target is hull down, the portion of the target that is turret rises to more like half (CM may give even more than that, perhaps 60-67%). If 40-50% of the turret-hit half, hit the vunerable areas, you'd expect 20-25% of overall hits to be effective.

What I actually saw instead was 1 deflection of turret area hits due to side angle, and 3 gun hits. 1/3 of the hits were to the hull area (most of the time the exchanges were hull down, but not all). Of the 15/22 to the turret area, 3 were gun hits - high, but probably just randomly so. Of the 12 that hit the turret front proper, 1 deflected, 1 partially penetrated, and 10 fully penetrated.

So the chance of a kill with any hit on a hull down T-34 in my test was 50%. The chance of actually hitting the vunerable 40-50% of the turret front with half the tank hidden by ground should be 20-25%. The kill chance per hit was 2 to 2.5 times what it should be, if every 50L42 can be expected to go through the mantlet at those ranges (300-500m).

There is no getting around it - CMBB is modeling every shot going in, not anything about a vunerable mantlet. Even the high and outside turret edges are not showing up as randomly high angles from "curved", or randomly bad penetration "rolls".

With my proposal, which please note I said was "tweakable" and meant only to give the general idea of how to design a solution to give the right overall behavior, there would always be a 15% chance of a high angle, unless the shooter was very good or the range very close, or both. The red portions of my turret area diagram (edges, "high and outside" portions if the mantlet is a "strike zone") are about a quarter of the turret area.

If anything, comparing the turret diagram to my angle proposals, the variance in my proposal is too low, rather than too high. Arguably the high and low angle chances should be more like 30% rather than 15%, and the middling angles only about 40% (representing the "yellow" area of my diagram, roughly). I willingly revise my proposal to that, in light of the turret diagram examinations.

Also, incidentally, on the KV point, if you notice back on page 1 I explicitly said, right after giving the results of my "hail fire" test - direct quote - "Of course, a T-34 was not nearly as invunerable as the 1941 model KV I used in my test." So it is a complete mystery to me why Bastables continues to flagrantly misrepresent my actual argument.

[ January 30, 2003, 01:43 PM: Message edited by: JasonC ]

[Andreas]

Jason,

your test is very interesting, yet it still does not seem to take account of what I understand (following Rexford's postings on this matter) about hit chances.

Is it correct to say that you assume any hit should have an equal chance to hit any area that presents itself (within the confines of the BFC model)? I.e. if you only have the tank turret visible, any hit will have an equal chance to hit any point on the turret centre, or the turret edge? If that is the case, then it appears to me you ignore aiming. ISTR a statement by Rexford that the majority of hits would be within a very narrow spread left-right, because of the training to aim for centre mass. Since the vulnerable area is centre mass, a high hit percentage amongst frontal turret hits going into it would be a logical conclusion, it seems to me.

Note, I am not even talking about aiming for weak-point/area hits here. That would exacerbate the effect (especially with such a large weak point/area), and I am not sure whether CMBB does model this 'under the hood'.

[Tarquelne]

If that is the case, then it appears to me you ignore aiming. ISTR a statement by Rexford that the majority of hits would be within a very narrow spread left-right, because of the training to aim for centre mass. Since the vulnerable area is centre mass, a high hit percentage amongst frontal turret hits going into it would be a logical conclusion, it seems to me.

Won't the mantlet vulnerable area and the "centre mass" aim-point only overlap if the turret is facing directly toward the gunner? (And if it's the turret-center, rather than the tank-center, that the gunner is aiming for?) I think that would make the vulnerable area _less_ likely to be hit, since only "dueling" tanks (or perhaps dueling hull-down tanks) would be aiming for each other's mantlets.

[ January 30, 2003, 03:13 PM: Message edited by: Tarqulene ]

[Renaud]

Tarqulene:

I think that's correct. When trying to kill early (45mm welded turret front) T34's with 37mm, you want the turret to be pointed directly at the gun and within 3-400 meters (inside 300 seems best). That way your 37mm AP has a chance of impacting the turret squarely (0 angle of incidence), negating most or all of the slope effect. Of course if he's pointing at you this means you will be spotted soon and not survive much longer. But a 37mm pop-gun for a T34 is a great trade.

