Advanced Search

Hi, I have now made some changes and am not experiencing any more problems for the time being.
1. I had Ethercat running on two network cards.
I have now combined them into one card and made sure that it has an Intel chip.
2. I have made the following changes in ethercat.xml: 
appTimePeriod="1000000" >>> appTimePeriod="2000000"
sync0Shift=“25000” >>> sync0Shift="50000"
Since then, the errors have disappeared and the servos are ready immediately.
3. I have attached ferrite cores to the Ethercat cables.
The ferrite cores for the servos have been ordered.

 

As I understood open loop drivers and motors are existing and are kept in the system. (for the time beeing)

It is probably difficult to answer the question. What makes a controller better (best)?

• cheap? (two parallel ports (PCIe cards) in old office PC)
• least amount of re-wiring? (existing parallel port machine?)
• cheaper? (but not a bit of documentation) ( FPGA cards of unknown origin)
• Is wasted time cheaper than money spent?
• easy to set up? ( pnc config)
• upgradable? (linear scales, tool changer, rigid tapping, closed loop positioning, linear motors)
  
  I used an old parallel port machine, changing form mach3 to LinuxCNC was 90% figuring the "charge pump" out (no re-wiring, all in software), then the computer quit service and the replacement had no parallel port. (a 10$ PCIe card fixed that). 
  Having problems to get an encoder running a Mesa card (6i25 with daughter card) went into the machine. (solving the spindle encoder issue)   (it still is a Frankenstein mixed Mesa and Parallel port). 
  
  Greetings Timo 
  
  p.s. the UC300ETH is that working with linuxcnc? Some older threads stated it does not work.
  forum.linuxcnc.org/27-driver-boards/2982...moothstepper-to-mesa

Does anyone have experience with this grease gun?
www.sorotec.de/shop/Professionnal-Grease...set-120ccm-6796.html

At €100, it's not realy cheap!

Thanks. Now waiting on delivery of a 7i52 to test those serial ports, no luck via 7i42ta.

However, I've had trouble from the start with the 7i48 card - the 7i80HD-16 wouldn't read its encoders at all, and the 7i80HD-25 had some trouble, using encoder.00 to .03.

Last night we tried using the 7i48's encoder.04 channel with one axis at a time, and found that all 3 worked perfectly. Then we tried adding all 3 axes, at which point no aspect of the 7i48 works at all: the +15V, -15V and 3.3V lights come on, but there is no +/-10V from any AOUT port, and no encoders are read through any channel.

If a cascading fault is possible (caused perhaps by the one attempt with a 24v encoder) then I seem to have bricked my board, though I don't know if that's consistent with channel .04 having worked perfectly for half an hour of testing last night.

Please would you remake the above firmware for the 7i80HD-16? I would like to try to rule out the fpga board as the thing that's broken here, before we order another 7i48. I'd note that the reason I tried the 24v encoder was that the 7i48 wasn't reading encoder info at all to start with, though when I switched to the HD-25 it appeared to partially work, making me thing that the HD-16 firmware may have had an error.

If you have any other suggestions I'd be grateful to hear them.

To come back to your original question, if I were to look at the gcode running on those machines, would I see more code running in G61or G64 mode?

Generally, how important is constant velocity along the toolpath for your applications?  

Agreed it's outside the scope of the OP

Just once again thanks for everything.


I'll go back to watching and hoping you guys get this running nicely


Cheers andrew

Bear with me:
I feel like I have read every article known to man on this subject and watched all the videos on YouTube that I can find but after several weeks cannot start my spindle via software.I am using Linuxcnc v 2.9.8, with QtDragon_hd as my GUI.

I have a 2.2kw water cooled spindle hooked up to a Huanyang HY02D223B inverter. I have a Waveshare RS485 USB connector, wired D+ to RS+ , and D- to RS- on the VFD. The shielded ground is connected to ground on the PC and COM on the VFD. The adapter shows up as ACM0.

