Advanced Search

Hi Everyone,
recently I was trying to get started with EC500. I was able to flash the firmware and was able to ping the board on the 10.10.10.10 address. It worked nicely with a pico probe and pyocd.

However, I continuously get joint following errors on all my joints while jogging and while executing gcodes. I cannot get rid of them.

Also, the following error is very high, even at low speeds it is more than 1mm. I also tried to increase the P gain or to increase the spike filter in remora-eth.c component without success.
My computer has the following latency  values:
 

If I ping the board over a longer period of time, I get following values:
 
There are some outliers with high ping, but they are rare.

I already tried to reduce the jitter:
- Isolated two cores
- Moved kernel threads to the non isolated core
- Moved NIC interrupt to the isolated core

I can think there can be two source of problems now:
1. Latency jitter of the NIC
2. Latency jitter of the PC (Fujitsu Esprimo P556)

Can you give me some advice please how to debug this problem? How could I trace down which one is the issue?
Thank you for your help.

Yes, you are right ! That's why it is removable when you don't use it.
The idea is to allow our users to engrave the sheet of steel with the laser, remove it then cut the sheet with the plasma torch (with the right offset).

Hi,

I have the a similar issue but with one 5i25 and two 7i77. With Halshow, I see Parameters / hm2_5i25 / 0 / 7i77 / 0 / {0, 3, 4} (7i77 digital on P3, 7i77 digital on P2 and 7i77 analog on P2). I think I should have 0, 1, 3 and 4.

I work with a fresh install, linuxcnc 2.9.2
I used mseflash with 5i25_7i77x2.bit
The 5V is powered by the 5i25 and the leds are on (but I measure 4,7V  and 4,8V ) . Field power is 24V.
In the firmware tab of MesaCT, I see the Smart Serial channels 4, 5, 6 on P2, 1, 2, 3 on P3.
The config line generated by MesaCT (because I don't succeed in finding my two 7i77 I/O with PncConf, this may be related) is :
loadrt [HM2](DRIVER) config="num_encoders=12 sserial_port_0=00000000"
If I switch the cables, that's the same board wich remain without analog (so instead to be on P3 it's on P2)
If I disconnect the board on P2, I still don't see the 7i77 analog on P3.
If I disconnect all I/O cables on the 7i77 it's still the same.


Any idea to look at ?

Thanks

No interest ? No one got the time ?

hm2_spi uses the /dev/spixx device. So you have to tell it which spi device to use.

hm2_spi only works with the 7i90HD. So unless you are using this board it won't work out of the box. You would have to modify the driver to suit the board you are using. No that's not something I can do as I have no Orange Pi to test on and I know for a fact that after trying this driver on a RPi5 it is not suitable for running a machine.

All the information needed to load the driver is at the top of the man page. There is no special requirements that I'm aware of.

Here are some parts you need to be aware of:

hm2_spi is a device driver that interfaces Mesa’s SPI based Anything I/O boards (with the HostMot2 firmware) to the LinuxCNC HAL.

The supported boards are: 7I90HD.

The board must have a compatible firmware loaded on the board by the mesaflash(1) program.

hm2_spi is only available when LinuxCNC is configured with "uspace" realtime.

As of kernel 3.8, most or all kernel SPI drivers do not achieve the high realtime response rate required for a typical LinuxCNC configuration. The driver was tested with a modified version of the spi-s3c64xx SPI driver on the Odroid U3 platform. The patched kernel resides on github.

"However I've got a mesa board and I did synthesize on its FPGA."
What mesa board ? The 7i90HD have been out of stock for a while.

Or did you use a spartan-6 dev board and build firmware to suit ? Now this will be tricky if you don't have an RPi4 to test with. There's a few gotchas in rebuilding Mesa firmwares for dev boards, the first being making sure you are using the correct pin for your clock, then making sure the sources have the correct clock setup, unless of course you have a 50MHz clock on the dev board.

But the very first thing I would do is too confirm that you can get SPI working out of the context of Linuxcnc, this will at least confirm that the SPI hardware does work whether this is via C or Python or whatever you fell comfortable with, I'm sure there will be plenty examples of this on the internet.

Just be aware trying to go a cheaper route with hardware that is not very well proven may in fact cost more in the long run. Also it will be far more frustrating. It will also require lots of experimenting and "I wonder if this works" attitude.

It took me a while to work these things out using dev boards to work via EPP on a x86 platform and via SPI on the RPi4, experiments on the RPi5 with SPI proved to be fruitless.

Unfortunately these are the realities of going down the less trodden path. And honestly with out the help on the forum from Pete at Mesa I'd still be trying to get things to work rather than being distracted by other projects.

Actually I did read about it and I know it is not a generic driver.
However I've got a mesa board and I did synthesize on its FPGA.
But the thing is I run linuxcnc on my orange pi 5 plus and I don't know how should I use the HAL SPI component to be compatible with orange pi to communicate via SPI with that mesa board.
can u explain a little more about this?

Just putting it out there, that laser can not be mounted there while plasma is in use due to huge amounts of very fine dust created during cutting.

I found the PathPilot ATC stuff.

All sorts of custom hal components, python files, and .ngc subroutines.  Plus the ATC itself is sent commands via a PWM signal (i.e. freq = pocket number).

It's hideously complicated. Yuck.

There may be some helpful information in these threads:
forum.linuxcnc.org/plasma-laser/51727-be...th-qtplasmac?start=0
forum.linuxcnc.org/plasma-laser/52597-bu...nuxcnc-and-qtplasmac
 

If your machine does not have a built in arc-ok signal then this is an effective way of providing a signal:
linuxcnc.org/docs/2.9/html/plasma/qtplas...l#plasma:reed-arc-ok
 

Hi,

I'm configuring a kind of lathe with one 5i25 and two 7i77 and I'm facing the same problem : in the spindle tab of Mesa CT, I can't select anything; all buttons are greyed out.
Where does it come from ?

Jôrg, did you terminate the configuration by hand or with Mesa CT? You did'nt explained what's wrong with your spindle tab.

I've another problem with analog outputs but I just saw Peter's answer (forum.linuxcnc.org/24-hal-components/528...how-and-is-not-found) so it should be ok. In case it would be related... I would try it tomorrow.

 

You can use the THCAD2 with a parallel port based LinuxCNC system.
 

Setting steplen to 2 sets the step length to 2 ns but this will get rounded up to
1 base thread period since that's the shortest period between output changes
unless you use reset mode. Same with dir setup and dir hold, they will get rounded
up to multiples of the base period.


 

Thank you Peter. I will dig a little bit more on the DoubleStep as an option.
So, setting the steplen parameter to 2, it will set the steplen to 2 nanoseconds or it will set 2 periods for a step pulse?

I have a small robot arm that I am attempting to control with LinuxCNC. Ultimately, I would live to have full IK... but for now just being able to control it is fine. I used the PnCConf Wizard and even though all 4 axis are rotary... the best I could do was an XYZA configuration... it works... I can program the robot and tinker. Great.

Now I am trying to add a 5th axis. I have a cross-slide so the robot arm can move side to side. I made a copy of the X-Axis and renamed it to the U-AXIS. I then went in and changed it from AXIS-0 to AXIS-4, changes all the other "stuff" from "X" to "U" which worked fine when I did it on a parallel port machine in my murky past.

But this... it has a failure and I just can't figure it out.

Anyone here able to see what the issue is... I'm simply not seeing it.  

Error List (Debug Output)


Working 4-Axis Files
 
 

Failing 5-Axis Files
 
 

Thanks in advance,
Jerry