The same applies to 50l42 of course, although you can afford to hit much more off-axis and still penetrate.

I've also found that 50L42 fired from a slightly higher elevation at between 300 and 700 meters will regularly result in the close-to-0 angle-of-attack, negating the 45 or 30 degree slope of the hull. Also, the farther out you are the more the shell arcs downward and this seems to more than compensate for loss of velocity if, by the decreased angle of attack, you are negating the benefit of sloped but otherwise thin armor. The early T34 is heavily dependent on presenting a 30 degree or higher slope to incoming german projectiles. Negate that (there are many ways depending on situation) and the T34 suddenly becomes dramatically easier to pot.

Conversely, keep your T34 on a higher elevation than the kraut tanks, creating even steeper AOA for the 50L42 or 75l24 to defeat. Don't bother with hull down as your rounded/curved (forgot which it says) turret front is the vulnerable part from all angles of attack anyway. Present that nice expanse of well-sloped hull-front to bounce their puny AP.

Naturally this applies to early war only. After that all bets are off.

[Bastables]

Originally posted by Tarqulene:

QUOTE]Won't the mantlet vulnerable area and the "centre mass" aim-point only overlap if the turret is facing directly toward the gunner? (And if it's the turret-center, rather than the tank-center, that the gunner is aiming for?) I think that would make the vulnerable area _less_ likely to be hit, since only "dueling" tanks (or perhaps dueling hull-down tanks) would be aiming for each other's mantlets. [/QB]

And in game the present curved modifier and an oblique angle will result in non-penetrating hits. Remember the first up armouring the factories preformed was in rasing the Mantlet armour after experience in combat when the main German panzer gun was the KwK 5cm.

Good to see the new improved Jason argument allows for Turret penatrations at 500m, even maganimously rasing it above a "small chance".

[Tarquelne]

When trying to kill... you want the turret to be pointed directly at the gun

Jason: With regard to the above, and Andreas' post above (I'm wondering if I just missed the point of A's post): What was the T-34's turret orientation in your tests? If the T-34 is directly facing your test-guns then you might - should - be seeing far more penetrations than you would in "real life."

I think I'll try some 37 and 50mm guns vrs. "head on" T-34 turrets and slightly angled ones...

[ January 30, 2003, 06:18 PM: Message edited by: Tarqulene ]

[Andreas]

Originally posted by Tarqulene:

</font><blockquote>quote:</font><hr />When trying to kill... you want the turret to be pointed directly at the gun

Jason: With regard to the above, and Andreas' post above (I'm wondering if I just missed the point of A's post): What was the T-34's turret orientation in your tests? If the T-34 is directly facing your test-guns then you might - should - be seeing far more penetrations than you would in "real life."

I think I'll try some 37 and 50mm guns vrs. "head on" T-34 turrets and slightly angled ones... </font>

[Amedeo]

Just a quick note.

Was the right translation of the term actually used by Rune's grandfather eventually found?

If I remember correctly there was also someone that suggested it could be referring to the hull MG port.

Moreover, it's true that the T-34 was no Uebertank and that the soviets prudently considered 1000m or so to be the best standoff distance. But what still puzzles me is that fact that in the famous report by the 10. TD commander, frequently quoted on this forum, the rants about T-34 vulnerability stated that the weak spot were the lower hull sides (penetrated by 20mm) and the hull (penetrated by 37mm). This seems to contrast the theory of the turret front as the weakest point of the tank, or at least a weak pont that could be frequently exploited.

Regards,

Amedeo

[JasonC]

"from a slightly higher elevation at between 300 and 700 meters will regularly result in the close-to-0 angle-of-attack, negating the 45 or 30 degree slope of the hull."

LOL. Let's do a little elementary trig, ok? Say the shooter is 4 CM elevation levels above the target. That puts the rise over run at 300 to 700 meters at 1 in 30 to 1 in 70. The arcsin of 1/30 is 1.91 degrees. The ArcSin of 1/70 is 0.82 degrees. The angle of incidence changes only (actually, less than) 1-2 degrees for the elevated shooter, compared to the fall of shot that occurs with one firing from the same elevation as the target.

