Advanced Search

Again! Many thanks. I have deselected as many items as possible in the BIOS and selected 'disable multicore technology' so presumably it is now running on just 1 core. I also presume this will have the same effect as disabling core2 with the Isolcpus Parameter and that it will no longer be necessary to use that parameter?

The new stress test reading is 65738 a reduction of around 20000ns. The next stage is to install the Wheezy iso you advised. Just need to find out how I can have both Wheezy and Bookworm available in the GRUB start menu so I can select either OS.

Since Rodw's post I have done a lot of reading - I had not expected the learning curve to be so steep! But very interesting nevertheless, and if I am to get to grips with Linux and stop wasting your time, I will need to overcome my 'Windows induced fear' of changing system files and do much more reading.

If I may, I will report back once I have tried it with Wheezy. If this also makes a difference I will buy the new graphics card and maybe I will get the figure down to something useable. Finger Crossed!

I suspect that the future is hardware based stepping and so I will probably go down the Mesa card route in due course. Bu first, now I have started with this I will do my best to make this computer suitable for software pulse generation - it will be a useful learning tool
thank you again
Mitch

@Rkatts
I think I have identified the issue you're struggling with.  I have been struggling to get the Kollmorgen AKD servo drives to communicate with LinuxCNC 2.9.X ethercat and been stuck on version 2.8.X for a while now.

I have long suspected that the choice to make EoE default to On is the root cause, and believe I have confirmed this.
It appears you need to compile the IGH master yourself with the --disable-eoe option, once this is successfully installed, the problems go away.

From some additional searches, it sounds like if EoE is turned on you need to have the specific function handlers to handle the EoE packets, otherwise the drive will get bungled up trying to process EoE packets with the CoE handler (this is my hand wavy explanation in my head, not sure its exactly what is happening).

I am using the generic controller and using generic XML configuration for the run up script, so this would make sense.

I am also very green with Linux so this took me quite a while to figure out, but I have put together a install guide for those wishing to do the same, hopefully it is adequate:

1. Download the LinuxCNC Debian 12 iso (with LinuxCNC 2.9.3 preinstalled)
   image the iso onto a thumb drive using Rufus, install onto PC
   linuxcnc.org/downloads/
   
2. Install extra tools:
   $ sudo apt-get install git
   $ sudo apt-get install caja
   
3. Manually compile/install the IGH Master
   Following official guide: docs.etherlab.org/ethercat/1.5/pdf/ethercat_doc.pd
   
   Open a terminal and clone the master repo
   $ sudo git clone gitlab.com/etherlab.org/ethercat.git /etherlab
   
   Change directory to /etherlab
   $ sudo ./bootstrap
   $ sudo ./configure --sysconfdir=/etc/ --disable-eoe
   $ sudo make
   $ sudo make modules
   $ sudo make install
   $ sudo make modules_installMark the EtherCAT config file editable:
   $ sudo chmod 777 /etherlab/script/ethercat.conf
   
   Open EtherCAT config and set the adapter Mac ADDR
   Retrieve mac address of adapter you with to use:
   $ ip addr
   
   In the /etherlab/script/ethercat.conf, update mag addr and devices:
   Enter that mac address on the master0 device lineMASTER0_DEVICE="00:e0:4c:f1:b9:af"   (where the highlighted string is the mac address of your adapter
   Set the DEVICES = “generic” (do not leave blank)
   
   Enable / Start the EtherCAT service
   $ sudo systemctl enable ethercat.service
   $ sudo systemctl start ethercat.service Verify Master is running/working
   $ ethercat master
   
   You should get an output (not that no master can be found)
   
4. Install EtherCAT support for LinuxCNC
   $ sudo apt update
   $ sudo apt install -y linux-headers-$(uname -r) linuxcnc-ethercatTrying to install without installing the package “ethercat-master” but watching the above build progress it appears to install that anyways.
   At the end of this build process it will notice you have already installed EtherCAT. It appears the user needs to answer “NO” to this prompt to maintain the manually installed version without EoE.
   
5. Note Location of LCEC configuration      
   This process installed the lcec object/config file in a new locationNeed to update the LinuxCNC project hal file to use the new location: loadusr -W /usr/bin/lcec_conf ethercat-conf.xml
   
6. Update ethercat-conf.xml based on network/device topology and start LinuxCNC

I think that was all the steps, let me know if that helps out or I missed some steps

For anyone that may know the process of updating the official releases, I would make a suggestion that EoE is disabled by default due to the requirements of having compiled driver for your specific EtherCAT device.
If the goal is to make LinuxCNC work out of the box, I'd argue that the configuration released with 2.8.X was easier to use than the default configuration of 2.9.X (so far) since the enabling of EoE appears to require the use of non-generic devices.

