Advanced Search

Ok. The 7i43 looks like it is working.

But I have no luck with the PCI card. Right now I think I have to go with a PCI, because there is a PCI-e graphic card which seems have to be installed for booting.

Mesaflash sometimes works. Sometimes the pc freezes. Everthing ist grounded with separate cables.

Addr ist 0xD010. Whe mesaflash is working it shows 0xD010:0xD410. So high addr should be 0xD410 as I understand.

Linuxcnc shows error while startup and mesaflash with --print-pd shows an error regarding access to the flash.

Any ideas?

One difference stands out immediately.
My file uses syntax
" except Exception, e"
New version uses
"except Exception as e"

I'm beginning to think that I would be better to start with a new configuration. As you point out, probe screen has an updated version. That would just leave the spindle speed buttons to fix......

As a PS: I am currently running LinuxCNC 2.10 on Debian 13 (both pre-release versions). Although XFCE is the de facto standard Desktop Environment for LinuxCNC, in my experience KDE nowadays performs slightly better with lower latencies. Gnome, Cinnamon and others are usually worse.
As for latencies: please do not be too obsessed with low latency numbers. Only when you are using old style software based step generation or encoder counting, latency (base thread max jitter) can be a critical factor. When using MESA cards and similar solutions, only the servo thread max jitter is relevant and keeping that below 100us is usually good practice. I always keep it below 50us and never encounter any problems. Even without  full BIOS optimization it should be possible to have a very acceptable performance already on this N100 mini-PC. 

I found some code in tool_actor.py
 
I changed 
transform.RotateWXYZ(90, 1, 0, 0) to transform.RotateWXYZ(-90, 1, 0, 0)
The tool module's origin will change to bottom
 
and tool will rotate in right point
 
but, if G43, the path will draw wrong

I have poor Engilsh, and just have a little coding, forgive me

Tim
 

Maybe a look at the current version of 'stdglue' helps:
github.com/LinuxCNC/linuxcnc/blob/master...n-stdglue/stdglue.py

Hi Tom, please let me know how I can send help! More coffee and pizza, or perhaps some nice beers?   I actually have quite some time available, so if you are working on specific things for which you need help, just let me know.
As for Debian, yes of course it is a big plus to have an easy install method, especially after doing system upgrades. But that was not always the case. I remember the days that upgrading under Debian could also be quite a lot of work. After LinuxCNC was added to the  mainstream Debian repositories things became much easier. And I hope that once CL has added LinuxCNC to their collection of software bundles, the same easy install and update method will apply to CL as well.
To answer you question about compiling LinuxCNC on CL: yes, I tried that, but many of the dependencies were not available on CL yet, or perhaps they were available but in places that I was not aware of. So instead of wasting more time, I followed the advice from the CL team and submitted a request to have a bundle created by them, which is comparable to having Debian adding LinuxCNC to their mainstream repositories. It should result in a single command to install and update LinuxCNC without having to build things from cratch, similar to having LinuxCNC in the Debian repositories.
But again, for me it is not about creating an alternative to Debian. It is about finding out if LinuxCNC under CL can benefit from their processor optimizations and if so (which still remains to be seen), how we may benefit from that on Debian and other distros as well.

Another night with some progress. I am almost certain that my Y Axis problem is not the Tang or Rio.

I changed motors, drives, wires and pins and the stalling still happens. Then I thought measuring Direction and Step at the same time. And I believe I found the error. I measured the output of the Tang which is almost perfect. A tiny bit of noise, but really not a lot. When I measure Step and Dir at the drives Dir has a lot noise! See pictures below. I have the step and dir per motor in a single 2 conductor cable. I changed that and put the 2x dir in one cable and the left the step each in a separate cable. No change. I think I have so much noise that the driver probably thinks to change direction.

Anyone an idea how to reduce the noise? Channel B (red) is direction.

