Advanced Search

Changing the cable can help a little. You can also try to go even lower with the SPI frequency; it really depends on how fast your servo thread is and as long as your SPI communication is faster than the servo thread, it shouldn't be a problem.

I agree that the wiring would be a mess if you had to wire the TMC directly to the RPi. I was surprised when you said that the H7 version worked but not the F4, as they share the same remora-core repo, which contains the tmc module. I think I found out why: on the SKR (H7), the remora comms is on one SPI channel, and the drivers are on another one. Scotta decided to use SoftwareSPI to communicate with the drivers. On the Octopus pro (F4), the remora comms and the motor drivers share the SPI 1 channel, so my guess is that is causing the issue. To my understanding, the same SPI channel cannot be a master and a slave at the same time, and it is a slave for the Remora component. If you want it to work on the Octopus, you'd have to connect the RPi on a different SPI channel and adapt your platform.ini file accordingly. I don't have a board myself, but as far as I could see on the product datasheet you could use the SPI 3 channel on the J74 header.

Congratulations! Most well deserved. (Infact, you probably deserve much more.)

There was no need to even run PNCconfig. You can simply copy the whole config directory from the old machine to the new machine.

Hello,

since my problems persist and one Y-axis does not move during homing, and I don't give up so easily, I tried “dmesg” and “ethercat slaves -v” again.I see two errors in the output:[  459.958251] EtherCAT WARNING 0-1: Slave did not sync after 5000 ms.[   13.362825] EtherCAT ERROR 1: Failed to calculate bus topology.Otherwise, everything is OK and the drives go to OP.Where could the problem be?I also don't understand why I have to turn off the servos and Linuxcnc and restart them for homing to work. 

I'm putting together my third LinuxCNC system. Using Axis and a Mesa 7i96s

This one will be an almost direct copy of a previous installation I did for my small lathe, this time for a new lathe.

To get started I ran the PCNConf wizard skipping over some of the configuration just to get to the point where LinuxCNC will launch.

Now I'm hoping to simply substitute in a copy of the hal files from my old installation (basically replace the hal files PcnConf made with my old tried and true hal files).

Does this sound like a viable plan?

Possible wrinkles:
One possible wrinkle is that I needed more I/O on the previous 7i96s and I did some sort of "firmware upgrade" to the 7i96s to get more encoders via the expansion port so I will need to do that again. need to see if I can find my notes for that. Also not sure if the old upgrade file is still viable?

Both setups use a Lenovo Yoga, however the latest one  (Chromebook C340) requires a USB-C to dock to get an Ethernet connection. I picked up a dock from Amazon (UGREEN) and have managed to get the Mesa card recognized by LinuxCNC but the mouse (also connected to the dock) seems to go to sleep (often) and I have changed the power saving setting to stop this, but they don't seem to be preventing mouse sleeps. My concern is that the entire dock might be going to sleep including the Ethernet port.

Any advice is welcome.

Congratulations!

Looks like you have a print statement that prints the current axis position somewhere in your code.

Hi
I got tree Stepperonline A6 servo motors up and running in XYZ mode using the attached files. When I close LinuxCNC, however, all three drivers show error ErC 1.1, i.e. Sync Error. When I start LinuxCNC again, the errors disappear. So it is not a big deal, but it is ugly to see that the program was not terminated in a decent way. 

I suppose that there must be some appropriate commands in the shutdown.hal file. I tried a lot of things, but none of them worked. Dies anyone know what to put in the shutdown.hal file to prevent the sync error or is there an other way to prevent this error?

Thanks
 

 

 

Aciera wrote:

Frankly I don't see the point of deleting 1000+ posts and leave god knows how many threads disjointed even if I could think of an easy way to do it.

Same here, especially now that i know you were the one doing huge moral support when i built my second shop.

I've got a press retrofit in progress and this sounds great.  I've got quite a bit of work to do hardware wise, but I'd be happy to test when I get things moving.

rodw wrote:

 
ChatGPT is not particularly useful with Linuxcnc. I worry when I ask AI a question and it quotes into I wrote on this forum back to me.

And then knowing the ratio between "dangerous half knowledge" and "expertise" for a fact makes it even scarier.
Suddenly your own work hypothesis becomes a fact and "best expert advise" one can find on the internet.
 

I will try changing SPI cable with LAN cable. Which might work better in shielding external interference.

I have already tried STM32F4xx-PIO and STM32H7xx-PIO branches. H7xx works ok on SKR3 board.
But F4xx branch is not working with tmc5160 but working in step/dir mode. I have tested ben-jacobson branch github.com/ben-jacobson/Remora-STM32F4xx-PIO.git who is orginal developer of F4xx branch. I have posted issue on GIT about this but didnt got reply. Didnt tested remora-cnc branch which 28 commits less than ben's branch. and since it was forked out of ben's branch so didnt thought that it will work on remora-cnc branch.

for configuring tmc5160 via spi I have to change jumpers on board for SPI. Other option is, I should remain board jumpers in step/dir mode and run external spi wires connected to stepsticks (daisy chain) to RPI, which will make really mess.

All I can offer is to delete your user profile.
Frankly I don't see the point of deleting 1000+ posts and leave god knows how many threads disjointed even if I could think of an easy way to do it.

It is normal, that a HD doesnt have the exact 50 to 1 ratio. This is just a "rounded" number.

It depends on the "gears" and normally a 100 to 1 for example is 101 to 1. This is because the drive axis rotates in the opposite direction of the driven axis. So the difference is always +1

Best regards
Julian

cmorley wrote:

But if you made something in flexGUI, it's probably not that hard to re-imagine it in Gladevcp or Qtvcp afterwards.
 

That is wonderful news. I am familiar with Gladevcp. I was reading lately how glade is outdated and Qt5/6 related is the way to go. I had made up my mind this is the project where I'll learn flexGUI. My enthusiasm failed me after I dived into it yesterday. I will happily continue with Gladevcp. I will learn FlexGUI eventually, but not in time for this project.