Idea for high speed machining / trajectory
16 Aug 2013 01:34 #37718
by ygs
Replied by ygs on topic Idea for high speed machining / trajectory
Looks like G93 doesn't help. I've managed to convert CAM output for inverse time feed, but linuxcnc won't be able to break the limits set in configs (I mean MAX_ACCEL setting for rotary axis).
Lack of any trajectory planning for AC axis cannot be avoided...
Lack of any trajectory planning for AC axis cannot be avoided...
Please Log in or Create an account to join the conversation.
16 Aug 2013 04:53 - 16 Aug 2013 05:25 #37725
by BigJohnT
Replied by BigJohnT on topic Idea for high speed machining / trajectory
Why would you expect LinuxCNC to NOT follow the limits of an axis?
Sounds like you need to turn the max acceleration up on that axis until you can go as fast as you need. You still need to use G93 anytime you run a linear and a rotary axis.
JT
Sounds like you need to turn the max acceleration up on that axis until you can go as fast as you need. You still need to use G93 anytime you run a linear and a rotary axis.
JT
Last edit: 16 Aug 2013 05:25 by BigJohnT.
The following user(s) said Thank You: ygs
Please Log in or Create an account to join the conversation.
18 Aug 2013 03:28 #37785
by ygs
Sure it's already done. But it's clear that inability to look ahead is a big disadvantage of linuxcnc. I can't understand why devs do not pay attention on this?Yes, it's a challenge but... Solving this problem may push linuxcnc to industrial market of high speed machining (I mean industrial controllers with good look ahead) and help devs to make some bucks....
Replied by ygs on topic Idea for high speed machining / trajectory
Why would you expect LinuxCNC to NOT follow the limits of an axis?
Sounds like you need to turn the max acceleration up on that axis until you can go as fast as you need. You still need to use G93 anytime you run a linear and a rotary axis.
JT
Sure it's already done. But it's clear that inability to look ahead is a big disadvantage of linuxcnc. I can't understand why devs do not pay attention on this?Yes, it's a challenge but... Solving this problem may push linuxcnc to industrial market of high speed machining (I mean industrial controllers with good look ahead) and help devs to make some bucks....
Please Log in or Create an account to join the conversation.
18 Aug 2013 04:03 #37787
by andypugh
There are no "devs" as a separate breed. There are users of the software who want a new feature and add it, or who don't like the way things are.
There are certainly a number of "devs" who would like the lookahead to be longer, and who have some idea how to do it. But it hasn't got to the top of anyone's list yet.
Currently all those with spare time are working on making the software work with newer kernels, or be better with advanced kinematics, or handle more tools in the tool table.
It is very likely that the lookahead will become much longer at some point in the future. It is a known-issue. But it probably isn't as much of a problem as you seem to think.
Replied by andypugh on topic Idea for high speed machining / trajectory
can't understand why devs do not pay attention on this?.
There are no "devs" as a separate breed. There are users of the software who want a new feature and add it, or who don't like the way things are.
There are certainly a number of "devs" who would like the lookahead to be longer, and who have some idea how to do it. But it hasn't got to the top of anyone's list yet.
Currently all those with spare time are working on making the software work with newer kernels, or be better with advanced kinematics, or handle more tools in the tool table.
It is very likely that the lookahead will become much longer at some point in the future. It is a known-issue. But it probably isn't as much of a problem as you seem to think.
Please Log in or Create an account to join the conversation.
19 Aug 2013 00:47 #37814
by ygs
Replied by ygs on topic Idea for high speed machining / trajectory
now it's clear. Thank you. It's should be a problem for those who do high resolution precise work. Especially with multi-axis machines.
Is there any guys who would like to implement this? I've seen some thesis on the Net with some look ahead algorithms/formulas. Probably it's worth to find and group interested users to collect some money?
Is there any guys who would like to implement this? I've seen some thesis on the Net with some look ahead algorithms/formulas. Probably it's worth to find and group interested users to collect some money?
Please Log in or Create an account to join the conversation.
19 Aug 2013 02:19 #37818
by BigJohnT
Replied by BigJohnT on topic Idea for high speed machining / trajectory
While it could be a problem for a poorly designed machine I doubt that a well designed machine this would be a problem. I have a machine that cuts in excess of 300 IPM with hundreds of thousands of tiny lines and arcs with no slow down. It does run within the limits of the machine. When I designed the machine good acceleration was a primary design for direction changes. To my luck this also makes for a fast machine cutting tiny tiny tiny segments.
Money in more important for Mack users and usually does not motivate here. LinuxCNC is open source and everyone volunteers to work on the parts they are interested in. Maybe you can find some programmers and do a kickstarter thingy...
JT
Money in more important for Mack users and usually does not motivate here. LinuxCNC is open source and everyone volunteers to work on the parts they are interested in. Maybe you can find some programmers and do a kickstarter thingy...
JT
Please Log in or Create an account to join the conversation.
19 Aug 2013 13:23 - 19 Aug 2013 13:24 #37832
by ygs
Replied by ygs on topic Idea for high speed machining / trajectory
I agree with you that the machine has to be well designed. However there're situations when high acceleration settings cannot be used even the machine is of perfect design. For example when you cut hard material with thin tool. You have to keep acceleration low just to save your tool for for fast direction changes.
I do not say software should solve mechanical issues. The problem with linuxcnc is inability to work fast in more straight segments (composed from tiny G-segments) when acceleration is set low. We can't enable something like naive naive cam detector for ABC. Why not have something like naive cam detector for rotation axes as well?
I do not say software should solve mechanical issues. The problem with linuxcnc is inability to work fast in more straight segments (composed from tiny G-segments) when acceleration is set low. We can't enable something like naive naive cam detector for ABC. Why not have something like naive cam detector for rotation axes as well?
Last edit: 19 Aug 2013 13:24 by ygs.
Please Log in or Create an account to join the conversation.
19 Aug 2013 18:45 #37842
by BigJohnT
Replied by BigJohnT on topic Idea for high speed machining / trajectory
If this is something that interests you then you should work on it. Check out a copy of the source files and create a branch to test your ideas in.
JT
JT
Please Log in or Create an account to join the conversation.
Time to create page: 0.108 seconds