And the slope of the T-34 hull is 60 degrees, not 30 or 45. That was the entire bleeding point of the design, the thing about it that was revolutionary. But suppose you wanted to negate an angle of 30 degrees, or to reduce an angle of 60 to one of 30 by being higher than a target 300 to 700 meters away.

A 30 degree angle change in the direction of impact requires a rise over run of 1/2, not 1/30 or 1/70. You'd need to be 150 to 350 meters higher than the target - ~500 to 1000 feet in the air. Of course, you couldn't possibly depress the gun that much, unless your tank were itself pointing downward at a steep angle, but that is another story.

"But the shell descends with some curve in the arc" of course. It does with a shot from the same elevation, too. At 300 to 700 meters, however, that is extremely slight, because with any of the typical tank guns involved, the shell has been in flight less than 2 seconds, for the higher MV guns about a second.

Half that time it spent ascending slightly to its mid-course apex. At 700 meters with the lowest velocity gun likely to be involved, the 75L24, the shell might have been falling for all of 0.8 seconds. Which means a drop of around 3 meters in 350 meters, or about half a degree down, average, for the second half of the trajectory. It is zero at the start of it so it might be twice that at the end - still 1 degree. With the faster MV guns like the Czech 37, the time of flight will be half that, and the result even smaller.

Close range shots by tank guns are very flat trajectory affairs. The size of elevation changes involved in tank duels are also very flat trajectory affairs. They might alter the angle of slope experienced by the impacting shell by 1-3 degrees, in combination, but there is no way they can add up to angle changes 10-20 times that size. They therefore have a neglible effect on penetration questions. Certainly well within what CM already allows to random "can vary" determinations, and swamped by minor deviations of side angle and the like.

As for my tests, they did not feature head on shooting alone. They featured realistic dueling between platoons of tanks, some facing each other, some shooting at a tank facing a different opponent and so with side angle to the turret, etc. I saw all of 1 turret ricochet, which occurred with a substantial side angle, between 15 and 30 degrees. One Pz III was hunting forward toward the flank of the T-34 in question, having lost LOS to it over the curve of a hill, while it was dueling a different Pz III. LOS was re-established at 350-400m, off 15-30 degrees to the T-34's right.

As for the aim issue, the things are not lasers. Is the gunner's sight exactly centered on center of mass? No, certainly not always. Is there shot dispersion from the gun itself? Of course. He isn't aiming for the turret anyway - center of mass is the upper front hull, and turret hits occur because of already existing dispersion around the aim point.

When there is any side angle, the hull appears even larger than in the head on case (as some of the side appears, though at first at an extreme angle to the shot), while the turret, if turned to face the shooter as in a duel, remains the same apparent size. The center of mass also moves "backward" toward the exposed side of the tank, thus off to one side of the upper front hull, as the side angle to the hull increases. Which puts the location directly above the center of mass aiming point off to one side of the turret, not down its center-line.

All of these effects are smaller than the known existing rough approximation of turret size as 1/3rd of an exposed, and more like 2/3rds of a hull down vehicle. That actual figures for a T-34 are 1/4 (more like 1/5 with any side angle to the hull) and at most 1/2. CM players are already getting far more turret hits than they deserve. Pretending they all magically hit the mantlet as well is compound fantasy.

The idea that turret hits always hit the center of the turret is just silly, given all of these variations. You might well say that all turret area hits should be gun hits, because the gun is in the center of the turret.

There is no significant variation in CMBB behavior when the turret front is hit in duels, as though 1/3rd (or 1/4, if you want to center weight it somewhat) of hits are being modeled as occuring near the edges of the turret, 1/3rd in my "yellow" areas of middling angle, and 1/3rd (or 4/9ths if you want to center-weight it) to the flatter portions. Instead, virtually all of them go in.

28mm penetrates the turret front routinely at 350 yards. Czech 37 penetrates routinely at 420 yards. The Pz 35t 37mm, not the better L48, still penetrates at medium range, though less often and typically partially. A head on duel, hull down at 500m, between Pz 38ts and T-34s is an even affair. A battery of 28 PAK hidden in trees can destroy a platoon of T-34s from the front, with an ambush sprung as far as 500 yards to make return exact spotting of the small caliber PAK difficult.

