Axle Length vs Axle Stopper

[Gerald Lasser]

I am starting the thread based on the discussion that evolved here on the PT concerning part 72008, which has no through holes, but an axle stopper at the centre.

There are some other parts that have a stopper, but no matter how small it is, this is currently incompüatible with our axel lengths, which are multiples of 20 LDUs.

Here is Magnus' statment:
Magnus Forsberg [MagFors] made the following comment. 2024-01-10 16:53:49
This is perhaps a good opportunity to discuss the error in all similar parts.
The hole needs to be exactly 20 ldu deep, but in most cases it has been modeled with a divider at 19.5, 1 ldu thick.
That's not correct.

All Technic Axle are made using the axle.dat primitive. Or something more suitable, when needed.
That primitive is not correct. A real axle has a slightly bevelled, convex end. Making it possible to have the 2 axles pushed all the way into the stopper ring, wich has a coresponding concave bevelled design

The axlehole primitives in this part need to be 20 ldu deep, to allow a correct position of a mounted axle.
But we need to also have a stopper, 1 ldu thick. It doesn't add up.

When I reworked the axlehole primitive I found that many parts use a primitive that is only 19.5 ldu.
This will be a problem in connection seeking SW, that will not allow a correct position of the axle. The hole isn't deep enough.

And Max' measurement of the Axles in real life:
Max Martin Richter [MMR1988] made the following comment. 2024-01-10 18:24:00
All axles I measure are shorter than our models.
2L = 15.7 mm
3L = 23.4 mm
4L = 31.7 mm
5L = 39.4 mm
6L = 47.7 mm
7L = 55.5 mm
8L = 63.5 mm
10L = 79.4 mm
12L = 95.2 mm

[Magnus Forsberg]

I made an image of the problem.

   

[Max Murtazin]

Do we really need to go this way? It doesn't affect pretty much anything important about the models - they are meant to be used for digital use, not for printing or anything like that, and small clipping with a stopper doesn't hurt pretty much anything

[Orion Pobursky]

I think I agree with Max. I don't see any compelling reason to alter the axle for the sake of an unseen overlap. I'm open to change my mind with a good counter argument.

[Gerald Lasser]

Will Stud.io report an overlap?

[N. W. Perry]

Stoppers aside, there is also the question of whether we should more closely model the actual shape of an axle. The bevelled end is noticeably different than our flat version, probably more so than other less-subtle features that we already model.

And how thick is the stopper in real life? Has anyone been able to reliably measure it?

[Max Murtazin]

Will Stud.io report an overlap?

No, but Studio's collision detector is extremely flawed.

[Max Murtazin]

Stoppers aside, there is also the question of whether we should more closely model the actual shape of an axle. The bevelled end is noticeably different than our flat version, probably more so than other less-subtle features that we already model.

And how thick is the stopper in real life? Has anyone been able to reliably measure it?

Bevelled end would be a pleasing addition to see, but it's not the best move for performance

[Travis Cobbs]

Bevelled end would be a pleasing addition to see, but it's not the best move for performance

Ignoring the stoppers entirely, I feel that LDraw axles look wrong with their flat ends. Not having grooves in 2-long axles also very much looks wrong to me. Whether or not either is worth fixing is of course open for debate.

[Orion Pobursky]

I'm ok with rounding the ends of the axles as this won't require obsoleting anything.

[Philippe Hurbain]

Ignoring the stoppers entirely, I feel that LDraw axles look wrong with their flat ends. Not having grooves in 2-long axles also very much looks wrong to me. Whether or not either is worth fixing is of course open for debate.

You mean the pull out notches? they are there: https://library.ldraw.org/official/5541 (the notchless older version exists too https://library.ldraw.org/official/6761)

[Philippe Hurbain]

I'm ok with rounding the ends of the axles as this won't require obsoleting anything.

One problem is that the rounding is not scalable, so we need a scalable axle body primitive AND a non-scalable rounded end, introducing some more complexity/vertices. That said, I find too that some simple bevel would greatly improve axle look...

[Max Martin Richter]

Whats about making a 19 LDU (or whatever is the correct length) primitive with rounded end on one side and another scalable 1 LDU long primitive without ends?
Combining just 2 of the prims with rounding end will result in a 2L axle and the use of the scalable prim in the middle will give us the opportunity to create all other lengths of axles.
With the short one we can also model other parts like short axle pins, bricks with axle etc.
Maybe we can even do an even short version with rounded end, which can be used for the 2L notched axle and in the 19 LDU long prim.

