Advanced Search

Did you verify that

joint.2.home-sw-in

Goes true when the Z axis limit switch is triggered?



 

Coorrect Z is a NO switch, X and Y are NC

Is the Z axis switch Normally Open?

( a reversed axis and inverted home switch might behave like this as it would
back-off the activated switch instead of the search move )

 

Link to my current config: github.com/ccatlett1984/LinuxCNC_configs/tree/main/dmc_test 

I have attached a image of the machine, with limit switch locations and homing direction labelled.

I'm not getting past Z homing, it travels up (the correct direction), I can see the limit switch activate in hal show config, but it doesn't stop and mark the axis homed.

These are the limit switches: (I believe the not's are correct, as those switches are NC)
net min-home-x          <=  hm2_7c80.0.inmux.00.input-15-not
net min-home-y          <=  hm2_7c80.0.inmux.00.input-14-not
net max-home-z          <=  hm2_7c80.0.inmux.00.input-13

Hey Matt, I don't know where you are at with the retrofit but I thought I'd share some info about mine as it may help you.
I've just got the X axis going using closed loop to the controller this weekend. I am setting this up with Riocore, there is a very long thread on this forum about it, I'd recommend having a look.

I am using the original brushed DC servos, I removed all the resolver and tacho generator bits out from the end, mounted a perspex plate and fixed a CUI capacitive encoder. I did build my own H Bridge and controller board, had a couple of issues but it is now working.

The actual part doing the tricky stuff is a $30.00 fpga board (Gowin Tang Nano 9K)

There is something about turning the servo by hand without the motor drive engaged and watching the DRO update on screen!

Cheers.
Russell.

matn1 wrote:

Good day,
 Currently the Mesa is on a 5v3A wall plug

This is likely your problem. Replace with a Meanwell MDR-20-5 or similar.

Great. So now just work through the QT dragon docs and add the many enhancements to the GUI in a postgui file

Yes, there are generic Ethercat IO modules but many of them will require you to write a driver for them.
An alternative is to look at a Beckhoff EK1100 and add modules to it to unit your needs

 

These can be pretty expensive but man of them are supported in the Ethercat hal driver.
You can find them second hand reasonably priced on ebay

Linuxcnc's waypoints are defined in device units (mm or inch)
the STEP_SCALE defines the number of steps per device unit
All of the planning and waypoints are in device units.

The base thread is used to generate the steps. Think of the thread as being an interrupt timer routine that is called (cycled) at precise intervals (frequency). The simplest step generator is a square wave where a pin is turned on for a number of cycles and then turned off for the same number of cycles. (In practice Linuxcnc's step generator is more complex than this as it includes step length and step spacing parameters). But in our simple software step generator, the maximum step rate is the frequency/2. 

A Mesa card offloads this step generation to hardware and all Linuxcnc has to do is once every servo thread cycle is to update the frequency the card is generating. In this case the base thread is not required.


 

I have been using EC300 with Remora for the last four months, and it has been great. Once I got it up and running, I didn't have any problems. You can now get an EC300 with Remora firmware on eBay.

Do you have a Ethernet connected motion device?

If so, the latency test is relatively useless for determining the actual
LinuxCNC latency.

A simple ping test is much more informative:

sudo chrt 99 ping -i .001 -q -c 60000 [Device_IP]

(will run 1 minute and print out timing statistics)

Where device _IP is the IP address of your Ethernet motion control interface.
 

I have a EL2502 here but have not tried to install it yet. 
I think if you use ethercat sdos and ethercat pdos you will have all the info you need to create a generic profile.
I have written a couple of drivers for unnamed Chinese IO modules this way

From the dmesg log and tainting the kernel, I think you have secure boot is enabled. Try disabling that in your BIOS.

Failing that I did cover very briefly how to enroll the dkms module into the kernel when  you open a spoiler in my how to.
Installing a kernel module (DKMS in this case) under secure boot is denied unless specifically set as being safe. eg it taints the kernel

THCAD/THCAD2 frequency divide by 1 will not work reliably
with default 7I96 encoder settings, so as rodw suggests:
use 1/32.



 

Do you get latency warnings when you run LinuxCNC?
Are you still using the same PC discussed here on several topics?
Why do you keep doing the same things and expect different results? Stressing the system beyond it's limits will always fail.
Change the graphic card, get a used one for AMD, or get a used PC.
You had good latency at one point in time, so what happened?
LinuxCNC, one youtube video, no whatever stress test you run, nothing, if that works OK, move on to making chips.