GearHeads Forum

Hi Bill:

I have just uploaded version 1.16B which fixes the problem with the bracket and many other weirdnesses.

For the bracket, make sure you unlock the shafts for the bracket. If the crown rotates, it tries to rotate the brack on the crowns plane
which makes it blow apart.

Yes, fusing to the bench isn't in the mtass spec, as the mtass is meant to be fused to whatever a user is building. Version 1.16 seems the best
reacting so far, most of the weirdness is gone from testing, I still am searching for a couple of gremlins.
I also wondered why your shafts are named as numbers, till I found adjusting the shaft min or max was changing the name
of the gear or shaft. This is now fixed as well.

I know you'll be offline for a bit, so hopefully Ill have more additions by the time you return.Thanks for the
aid in debugging, and for the crown escapement. (I really like that crown, it shows how reactive those strange shapes can be. )

Soon
Art

Art
I decided to tweak the escapement one more time before going offline for a while.

I've simplified the experimental escapement and have attached a zip file with the model... it is now version-4. I also liked your idea of creating a functional mtass file with a working escapement. I've created an mtass from version-4 model and have now named it Crown-Verge Escapement V0.

When I first imported the new mtass file, the escapement blew apart. After looking at various settings I realized the the mechanical asset did not fuse the pallet shaft to the workbench. Once I fused the shaft to the workbench the model did run.

BillM

Hi Bill:

I opened these and they ran terrible. I was cursing myself , till I realized I haven't updated my desktop Ticker, so I ran
my compilers' version and it ran great. A really nice example of a crown verge. The bracket example is a perfect example of
a bug I must have introduced in my changes. Ill have to add something similar to my test collection that I use as a final test.

I will create likely this weekend a new topic for "Ticker - Escapements", where things such as this escapement can be shown
and discussed. Any working types, such as this crown model will be added to Ticker's release so that people dont have
to navigate a forum to find them.

Ill let you know what I finds with the bracket and why it seems so unstable. Seems designed properly, so it
should do fine.

Art

Art

Good Job! I loaded V1.15B. The crown-verge escapement is now stable. Ran for a long time without the shaft becoming unstable.

I've included an updated version of the escapement. The project is saved at a half-speed time scale. I've modified the pallets and have been experimenting with the phase angle between the two pallets. Keep in mind I'm a retired electrical engineer and not a clock maker. I've made some reverse engineering guesses based on photos and videos of such an escapement to build the model.

At this point I think the "conversational" physics look correct. At slower speeds ( ~.1) you can see the recoil of the crown gear when it hits a pallet. I played around with the mass of the balance wheel and the operation of the escapement did change.

The second attachment is the crown wheel supported in the vertical position by a right angle bracket. This one shows instability of both the bracket and the shaft that the crown wheel is attached to. My thoughts behind the bracket is that components in a design might be mounted off a bracket rather than simply floating in space. Perhaps this type of mounting should not be bothered with.

Do you think I should create a new topic about this escapement? Perhaps some other member might want to experiment further with this third version of the crown-verge escapement.

BillM

Bill:

V 1.15b is online. Its changes may seem subtle, but the orientation and position locks now apply a force to correct and are
constraints of their own. This at least means the total force in the physical system is as expected and therefore spreads properly
to other objects causing a higher level of stability and fewer "weird" response variations. I see see some errors in some of my tests so I still have work to do, but they appear mostly object specific, ( the bevels have a strange torque on them Im looking into for example).

Playing with this escapement of yours in the new system , I can see its really close to doing as its supposed to. That crown and lever
I think will end up as an MTass in the release. I really like the various ways of making a physical 2-5 piece mtass file that is an escapement
that stands on its own as a device. Thanks for the idea's and the items, I suspect they'll end up in a release and video one day. I'll name
any MTass'es from them as "BillM Escapement Devices". Ticker could end up being a nice clock designing tool with a lot of work
and additional math tools.

Art

Thanks Bill:

Ill investigate that. While a shaft will not collide with its master, I have seen where an attached gear can collide with its mate shaft
in certain situations. Im hoping to see some improvement as I change the "lock Position" and "Lock Orientation" code.
In the past, these were done by setting the angular or linear factors to zero. (This means all forces, angular or linear, were
zeroed on an object every frame.). I never liked it as it violates physics, all natural forces should be used really, so I'm recoding
so that if locked in position , for example, the forces are not zeroed, but symmetrically opposite forces are added instead.
This allows for these forces to be propagated properly through the model, Im hoping this will improve things markedly so that perhaps its not necessary to put a gear on a shaft to stop an object from moving inappropriately. It may make a "position locked" object more susceptible to motion from immovable object meets unstoppable force type situations, but Ill try it as I think it is more accurate to nature and
perhaps more stable. We'll see. :)

Art

Art

I've attached a copy of the crown gear asset in case you'd like to play with it. The one in the model that I sent previously had the crown gear fused to the bevel gear in an attempt to stabilize the pair.

The shaft through the Crown and the two bevel gears does seem to wander as the simulation runs which raises havoc with the mating of the pallets. It's almost as if the shaft and bevel gears are colliding with each other.

In some of my other experiments, I've found similar interaction between the shaft and its object. Turning off shaft collisions sometimes corrects the problem.

BIllM

Hi Bill:

Ahh.. I see where your going there. That works quite well as a start. I think Ill add a button to export an object as an asset
when loaded within a project so I can reuse things like that great crown gear you made.
Yes, adding a gear to a shaft suddenly makes the orientation work properly as a workaround.. I'm working on a more permanent solution.
I have already modified the gear orientation and position locks, next I have to do the objects in the same way.
Ill let you know when its ready, Im investigating a few other oddities I dont like in there.

Art

Art

I started to play with the vertical crown gear but created a bracket to mount the crown in a vertical position. That attempt did not work too well, the secondary shaft mounted to the bracket was not stable when the crown gear was mounted on it. I then read your comment about bevel gear and that gave me a work around for the craziness of the horizontal axis of the crown gear.

The attached zip file uses bevel gears as mounting points for a verge-like construction.

The pallets need some work probably are too thick and positioning is still tricky. Movement is a bit jerky but not bad for a first attempt.

There was some rotational movement of the crown relative to the horizontal bevel's shaft so I fused the two together. I added a bevel gear set to the end of that horizontal shaft in an attempt to provide some stability. I notice that the horizontal shaft starts to move around on the right hand bevel gear set. That motion causes trouble between the crown and the pallets. This is definitely a work in progress.



BillM

Hi Bill:

First..awesome crown. That is built very well. The problem isn't the crown itself. I noticed you have a shaft inside a shaft where they
join to the vertical shaft. (if selected you can see there are two there.) If I erase that shaft the physics start to work more normally
but a swinging motion gets created.

I do see the real trouble is the orientation locking and position locking. They are causing havoc because of the way they
are implemented. It seems Orientation lock doesn't like anything off of the Z-- axis. Rotating the crown to the Y-- axis makes
the orientation lose its mind. I will recode this to a new constraint type as Im not happy with the way those locks
operate. The crown itself seems fine, if a bevel were used to get that 90 degree change to the Y it likely would work good,
but Ill let you know when I manage to fix the code on position lock and orientation lock.

Art