Advanced Search

Hello dear forum, (Translation GER-ENG Google)

Short introduction:
I've been a fan of LinuxCNC for over 15 years, and unfortunately due to a lack of time (I'm sure many of us have) I've neglected my hobby a bit in recent years. I'm a bit out of programming in HAL and am looking for some food for thought or a solution to the following problem: 

Compensation between B and C axis required. 
I built a rotary/tilt table for my milling machine. Since my installation space is very limited, the whole thing had to be relatively small.
Harmonic drives were used for this purpose, the shaft runs through the HDrive of the B axis to rotate the HDrive of the C axis. The whole thing is solved via belt drives. Since the motor of the C axis does not pivot, it rotates when the B axis rotates. Now I need a compensation that compensates for the position of C every time the B axis moves. For one revolution of the B axis I need about 1/51th of compensation at C. My idea is to compensate for the whole thing using moveoff. 

Are there already solutions that might be better, or has anyone done something similar?
I would appreciate any help and tips.
Kai

my hardware:
Mesa7i92, 2xMesa7i76, Mesa7i84, Mesa7i73
Homemade SK15 ATC spindle, Lenze FU
Homemade 3 axle + BC,



Hallo liebes Forum,

Kurzvorstellung:
Ich bin sein über 15 Jahren Fan von LinuxCNC, und habe leider durch Zeitmangel (haben sicher viele von uns) mein Hobby die letzten Jahre etwas vernachlässigt. Ich bin etwas raus aus der Programmierung in HAL und suche einen Denkanstoß bzw. eine Lösung für folgendes Problem: 

Ausgleich zwischen B und C Achse benötigt. 
Ich habe für meine Fräse einen Dreh/Kipptisch gebaut. Da mein Bauraum sehr begrenzt ist musste das ganze auch relativ klein werden.
Dazu wurden Harmonic Drives verwendet, durch das HDrive der B Achse läuft die Welle um das HDrive der C Achse zu drehen. Das ganze ist über Riementriebe gelöst. Da der Motor der C Achse nicht mit schwenkt, rotiert dieser wenn sich die B Achse dreht. Nun Brauche ich einen Ausgleich der bei jeder Bewegung der B Achse die Position von C Ausgleicht. Bei einer Umdrehung der B Achse benötige ich an C ca.1/51tel an Ausgleich. Meine Überlegung ist das ganze über moveoff zu kompensieren. 

Gibt es dazu bereits Lösungen, die evtl. besser sind, oder hat jemand ähnliches gemacht?
Ich würde mich über jede Hilfe und Tipp freuen.
Kai

meine Hardware:
Mesa7i92, 2xMesa7i76, Mesa7i84, Mesa7i73
Eigenbau SK15 ATC Spindel, Lenze FU
Eigenbau 3 Achs + BC, 

Having completed a new CNC build, I am fine tuning it. It is a dual X-axis machine (XXYZ), and I was looking at using matrixkins to square the X and Y axis. The document man9/matrixkins does not reference dual-axis. However I found a note (written in 2022): 
"Dual-axis gantry machine support might be worth adding, but I don't have such a machine myself."

Was there an update related to dual-axis use? Where can I find the source?

Thanks,
Robert
 

No one has asked for a specific facility for manual tool setting, which would start with specific procedure. There are a few common tool length strategies.

'Touch Z' usually implies _coordinate_ touch off which QtDragon has buttons for (REF and ZERO).
For tool setting, you can of course manually change the offsets in the tool offset page but also when you are on the tool offset page, the 'HOME' buttons change to 'TOOL' buttons.
Pressing that button will give you several options for setting the tool offset.

If you have errors with a base period (not frequency) of 100000 reducing it will not fix it, it will only make the issue worse.

What you will need to do is run a latency test and see how things are going.
The first error that shows in your picture is a latency error, that need addressing first, you will need to set a very high base period and slowly work backwards.

