Howa VMC retrofit advice, please!

Ah, yes - ATC motor is an HA-FH23. It's a 200w version of the HA-FH13, I believe.

aha, ok.

i don't have any MDS-A/B-V* drives to verify, but i've checked the MDS-B-V2 firmware (BND513W000A6) and it *may* support those encoders. along with the usual OHA/OHE/OSA/OSE 25/104/105 there's a OAM-13B - the name suggests it's a 13-bit one. good chance that it's the thing.
i would try to connect HA-FH motors to one of your MDS-*-V2-0505 drives and set SV025(MTYP)=xx7d for HA-FH13 (HA-FE is the same motor)
SV025(MTYP)=xx7e for HA-FH23
xx could be 11 or 22 - not sure
RNG1=RNG2=8 or 10 not sure either, need to try

if it works, cheap used MDS-B-V2 drives appear on ebay regularly. you already have the power supply feeding XYZA drives and pretty hefty spindle - i'm sure it can handle one more 2-axis amp. and it will be an easy installation

thus you could implement all tool changer logic conveniently in the linuxcnc - in ladder, python, c or g-code. without messing with some ancient proprietary mitsubishi software...

dm17ry -

Looks like the best bet from both a cost/availability perspective and ease of installation in the existing cabinet would be to use a pair of MDS-A- SVJ drives. The HA-FH13 motor on the Magazine is 100w, but currently paired with a 300w rated drive. The HA-FH23 motor on ATC is 200w and currently paired with a 600w rated drive. Is that much headroom on the drives necessary?

It appears like 1 each of your YMDS2P interface adapters would allow me to control my 7 Mitsu servos with their existing absolute encoders via LinuxCNC. The other big question is I/O. Would it necessitate using your YIO Modules, or would it be feasible to use I/O boards from MESA or others?

i have never seen mitsubishi servo setups using "overpowered" drives. normally most drives even refuse to work with smaller motors... no idea why they did it like that. i would use matching capacity replacement drives.

using mesa i/o boards - though i have not tested it, i think it should be ok if they don't take too much CPU time.
the servo thread has to run with 888888ns period, make sure hm2 is ok with that
the read function should be called more or less at same time within the servo cycle
the time between the read and the write should be <300us