Advanced Search

  I got the water tray into an acceptable condition.
I removed the problem welds and
welded some slats underneath to straighten things out. (The sheets had a radius when they came back from bending)
Then I re welded them..

Next time I would cut them in such a way that seems will be shorter and all in the same direction. I hope this post will help some others hobby welders avoid the same problems.

I must say viaconstructor works quite well for plasma. Only thing you have to do is choose laser cutter under machine and set the tool with.
I cut some complicated shapes like an Elephant.

However on one file a single shape I could not choose a starting point. Is there a bug with the software.

However I found an easy work around I just rotated the part and moved tho cut on a different part of the metal sheet.

Today I removed the graphics card and ran the machine.

Latency data in attached screen shot. Latency test and Linuxcnc run together.

I increased the servo period.

but "error finishing read" continues.

What to do?

Hi all

I am trying to setup an external usb pendant using a dragonrise USB zero delay board.
The board is showing up in linuxcnc but I'm having a but of trouble with the pin connections as follows.

#

---

# USB Buttons
#

---

net Jog-X-Plus         <= halui.axis.x.plus         <= input.0.abs-x-is-pos
net Jog-X-Neg         <= halui.axis.x.minus         <= input.0.abs-x-is-neg
net Jog-Y-Plus         <= halui.axis.y.plus         <= input.0.abs-y-is-pos
net Jog-Y-Neg         <= halui.axis.y.minus         <= input.0.abs-y-is-neg
net Jog-Z-Plus         <= halui.axis.z.plus         <= input.0.btn-base6
net Jog-Z-Neg         <= halui.axis.z.minus         <= input.0.btn-base5

When I press the button both the input.0.abs-x-is-pos and the halui.axis.x.plus pins light up but I get no motion.
I have tried with the machine homed & unhomed.

Is there a QTdragon pin that would be better to use?
I am testing this on LINUXCNC - 2.9.2-168-ge7cd5217d9

Thanks
Alex


 

Mae22 wrote:

Thank you to everyone who dealt with my problem homing, we are now up and running and calibrated on X, Y and Z axis. My question here is the A axis and Scale in the ini file, i really do not want to back into stepconf as it wipes out everything i have altered in the ini and i have to start from scratch, which is frustrating lol)
Our rotary axis is a small one for making my wood jewellery, it has a ratio of 4:1 and we run on 8 microstepping we worked the steps out to be 17.78 but i don't think this is what i put in the scale on the ini file, or is it????
Could anyone who runs a 4th axis give us any information on this as we are used to Mach4 and have only just started with LinuxCNC, any info is usefull such as calibration and how you do it, working out the steps per degree (just in case we have done that wrong) would greatly appreciste the help snow rider

Great to hear you have your X, Y, and Z axes up and running! When setting up the A-axis (rotary axis) in LinuxCNC, particularly when switching from Mach4, the key is ensuring that your steps per degree (or per rotation) are properly calculated and input into thefile under thesetting for the A-axis. 

I was experiencing problems again yesterday... I will be experimenting in this space next week.
In qtplasmac when you start getting problems and your half way through cutting a program you are just stuck.... you can't change many things without closing the program and reopening and reloading everything. For this reason I stuck it out and kept trying to cut with faults. (Torch wouldn't fire. The pulse function was working fine. Ohmic vaults were hovering on mid 23's ) . It didn't want to work even with ohmic sensing turned off. Eventuality I finished the last few cuts. Productivity was not good.

I tried lots of different things and in the end I have changed to sheet cam.
I use the nestmaster to create a nest and then use sheet cam to create the gcode for qt plasmac... seems to be a good solution.

I did the first proper test of the new spindle servo motor today milling some 7050 aluminium.
The test was done using a 80 mm face mill at 100% width of cut, 0,5 mm depth of cut, 2000 rpm and a feedrate of 600 mm/minute. This worked splendid.

Not only is the machine much quieter but the finish at high feed rates is noticable improved which was unexpected. I suspect that this is the result of removing all the gears and shafts from the transmission as there are much less vibrations from the head now.

Thank you and apology again for dump questions

It is working now beautifull

I had to issue sudo apt --fix-broken install

and sudo apt-get install python3-pyqt5.qtwebkit



 

Those temperatures are all fine.

I would suspect failing PC hardware with the 49 ms latency spike

Can you try without the graphic card installed?

@briansm The 6.1 kernel is a bsp kernel so I doubt it patched successfully, you might have more luck with patching the 6.11 kernel including these patches and use the right config:
github.com/armbian/build/tree/main/patch...rockchip-rk3588-6.11

But some things might not work yet. If you can't figure it out, maybe I'll take a look at it sometime. But for now I don't see much added value in it.

Hi!

I've made my LinuxCNC work with SERCOS II (fiber optics). Now my machine (Trumph TC2000R) has been working well for almost a year, and I'm positive that there are no bugs.

My steps were:
1. Bought a cheap SERCOS II PCI extension card from Beckhoff (FC7501, ~$150), plugged it into the free Beckhoff 3 suite
2. Sart sniffing the digital signals using my $10 logical analyzer from aliexpress
3. Set up all drives using the Beckhoff, so I have a full log of startup messages and synchronization process
4. Parsed it
5. Captured and analyzed a lot of command messages, so now I understand how messages are built
6. Made a dummy STM32(~$200) device, which takes a SERCOS message from the Ethernet and transmits it into the IO pin. You don't even need a real-time ethernet card, because SERCOS II is synchronous and the period between timings is fixed. My module even operates in user-space!

To sum up, the whole Sercos II protocol was re-implemented, and I developed a Linuxcnc module that prepares a message. The code is available here: github.com/arabel1a/sercos2-linuxcnc

However, it is more proof-of-concept than a plug-and-play solution. The main drawbacks include:
1. You need a capture synchronization messages (first few seconds after startup), I wasn't able to reimplement this stage.
2. STM32 hardware code (which just sends a message received on ethernet) can't be published due to some legal reasons.
3. The code is a bit messy, and not any configuration is supported. I used my drive in velocity command mode with feedback other modes are untested.

Anyway, my machines have been working for almost a year, and you can reproduce it with the code above. It costed me <$500 and a nine months of work, which is much less than a brand new NC unit from Trumph ($10,000!) If someone would be interested in making a more user-friendly solution, I would be honored to assist you -- contact me via github.

Today I was sitting with a frequency generator
my notes, if you enter only channel A in txt, the counter counts 0,1,0,1,0,1 instead of 0,1,2,3,4,5, there is no acceleration, and velocity = 0
if you connect A and B to the same pin, then at a low frequency the counter is 0, apparently the board understands that the pulse comes at the same time, if you increase the frequency above 1 kHz, errors begin, and the counter begins to increase or decrease non-linearly, well, it seems the simplest thing is to exclude channel B from the calculation, but I don't have the knowledge at all to write such a commit...

and I also need to use the PRU encoder module, not just an encoder, right?

@nhanpan You probably need to pin the ethernet irq to another core, preferably an isolated core. The irq you can get by entering the command: cat /proc/interrupts and find your ethernet adapters. In case yours also start with "enP" you could enter:
grep enP /proc/interrupts
To pin them to cpu5 with the above result as example you could add to /etc/rc.local in a line above "exit 0"
sleep 15 && echo 20 > /proc/irq/185/smp_affinity & echo 20 > /proc/irq/186/smp_affinity

It seems that the error tells that the kernel doesn't support that, perhaps a missing module? But with the irqs pinned you can ignore the message.

If you have used the PnC-Configurator to create the config then its called 'STEP_SCALE' , in the relevant [JOINT_n] section. (Presumably [JOINT_3])