A base period of 500 would yield a base frequency of about 2Mhz, a 250000 base period would yield a base frequency of 40kHz. The larger the base period the slower the base frequency.

If you were getting errors at at base period of 100000 i would double this to see how things are, you may reach a stage where the base period is so high that the machine may not be usable with the hardware\OS combo.

At some stage you may have to get access to the BIOS.

Be aware that when timings are talked about they are in nano seconds, it does make things easier if that convention is kept.

Hi,

Has anyone done a plasma interface with Flex GUI?

Just a short reply to end the thread.

At the end of the day I used external-offsets and it worked like a charm. No g-code has to run at all in my current setup.
linuxcnc.org/docs/html/motion/external-offsets.html

It's very similar to the jog-counts/jog-scale approach that was suggested here too, but I wasn't sure whether I would be able to control acceleration in that case. So I started with external-offsets and that was enough.

Thanks everyone for the advice!

And the slotted disk can sometimes rust.

I am 100% convinced it's a software issue. I was able to create a basic application using the Modbus library, using the same basic logic as the hy_gt_vfd hal driver, and it works. Meaning no communication error, and I can start and stop the spindle fine. So that along with the mbpoll that also works, it's in the hy_gt_vhf hal module.

I think that there is a memory allocation that stepping over the modbus context (where the port settings are)?
 

From what I can tell it looks like it is attached in Hal?

Hi langdons! Thanks for your question!
Yes the VFD has more than one output. One +24V for maybe switches or relays, two relay outputs and one +10v output that puts voltage to the electronic potentiometer on the 7i76e. The mesa doesn't have such a voltage output on board, so you have to give that from an external source (eg the VFD 10V out) to the spindle+ input. The spindle-output then gives the 0-10V voltage correspondingly to the set rpms back to the VFD's analog input and, if the FWD signal is on, the spindle turns.
In my case, the rpm output voltage AND the spindle.enable = spindle.on signal is only given out from LCNC if I do this from a file. Normally - and any time before, this works even with any MDI speed command.

So - the VFD (now) works as it should but LCNC has problems to give the right MDI Commands as ordered to the outputs of the MESA card.
The VFD had no +10V on its H output as it came to me, from the dealer, new product, no NOS, freshly produced 3 years ago. I wasn't ready to build that thing in because I was busy with the mechanic part of my CNC. I put some parameters in the first weeks, but no real testing with any controller, so I didn't mention that problem. I found that few weeks ago, as I installed the VFD and tried to get the spindle turning. And found that there is no +10V at Out H. Then, the external 10V source, but I wanted the VFD beeing allright, so I sent it to Hitachi for Service.

I wasn't able to go down to my machine the last days, but I will make a video of that process the next days.

Kind regards!

Hello atrex77,
this is a very nice project. I found this video , it looks fantastic.
A very fast LinuxCNC-Board for low cost.

You made a very good job, but is it useful in practice?
I think there are not enough input and output free.
In my opinion you need for a machine:
- for each axis 2 inputs (one side switch for reference and limit together, other side only limit)
- one input and one output for e-stop
- one output for enable the spindle additional to the PWM
- one output for the coolant
- if you have a machine enclosure, you need one input for the switch and one for enable the drives.

What do you think about this? Do you use it in an other way?
Do you already have a solution in your mind?

One thing about optical encoders, they really don't appreciate contamination. If oil or coolant get into them, it will generally really screw them up. Even if none of the electronics with are hurt, a little dirt in the wrong spot or on the wheel can be very bad. Also when they get really old sometimes the LED source can get weak or fail.

Latency is not really an issue with Modbus as it's not real time to begin with...

andypugh wrote:

M62 and friends need connections in HAL to have any effect. Do these connections exist? (motion.digital-out-NN etc)
 

Im not sure they do right now

This seems to be ocuring for other people too like this thread.
www.forum.linuxcnc.org/plasmac/42445-hel...s-qtplasmac?start=10
The feed reduction commands in gcode were not working for them but it never got solved. Other than using the post to generate the small hole fix.