Advanced Search

Hi all, its been awhile sorry, not had much time for anything of late and I've just been quietly using my linuxcnc converted bridgeport, but it seems I finally will get to play again.
I decided I have need for a cnc plasma table, so I started to put together a list and get parts.
So far I have:
Usual old PC for control,
Mesa 7i76e for control,
Mesa THC-5,
3x 3Nm closed loop steppers with drivers + psu's, the closed loop function is handled inside the stepper driver so they look like normal open loop steppers to the mesa, I hope they dont get affected by interference...
A 200mm travel z axis
Various limit switches
4 HGR20 rails with trucks and ball screws for x & y
P80 plasma torch suitable for z axis mounting
 

For a table, I have a used very old JCB backhoe chassis that had become damaged in a roll over incident and had to be cut up
I rotated the backhoe chassis and started to cut off extra bits, and put it on heavy casters to roll about already, and am in the process of cleaning it up to mount the rails directly onto it via tapped holes then I'll shim the rail->chassis where needed. 
A chassis for a digger weighing 8t is going to be plenty rigid & I can move it round with the fork truck if needed without worrying about it distorting. I'll mount the control electronics inside the section of the frame at the front that used to form the machine's diesel tank, so it will be good for sheilding + heat dissapation.
 
I'm posting this because I'm a bit hazy on the details how to connect a THC to the plasma unit.
I have a mesa THC-5 ordered, but I'd like to use my Stahlwerk Cut60 because its 380v tri-phase and I have a amp limit to run the table + compressor.  Someone from Stahlwerk told me when I enquired that it was scratch start and I'd need to buy a later model to get HF start, but if you hold the tip near the work you can see a small spark jump before it establishes the plasma and it doesnt need to touch to establish so perhaps something was lost in translation.
I read that the stahlwerk plasma cutters need modification to get some of the signals out for cnc use, my search is failing me for finding more info on what's specifically needed to be done inside this plasma cutter itself, any links?

Also I have only one ballscrew for the table long axis (2m), is moving the gantry axis from one side only going to be ok? I'm a little worried that rapid moves will cause it to go out of square. I dont mind slowing it down a little if that can help eliminate this, I dont care if I have to wait a bit longer for parts.

Looking forward to finally making progress on making something again, and your comments.
 

Which of these components can negatively influence the performance of the constellation Linuxcnc with Ethercat?

I tried the donker optocoupler for starters. It can't keep up with the speed of the servo. With the isolation, oddly feedback still occurs.
The 7i77 isol I have to shoehorn into my enclosure. I'm holding off on installing it for now.
When switching the drive to current/torque mode, where the drive no longer reads in the encoder, the feedback problem goes away. Probably still there just not traveling back / assuming influencing velocity feedback through the encoder in the drive. How is that possible?
I struggle tuning in current mode. I can't really get the motor shaft locked and slow speed seems very coarse/rough running. 


Here is a link to what's going on prior to switching to current mode. The analog feedback from mesa is disconnected shown at the end to illustrate that the interference is not coming through the 7i77. 

 

 

You were probably running base period at 10kHz, not servo period, although it might be, there is no need for it.
I have big industrial machines running 1kHz servo periods with Mesa hardware, never had a single issue in probably more than 10 or 15 years (getting old...).
A single Mesa 5i25 (PCI) or 6i25 (PCI-E) or 7i92T (ETH) can replace both prallel ports, and can do steps at MHz rates on any of 34 pins, while parallel is limited to 5-20KHz.
A Mesa 7i96S would be a better choice, much simpler wiring, and if you do need more I/O, you can use one of the old BOB with it, just needs an IDC26-DB25 adapter cable.
As for the modbus stuff, they are simpler now, but do require some editing, although it might just work depending on what was involved in it, probably spindle pins will need adding a spindle.0. anyway as later versions have support for multiple spindles.

The 7I96S at 1 KHz should work fine unless you need extreme acceleration
(say > 5G)

Here is a plot of 1 KHz stepgen errors on a 7I96S at 20 IPS (= 508 mm/S)
with 240 IPS^2 accel (6096 mm/S^2) Note that the error (pink) scale is 10 uInch
per division ( 0.254 u per division)


 

Hi, I'm running 2x lot port +bobs but add on lpt card died on me and I can't find  a replacement.

I'm tempted by Mesa Io cards and anticipating my pc dying too(it's core duo lga775 dell antique) I'm considering thernet cards 7i96s to be specific.
I like all on one solution with one cable for pc and not having to deal with chinesiumn BOBs.
Now, I was running 10khz servo thread, but I read that 1khz is "standard for Mesa ethernet 2khz possible and 4khz extreme, reported by one user here.
 I'm afraid that even at relatively slow feeds/rapids and accelerations I'll run into overshooting problems, in the end 1khz servo thread and 100mm/s rapid moves 0.1mm between updates. 
It's been a while since I went through the set up, so probably I'm mixing something important here but wouldt 6i25 be better at that respect?

If going 6i25 is there a DIY solution for getting analog out of this? I don't think I have it in me to rewrite after reinstal(I'm on lcnc 2.3) spindle vsd modbus module, and I suspect copy-paste may not works in later build..

​​​​

I'm having the same problem as well. Seems this is a common problem with no obvious solution. Step count stays = 0 in halshow. I started with the pendant config file from linuxcnc.org/docs/stable/html/man/man1/xhc-whb04b-6.1.html 
and modified it to fix the duplicated pin mappings (there were not too many)

I've tried the [HMOT](CARD0) change suggested. That doesn't work.
I've also tried all of the various other fixes from people posting on the forum to no avail.

I can see the jog dial send messages when I run xhc-whb04b-6 -ue so I'm certain the pendant is not a dud.
However when I run xhc-whb04b-6 -p I don't see any messages from the jog dial. Is this expected?  I'm beginning to think maybe there is a bug in this version of the driver.
Linux 2.9.2 using 7i96s Relevant hal files attached.

 

Thank you so much for your prompt reply. Great project indeed.
You have still the DC motors which I replaced on my machine already. I will take pics tomorrow and post them here. The hard ware from the 1. retrofit is still like brand new, the target will be to use much of it.

Yeah, that's a normal (non-RT)  kernel

Yes, I'd say you are lacking some feedback signals.

Try to add these to your hal:

net tool-prep-loop iocontrol.0.tool-prepare iocontrol.0.tool-prepared
net atc_zylinder iocontrol.0.tool-changed

[edit]
Note that the last line is probably not correct as you would really want a signal that reflects the tool change is complete. The usual manual toolchange has a popup window with an 'OK' button so the user can confirm that the toolchange has been completed.

[edit2]
Given the title of your topic I presume you are aware of this component but I do not know if you are actually using this along with your hardware button, which may be a good way to confirm tool change completion:
linuxcnc.org/docs/html/man/man1/hal_manualtoolchange.1.html

Good stuff.  Along similar lines, one of my friends designed custom keys with tiny displays in them so he could make a fancy CAD macro pad for Fusion 360.



I can't find a good photo of the current rev of his custom keys, but here's one of the earlier prototypes: