Advanced Search

Thanks for that, I was not able to get into the forum for the last couple of days.

I had similar advice from AI, and tried both the solution yours and mine gave and it has not corrected it.

I even used visual code studio, and rewrote the whole program be hand, to ensure there were no secret character codes in there. (See attached file)But the result is still the same.
hal:turret:1: character unknown '  '

Any other suggestions much appreciated. 
Cheers,
Mark

spumco wrote:

Before you make any decisions, you've got to sort out power
 

THat's what I did today:
The VFD takes 480V directly from source (I am going to replace VFD and spindle, so it doesn't matter).
A 480V single phase line powers the DC bus which in turn supplies power to all 4 drives.
The transformer ties two of the lines into a neutral-ground, and outputs 120V to power the controller and peripherals. I replaced this with 120V AC to test if the controllers are working. Was told by seller that it takes a while to boot. It's abootin as I write this.

The servos are all 140V 65amp max. I do not need the maximum capacity (I am carving wood, and I am doing low throughput. I was thinking to sell the drives (they are out of production and I imagine someone who is running this machine in a production setting would want parts) and replace them with step/dir ones as someone else had recommended. The VFD takes 3 phase power at 360 to 480. I was wondering if I could just supply 220V power 3 phase to the VFD and have it power a 220V 3 phase spindle. This would make the work a lot easier, or just replace it with a single phase 220V VFD to power a 4ish HP spindle. That would work too. Either way, I really don't see why I should stick to 480V single phase since the part that is drawing power is the spindle which I really do not need.

placeholder

Aciera wrote:

What would be needed here is to have the _interpreter_ reject the axis word which was disabled. 

So the NML description in the LCNC manual isn't accurate? Or perhaps not exactly clear what happens?

I would have thought 'turns off all the calculations..." would have removed that joint/axis from the TP calcs entirely.  

On second thought it probably wont work like that  I can't have "v" defined in case-0 as well as an 'unused coordinate.'

But maybe there's a way to define the motion math of case 1  pos->v= j[0]; so that no motion results regardless of input.
 

I wonder if this could be done through the switchkins component.

Do you think LCNC would throw an error if I just left joint-0 out while operating under Case 1? 

Instead of this:
Maybe this:
And do something similar for the inverse later in the comp.

​​​​​​​

What would be needed here is to have the _interpreter_ reject the axis word which was disabled. So just like if the operator used an axis letter that was not configured in the ini file. Any MDI or gcode program line containing the disable axis letter would create an error 'Bad character x used'.

Unfortunately I do not know how to dynamically pass the disabled axes to the interpreter during runtime.

Not sure if there is an easier way of trying this but I inserted the the actions of 'EMCMOT_JOINT_ACTIVATE' / 'EMCMOT_JOINT_DEACTIVATE' (ie set/reset 'SET_JOINT_ACTIVE_FLAG' for a specific joint :
 

into 'EMCMOT_SPINDLE_ON' / 'EMCMOT_SPINDLE_OFF':

 
 


Which lets me clear 'joint.1.active' using the spindle control but it does basically the same as disabling the drive as 'joint.1.pos-cmd' is still updated as the axis is jogged or commanded through the interpreter:

 

The problem with the following error when reactivating the joint remains the same as if it was done in hal or the drive disabled. 
 

After you set it the first time, does it disappear from the sdos listing or does it just become read-only?

A servo of this size is not a toy.
The “Servo On” or “Enable” signal comes from the controller (PLC / lcnc).
The emergency stop and limit switches are hardwired.

Have you tested the module in Twincat yet?
If it runs without any issues there, you could use Wireshark to monitor the communication. Maybe that will turn something up.

I imagine that you will be satisfied with the updated xhc-whb04b-6 component by hdiethelm, which allows the pendant to reconnect in case it disconnects (as well as several other improvements). But you need to compile it from source, same as building LinuxCNC from source (which is not all that difficult). He made a backport to LinuxCNC 2.9.8:
github.com/hdiethelm/linuxcnc-fork/tree/...x-v2_backport-v2.9.8
So basically git clone that repository, checkout xhc-whb04b-6-fix-v2_backport-v2.9.8 and then compile. I don't know about exact version compatibility, but I don't think you actually need to use v2.9.8 to run the component. You can just give the full path to the newly built compnonent binary in the loadrt command.

I made a custom firmware for the USB dongle that I use myself, but I can't really recommend that for others. There is a decent risk that it would irreversibly break your dongle in case the hardware is slightly different than what I have.

I don't know your proficiency with git and compiling LinuxCNC from source. Just ask if my somewhat terse instructions are not clear to you.

desoler , je ne comprend pas ,car j 'ai bien le signal dans la config de hal

Aciera wrote:

I'm not aware of a way to block the operator from commanding a move of a particular axis. Would be a handy feature to have.

Appreciate you having a think about it.

I'd like to experiment with that NML message I found (DEACTIVATE_JOINT).  If I can deactivate joint-0 near the beginning of the sub-spindle transfer process I can simplify the parting-slide jogging disable (which happens later in the overall sequence).

Any idea how I can send an NML message to LCNC?

Maybe it's done via python interface, but I don't see anything obvious in the LCNC manual section on python-interface that includes that specific command.
 

Any ideas how to no-kidding stop an axis from being moved without disabling the drive? like DEACTIVATE_JOINT or something?

I'm not aware of a way to block the operator from commanding a move of a particular axis. Would be a handy feature to have.

- I use Gmoccapy where the onscreen jog buttons can be hidden and only physical jogging through hal pins needs to be handled.
- MDI entries are more involved as one would need to parse all commands that an operator enters, still possible without too much effort.
- Rehoming can be handled using 'motion.homing-inhibit'.

Still leaves program execution and any mdi commands that are not coming from the gui entry.