I've noticed others having better luck than I had with EtherCAT on LinuxCNC 2.9.X, so maybe I just need to learn how to compile those device drivers, but my lack of experience in linux is holding me back at the moment.

Best Regards,
Paul

sorry, i forgot the report file 
I install the stable Version from PB. I follow the guide. 
Don't know if this is the problem
emcMotionInit: emcTrajInit failed
emc/task/emctaskmain.cc 3054: can't initialize motion
or this part. But the feed override i copied from the guide. 
INIFILE: ERR_CONVERSION, section=DISPLAY, tag=MAX_FEED_OVERRIDE, num=1, lineNo=15
 

Noooooooooooooooooo!!!
Never use those power supplies for anything, ever, except powering LED lights maybe.
That is a non regulated transformer power supply with no output control at all, so voltage varies wildly with load, and is way above 5V at low loads when set to 5V.
Or look at it the other way, never use a 3$ power supply to power 1000$ equipment.
Get a MeanWell or Delta 5V power supply, at 20-30$ they are still cheap and can be found easily online.
In the meantime, if you have an old Samsung 5V charger, they were really good, the original ones, not the counterfeit ones.

what error are you getting? need more information from your install also, without it we have no way to know what the issue could be.

I would say that power supply is marginal

I would get a DIN mounted PS say a MDR20-5

Older encoders may draw considerable current

Hi, 
i have some problems to convert my *ini and *hal file from QT-Dragon to Probe basic. Have some errors when i start it. I screenshot my CNC Folder. 

i cant find the error. please help me that i can start it
 

well, I'm trying to locate the trouble, it's with the encoders: when I unplug them, there doesn't seem to be any restarting, and when I plug them back in, the default comes back from time to time.

my encoder is power by 5V

Could it be the power supply or the encoder ground?

below, is the power supply for the 7i97t card too weak?

 

I don't think so. It appears to just be a three phase brushless motor with integral encoder. Here is how the schematic shows the y-axis motor:
 

Looking up the part number, it is described as:
MHD041B-144-NP0-UN by Rexroth, Indramat, Bosch is a Synchronous Motor in the MHD Series. The MHD041B-144-NP0-UN features a Motor Frame Size of 82mm Flange Size, 50mm Centering Diameter and a Motor Frame Length of 2.7 Nm. This Motor has a Motor Encoder of Digital servo feedback and comes with a Output Shaft of Shaft with keyway according to DIN 6885-1 (with shaft sealing ring).

Thanks

 

no-ops currently
 

Did you add your new pin file to the project?
 

Do the motors have tachometers? A dirty tachometer commutator might explain the behaviour

Look like more than just EMI causing lost Ethernet communications.

It actually it looks like the 7I97T card lost 5V power briefly,
since it does a full reset. I would check the 5V supply voltage,
current capability and connections.

In addition to tommylights suggestions, you might check that the contactor
contacts have arc suppressors, especially if they switch inductive loads.

Hi Guys,

I've re-controlled a Homag CNC router to use Linux CNC with MESA controls. It's was a complicated project, but for the last year it's been working well enough for me to be running commercial parts on it.

All of the motion hardware (mechanical / servos / drives) is what the Germans put in the machine 20 years ago. It uses Bosch ECODRIVE servo drives and has ball screws on x-axis and z-axis and a rack and pinion on y-axis.

The x-axis and z-axis work fine. The y-axis is "jumpy" when moving in the y+ direction. And smooth in the y- direction. I can clearly the position errors ramp up in halscope and then it appears to catch up and get back into position. With my hand on the gantry I can feel it jump / jump / jump going in the y+ and you can clearly see it in the tool marks. If I rapid at anywhere near full speed in y+ it will get wildly out of control and start slamming back and fourth and attempting to rip the whole 8,000 lb machine apart. But it rapids super smoothly in y-.

Mechanically the axis seems fine and there isn't anything like a dust collection hose that drags differently when going in a different direction. I've manually excessively lubricated the rack and pinion and it perhaps helps a little bit, but certainly doesn't solve the problem.

I recall probably spending about 10 hours attempting to tune the PID parameters and was never able to get it moving smoothly in y+. But I'm at the point now where I need to get it cutting smoother. So I am looking for any suggestions of things to try?

Thank you!

If I recall correctly I've PLL a fpga gpio at 50meg so to rid of that wizzy OSC extra part (im cheap, I know lol). If not 50M SPI then for sure it's full on 25Mhz and plenty capable kit. 

Doesn't matter, what's important to pay attention is something ain't right at 4Mhz, I suspect Lcnc won't be happy about that in fact. Perhaps the raspi is to be avoided if running on such a lazy tick?