Advanced Search

Dear linuxcnc users,

There is developped a new trajectory planner with help of several forum members over the past weeks.

- Credits : Joco, Lcevette, Turboss, Linuxcnc8, connor, Grotius, and more.

The new planner is called tpmod_scurve.so

This new planner can be installed alongside the original trajectory planner tpmod.so
by specifying the used planner in the config .ini file.

Existing linuxcnc (RIP) installations can be patched using a small script that replaces a few files.
There is no .deb package patch available.

This codebase uses miminal source code to achieve an S-curve motion profile. No heavy dependencies are required.
Note there are several HAL pins and parameters that must be set at runtime to tune a specific machine to its inertia limits.

Important: Alway's exersise caution using new software, especially on high-speed machining operations.

repository
readme
licence  
disclaimer

Automated installation example:

github.com/ozzyrob/Linuxcnc-FPGA/tree/main

Spartan 6 based project.
Supports Linuxcnc-RIO either with direct spi connection or with the UDP Bridge.

Based on a common Spartan 6 dev board available via aliexpress.
Easily assembled, mostly through hole components only SMD parts are a couple of chips.
Modified 7c81 firmware spi mode
EPP interface also can be implemented.
Have experimented with firmware that supports upto 60 i/o (compared to 57 with the 7c81) requires a dedicated driver, easy enough to implement based on the 7i90 driver for EPP interface.
Easily interfaces to your basic Parallel Port breakout board, up to 3.
using a Aliexpress RS-422 interface baords allows SSerial connection, tested with a 7i73 device.

If anyone in Australia is interested I have a few unpopulated pcbs available.

Read the whole, if you haven't, have a look through the github repository. Be prepared to get your soldering iron out. I guess you didn't come across the DIY 7c81 clones or other FPAG dev boards running Mesa firmware.

Linuxcnc wont work in a VM, you have to install on some actual hardware.

In all honesty think how you want to use your lathe, if you haven't had a lot of exposure of using a lathe I'd spend some time (12 months) learning how to use a lathe.

If you are a machinist, you know what you're doing.

我使用的是sv660n

Might be best to raise the issue on github, more devs there.

Yes I agree that this will put radial load on the ballscrew, I didn't think of that. Ok, I'm open to other ideas please do let me know.

minimal changes allow spiral interpolation with large radius differences
 

Hello everyone! I bought myself a lathe, decided to convert it into a CNC machine! While searching for a control board, I came across this forum topic! Guys, can you advise or somehow give me more detailed tips! Since I'm not particularly friendly with Linux! I can install Linux for debugging on a virtual machine! And then it's not entirely clear what to do. I read the forums, it says that you first need to flash the board! But maybe someone will give me links for a more understandable course of action! I'll be grateful!

Hi, I'm building a high speed rDelta robot arm.
I have a test program that just repeats a circle move twice at nearly 2m/s.
This works great except i noticed a weird deceleration artifact. It stops slowing down for a moment, speeds up and then decelerates into its last point.
This is also visible on the robot itself.
The blue line in the plot I attached is velocity from the hal pin motion.current-vel.


As I've been playing with accelerations and velocities in the .ini file I suspect something in there might be the cause. Any ideas?
I also attached my .ini file  

ps I'm building some custom plotting software but the same thing is visible in hal-scope

Thank you cornholio. As a matter of fact, they did run with:

--enable-non-distributable=yes

 

I thinkswitch when running configure may help.



Here's my opinion on latency.
For mesa cards epp\pci\pcie\ethernet\spi (rpi platform) or similar motion cards (seems to be a few more of these projects around now) where a base thread is not required it a little say loose, up to a certain point. I may be one of the few Luddites that still considers EPP an alternative .
It's when the Linuxcnc machine generates the step\dir signals and read encoders requiring a base thread that things get tight.
Even a i5-3470 machine from about 2012/2013 can perform fine when having to produce step\dir signals. (btw I found this machine on the side of the road, someone had put it out for "council" cleanup.)

@Tommy Slackware was my goto disrto for years, but I got a little tired of having to build missing libs and such to install a new piece of software. The last time I used Redhat was version 9, back in the days when you had to do some weird stuff to use wifi, some of which was using windows drivers, in some way or another, that's right ndiswrapper.

What a coincidence! 
Just yesterday I was experimenting with this parameter and noticed that the value cannot be increased.
It's really only in the range of
#define CENTER_ARC_RADIUS_TOLERANCE_MM (2 * 0.01 * M_SQRT2)
#define MIN_CENTER_ARC_RADIUS_TOLERANCE_MM 0.001
rs274ngc_pre.cc
0.001 < tolerance < 0.0225


I wanted to disable the ARC_RADIUS_TOLERANCE constraint and see how interpolation works with a large difference between the start and end radius. This could be used to do spiral interpolation by extending the g-code to g2.1 / g3.1 - just copy interp_arc.cc, but there would be no radius difference check.

 

Alright, I will list some disadvantages that I see. There are more moving parts and more expensive bearings. The ball screw needs to have a large diameter and be of a high grade. The biggest drawback of this drive system is that it will create not only axial loading on the screw but also radial loading. In other words, the play will depend significantly on the size of the screw.
If you decide to go another route I can give some ideas