Advanced Search

Thanks for the reply ISS, 

In the end i went back to LinuxCNC (I had to go on a literal tolkien level adventure into the Australian outback involving getting l lost and losing all phone signal and having to climb a tree to make a phone call to meet up with a strange hippy dude who sold me a Dell Optiplex 980 with a built in par port)

I hooked it up and straight away got the parport working, and then the oxford parport also worked via PCI (NO ADAPTER) and to ad icing to the cake the max jitter after disabling multi-core, front USB ports and Audio is sub 8000

Very happy at the moment, wishing it was the weekend so I could built the rest of the cabinet out, 
Thanks again for all the help guys If I have any other issues or updates I'll start a new thread as it will be off topic. hope it helps people in the future though.

 

With parallel control you are interested more on the base thread jitter which is ~37000. Set up step timings based on that, not 8000.

I suggest checking lspci now for the oxford card, does it show same or different ports than with the adapter ?
Also parport driver messages (same port or different) ?
Just to be sure what the card tells...

 

Maybe this helps as an example:
forum.linuxcnc.org/38-general-linuxcnc-q...h-ui?start=20#299957

Note the above thread as such is a write off but the example component I presented there might be of interest.

Lately I've been thinking about the Ethernet <> FPGA interface in this project.My current setup is using WT32-ETH01 (basically an Ethernet capable microcontroller) with UDP2SPI bridge firmware has a RTT latency of 0.6ms- however, by pinging it I'm seeing ~0.3ms of RTT time. SPI communication (18 bytes at 2MHz) takes 72us, so perhaps by using lower-level SDK instead of platformio I could shave it down to around 0.4ms in total. Now, just need to find time for this.

Also, I noticed there's Wiznet W5500 support, but it comes with a caveat:

!!! experimental !!!still has an error rate of 0.02%,but only the 0.2ms duration, thus much faster than the UDP2SPI bridges

@meister: do you know what is the source of the error rate?
By looking at the code, it seems that the SPI interface from FPGA to W5500 seems to be clocked at half the FPGA clock- maybe it's just signal integrity issue?
 

Thanks, I'll give that a go and report back.

Attached is a revised homing component that includes the option of choosing between DEFAULT and CUSTOM homing for each joint (I use joint '0' as an example, same goes for all other joint numbers).
Not sure what signal you want to use to call DEFAULT or CUSTOM homing (see question in my post above):


The CUSTOM homing procedure expects the following hal connections:

 


The component compiles for me and switching between DEFAULT and CUSTOM homing procedures works. The CUSTOM homing procedure works for me when I manually switch the hal signals pointed out above and 'HOME_OFFSET' values are set as defined in the [JOINT_n] ini section.

Thanks for the reply ISS, 

In the end i went back to LinuxCNC (I had to go on a literal tolkien level adventure into the Australian outback involving getting l lost and losing all phone signal and having to climb a tree to make a phone call to meet up with a strange hippy dude who sold me a Dell Optiplex 980 with a built in par port)

I hooked it up and straight away got the parport working, and then the oxford parport also worked via PCI (NO ADAPTER) and to ad icing to the cake the max jitter after disabling multi-core, front USB ports and Audio is sub 8000

Very happy at the moment, wishing it was the weekend so I could built the rest of the cabinet out, 
Thanks again for all the help guys If I have any other issues or updates I'll start a new thread as it will be off topic. hope it helps people in the future though.
 

I am getting the impression there is something not set up properly, so can you post some pictures of the machine and where the home switches are located in regards to machine 0.0. Might be homing from the wrong side, perhaps?
When set up properly, there is no possibility of moving the machine outside any limits, be it soft or hard.

I do have a program loading on startup (3D_Chips). Weirdly, I do not have a pd_examples file in my nc folder, yet sim still loads fine referencing blank.ngc.

You could use different switches for homing and for limit with the homing switch being mounted before the limit switch. Once homing is done the homing switch has no further effect.
Usually though one limit switch doubles as home switch and the home position is offset using 'home_offset' values.

Ideal solution to me seems to alter tool table to allow a max rpm.

github.com/LinuxCNC/linuxcnc/issues/2949

0rpm for a probe (or have another field that allows spindle to be enabled or disabled).
Face mills / larger tools would benefit from max rpm function.

(I am not capable of doing this sorry).

Just now i realized you are using gmoccapy, but i am using axis. There you can use axis.rc or usercommand.py file to do some modification.It needs an entry in ini file, DISPLAY section.
I don't know if gmoccapy has this option also, or how to do it there. Maybe just try, or someone else knows about that.
The manual doesn't tell much

linuxcnc.org/docs/2.9/html/gui/gmoccapy.html#_user_command_file
 

Having spent some time this morning looking at the ms300 driver within linuxcnc-ethercat/linuxcnc-ethercat and also while reviewing documentation.

Looking a my screenshot from last night, I realize, that the drive does not pass into Ready to Switch on State. That however appears to be less of an EtherCAT Issue than a drive isseu, right?

Any ideas?

Thanks for explaining that. Is a bit different approach than what I have on the other machine.
But what is not clear, how can I sent home switch outside soft limits. I tried but get the soft limit error immediately. This is what the home offset is for?
My home swithes are at the very end of travel.

Oh that is good. Thanks. Why do you have two different probes in your tool table? What are you using?

So I just save this as usercommand.py? Where do I call the function user_live_update(): do I need to add something in the Ini, or is that a standard already built into LCNC?

Cheers,
Mark

I'm going to assume you are using Debian as you haven't said you are using something else.

sudo apt-get purge package-name

Generally if you right click on a deb file and select "open with gdebi package manager" it will install over the previous version.

This page may of be of help as well.
www.wikihow.com/Install-DEB-Files