Advanced Search

The third CPU (Number "2") of the Raspberry is not running.
LinuxCNC ("realtime stuff") only uses Core Number 3 (the fourth)

Is there something that can be done to use both Cores (Number 2 and 3 = third and fourth) for LinuxCNC?

2 cores run the day to day stuff
2 cores run the real-time stuff

Without isolating 2 cores latency takes a hit. It’s a common fix for latency on all platforms.

I got 100 of 2n2222 as i use them often for testing and some stuff, they should handle quite some current, so when one went belly up with 70mA going through it i had to do some testing, 30mA is OK, 50mA they heat a lot, 65 to 70 they are dead in about 10 to 40 minutes, so for sure not original. All this at 24V driving LED's, not inductive load < this woud probably end them at 30mA.
I think the originals should do something like 500 or 600mA, forgot exactly, been several years since that.

Dear All,

I've set up Mea 7i95 + mesa_modbus with Omron Q2A VFD. Wiring is OK, communication works at physical layer, but I'm in stuck, because:

Omron Q2A VFD expects the following Modbus commands:

CW run: Func code: 0x10, address 0x0001, value 1 (last bit)
CCW run: Func code: 0x10, addres 0x0001, value 2 (last two bit)
STOP: Func code: 0x10, addres 0x0001, value 0
Freq: Func code 0x10, address 0x0044

But spindle.0.forward and spindle.0.reverse are binary outputs, how can I link them to Modbus?
I mean modbus would write "1" when spindle.forwad=true, and "2" when spindle.reverse=true at address 0x0001
This was easy with vfd_modbus, but how to do with mesa_modbus? I couldn't figure it out.
+ cw and ccw and stop register address is the same. It seems it is not supported if I link two hal_pins to the same address in mesa_modbus conf file.
+ func code 0x10 is not supported ... ? btw it seems it works with 0x06, however based on the official manual 0x10 is expected.

Would be great to get rid of vfd_modbus due to lack of rs485 ports in my new industrial PC.
Thank you for your ideas!
Tamas
 

Hi,
I used this image: rpi-5-debian-bookworm-6.1.61-rt15-arm64-ext4-2023-11-17-1520.img.xz
to set up a Raspberry Pi 5 with LinuxCNC.
So far it worked fine. Thanks a lot to all the people that put this image together!!

Now I discovered, that the third CPU (Number "2") of the Raspberry is not running. When LinuxCNC is started, using "htop", I can see, that the third CPU has no load, it is idleing. In the cmdline.txt there is this entry: isolcpus=2,3 and it seams, that LinuxCNC only uses the fourth (Number "3") CPU.
Is there something that can be done to use all the computing power of the Raspberry Pi 5?

I've wired the A and B outputs to two drivers (so according to the breakout board I now have X,Y,A and . I've made an attempt to configure the ini and hal files - X and Y are working, but a U or V G1 does nothing

Balena etcher is multiplatform and the only one I use. The ISO's were tested with it when they were built.
Win32DiskImager has never worked for me for years.
Rufus needs to download stuff of the internet whwn it runs which is disconcerting...

also disable Enhanced Speedstep

did you sort this? i got the same mpg working with a 7i76 mesa card and can send you the hal file if you want

regards

Andrew

Hi Becksvill
Could you send your hal?
I also have this MPG and 7i76 and I would be happy to run it.

HI
I looked at the .hal and .ini files you posted
I think the PID component is missing



sorry, I think with Cia402 component it excludes PID component I thought about my standard configuration.
Excuse me

 

Thank you very much for your reply and sharing your information
I implemented the configuration you created for the RV-6SDL robot and worked with it in the simulated environment and the result was almost satisfactory. I think I can do what I want with the configuration you have created.
Then I edited the HAL file based on the hardware I had (litexcnc).
There were problems which I will describe to you later.
I have two solutions to generate the gcode file, the first one is to use Axis Position Logger.
In this way, I move the end part of the robot to the desired places and save the position and then move between the saved points.
The problem with this solution is that I don't know how to do circular movements in 3D space.
The second solution is to create complex paths using gcode generation software and place these paths relatively in the final gcode file, which, of course, requires simulating the robot body in software such as robodk.
Please watch the video I linked.

Do you have information about determining the orientation of the end effector of the robot?
And also if you have any tips or suggestions let me know
 

Read the tool file directly with python and make HAL pins for them.
The only problem is that the info can become stale as the motion controller doesn't save info immediately.
This is how the GUIs do it.

Playtime:

www.youtube.com/shorts/Ki7himYylcY

How would one go about building a hal component that would read the values of several Z tool offsets from the tool table, then place those values on individual hal pins?  I need to get the values saved for the Z offset of tools 1-8 and have them exposed as hal pins. 

This is for an 8 spindle gang router that has individual Z axis joints for each of the 8 spindles.  I want to be able to apply, save, and change the a tool offset for each spindle.  The tool offset will not be applied using G43.  They will be permanently connected in hal with each spindle joint along with the Z axis command.

At this point the only thing I can do is load each tool offset one at a time with G43Hn, and parameter #5403 then save that value to a hal pin using M68.  But I don't want to have to go through that for each spindle joint every time the machine is started or an offset is changed. 

It looks like you used the wrong dbname in the mesaflash command line.
 is should be something like:

mesaflash --device 7i96s --addr 10.10.10.10 --dbname3 7i89 --readhmid

Pncconf cannot do anything harmful

Using the wrong firmware that does not match the daughtercard can theoretically cause
damage (excessive current from conflicting outputs) but I have never seen this happen