The historical participants speak of bringing up 88 FLAK and 105 howitzers to fire direct, or closing to point blank range from the flanks using Pz Gr 40 from 50L42s, or using hail fire to drive them away even without penetrating them. Something is very wrong with this picture. Way too wrong for quibbles about minutae like 1-3 degree deviations and slight center-weighting of aim points to make it go away.

In the real historical fighting, the typical angle produced by "curved" on the T-34 turret front was often well above 30 degrees. Which should surprise no one who so much as looks at the tank.

Would some portion of shots occasionally get flatter hits? No doubt. But all of them under 30 degrees is just flat wrong. 60-70% of them should be significantly higher than that, and about as many should result in extreme angles as in small ones.

CMBB simple got "curved" wrong, undermodeling its typical effect in the case of the T-34. The variance is too low and so is the typical angle. That is what has opened the wide gap between historical accounts and tactics, and CMBB experiences and the tactics they allow and encourage.

Incidentally, it is also possible that CMBB armor models are too generous to small high velocity rounds, and that the weighting given to the shell diameter to plate thickness factor is too small. Because the 75L24 seems to have been thought a better performer than guns like the better 37s and the 28 PAK, while in CMBB they perform about the same at 500m, and the higher MV small guns outperform it at closer ranges. I've never seen a report of "busting the weld seams" with 28/20 APCR.

[ January 31, 2003, 11:41 AM: Message edited by: JasonC ]

[Andreas]

Originally posted by JasonC:

The idea that turret hits always hit the center of the turret is ludicruous, given all of these variations. You might well say that all turret area hits should be gun hits, because the gun is in the center of the turret.

For all your verbosity, you did not get my point. Reread my post before making an ass of yourself.

[JasonC]

I understand your point completely. You want some tiny variance tight bell curve of probable dispersion centered on the gun mantlet, to boost the probability of flat hits and reduce the possibility of turret edge hits. Which is simply grasping at straws trying to save an indefensible CMBB model in which *everything* hits 30 degrees or less and goes in.

You might shade the weightings slightly, though side angle to the hull would shade them the other way, which is probably a wash overall. But to put the entire variation within the size of the mantlet is crazy when half of the shots flat miss. The dispersion pattern is much bigger than the whole target tank, which means you are in the center of the bell already. The low tails of the distribution of shot are misses, not "high and outside" hits.

[Andreas]

Originally posted by JasonC:

I understand your point completely. You want some tiny variance tight bell curve of probable dispersion centered on the gun mantlet, to boost the probability of flat hits and reduce the possibility of turret edge hits. Which is simply grasping at straws trying to save an indefensible CMBB model in which *everything* hits 30 degrees or less and goes in.

No you did not get my point. Thanks for making that clear.

[Tarquelne]

Ok, ignoring the last few posts...

50L60 ATG vrs. a dug in 45mm T-34 in various orientations at 700m:

It seemed to boil down to: Hit on turret front = penetration, hit anywhere else = deflection. I think there was only 1 hit to the front turret that didn't result in a penetration.

50L42 PzIII vrs. same T-34:

Same as above, except that turret front hits that didn't cause a penetration were merely "very unusual" rather than "quite rare." It seemed that only within, say, at most 15 degrees, was there much of a chance for a deflection.

Curved seems to mean "It'll bounce a shot if you're very lucky, but you should think of it as flat."

[JasonC]

Andreas thinks it is clever to pretend mysteriousness. If you think I haven't understood something, you should say what it is.

Your post alleged two things. One, that I was only testing with the turret directly facing the shooter. Which was completely false, I was not only testing with the turret directly facing the shooter. I dismissed that portion on that basis - it was a statement made under a false impression of the conditions of my test.

Two, that if the turret was facing the shooter, then aim point considerations should enter, and make hits near the center-line more common than hits farther away from the center-line. If you can't tell that that means a "tight distribution centered on the gun mantlet", then you just can't understand what I am saying.

