Advanced Search

One of my older HAL files using COUNTS mode is attached.

Some differences from my earlier INDEX mode summary:

• Stepgen is in position mode
• carousel.N.scale is stepgen pulses per pocket
• carousel.N.counts is connected to stepgen.N.counts
• carousel.N.counts-target is connected to stepgen.N.position-cmd

As far as the difference between stepgen modes (velocity vs. position), the LCNC user manual has a description of these:

The section on Mesa card stepgens is here:
linuxcnc.org/docs/devel/html/man/man9/hostmot2.9.html#stepgen

But I think the generic stepgen component description is a little easier to understand.

If you read it like that you think its modmot loading threads. But reality is different.
rtapi creates the threads.

 

You might want to break the links between iocontrol and probe.. as I suspect that is prompting the popup.

Maybe try looping the tool change like the tool prep is.
(but maybe with no tools in the carousel, just in case....)

It would be very nice to have new forum software. 
Members like Joco have ip ban. This is a very bad situation.

How faster how better.

Maybe JT can give volunteers the 
information to at least setup a new software platform and do migration work.

Now we are all using this outdated nasty
text editor.



 

motmod does create threads, in the sense that "loadrt motmod ......" will create the base and servo threads with the given rates.

Motmod is a component like tpmod.
rtapi creates posix threads. Rtapi does not care about thread speeds. Each posix thread calls a list off components. Then rtapi calls a function order into the component itself with dlopen...

motmod does not create threads.
motmod is a trajectory planner coordinator and supervisor.

 

PCW wrote:

That error (and others I have seen) seem to be a bug in the mesa-modbus code
 

Can you add these to an issue on the issue tracker? 
Just so I don't immediately forget. 

I think that this is a perfectly valid approach, and as you have noted, it was used by Centroid.

The "right" way to do this is probably with a custom kinematics component. That way it is solved by mathematics.

linuxcnc.org/docs/stable/html/motion/kinematics.html

An Arduino is just a board with a microcontroller on it, there's nothing wrong with the Arduino itself. Plus the fact that you can get into much faster microcontrollers with the same Arduino framework makes it alot easier to deal with.

The problem with USB isn't the speed, it's the fact that it has to run in userspace. A USB serial port isn't using a fraction of the speed of USB anyway and the 10/100 ethernet that Mesa uses isn't exactly fast either. USB is always mentioned with this but realistically it could be any serial port. That's where the more productive part of the conversation was going, the fact that a serial port technically can be made to run on the real-time layer. Nobody has ever bothered to make the LinuxCNC component for real-time serial which baffles me since LinuxCNC loves legacy interfaces.

The PC side of this is written in Python anyway, it can't even be real-time from that angle. It'd have to be reworked into a C component. Alex mentioned the Mesa card but a Mesa card for something like this wouldn't be doing anything special that an Arduino can't. The only draw for the Mesa card is the fact that it has a ready to go real-time environment already setup. The Mesa cards intended for use with a control panel aren't even FPGA cards, I believe they run a PIC MCU so hardware wise they aren't really special other than just being well designed.

The best option is a worm gear like 4 axes do. The other option is something similar
www.haascnc.com/service/troubleshooting-...set---ngc-lathe.html

 

 

Is there any chance that this corresponds to something in the kinematics? Maybe an inflection point in one of the joints?

No, rotation about Z is a special case.
Rotation of the other planes (including uvw) might be nice, though there doesn't seem to have been many requests for it.

The num items in pncconf are the channel numbers for things like stepgens,
pwmgens, and logical pin number for things like inms, outms. This is how
the actual hardware is addressed in the hal file.

For example, your configuration has 8 stepgens, 4 on each connector
and numbered 0..7.  The actual pin  assignments are listed in the pinout file
for each of these  8 stepgens.

One exception is GPIO pins which are listed by GPIO number in pncconf.

It probably would be helpful if pncconf added a physical pin column, but note
that physical pin numbers are not used at all in hal files for LinuxCNC configuration
with Mesa cards.

Configuration pin-out:

IO Connections for P2
DB25 pin#       I/O   Pri. func   Sec. func        Chan   Sec. Pin func   Sec. Pin Dir

 1                       0   IOPort       StepGen          0        Step/Table1      (Out)
14                      1   IOPort       StepGen          0        Dir/Table2        (Out)
 2                       2   IOPort       InM                  0        Input0_EncA0  (In)
15                      3   IOPort       OutM               0        Output0           (Out)
 

This isn't an answer to your question, but it looks like you can override it with an environment variable:
github.com/LinuxCNC/linuxcnc/blob/master...ce_rtapi_app.cc#L984
It might be interesting in your case to do that and see if latency improves.

Are you using Linux and not BSD or similar? (I see a comment about it only working with Linux in one of the commits.

Otherwise, looking at the code, I have no idea how it works, it looks very odd.

Hi everyone,Apologies for not getting back to you sooner.If I understand the current situation correctly, here’s how things stand:

• We have access to fairly large web storage, but we currently lack the permissions needed to perform repairs or set up a new installation.
• The current operator continues to finance the website, the storage, and the domain.
• There’s no possibility at the moment to request root access, which would allow repairs or a new installation.
• The domain linuxcnc.org is owned by several people who also finance it.

The main issue, therefore, seems to be the lack of root access to maintain or reinstall the system.I have an idea on how we might proceed:
If I were able to find an investor or sponsor who could provide sufficient web space, we could rebuild the forum and website, then link the existing URL to the new space. For this, it would be helpful to know how much storage we currently use and since when, to estimate the long-term requirements.Would such a proposal be of interest to the community?
Just to clarify: I don’t currently have a sponsor, but I’d be happy to look into it if there’s interest.Best regards,
Mücke