Singularity

Started by mowens, April 13, 2021, 11:54 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Go Down Pages 1 2 3

Tim Johnson

#15
April 14, 2021, 05:20 AM Last Edit: December 31, 1969, 04:00 PM by Guest
Quote from: Newbeeee™ post_id=9572 time=1618397935 user_id=157Mike
Out of interest...is this a MTB "fault", because they never had logic to (somehow/PLC/Parameter?) avoid the situation?
Or is this now a post builder/kinematic issue which needs building into your CAM system, so you cannot 'gram a part using the axis at those limits?
And who's footing the bill...


This is what I was leading to. If there were previous meltdowns then the software should have been adjusted to prevent the issue or at least go into e-stop.
FJB

gcode

#16
April 14, 2021, 07:12 AM Last Edit: December 31, 1969, 04:00 PM by Guest
I believe Mike worked hand in glove with CGtech developing a VCM for this machine.
It shouldn't be too hard for the guys at CGTech to add tests that warn you when these conditions
are present.

TylerBeer

#17
April 14, 2021, 07:20 AM Last Edit: December 31, 1969, 04:00 PM by Guest
I've added singularity checks into posts because machines leave bad surface finish near it due to rapid accel/decel or they fault when they hit it (all much like a robot) but I've never heard a crash because of it unless it swings the wrong way into a fixture, that sucks

gcode

#18
April 14, 2021, 07:33 AM Last Edit: December 31, 1969, 04:00 PM by Guest
Quote from: TylerBeer post_id=9585 time=1618410007 user_id=116but I've never heard a crash because of it unless it swings the wrong way into a fixture, that sucks


agreed, this is a fault with machine parameters IMO
I have never seen a machine that doesn't alarm out if told to exceed it's travel limits.

mowens

#19
April 14, 2021, 07:41 AM Last Edit: April 14, 2021, 07:53 AM by mowens
Quote from: Newbeeee™ post_id=9572 time=1618397935 user_id=157Mike
Out of interest...is this a MTB "fault", because they never had logic to (somehow/PLC/Parameter?) avoid the situation?
Or is this now a post builder/kinematic issue which needs building into your CAM system, so you cannot 'gram a part using the axis at those limits?
And who's footing the bill...

I think it is both. The machine builder is not one the OEMS you normally think of. Their background is in assembly automation; automatic riveting machines, etc. This machine was one of the first they built. I mean, it's an amazing machine.
CGTech wrote our programming software. This is their first foray into the programming side. We were testing a new option they had given us, which caused us to run in a different configuration than normal.

The way I see it we had three opportunities to catch this. Machine, software and programmer. I do think that if this is a well known issue they should have put limits on B axis to keep A and C from approaching parallel.

There is lots of activity going on. I don't think payment  has been decided yet.

It did alarm out but had too much momentum to stop.
"I would gladly risk feeling bad at times if it also meant that I could taste my dessert." - Data

neurosis

#20
April 14, 2021, 07:49 AM Last Edit: December 31, 1969, 04:00 PM by Guest
This sounds like the sort of crash that needs a video.
I'll go back to being a conservative, when conservatives go back to being conservative.

mowens

#21
April 14, 2021, 07:54 AM Last Edit: December 31, 1969, 04:00 PM by Guest
It would have been pretty spectacular with all of those ball bearings spewing out.
"I would gladly risk feeling bad at times if it also meant that I could taste my dessert." - Data

gcode

#22
April 14, 2021, 08:00 AM Last Edit: December 31, 1969, 04:00 PM by Guest
With a crash of this magnitude, the machine may not be repairable.
You may get it back together to discover it does not perform like it used too because the frame is bent
or it's structure is knocked out of alignment.
On the plus side, I believe this machine is used to wrap carbon fiber sheets on aircraft structures and
± .0001 accuracy is not required.

mowens

#23
April 14, 2021, 08:44 AM Last Edit: December 31, 1969, 04:00 PM by Guest
You are correct about the tolerances. Maintenance has the A axis torn apart so we should know before too long the extent of the damage.
"I would gladly risk feeling bad at times if it also meant that I could taste my dessert." - Data

gcode

#24
April 14, 2021, 08:46 AM Last Edit: December 31, 1969, 04:00 PM by Guest
QuoteMaintenance has the A axis torn apart

an easy job as most of it was already self disassembled  :cry:

Matthew Hajicek

#25
April 14, 2021, 09:04 AM Last Edit: December 31, 1969, 04:00 PM by Guest
The Apollo crews had to contend with singularities on their navigation gyroscopes.  They called the phenomenon "gimbal lock".

TylerBeer

#26
April 14, 2021, 09:10 AM Last Edit: December 31, 1969, 04:00 PM by Guest
Quote from: mowens post_id=9566 time=1618375210 user_id=74
QuoteI smiled at him and Said. "I Know a Guy who just blew up a super expensive Machine"


Glad I could cheer him up!

Yes this machine was somewhere around 8 to 10 million when it was installed in 2009.


holy spamoly  .. too bad you didn't just need, uh, squares:

">

mowens

#27
April 14, 2021, 11:30 AM Last Edit: December 31, 1969, 04:00 PM by Guest
I talked some more with the operator. I guess I was mistaken. It didn't give us an over travel alarm. It did, however, give us an excessive speed error of some kind.
"I would gladly risk feeling bad at times if it also meant that I could taste my dessert." - Data

gcode

#28
April 14, 2021, 11:39 AM Last Edit: December 31, 1969, 04:00 PM by Guest
QuoteIt did, however, give us an excessive speed error of some kind.


that was the poor machine screaming in terror as it blasted through the hard stops  :shock:

Do you have more than one of these machines now, or are you dead in the water right now ??

TylerBeer

#29
April 14, 2021, 11:41 AM Last Edit: December 31, 1969, 04:00 PM by Guest
Which usually happens close to singularity, unfortunately all inverse kinematics breaks down when you have two possible solutions, so one joint could move "instantly" while the other doesn't, buuuuut since it's not at 0 yet, it tries to get to 'instant' speed, and yavvo. I've seen many a robot cable broken this way, but not a machine crash like that
Go Up Pages 1 2 3