And I gave two reasons why it is wrong. First, because even with the turret facing the shooter directly (as in a head to head duel), you cannot assume the aim point is directly on the turret center-line. If there is any side angle to the hull, center of mass is not on the center line, but off to the side toward the exposed hull side.

The aim point is of course directed at the hull, which becomes larger as the side angle increases. Thus, the likeliest portion of the turret (the vertical "highest probability line", if you follow), moves across the turret as the side angle to the hull changes. It does not remain fixed on the centerline and mantlet.

Second, only hits that are already away from the aim point hit the turret at all. Only shots that are close to the center of the overall distribution of expected shots, hit the whole tank at all, anywhere - the long low "tails" of the distribution of shot, in probability terms, are misses not hits.

If it is not obvious, the top-center of gaussian distributions are pretty well flat. If the overall hit chance is under 2/3rds, everything more than a standard deviation out is already gone, "misses". As the hit chance falls, the portion of the overall distribution that remains gets to be a tighter and tighter portion of the top of the bell. Which means, the distribution of shots within the remaining sample, is getting closer and closer to linear-flat-even.

With varying side angle, where this flatish probability curve is centered shifts slightly from side to side across the turret front. As an average over many engagements at different angles, you have to sum all of these distributions into one overall one. It will be flatter again than a single distribution on the center-line.

Aiming considerations are therefore not going to give you a huge difference in probability between a point on the center line of the turret, and one at the edge. There may be some deviation from equality, but it will be small. I called that "center-weighting" the hits. I allowed that a little of it might up the center portions from 40 to 50, or reduce the side ones from a third to a fifth.

What no such explanation can do, is make *every* hit turn into "right on the mantlet". And that is what actually happens in CMBB. It is not a difference in shading probability distributions. It is "they all go in, not a portion". A tiny number, under 10%, may deflect due to "curved" or side angle, or both in combination.

Which is just plain wrong. I wish somebody - any one honest person on the other side of this debate - would admit as much. Bastables is happy that I agree some hits might go in at range, when I proposed that in my dispersion of "curved" angles back on page 2. But I haven't heard him - or Andreas, or anyone else on the other side of the debate - admit that any of them might *not*. So far, there is complete silence on that score.

Meanwhile the test results that in CMBB they virtually all go in, accumulate. I've taken out platoons of T-34s with a battery of 28mm PAK. 50mm PAK should do that, and do - but the 28mm? I've taken out a platoon of them with Pz 38ts; others have reported regular penetrations by them out to 580m. I've gotten occasional kills on them now with 35ts and Pz IIIFs.

Will anyone admit any of this is a problem, or is the blind stonewalling permanent as well as complete? Is there any point in arguing with people who will not admit anything wrong with any of the above? I put it to Andreas and Bastables directly and by name; please answer the following direct question.

In your opinion, is CMBB modeling of the T-34 turret front *perfect*? If not, where is it wrong? If so, please produce historical reports of 28 PAK and Czech 37s killing T-34s from the front at 500 yards.

[ January 31, 2003, 01:25 PM: Message edited by: JasonC ]

[Jarmo]

Originally posted by JasonC:

In your opinion, is CMBB modeling of the T-34 turret front *perfect*? If not, where is it wrong? If so, please produce historical reports of 28 PAK and Czech 37s killing T-34s from the front at 500 yards.

Wrong question to ask. Nobody's going to call CM "perfect". Just as close as can be hoped, ie not easily correctable.

I at least agree there's a huge* problem in representing all turret shapes as having just one angle. Or being "Rounded", which only goes a tiny step towards fixing the modelling limitation. It doesnt seem to have nearly as large effect as the turret shapes of many tanks would demand.

* huge, but only for anal grogs, insignificant to normal people.

[Andreas]

Originally posted by Tarqulene:

50L42 PzIII vrs. same T-34:

Same as above, except that turret front hits that didn't cause a penetration were merely "very unusual" rather than "quite rare." It seemed that only within, say, at most 15 degrees, was there much of a chance for a deflection.

Curved seems to mean "It'll bounce a shot if you're very lucky, but you should think of it as flat."

Err, nice bit of testing, and I just love the precision in the conclusions.

I just ran the following tests, each with 6 III 50L42 facing 6 T34obr40 in shooting ranges, head-on.

