Advanced Search

Probably need the drive manual for that...

If it's like a classic motor drive, there will be three nested PID loops:

1. Torque (current) command/feedback --> torque (current) out
2. Velocity command/feedback --> torque out  
3. Position command/feedback --> velocity out

Is there any tuning program available for the drive?





 

And you can build a Linux system entirely from source, if you are into that kind of shenanigans.
Boot from a tftp server and run the OS entirely from ram. Or boot from a tftp server and mount the rootfs from a third machine.
A lot of home routers/access points/modems run Linux.

Running a Linux desktop is quite boring and mundane when one realises what Linux is capable of, and what a user can do with it once they read & experiment.

That's correct. I'm going to try to adjust the gain settings and see if I can get this tighter. Any suggestions what settings to look for? Many thanks

As had been suggested before, it looks like the drives themselves need tuning.

Not a firmware issue.

I would not expect motion until you enable the drives.

If there is an analog output issue, it typically shows up as a non-zero
output voltage when the 7I77 is powered.
 

Hi tommy, 

The voltage was the same at both drivers. I have dug a bit deeper and used Halscope and probed 

Joint.0.motor-pos-cmd against hm2_7i95.0.encoder.01.position

For both X and Y, I cut a few circles using arcs, and this is my result. Looks like the motor isn't keeping up with the encoder?

I wasnt sure what /k to use, so I have attached two screenshots, one all even /k and some scaled to fita better curve. This is all new to me so apologies in advance. 

I'm using Step/Dir Servos, I have just temporarily fed the endcoder back to LinuxCNC for testing. 
Im using 1kw ac servos by Litchaun 4NM the machine is a Haas mini mill. any thoughts?
   

  It works perfectly, will attach my hal for others to use

weedsmoker wrote:

that things sick, i wish i worked around something that epic and complicated,

And terrifying !!!
You see those big things on both ends of the machine with the rotating parts? Those move at 60m/m when at full tilt so the whole concrete floor shakes so much that the left servo drive errors out with over current!!! No errors when moving one or the other, just both. At 50m/m all is fine.
15cm of concrete is not enough rigid for all that monstrosity, but no one ever told them. I would venture a guess at least 30cm or 50cm just to be on the safe side, but i am no expert.

This is the third time i am reading this and still makes no sense.
If you browse a bit through this forum, you can see i have/use/built over 20 CNC plasma machines, so it is very hard to understand why would you change something that works perfectly for something else that will not work at all, or will work ONLY if you have a perfectly flat table and perfectly flat material.
Makes even less sense if using THC as that takes over control of Z axis, so does not matter what you put in the tool table.
Lack of replies should have alarmed you to something being off as there are plenty of Plasma machine users here willing to help anytime.
Material will warp during cutting, always, some more and some less, so using a plasma with no THC is possible but not enjoyable at all, and causes messed up nozzles and bad cuts.
I am pretty sure QtPlasmaC has a feature to skip probing for a given distance to reduce probing time, but i never use it.
I want to press "run" and go do something else while the machine does it's job.

Pictures of the machine?

Did anyone contact you about this?

That means if a PC running Linux happens to break, you can just take out the HDD and plug it into another PC and it will work no fuss, no muss.

Yes, for the last 20 odd years Linux could do that, even through external USB cases and/or CD/DVD rom's.
That is very hard to explain to "PC experts" who used windows only as that never worked, till win10 had that haphazardly implemented and intermittently working about 5% of the time.
"""ntldr not found"""

Just guessing that the port migration is because USB communications were
lost due to EMI as soon as the VFD started.

Does the VFD have a line filter?

I would keep the USB cable as short as possible and try wiring the RS-485 connection
through a common mode choke (a large ferrite bead)

Also check the VFD frame grounding

VFDs radiate EMI like a radio station, and USB is typically quite poor
in EMI resistance.

 

If the ferror is in the opposite direction of the velocity plot, FF1 is too high (first plot, FF1 = 1.8 is too high)

If the ferror is in the same direction of the velocity plot, FF1 is too low (second plot, FF1 = .027 is too low)

On the polarity, it should be checked with a voltmeter and the most positive lead goes to the OUT+ pin.

Spoke too soon.

After verifying the VFD parameters and the loadusr arguments to insure matches, I was able to issue a M3 S6000. The spindle enable relay activated and the spindle ramped up to 6000. However, M5 did nothing and I had to stop the spindle with the VFD control panel.

Next I tried to rerun the M3 S6000 command. I heard the relay activate, but the VFD did not respond. I shut down and restarted LCNC this time it failed to restart complaining that about not being able to communicate with the VFD. Checking dmesg I found that the dongle had migrated from ttyUSB0 to ttyUSB1? I updated the custom.hal and restarted LCNC without issue. This time I issued a M3 S10000 which worked perfectly. but again M5 was ignored.

I've now gone through this sequence several time and each time the VFD starts at the requested speed, then fails to stop. Each time the dongle migrates to a different ttyUSB1/2 port. The last time I issued a start command I noticed that the pyVCP display did not show the speed nor at speed and as I was monitoring the display I saw the OS notify that the dongle had again reattached at a different port.

At this point I'm confident that I can communicate via hy_vfd to VFD, but I'm frustrated that the dongle keeps toggling between ports on its own. Any ideas what is going on and how to resolve the issue?