Advanced Search

VNC should not be included with official releases, it is a huge security risk.

I  only had a few minutes to contort myself into position to take a few pics of this CNC mill.
If the wiring diagram is a match it appears to use 10V analog control for the DC brushed servo motor drives.
Can anyone confirm from the pics attached before I have a chance to go back with a scope and move a machine out of the way to investigate the control cabinet innards further.

I've migrated the Remora for RP2040 to now use the Remora-eth-3.0 component. 

Download the latest firmware

github.com/scottalford75/Remora-RP2040-W5500/tree/main/firmware

Upload the remora-rp2040-2.0.0.uf2 file to your pico.

To configure the linuxcnc hal module and configs use the files here:

Hal Files (run halcompile on these)
github.com/scottalford75/Remora-RT1052-c...omponents/Remora-eth

and the configs:
github.com/scottalford75/Remora-RP2040-W...onfigs/remora-rp2040

The rcpwmgen should be much easier to setup
as it does not require adding external components
to set the offset. It addition it has higher resolution
 

 I reverted back to using PWMgen vs RCpwmgem, I just found more examples to reference.In my rservo_2.hal I commented the last line out as it kept posting an error.I'm close but I think I missing HAL pin assignments in the Main.HAL for the BOB PWM channel.  Or is it more than that?As always any guidance is appreciated. 

Ok, some more investigation. Using the publish uf2 file and your config I confirm the issue. I've recompiled my master branch which had some unpublished clean-ups (debug info) from the ping-pong buffer implementation, and your config happily loads and runs.

I've also added a firmware folder to the repo to make finding the uf2 file easier.

github.com/scottalford75/Remora-RP2040-W...ora-rp2040-1.0.1.uf2

I'm working on making the RP2040 version compatible with the latest LinuxCNC Remora ethernet component, but currently this version still uses the legacy remora-nv component.

PCW wrote:

1. step_length+step_space = step_period
Maximum frequency = 1/ step_period
(so 500 KHz for 1000 ns step_length+1000 ns step_space)

2. They suggest 50% duty cycle. With the mesa hardware this would be true
only at the highest step rate, though this is probably not important.

3. Looks correct

4. Max linear velocity is after all reductions and ballscrew conversion to linear motion.

5. The shortest step_length on a 7I95T is 10 ns but you are limited to 1000 ns minimum by the drives
Note that it is better to leave some margin and not use the minimums.

6. If the speed was limited by the step timings, yes, 250 mm/S would be your maximum linear velocity.
at those microstep settings.
Practically, the maximum velocity will be much lower, as it will be limited by the step motors maximum
RPM which will be closer to 1500 RPM (= 50 mm/S) than the 7500 RPM that 250 mm/S requires.
This should allow you to increase the step_length and step_space to say 2000 ns so you have plenty of margin.

7. Really depends on the motor shaft/ballscrew/axis mass. A good guess to start is 5x the linear velocity.
If that is 50 mm/S, an accel of 250 mm/S^2 would be a good start. Too high will cause stalls, too low
will limit speed on curves.

 

Thank you for the help PCW  

That's already there, but you'll need to write the image to your nvme drive.
Disable EEE on boot, will need to be setup as a service, I'll see if I can find the info I found the other day. I'll also check if the driver has a switch to do that as well.
I'm beginning to think the RPi5 is a bit of a lemon for Linuxcnc is some cases.
Honestly at its price point a mini x86 system is a better choice.
The basic Odroid H4 runs in at $99 USD, a little bigger front print.
The only advantage the RPi4 had over an x86 system in terms of use was the ability to use the SPI interface. With the RPi5 that's not an option. And using the GPIOs is more or less on par with using a Parallel port.
In saying that I'll still try and assist rod as much as possible .

Update- disabled uefi in the CMS and it all loaded. Always the dumb things..

I see it. copy paste error - should be a bit pin- I pushed the fix.
The preference file doesn't erase unused entries. You can erase the [VERSA-PROBE OPTIONS] header and the single entry when qtdragon is not loaded.

Chris

I tried to download the latest version and it said 2 days to download. I already had this . It's newer then 2.6. whatever I'm using right now. Id probably have the same issue with the latest.

I agree with vncserver …and the ability to run/boot on m.2 media
 

shasse wrote:

That looks good Rod, thank you. I had previously just jumped to the driver section.

For the record, it appears that both updating the driver and setting the kernel parameters are needed. I did a fresh install and just changed the kernel parameters and still encountered the problem.

Once I also updated the ethernet drivers the issue was resolved.

I think that would be right because the 8168 driver is not part ot the default kernel. The dkms driver compiles and installs kernel source. THis is why you need the linux-headers for your kernel. dkms should also recompile the driver if you install a different kernel.

Tommy,

This is alternative example : codeberg.org/tok/libreStove
Looks ok to me, and it's free.

It seems ddns is free. Then you can setup a nas to run a public git server, or a website, forum etc.
Then the url would be something like : 

That looks good Rod, thank you. I had previously just jumped to the driver section.

For the record, it appears that both updating the driver and setting the kernel parameters are needed. I did a fresh install and just changed the kernel parameters and still encountered the problem.

Once I also updated the ethernet drivers the issue was resolved.