Advanced Search

Another option is to use the 7I96S's parallel expansion connector for the MPG
(there's firmware that puts a second INM module on P1 so adds 4 MPG encoders
and debounced inputs)

As far as my testing has shown, on my cobbled together 7c81 clone SPI works on the RPi 5.
I received my RPi5 around Nov\Dec 2023 and can't say I've noticed any hardware issues. Booting from a NVME drive has shown no issues, I've run octoprinter on it, once again no issues.
Speed wise even when running via a SD card is night and compared to my RPi400, which has a slightly faster CPU than the RPi4. Actually to be fair it is a bit of a beast.
No trouble building Linuxcnc packages, even with 2 cores isolated, tho I suspect this may not be needed when connected to a Mesa board.
Bang for buck a PC from the last ten years is far better value.

Personally I'm still undecided whether to use my clone with an EPP interface with a PC or with the RPi5 via SPI.

To tell the truth I'm in 2 minds whether I actually want to CNC the lathe, the only reason I started to convert it is the AUD $600 to get hold of a new screw, maybe it would be worth it, if only to have one of the few Myfords with a leadsrew that isn't worn.

Back to topic.
Whilst the RPi is able to run Linuxcnc not a lot of info for the official Raspberry Pi foundation images are applicable, as it is only really the RPi kernel sources that are used, needed for use of loading overlays, the rest is pulled from Debian arm64 repos. Config.txt must be edited to enable the SPI interfaces. And Linuxcnc will need building from source to have the hm2_spix driver. Last time I looked the pre built packages pre dated the new driver.
Re Tommy's comment regarding a small SBC controlling a large CNC machine, how would you feel riding a bike at 220-240 kph ( on a race track of course) that's ecu is controlled by a Motorola microcontroller ? Cos that's the brains used in a lot of Ducati's and Moto Guzzis around 2000 vintage. Very hackable, very easy to download the operating code off and disassemble. Even a fully unqualified shit kicker like myself was able to do, the one and only time I did anything with C Sharp.

Deckel FP4A Retrofit
I just bought a used mill for 330 Euro and now the fun begins
I have planned to do a full EtherCAT Servo build, with the Bechhoff moduls or similar. 

More will be posted soon when the unit is placed on the floor and cleaned. I have already stripped some of the guards, panels away to get more room for cleaning. Also the control unit and the "electrical cabinet" are removed. 

Best regards and Happy New Year From Denmark.
 

Wittle ATC magazine. Probably going to mount in on the front frame though, on a small air powered pivot that just flops it out of the travel area.
I will develop a more advanced atc later.

 

Spindle arrives tomorrow. Little laser tool setter is already here. 

Yeah i think its fine. my main worry was wasting all my IO, but it seems i have enough.

Hello,
I have similiar problem.
I was running latest version of linuxcnc(complete installation from usb drive), and cut recovery function was working correctly.
Because I am in the process of creating localisation for the qtplasmac, I did the RIP install (v2.10-008.054). And now the cut recovery is not working.
The cut recovery window shows up ( when paused or triggered by arc-ok loss), Z goes up, but forward and back buttons are not working. I mean they click "yellow", but axes doesnt move on the contour. directional buttons on the screen are working.

I did two of these RIP installations on different computers, and behaviour is the same.

Even in the qtplasmac-metric simulator it is not working.

M52 P1 is called in the program.
Do you have any idea what can cause this problem? thank you.

Your remedy is working as it should and now i can set correct value, thank You. But I forgotten to tell that linearity of voltage 0-10V is terrible, I used scale for make it "more linear" and that works like a charm. Before on raspberry pi 4 i didn't noticed that strange behavior, it was simple as that to calculate resolution etc, but I also didn't use qtdragon display, when I switched to rpi5 then I noticed those strange things.
Now I bend it little by cheat real max rpm so its almost fit, but I don't like doing things in a slapdash way.
Below table how analog output voltage works for me...
value [RPM] / VFD frequency [Hz] / Uout [V] .
3000 30 0.8
6000 93 2.36
12000 220 5.6
18000 350 8.6
20000 396 9.6
21000 400 9.7

It does look like using the non-interpolated position
(perhaps because of the low resolution)
triggers some kind of bug with the tracking.

Its a bit disappointing that LinuxCNC does not cut the same
thread at different speeds, that really does seem like a bug.
(and a bit unexpected since mechanical threading doesn't have this issue)

Not exactly sure what this is all about, but
1. RPI is not a good choice anymore, there were several hardware versions of RPI5 with many issues
2. There is a ready made image for RPI5 that works out of the box
3. SPI does/did not work at all on RPI5, but Scot has done some work on it so by now it probably works
4. To expensive for what it is
and personally (after buying 3 of RPI3)
5. having a very small SBC controlling a big machine is ...
6. to damn weak for comfortable use
7. to much hoops to jump just to get things working, so utmost respect for whoever made the images
8. for the same price a good used PC or laptop can be bought, and everything works.
-
As for making user experience better, utmost respect, see with ScotA if he needs help with SPI

7i84 has two separate V-Field power supply inputs, so it should be easy to have one set at 24V and the other at 5V, granted everything has the same ground/0V.
Not sure that affects the inputs, but it does outputs as far as i can recall i have one such setup i did many years ago, so a bit fuzzy on details.
Either way, the 4 inputs should work as is with 5V when set to MPG mode, again if i recall correctly.

I ran a couple of tests this morning to see if I could figure out why the velocity started working OK. These two scope captures are both at 200 RPM, running the same threading program. It shows me that 10 teeth is not adequate for some operations, and that I should modify the spindle with a new gear with more teeth. I'm thinking I'll do 180 teeth/rev. It'll take me a while to make the modifications.

I don't expect this to fix the original problem of threads not aligning at different speeds, but it should fix the slightly varying thread pitch..

HAL using: net spindle-position encoder.0.position => spindle.0.revs



HAL using: net spindle-position encoder.0.position-interpolated => spindle.0.revs



 

The scale is odd because you have the .00035 scale factor (so the full scale value (to get PWM=1) is 2857 )

Do you need the scale component? Eliminating it and setting the output scale to 24000 would certainly
simplify debugging.

With the existing scaling, setting the output scale (and max output) to 8.4 should be about right:

MAX_OUTPUT = 8.4
OUTPUT_SCALE = 8.4
OUTPUT_MIN_LIMIT = 1.75
OUTPUT_MAX_LIMIT = 8.4

But why would you do that?

MAX_OUTPUT = 24000.0
OUTPUT_SCALE = 24000
OUTPUT_MIN_LIMIT = 5000
OUTPUT_MAX_LIMIT = 24000

and no scale component is 100% easier to understand


 

Thank you cornholio.

I'm looking to cut wood for building boats.

If I run into issues doing the above well, I earned a living doing Unix driver and kernel work a while ago; I would like to contribute now if I could. I have experience with a number of device driver debugging and kernel development techniques.

Let me get some hardware and software in. Again, if there is a certain feature that doesn't work that should work, let me know.

Thanks,
Tom Alborough

No, the host Ethernet address should be say 10.10.10.100

 

This is a dedicated PC for LinuxCNC for my lathe. It has built in WiFi that connects to our home network. The Ethernet port is dedicated to the MESA card. I don't plan to connect with anything else via that Ethernet port, just the Mesa card.