VFD jumper has been switched to VI, and programmed as follows:-
PD001: Set to 2 (Source of run command = RS-485 communication port).
PD002: Set to 2 (Source of operating frequency = RS-485 communication port).
PD163: Set to 1 (Communication address of the VFD).
PD164: Set to 1 for 9600 baud 
PD165: Set to 3 (Communication method: 8 data bits, no parity, 1 stop bit - 8N1 RTU). 

While linuxcnc loads, if, after homing, I select MDI and type M3 S1000 (or any other value), nothing happens.
For a while I was questioning the wiring to the spindle/mains and had the awful thought that either the spindle or VFD were duff.
Last night I found this thread forum.linuxcnc.org/24-hal-components/310...anyang-speed-control and when I followed these steps in Terminal, voila! my spindle runs.

I have attached my HAL file and a screenshot of my procedure in Terminal in the hope that someone can explain why I can't operate the spindle from within my GUI. 

Thanks in advance.

Do you think it is even possible to make a general homing component?
The drives I have seen have numerous different homing options.
Will all of them work?

The cia402 component has no easy task.
The device runs the cia 402 state machine and the cia402 component has to figure out what state it is in. 
From that, rebuild the control word to match what you want it to do next.
You can't let the cia402 component be the master of the state machine, the device is.

I don't know much about how homing is supposed to work,
otherwise I could try to help more.
 

This looks great, have anyone been able to port it for LCNC?
www.pibot.com/pibot-cnc-pendant-v4

We make machines, on our factory floor, we always have 3-4 machines on the floor I can test, some complete some partially because in construction. our customers have multiple shops running LinuxCNC, but it is in the background, and they don't really know, they only look at CAM, CNC for them is mostly open gcode start, regulate feed, stop if have problem.
The biggest changes we made you have them now in master, not sure how my implementation can help, and it is not in the scope of the OP.

Hey guys have a look at the link to the YouTube channel

These guys have a whole factory running linuxcnc I think.

Is awesome.

I have 7 machines here in NZ but these guys have way more stuff for testing.

Grandixximo.

Can you take some more videos of your machines please
Even a non promotional video for us linuxcnc guys where you can really dig into how your have implemented it.

This is perfect for testing 5 axis stuff and your jerk control.

And I just want to say thanks for trying to help the open source community.


Not everyone does that.

Cheers

It does seem something is not set correctly as setting deadband should not trigger joint error, at least i do not recall it ever happening to me, but i do use lower values:
DEADBAND = 0.0001
DEADBAND = 0.001
as a starting point, and go up or down a bit till i get soft noiseless motor standstill.
For a Delta drive, i would try with 0 I and 0 D values, find FF1 where machine would run at full speed properly, then set a very low value for deadband if you end up needing it.
First and foremost, find the P value where the motor does not lag to much at full speed and does not vibrate at standstill.

my1987toyota wrote:

And even if it doesn't go anywhere this time governments will continue to try again. It's how the American
Government has gotten away with mandating kill switches into all new road vehicle
(2026 forward if I remember correctly). There is another bill that is suppose to repeal it (H.R. 1137) but 
I don't know if it passed or not. Given all the other crap that has gone on I wouldn't be surprised if it
got lost in the noise. 

 

As a side note from an earlier post, NO they have not repealed the stupid vehicle kill switch thing. It's just
the law hasn't been implemented yet.

 

Inadequacies in the milling path should not be compensated for by violations of the limits of jerk and acceleration 

It is not the case that we try to fix inadequacy of the path by spiking the jerk, is that we cannot control the jerk spikes period. This is not because we don't want to or lack of trying, is IMO fundamentally a limitations of the current tp architecture. YangYang is trying work arounds...

Yes, most machines you see on the website do, the one I have most problems with is the sawjets, because of the huge kinematics pivot.
The only issue I found that needs fixing with the latest code, is when I do codes to mill down the table, zigzag style, YangYang will take a look next week, he keeps patching that up. And then it comes back again...