Advanced Search

Hi No I have not make any progress but i conekt a home switch to my Gpio pin for spindle index and when i click manualy on the switch the reading moved a bit , when I click fast i reach 60 rpm o the display so I think the problem is in the filter for the signal , i have to optical sensor and there are signal from both index and fase a, so my problem is to find the value for this part and if anybody can help me sort it out i be very happy ./krister

The 7I76 is not detected

This likely means (assuming you have the 7I76 connected to 7I92 P2):

1. You do no have the correct firmware on the 7I92
(check with mesaflash --readhmid)
2. You have W1 or W3 in the right hand position
3. You do not have field power supplied to the 7I76
4. You do not have 5V power supplied to the 7I76
(either from the 7I92 or externally)
5. Mis-matched 5V power supply options
 valid options are
7I76 W2 left and 7I92 W4 up (7I92 supplies 5V power to 7I76)
7I76 W2 right and 7I92 W4 down (7I76 gets external 5V from TB3)
 

Download the LinuxCNC 2.9.2 ISO from the downloads page, burn to a USB, boot the PC from it or install it to HDD/SSD, use the included Hal Scope to watch the input pins.
See how to use the scope without starting LinuxCNC if it bothers you to start the scope from there (machine menu, hal scope).
Normally you have 5 inputs on a parallel port, but almost all built in and some add on parallel ports can do 3 modes:
Mode out = 12 out 5 in
Mode in = 13 in 4 out
And mode X = 9 in 8 out
So using mode in you have 13 data pins to play with.
All this runs at quite slow rates compared to normal scopes, but is usable for a lot of things.
-
All PC' have a Line in and Mic in analog inputs, so 4 inputs in total, there was an open source software that used those inputs as scope and worked OK, no idea what it was called.
I have used both methods for long time before i got the small Miniware scope.

To check for an EMI issue, you could monitor
hm2_7i76e.0.packet-error-total in the watch tab of
halshow to get an indication of how often packets
are dropped.

Also you might try pinning the Ethernet IRQ to the
last CPU (the one LinuxCNC uses for real time) 
This requires that you isolate the last CPU with
an isolcpus statement in your grub file
(isolcpus=3 for a 4 core CPU for example)
And then pinning the Ethernet IRQ to that CPU

 

Have a look at PlasmaC and QtPlasmaC, those have a lot info about number of pierces, torch on time, program run time, etc, so yes a lot can be done, and most probably all that info can be also monitored from other networked PC's by running HTTPD on the machine PC's.

Here is the output of the console:

But in this case the Spindle Drive is off and Axis-controler are in STO mode / Standby.

I still have my servothread set to 2ms.
In the .ini file there is P = 744.158356898348. Sometimes I read it should be adjustet?

Trying 2.8 Version could be possible. But the hdd is hard to access. Maybe it´s easyier to do an image of this version and reinstall 2.8.

Are there any options to see if it´s an EMI issue? Because I feel the errors are popping up more, when the spindle is on and changing speed etc.


 

I need to write a single-use application to use a parallel port as a "poor man's" logic analyzer.  This will be used to look at the control protocol for an old 8" floppy drive that has a parallel port interface.  I'm not sure of the speed that I need, but it will surely be at a snail's pace compared to today's hardware.  I'm looking for work that I can "go to school on" that I may adapt for my needs.  It would be a time saver compared to programming from scratch.

I realize that this is not a CNC question.  If I don't belong here... 'my apologies.  Advice on a more appropriate forum would be welcome too.

  Thanks

Maybe a somewhat hack would be to have a relais which activates the servos after some delay once LCNC has started.

Thanks for the pointer - updates are disabled

In the meantime, I managed to put a new RT kernel on: looks much better, right?
I followed the github.com/by/RT-Kernel instructions - probably not ideal as you, but I avoided a reinstall (and it's kernel 6.9.1 :D)...

ok, thank you very much. My counter questions because certain things are not clear to me.

1. this is ready and ok so far, please see pictures

2. which is the "right" input for you on the 7i76?
I have a total of 3(+1) PNP sensors connected to the 7i76 as shown on the data sheet "7i76 contact assignment" (please see post from May 27, 2024 14:24)

Check that the 7I76 input pin in halshow (hm2_7i92.0.7i76.0.0.input-xx)
follows the switch state and has the correct polarity (true = yellow when activated)
if the polarity is incorrect, you must invert the checkbox for the input in pncconf.
My question:
I do not see the 7i76 in the neck show. Or am I not seeing in the right place?

3. input pin connected to the correct function in hal (from pncconf setup)
My question:
Shouldn't this be the logical consequence of point 1 and 2?

if you can change things on the sindle (rpm, direction, etc) without turning the spindle on.....and your arduino works perfectly. Then when you turn the spindle on nothing works or works poorly, you likely have an emi issue not a software one.

lookup emi protection for more details but in general. shielded wires grounded on one end only. avoid running wires parallel, when they need to cross cross at 90 degrees, ferrite chokes, use different power supplies. make sure no ground loops (computers are often grounded), etc

In general that's why arduinos are rare in the industrial world. not because they cant work, but because hardening them against emi is much harder and more time consuming than a board designed to work in those enviroments. e.g. mesa, plc's, etc 

jimmyrig wrote:

On another note what's the process for editing the UI? seems like the files are under the /lib folder. load them up in qt, edit and restart? I'm sure its more involved than that but not sure where to begin.
looking at adding an axis for the rotary (A) and bar feed (U) and need corresponding home buttons and read outs.

The A axis dro and buttons will appear if it is a valid axis in the config, currently XYZABC axes are catered for.


For customising the UI, it is best to do it with a USER_COMMAND_FILE in python if possible as this will not get overwritten by any future updates. If the update issue doesn't bother you then the UI is a Qt Designer file and the handler file is a python file.

linuxcnc.org/docs/devel/html/plasma/qtpl...#_custom_python_code

was also thinking of playing around with blue line thickness that shows torch travel between cuts. Sometimes they can be hard to find if needing to restart.

The preview uses the QtVCP GCodeGraphics widget, I think you can change the colours but not the thickness. You may need to dig deeper for that.
linuxcnc.org/docs/devel/html/gui/qtvcp-w...idgets:gcodegraphics

Could it, yes! Does it now? not sure....prob not

linuxcnc being open source gives so many opportunities for advancement and tuning that cannot be done with closed sourced machines.

Your idea is a great one, no reason its not possible, someone just needs to code it. With AI now adays it couldnt be a better time.

id start with a simple python script to record the values every 10th of a second or so (outside of linuxcnc) then tell linuxcnc to stop if those compared values are different (+ some perfentage). once it is working the integration to linux cnc wont be too bad.

Hello PCW,
thanks for your help. I will try another PCI slot soon.

will give that a shot and report back. updated the issue as well.

Was wiring things up today and bricked my 7i96 when uploading firmware. Luckly i had a 7i96s on backup which i would have needed anyways for the BOB. Sending it back to Mesa, they will flash it for free (+shipping). Thanks Mesa!


On another note what's the process for editing the UI? seems like the files are under the /lib folder. load them up in qt, edit and restart? I'm sure its more involved than that but not sure where to begin.
looking at adding an axis for the rotary (A) and bar feed (U) and need corresponding home buttons and read outs.

was also thinking of playing around with blue line thickness that shows torch travel between cuts. Sometimes they can be hard to find if needing to restart.