BTW what will happen, if we create the rounded end prim in die \48 folder and another one without rounding in the prim folder? (Yes, it's a misuse of the \48 folder, but this will result in faster rendering, when "48-substitution" is not activated.

[Gerald Lasser]

One problem is that the rounding is not scalable, so we need a scalable axle body primitive AND a non-scalable rounded end, introducing some more complexity/vertices. That said, I find too that some simple bevel would greatly improve axle look...

There is an axlecap primitive which looks like the beveled end



I also like the proposal of max

Whats about making a 19 LDU (or whatever is the correct length) primitive with rounded end on one side and another scalable 1 LDU long primitive without ends?

[Willy Tschager]

I am starting the thread based on the discussion that evolved here on the PT concerning part 72008, which has no through holes, but an axle stopper at the centre.

Short answer: who cares. Long answer: who cares.

w.

[Max Murtazin]

BTW what will happen, if we create the rounded end prim in die \48 folder and another one without rounding in the prim folder? (Yes, it's a misuse of the \48 folder, but this will result in faster rendering, when "48-substitution" is not activated.

I think, if rounded caps is the way we will go, it's better to make this something up to let software developer handle on their own, similar to stud primitives, which have a range of use options

[Magnus Forsberg]

I think, if rounded caps is the way we will go, it's better to make this something up to let software developer handle on their own, similar to stud primitives, which have a range of use options

Not possible, since it require a 1 to 1 replacement switch.
One stud design is replaced by another design, prefered by the user.

All axles are made using only one primitive. 
To be correct it needs to be replaced by two endcaps + a scaled axle prim with open ends.

[Max Murtazin]

Not possible, since it require a 1 to 1 replacement switch.
One stud design is replaced by another design, prefered by the user.

All axles are made using only one primitive. 
To be correct it needs to be replaced by two endcaps + a scaled axle prim with open ends.

Simplified axle cap could be a file officially included in LDraw library. And the software can choose to use it instead of beveled cap

[Magnus Forsberg]

There is an axlecap primitive which looks like the beveled end

This prim will not allow a design of what is seen inside these axleholes.

Are you all suggesting that we ignore the inner stopper ring design, inside these parts, and only redesign the axle ends?

I guess all axles need to be changed, and I don't want to do that.
Not only the regular axles using axle.dat, but all parts with an axle end, would need editing.

What I see in LDCad, and perhaps in other connection seeking SW, is that the axle hole is is only 19.5 ldu deep, perhaps due to it being linked to a scaled primitive, instead of it to always have a depth of 20 ldu.

Why does the axle parts sometime snap into a bad position?

[Roland Melkert]

What I see in LDCad, and perhaps in other connection seeking SW, is that the axle hole is is only 19.5 ldu deep, perhaps due to it being linked to a scaled primitive, instead of it to always have a depth of 20 ldu.

It is in LDCad both problem part use a scaled axlehol4.dat
Can be corrected by dropping the id=axleHole connections in higher shadow dat's though.

Why does the axle parts sometime snap into a bad position?

It will only go up to the defined space, it might also be limited by the current slide grid (4th grid value)

[Gerald Lasser]

I just pointed to existing prims that have the form, we could merge this with a 19 or 19.5 LDU long prim that represents the end of an axl (like Max suggested)

The seperator we have in 6538b is .5 LDU wide

Here's a picture adding the endcaps scaled at x=75%

   

[Max Martin Richter]

I just pointed to existing prims that have the form, we could merge this with a 19 or 19.5 LDU long prim that represents the end of an axl (like Max suggested)

The seperator we have in 6538b is .5 LDU wide

Here's a picture adding the endcaps scaled at x=75%

I just made a measurement for 6538b. Total part length: 15.95 mm.
Insert length per one side: 7.8 mm. Which means, that with an idealized part length of 16 mm the stopper is 0.4 mm thick in total.
I have not checked how many parts are using the axle prim. But at least the plain axles could be changed quickly to a design using 3 prims. And parts with an 1L axle stub could be done as well, if we decide to make a 19 or 19.5 LDU prim with beveled end on one an open end on the other side.

Are we sure, that there is a special design for the stopper ring? I can cut on joiner, if we want do see this exactly. I cut an 6538b into a quarter. It is just a plain ring without any beveled  edges or anything else.

/Max

[Max Martin Richter]

As stated in Billund, I'll rework all the axles, when I finished the 9V and PF train tracks.