LPKF ProtoMat S62 Retrofit Advice

Hi, I picked up an LPKF ProtoMat S62 and am hoping to get it back up and running using LinuxCNC. I've done a bit of reading but think I need a push in the right direction. I’ve listed some information below but my hope is to run in with a Raspberry Pi and a Mesa 7c81. I’m not sure if I can interface directly with the original controller or if I’ll have to add in new drivers etc.

Can anyone suggest a good way to go about this?
Details below:

The original controller (LPKF SMCU II) drives the X and Y axis, receives input from the X and Y limit switches as well as the main door safety switch. It also has outputs which are designed to control a dust extractor.

There’s also a connector that goes to a DAC1005 (I think runs the spindle, Z axis and ATC motor) labelled LOCAL COM and pinouts:
CAN-H
CAN-L
GND
SYNC-H
SYNC-L
GND
PE/EARTH

Image of the controller

I’m not at all familiar with CAN bus but I think that’s the communication protocol it uses.
Similarly the SMCUII has an RJ45 connector (labelled CAN-A/CAN-B/RS232) with pinouts:

CAN-H
CAN-L
GND
NC / (RS232 TxD)
NC / (RS232 RxD)
GND
GND
NC


As a quick background:
The machine was designed to cut prototype circuit boards and has an automatic tool changer setup with 10 slots for tools, vacuum hold-down/extraction combination, 62K rpm spindle, fiducial recognition camera, all in a sound proof (maybe) box.

The original setup:
Brochure

X-axis and Y-axis
Schneider Electric BRS366HS0005

Z axis - Vexta 2-phase (model c9641-9212KGM) driven through a gearbox - 2.8V, 1A, 0.25deg/step
Driven via a DAC 1005 ?using RS232

ATC/Clamping motor
Maxon 114954 (24V DC) again controlled via the DAC1005

Camera
uEye UI 2410se-m-gl with a USB interface info

Controller
LPKF SMCU II

It looks like your drives use CANopen. Linuxcnc does not support CANopen.
At least not directly, apparently there have been users who have used EtherCAT as a go-between to CANopen. That path however will not be an easy one as there will be limited support for something as "exotic" as that :
forum.linuxcnc.org/24-hal-components/223...iver?start=110#49794

A much easier route would be to replace the motordrives. Although with 3phase steppers the choice will be somewhat limited. I'm also uncertain about the kind of encoder feedback in your system (seems either incremental or resolver).

Thanks for the advice! I thought that might be the case but didn't want to throw away the existing controllers in case I could interface with them directly. Guess it's time to figure out how to communicate with the spindle and ATC motor!

I suspect that the only difficult part of getting CANOpen working with LinuxCNC is the hardware interface.

It would need to be something realtime. In practical terms that means connected by PCI, maybe the Mesa UART or (though I hate to suggest it) the Parallel port.

All can be driven directly from HAL in realtime.

It seems that there was an EPP to CAN bridge on the market until a couple of months ago:
www.computer-solutions.co.uk/gendev/can-dongle.htm

PCI and PCIe CAN interfaces exist, but are definitely not cheap.

USB to CAN is very easy to find (I use one extensively at work) but that won't work with the LinuxCNC realtime layer.

How much of a project to you want?

andypugh wrote:
How much of a project to you want?

Let's put it this way... I was expecting to have to completely gut the machine, replace the spindle, drivers, leadscrews... so if I can keep everything and just communicate with the existing controller that's be great!

andypugh wrote:
PCI and PCIe CAN interfaces exist, but are definitely not cheap.

Probably cheaper than my original plan! Do you have any suggestions/recommendations?
It looks like controltech.uk.com/collections/peak-syst...nterfacesControlTech UK has taken on most of the Computer Solutions products (or similar enough stuff)



I think my current barrier in the project is not understanding how the communication is going to work. I'm hoping that's just ignorance about the CAN protocol. Is there a standard way in which 'actions' are sent like in g-code or will I have to reverse engineer everything?

maltandmake wrote: I think my current barrier in the project is not understanding how the communication is going to work. I'm hoping that's just ignorance about the CAN protocol. Is there a standard way in which 'actions' are sent like in g-code or will I have to reverse engineer everything?

Having found potential hardware, the next thing would, indeed, be the protocol.

Do you have the documentation for the drives? I would rather expect that all that is needed is a CAN packet to be assembled with ID for the drive and the position to move to.

Do you have the drive documentation?

There has been some recent discussion on the mailing list (but not in the context of realtime control)
sourceforge.net/p/emc/mailman/message/37203425/

And, from a long time ago:
sourceforge.net/p/emc/mailman/message/26045690/
But it seems he got no replies.

It might be worth asking General Mechatronics if they have anything: www.generalmechatronics.com/linuxcnc.html