Mitsubishi SSCNET

Thank you for this great contribution! I will buy 10 SSCNET-III cards when the final release. As a teacher, recently I have been assigned to explore a new course to teach students how to make things. Buillding machines is a synthetic and exciting topic. Personally I would like to learn more and explore the LinuxCNC to make real things.

Does the new card support Master-slave operation function? As manual says, The slave axes use the control command from the master axis. Therefore, the controller mainly controls parameter settings, servo-on command, acquisition of
monitor information from a servo amplifier, etc. The commands regarding absolute positioning such as setting absolute position detection and requiring home position setting from the controller to slave axes must not be made. Is this function non-trivial for new card?

I mean handle the grantry machine with master-slave function of Mr-j4-b.

the master-slave function seems to need a support from the controller:

Use the master-slave operation function with the following controllers. Refer to the manuals for each servo system controller for compatible software versions, and other details. RD77MS/QD77MS_/LD77MS_ R_MTCPU/Q17_DSCPU Q170MSCPU

so i'm afraid it will not be supported, at least for now, sorry.

but it is perfectly possible to drive a gantry-type machine the usual way controlling servos in position mode. done that a while ago using a patched gantry.comp component on a plasma cutter table. servos were using absolute encoders.

andypugh wrote:

No need for gantry.comp any more with the development version of LinuxCNC. Gantries are now handled natively (and well)

yeah, i've seen that. i should finally switch to the dev branch...

about mitsubishi master-slave function. reading the docs i've got an impression that it is not suitable for precision positioning because slave motors are torque controlled. or the gantry must be absolutely rigid, otherwise the slave end will be basically flapping in the breeze. imho

Thank you very much. I got it. One more thing does the new j3-card support handwheel and tool touch setter?

wu-010@163.com wrote:

One more thing does the new j3-card support handwheel and tool touch setter?

yes, same as the old one - it has general-purpose inputs and a hardware quadrature decoder. a touch probe/tool setter can be connected to GP inputs, a handwheel encoder A,B phases: TTL to expansion connector directly, differential - via a mezzanine board. the rest is handled by linuxcnc...

put together a small xyz gantry out of parts which were laying around for years and SSCNET III MR-J3-B servos. and tried to kinda push the limits so see what they are capable of:

a little more reverse engineering. recently got a CPU board out of a mitsubishi MDS-DJ-V1 servo drive. those are CNC series drives but they use the same hardware as general-purpose MR-J4-B. the control protocol is different though. don't have access to a working mitsubishi control to snoop the communication. but as i previously spent a lot of time playing with other mitsubishi hardware it wan"t too complicated to find a way into the drive's debug monitor, dump the memory contents and locate the code for control protocol processing. and with a bit of trial and error got the drive running. the protocol turned out to be very similar to sscnet-iii, with differences like 64-bit position command/feedback for "nano control", i presume
i doubt anyone will want to ditch an M700 control for LinuxCNC retrofit anytime soon but as readily avaliable surplus MR-J4-B drives are "convertable" into MDS-DJ with firmware reflash it would be interesting to do it and compare them in some CNC application..