Advanced Search

Got the new version, built and tested. This one works as well, no big difference from the previous version.
Maybe the linearity is a bit better, it goes to zero Volt now when there is no voltage.
 

View from initial test.
 

I just upgraded from a 2019 version of PlasmaC to the current QTplasmaC on a new SSD. I now am having massive issues with latency/jitter. I’m running a BRIX mini pc that’s sporting a J1900. In the past it was fine but now it wont even run. I went into the bios and disabled everything like I did in the past and disabled Wifi. I think befor my max jitter was like 20k but now the max jitter is Max Jitter (ns) 129316 / 106580

Am I just out of options with this older mini pc? Is it time to look for a more powerful cpu mini PC? 

thanks for all your help. 

For your coolant flood, you would use the "toggle2nist" component.
It has 4 hal pins:
toggle2nist.N.in Would connect to the output signal from your button input.
toggle2nist.N.is-on Would connect to the output signal from the hal pin "halui.flood.is-on".
toggle2nist.N.on Would connect to the input signal to the hal pin "halui.flood.on".
toggle2nist.N.off Would connect to the input signal to the hal pin "halui.flood.off".

There is not any pre defined pins designated for lube or spindle release. You must build your own logic to handle these functions.

Often machine integrators will set up custom M-codes (M100-199) that turn the required hal pins on or off. Then implement special logic (Hal or PLC) to prevent or enable the states as necessary. Such as preventing the tool from being released if the spindle is running.

The issue is that the disk is not listed there.
Motherboard is MSI. I will try something else later today, but does not look promising.

I would consider integrating a drive using RS485 a more advanced subject.

Start with reading the Linuxcnc documentation, and begin with a basic configuration for running your machine without spindle control. Then after you have a better handle on Linuxcnc, and the basics of Hal programming, try to integrate the modbus spindle control into your configuration. There may be an existing modbus driver compatible with your drive. (I don't know, never used one.) If so that would probably make integrating RS485 control much simpler. But if not you can always make the generic Mb2hal driver work, but this is a bit of a chore to set up and configure. (I have done this before.) The generic RS485 drivers are a lot of work to set up because you must research all of your drive's modbus settings and communication parameters, and every drive (even different models from the same manufacturer can) have different parameter arrangement, there is no "standard" way drives do modbus.

If you are following steps which I have desribed somewhere here at the forum... It has to working properly every time on startup.. I am running profibus, profinet and canopen on single koppler in same time over ethercat

After koppler restart have a look at command -

sudo systemctl restart ethercat.service

... But U are decribing other type of problem... I am currently working on C# console app which will generate us configuration from ethercat slaves && ethercat pdos commands ... I have beta version but it is not too adult to outcome on public... Do U have a TC3 project?

@Lcevette,

that is awesome!!
We hope so.

I had to add extra logic to the arc-helix.
When the arc-helix is trimmed at the front, i had to update the center point.
As the old center at the front is off plane after trimming.

The result looks ok now. Helixes in plane G18 & G19.
Deviation = 5.
 

If there are shorter segments, the clothoid fillet previews to be tinyer.

Also working on the look ahead now.

Press F11 or F12 or on some PC's F8 after power on, see if you can select that SSD to boot from and also notice if there is more than one entry for it, usually the name of the SSD as one entry and another with UEFI SDD, try both.
-
Edit:
Is it an ACER?
those require setting a BIOS password to be able to disable secure boot.

that is awesome!!

Thank you for the reply.
I did what you suggested, before my post, and the motherboard does not recognize the SSD. Strangely it displays a very fast message that I was able to see:

There is no vBIOS support detected in this card.
'BIOS CSM/UEFI Mode' settings in BIOS will be changed to 'UEFI'

and automatically it moves to the BIOS screen, and the previous changed settings are converted to UEFI again.
Note that I do not have a separated graphics card.

Disable "secure boot" in BIOS.
Also enable "legacy" or CSM or both.
-
Changing from MBT (should be MBR or Master Boot Record) to GPT will destroy any info on the drive, so to do that would require cloning the partition (not the drive) to another drive, switching partitioning and then cloning back that partition. Either way you look at it, very dangerous and not much chances of working properly.
And more importantly, that is not the reason for failing to boot.

hello,
Thank you for your answer.
Sorry for the late reply. I've been a bit ill and didn't get a chance to test it.

I have changed the line as described.
It actually seems to be working better now.
I still have the problem that when I restart the coupler it doesn't work right away... it only works after I have loaded the configuration once with TwinCAT. I think I'll have to see which CoE register might have a different value.

@Konrad,

Thank you!

Besides that can I ask if the points raised by hmnijp ... are addressed?
A few are adressed now.

The algo is still using local smoothing. as shown on page 3.
But now it uses a 3d-clothoid instead of a spline, and uses the correct path deviation to fit the clothoid.

I guess to really have jerk limiting you not only need to be smooth in the time domain but also in the path, thus the use of clothoid(?)
1. The 3d clothoid fit's the trajectory with G3 continuity, as explained in the abstract.
2. Then we have scurve motion profile, wich is used for jerk limiting.
3. Then the trajectory can use different algo's for path optimalisation, like flattening out a zigzag motion. This will also contribute to jerk.

The ringbuffer is working. And to my surprise i did'nt need a seperate thread so far to fit clothoids.

 

Good morning.
I had a working system using a old linuxcnc version (+- 5 years old). It was running fine, but the motherboard needs to be replaced.
I was unable to boot the same SSD on the new motherboard, so I suppose that from what I have read that the issue is more likely because the current installation was done on the old MBR.
The new candidate motherboard only has UEFI so for what I have read, I must convert the current disk to use GPT.

I have read some online information, and there are many steps involved on the conversion. Should I try to start from a new installation or still the conversion would be less trouble. What is your opinion?

Thanks for your reply, andypugh.

Unfortunately, the counter and position values ​​of the EL5101 being used seem to be reset the moment index-enable is turned OFF. The position where it stopped at the end of homing is the correct position. The coordinates of LinuxCNC are 0, and the POS position of the EL5101 is the same as the HOME_OFFSET value. It seems that it is not possible to avoid the excessive PID deviation when index-enable is turned OFF. This phenomenon does not occur when homing is performed again.

Until now, I have tried converting the Count value instead of the EL5101 POS and using it for Feedback, but it did not work.

It worked fine with the EM7004, but I'm starting to think that an external torque mode amplifier cannot be used with the Ethercat driver.

Thank you.