5i25T Setup with PathPilot

Hello,

I've done a bunch of lurking trying to figure this out before I asked, but I am trying to upgrade my PCNC 1100 Series 3 to pathpilot. Tormach no longer offers the 5i25 upgrade kit and instead offers an expensive box that connects with eithernet instead. I purchased a 5i25T reading that it is intended on being a drop in replacement for the 5i25 for LinuxCNC versions 2.10 and up. I believe PathPilot is derived from a version of Linux CNC at least past that version but that's a guess.

I believe the 5i25T uses a different FPGA due to chip shortages. Due to this it needs a different bit file than the 5i25, but after flashing it, it's identified as a 5i25 by the OS and functions identical.

So the only thing I think I need to do is replace the default bitfile with a comparable one for the 5i25T in PathPilot so it doesn't overwrite it on startup. Correct me if I'm wrong here, but once it's flashed the first time, PathPilot should automatically recognize it as a 5i25 and load the profile like normal. 

How can I take the bitfile provided by Tormach for the 5i25 and make a comparable one for the 5i25T? Is it as simple as installing mesaflash and flashing the default profile?

I appreciate the help! Thank you.
​​​

I don't know if PathPilot will work with the 5I25T as it has a different PCI ID
I can make a Tormach PNC1100 compatible configuration file for the 5I25T,
but I am not sure if it will work with PathPilot.

If PathPilot does not recognize the 5I25T,  a fully compatible option is the 6I25

Hey PCW,

I can give it a shot at least. Since it'll show up as a 5i25 after flashing I think it should work with the correct bit file.

Thank you!

I was faffing around with Linux a bit and managed to get a wifi driver working (hurrah) and then I thought to myself, why isn't PathPilot recognizing the 5i25T if it's supposed to appear to the OS as a "5i25". You were right PCW that it has a different PCI ID and path pilot will not normally recognize the card. 

PathPilot's card decision logic is in a python file shoved in a bunch of folders. It figures out if there's a 5i25 installed by matching the PCI ID for the 5i25 with the output of lspci. If that python file doesn't find PCI ID (I think this is right, recalling it by memory) 2362:5125 it falls back to checking the ethernet connection. Simply changing the logic to look for 2362:5925 (PCI ID of the 5i25T) instead allows Pathpilot to move forwards to booting. 

The next problem I run into is that I believe the HOSTMOT2 PCI driver is supposed to flash the card installed with the firmware provided by Tormach. I didn't find mesaflash installed on the system so I assume the driver handles the flashing process itself. I know that mesaflash requires you to specify the T when flashing the card or it just returns "card not found" and hopefully it's the same with the driver. Going to dig through more files to see if I can find where it's being referenced

It seems like if you modify the setup python file for the card you could pretty much have it recognize any Mesa card as the logic isn't that complex. I could probably write up a little guide on how I changed the python files to recognize a different Mesa card than supported if you think that would be helpful? I don't see a lot of information on the guts of PathPilot in the forum.
 

Would you happen to know what changes in LinuxCNC 2.9.2 were required to support the 5i25T?

Changes in the hostmot2 driver to recognize the PCI ID code (5925 vs 5125)
I f you had the Pathpilot source, it would be easy  to change.

Here are the changes to the LinuxCNC driver:

github.com/LinuxCNC/linuxcnc/commit/c84a...c4da26fee531569f07a2

Hey PWC,

Thanks for the info. I'll see if those changes are already in Pathpilot and if not I'll figure out how to implement them.

Did you have a chance to look at creating the .bin file for the Tormach? If you need any info from the control, just let me know

Here's a pic of the .hal file if it's helpful 

I'll try and do that tomorrow.
 

 

Note that I'm not sure about the encoder inputs. I am assuming both are A inputs
and that the rapidturn encoder is not correct as Tormach has special interpolation
code for this (1 PPR encoder)