Anyone looking at CS3E-D1008 drivers from Leadshine?

Awesome you got it  going. Did you use my method in the end?
On bullseye (5.10 kernel), I was at a total roadblock.  I could not use Buster as I needed hardware drivers in Bullseye. Grotius kindly made a deb for me that worked but I still had issues getting it all to auto start
So then I started to work with Bookworm that has linuxcnc packaged in it. Combined with their repo, I was in business again so documented it on the forum here. The only issue I have had with that approach is that when you end up with a few kernels, debian can choose the wrong one to install linux-headers into. Thats why I say to use $(uname -r) in the installation and it just seems to work.

the changes to the .mak file were good until just recently so I made that bit optional and hid it under a spoiler.

ec-debainize is well and truly redundant as etherlabmaster migrated to gitlab so the repos used in the old approach are no longer maintained but the debs are built daily. Thats why I prefer their repos.

 

Awesome you got it  going. Did you use my method in the end?

Yes, absolutely! Your sticky post at forum.linuxcnc.org/ethercat/45336-etherc...-step-by-step#237098 was exactly how I was able to do it.

On bullseye (5.10 kernel), I was at a total roadblock.  I could not use Buster as I needed hardware drivers in Bullseye. Grotius kindly made a deb for me that worked but I still had issues getting it all to auto start

I did see that post, but I was a little reluctant to use that same .deb as that seemed to be specifiic for your machine. The names of my NIC are also different. That could be a potential blocking issue for me. I wanted to do as much from source as I could, mainly to learn.

The only issue I have had with that approach is that when you end up with a few kernels, debian can choose the wrong one to install linux-headers into. Thats why I say to use $(uname -r) in the installation and it just seems to work

I also always do it that way to install the kernel headers. Keeps me from making dumb mistakes

the changes to the .mak file were good until just recently so I made that bit optional and hid it under a spoiler.

Well, that's where I may have gone wrong. I didn't replace the contents of the .mk file and all compiled without errors (I doublechecked the terminal logs to be sure). But when running linuxcnc, I still got an error "lcec: dlopen: /usr/lib/linuxcnc/modules/lcec.so: undefined symbol: ecrt_slave_config_sdo" ( forum.linuxcnc.org/ethercat/47555-lcec-s...ve-config-sdo#258429 ). This was solved by replacing the content of the .mk file. So this may not be optional?

ec-debainize is well and truly redundant as etherlabmaster migrated to gitlab so the repos used in the old approach are no longer maintained but the debs are built daily. Thats why I prefer their repos.

yeah, that's what makes a little hard to get started. The forum is a really really great help, but it's sometimes a little unclear as to what will still work and what not without knowing the history. I understand this now, but last week wasn't easy to get started.  Anyway... I evolved from a user downloading an ISO to slightly tinker with to compiling everything from scratch. I also tested out bookworm a while ago (but dropped because there where some issues I did'nt know how to get around) and the I even learned what it was like to patch and compile the kernel. Glad I did this as I learned soooo much, but also afraid it's just the tip of the iceberg .

 

Great I could help.
A user recently reported that he had to use the default .mk file to get it to compile
I tried again and found the same thing. But I did not have any ethercat hardware to test on.
So its optional for some and mandatory for others!

Hey Peter how are you? I am really interested to know your opinion on the leadshine ethercat drives and steppers, I plan to build a small cnc with mesa card and JMC 140w servos but the simplicity of wiring of ethercat systems is really cool, could you share a pic of what your setup looks like wiring wise? If I understand correctly is posible to wire limit switches directly to each axis drive?

Thanks a lot in advance!

I haven't gotten around to installing them yet on the CNC but so far they seem to be working great in my testsetup. They really are a substitute for servos if you don't need to go very high in RPM (2Krpm is a realistic limit for these steppers although I was able to push them up to 3Krpm).

As for cabling: it's massively easier to connect than the mesa cards and I would definitely recommend going for the closed loop steppers. You only need 1 ethernet cable to the first drive and connect the drives together in series using a small ethernet cable.

The drives have some in- and output you can use eliminating the need for the GPIO's on the mesa cards. It's indeed possible to have the drives perform the homing autonomously, but I prefer to forward all in- and outputs to linuxcnc so you have control there. But that's just my preference and does have the drawback of having slightly less accurate homing position. I have a testsetup here with 3 drives/motors/inductive switches and connecting the inductive switch for homing really is just connecting it to input 4 of the drive and reading the IO's over ethercat.

They only drawback to these drives is the quality of the configuration software. It only runs on windows, so reconfiguring the drive requires a windows machine (or a vm) and it has quite a few bugs and untranslated chinese stuff. Software also states is an "alpha release for internal use only". I've contacted leadshine about this and they promised an update last month but haven't received anything yet. But on the other hand, once you configured the drive you do not need the software anymore. Not a showstopper though...

But bottomline: If I had to choose between a setup with the JMC's and the closed loop steppers I tested -> I would without a doubt go for the closed loop steppers.
Mesa has great hardware and I've been using them for years, but the ethercat drives are just sooo much easier to work with.

Very interesting thanks for the lenghy answer.

My first idea was to get the JMC servos because are cheap and having the integrated drive would result in a smaller control box that I really appreciate, but now discovering about Ethercat the wiring is much simpler and my Omron VFD for the spindle I believe has a Ethercat option module so that's interesting.

Also for my small mill build I don't need crazy speed on the axis but I am looking for accuracy, so after doing some research is not clear to me yet who wins, the servos need tuning and steppers are easier to work with and don't need a brake for Z axis.

In the Marco reps video that he talks about the JMC servos he says that likes them more than the leadshine closed loop steppers, but if speed and torque at higher rpms are not as important I would like to see his opinion then.

So I think in the end considering cost the JMC and mesa route is less money but more work, so we will see..