   

Hi Wrightys99, if you want the same mini-PC that I have, go to Aliexpress and seach for "T8 N100". The result will show many of these mini PCs, sold under different brand names, like Chatree, Firebat, or no name at all. Mine has no name. As far as I can tell they are all the same. Just pick the cheapest with your preferred configuration. I bought mine for around €150,- with 16GB RAM and a 256GB SSD. Make sure it has the N100 processor, not the older N5095. Although I ordered it without an OS, it came with Windows 11 pre-installed and activated at no extra cost. So I just made the system dual-boot where I can use Windows for tuning my servo drives, as many servo drives come with Windows tuning software.
Below are some pics.
As for the most optimal settings, I will have to check how I modified the BIOS setting. Please allow me to follow-up with a separate post on that. As for isolcpus (boot config parameters) using isolcpus=1,2,3 gives by far the lowest latency, but at the expense of a slightly affected responsiveness of the system in case you want to run other tasks in addition to LinuxCNC, like YouTube or so. But since this PC will be a machine controller, it is recommended not to use it for anything else than LinuxCNC and not have it connected to the internet, certainly not during machine operation. What I typically do is use isolcpus=2,3 during system configuration, e.g. when building a new machine or modifying an existing machine, where I sometimes have to check online resources, and once the configuration is done switch back to isolcpus=1,2,3. But using isolcpus=2,3 still gives latencies that are low enough for using LinuxCNC, so setting it to 1,2,3 is not even needed. But many of us just love the fact that the system is running with the lowest latencies possible   Just make sure that the latency settings (max jitter) in your machine configuration matches the correct isolcpus settings. If the latency (max jitter) was determined with using isolcpus=1,2,3 and you run the machine with isolcpus=2,3, you could potentially encounter some issues. So if you want to run the machine with isolcpus=2,3, then also determine the max jitter with that setting and enter those results into your machine configuration.
For easily modifying boot parameters such as isolcpus, I typically use the program grub-customizer, which I find a bit more convenient than using the command line (sudo vi /etc/default/grub).
    ​​​​​​​  ​​​​​​​ 

Good morning, it is 1:47AM here.

Hi, just been reading this post and I’m looking for a linuxcnc pc to use with my 7i95t. I kind of get the gist of what is going on, but could you provide a link to one of these pcs please.
Could you also provide steps I would need to take to turn off all the unnecessary options or settings, to make it as streamlined as the thread is indicating please.
I am a noob to this cnc hobby and I’m currently in the learning a steep curve process of cnc.
I’ve just bought a cnc of fleabay that I am planning to upgrade and learn the ropes.
The current board in their is the CPO-10V which I am going to change to the 7i95t and run linuxcnc on this new mini industrial pc.
Like many most likely before me I have great plans for this cnc, only time will tell if most of them come off

if you wire the MPGs to normal encoder inputs,
you should be able to monitor the position or raw count
pins of the encoders in the "Watch" window of halshow.

something like:

hm2_7i95.0.encoder.03.rawcounts

Note that if the MPG has single ended outputs you will need
to jumper the 7I95T encoders for single-ended/TTL mode.

Note that there is an advantage to using the 7I95T isolated
inputs for MPGs in that they support a 1X mode which works
better with 100 PPR MPGs that have a detent, as it gives you exactly
one jog increment per click.

Good morning tommylight, Thank you for your suggestion. I appreciate it.

Best regards,

...behai.

Its not clear it's just one axis with the error.

Random uncommanded moves suggest a hardware issue to me,
either in the drives (encoder feedback error?) or the step/dir
interface (cross talk? EMI? marginal levels? bad connection?)

Yes, all this will work on a parallel port.

something like this:

loadrt siggen
addf siggen.0.update servo-thread
setp siggen.0.frequency .1     # 10 second period)
loadrt oneshot
addf oneshot.0 servo-thread
setp oneshot.0.width 2           # 2 second on time
loadrt not
addf not.0 servo-thread
net cool-trigger siggen.0.clock oneshot.0.in
net coolant-out oneshot.0.out [some_parallel_port_pin]
net coolant-on not.0.in
net gate-coolant not.0.out oneshot.0.reset siggen.0.reset  # when coolant is off, reset both siggen and oneshot

 

Thank you all, it's motoring up and down et al now.

The next trick will be to cut things into my DRO but I'm worried about over loading the sensors if I
just wire it in parallel - the impedances are very low. Do I need to buffer things? Like a bunch of non-inverting gates?

Also I'm looking at an MPG wheel I have with A/B quad outputs.
Since I'm only using 3 axis and I have 6 encoder channels should I be able to wire it into one rather than use the input pins
and get numbers. I tried and got nothing on the hal scope but I don't have anything in the mesact setup
for it yet.

I'm trying to read the manual but at 1300+ pages it's a bit intimidating.