Range:

740-750m (5 tests)

520m (1 test)

1020m (1 test)

Results:

740-750m

Turret penetration ~60%

Ricochet ~21%

Partial penetration ~10%

Gun hits ~10%

After max 3 turns, the duels were over. Total results (kills for 4 tests only): 9 III dead, 2 III shocked, 8 T34 dead

1020m

Penetration 27%

Ricochet 13%

Partial penetrations 53%

Gun hits 7%

3 turns, 4 PIII, 2 T34 dead

520m

Penetrations 84%

Ricochets 11%

Partial 5%

Gun 0

2 PIII and 4 T34 killed, 2 turns.

Interestingly, first round shots rarely hit for the PIII at any range. From the second volley onwards, hits were following the general pattern. This to me indicates that aiming may indeed be simulated to some degree.

I have all three files, interested parties can DL them here:

50l42 test

Happy testing.

I am still waiting for Jason to answer the question by the way. Does his model take horizontal dispersion into account? If not, instead of droning on about how I am clutching at straws, by how much would it increase the penetratable area?

Edit: T34s dug in, to prevent them from bugging out and preserve range integrity. This of course leads to more turret hits than you would otherwise get.

Also, I ignored the 50L60 in testing. The 50L60 has no trouble getting through the T34 and is not the issue here.

[ January 31, 2003, 02:37 PM: Message edited by: Andreas ]

[Andreas]

Originally posted by JasonC:

In your opinion, is CMBB modeling of the T-34 turret front *perfect*? If not, where is it wrong? If so, please produce historical reports of 28 PAK and Czech 37s killing T-34s from the front at 500 yards.

No, it is not perfect, but as you well know, that is not the issue, unless you want to pretend that your suggested model makes it *perfect*.

Regarding the sPzB41, which you mistakenly refer to as 28 PAK, Etterlin gives it a penetration of 52mm @ 60° @457m. Sounds like it should get through curved 45mm at 500m with little problem to me. CMBB does disagree with this value though, and has a lot lower penetration. Etterlin's judgement is 'at the time of introduction 1941 a splendid (ausgezeichnete) gun'. I leave it up to you to interpret this judgement. I look forward to your creative efforts.

I think I have shown or at least given a good indication that your 'tiny number deflection' 10% claims are wrong at the range where it matters.

Regarding 'only hits that are already away', again, you are messing up lateral and horizontal dispersion. IIMU that lateral dispersion at this range does not matter. You will hit something. Horizontal dispersion does matter. Now, I am still at a loss why you find it so difficult to accept that German gunners will aim at a point that they can penetrate, instead of aiming at a point that they can not penetrate. Please elucidate.

Edit: because I got my numbers wrong on the deflections.

[ January 31, 2003, 03:05 PM: Message edited by: Andreas ]

[Tarquelne]

Err, nice bit of testing, and I just love the precision in the conclusions.

My mother always said "Drink your moo juice, and if you're not willing to perform enough tests to eliminate reasonable doubts about sample sizes, don't give any figures at all, sweetie."

Results:

740-750m

JHC - I tried the above test and the first time through three T-34s scored first shot kills.

Speaking of statistical weirdness: 50-100% more ricochets at 750m compared to 520m or 1020m... is that an expected reflection of "real life" behavior, or should Andreas have another glass of milk?

[ January 31, 2003, 02:46 PM: Message edited by: Tarqulene ]

[Andreas]

Originally posted by Tarqulene:

</font><blockquote>quote:</font><hr />Err, nice bit of testing, and I just love the precision in the conclusions.

My mother always said "Drink your moo juice, and if you're not willing to perform enough tests to eliminate reasonable doubts about sample sizes, don't give any figures at all, sweetie." </font>

[Tarquelne]

Draw your own conclusions about reliability

Well, "poor" of course, until someone(else) runs the numbers or Treeburst does hundreds of trials.

or better yet, perform more tests and let us know the results.

That's just what I've been doing, since no one stomped on the test I outlined.

Jason, counting my tests, ricochets are from around 5-30%, and seem to be clustering above 20%. (I had ~5%, ~16%, and 25% over all ranges) How's that?