Howa VMC retrofit advice, please!

21 Jul 2021 22:42 #215529 by tropostudio
Hello - new to this forum and LinuxCNC

Background:   I build interactive museum exhibits and a decent grasp of machining, electronics, and some microcontroller and PLC programming experience.  I've gutted and retrofitted an old Gerber CNC router that now runs on an Eding controller and have scratch-built a CNC plasma table running on same, with an Arduino THC. 

Retrofit Machine Candidate: My son and I recently acquired a 1996 vintage Howa 320H-VCJ vertical machining center.  Mitsubishi MDS A/B servos for X, Y Z, and A (jig-changer table).  Mitsubishi MDS-C1 spindle servo up to 15k rpm.  Single-axis Mitsu Model E amplifier/control drives for the ATC and 16-tool magazine.  Has a tool switcher which activates within the cam sequence of the ATC servo travel (much like the shifting of a dog drive motorcycle gearbox). A  pneumatic tool pot exchanges tool holders between the magazine and switcher arms.  Main CNC control is Mitsubishi Meldas M520AM, and the Operator panel allows me to a access parameters for 4 primary axes and to read or write to the PLC ladder for those axes if necessary.  BUT, I have lost parameters and User PLC programs for the two Meldas Model E controllers at the ATC and Magazine ( dead batteries).  Those two controls sit at factory defaults, with standard PLC. Have asked Howa for copies of parameters for the 2 drives and have asked Mitsu if PLC programming software is available so I could take a whack at bringing them back to life.  Not looking good - hence thinking about a retrofit - fully aware that it could be a long process

Retrofit Considerations and Questions: 
  1. I'd like to keep all the existing servo motors, which are Mitsubishi AC (including spindle) running at 200-240VAC.  All have absolute AB/Z encoders and  use battery back-up.  I get drives would have to be replaced.  Would DMM servo drives work?  If not any other cost-effective recommendations?  I want to close the loop in the control, not the drive.
  2. As mentioned above, current system has absolute encoders on all 6 servos.  Homing is relative and automatic.  There are urethane bumpers at hard stops, but everything uses soft limits. There are no physical limit switches or hard-stop homing procedures.
  3. There are a lot of physical switches/keys, indicator LEDs, and jog encoder on the operator panel.  The 9" CRT stinks, of course.  I'd like to have a keyboard/trackball with a mix of hard buttons/switches, and  decent LED screen.  Small screen is fine.  Not a huge fan of touch screens.  I can deal with re-wiring...
  4. Looks like +/- 90 x contacts/inputs and another +/- 90 Y outputs in the current main control ladder.  A system that can add a lot of I/O would be essential.  I don't want to run a half-done machine in the end.
  5. I envision starting with Fusion360 HSM for CAM. I believe Fusion has generic posts for LinuxCNC?  We've worked through post customizing on the CNC Router and Plasma, so that wouldn't phase me. 
  6. I have current ladder for the Mitsu M520AM. The initial thought was to use that as a basis for emulation in a retrofit.  Is that crazy? The wiring and labeling in this machine is impeccable, along with printed documentation for most of it. It makes it relatively straightforward to figure things out, despite how complex the system is.
Recommendations and Advice:
  1. Are we nuts to be looking at a retrofit using LinuxCNC?  (perhaps a rhetorical question).
  2. What hardware would you recommend based on above system description and desire to save original servo motor and electrical (sans drives & controller, of course).  Thinking mainly in terms of servo drives, encoders (keep existing encoders if possible), lots of I/O, and buffering.  Don't mind ladder, OK with structured programming, pretty poor with OOPS.
  3. Any deal-breakers going between Windows based CAD and CAM and Linux control system?
Enough to start?  Thanks for any help.


Please Log in or Create an account to join the conversation.

22 Jul 2021 11:46 #215564 by arvidb

Are you sure the drives need to be replaced? What's their control interface (digital bus, step/dir, analog)? The "typical" solution for a retrofit like this seems to be to replace only the control and keep as much of the rest as possible.

Please Log in or Create an account to join the conversation.

22 Jul 2021 15:46 #215597 by tropostudio
The Mitsubishi drives use  proprietary serial bus control.  The Mitsu M520AM CNC control and 5 of the drives (Spindle, Jig Changer (A), Z, Y, and X) are still operational. I have the ability to access and read write to the PLC ladder via the Operator Panel for the main control. 

The ATC and Magazine each use their own Mitsu FCUA-MP10 combined drive/control which normally communicates to the main control via a Mitsus FCUA-DX100 I/O expansion module.  I say 'normally' because these 2 drives have their own small, separate User PLC programs written by the machine manufacturer.  Those are gone, along with other parameters.  Both are back to Mitsu default parameters and PLC program.  I can write parameters using a Teach Control, but have no way to read or write a new PLC ladder for these 2 drives/controls. It would take software and cables that I'm not able to source (I've been asking Mitsu USA).

All encoders are absolute and interface to the drives in a proprietary manner.  I plan to put a scope on one and see what the output looks like.  Perhaps they could still be used with another brand of drive (DMM?).

Assuming there is no hope of obtaining Mitsu PLC software for the ATC and Magazine drive/controls, my next thought was to see if something like 2 DMM DYN4 drives could be installed and interfaced to the Mitsu M520AM control with a new PLC.  Kind of a 'FrankenControl', I guess.

Final resort was 6 new drives that I can program, a new control and PLC (linuxCNC?), and hopefully save the existing servo motors and encoders. Does that sound right?

Please Log in or Create an account to join the conversation.

22 Jul 2021 16:49 #215608 by Henk
Maybe this could help?

Please Log in or Create an account to join the conversation.

23 Jul 2021 01:05 #215627 by tropostudio
Interesting item. Looks like they have an interface card for the drive used on X, Y, Z, and A axis (jig changer) for my machine, but not the two axes running the ATC and Tool Magazine. I'll inquire, though.

Please Log in or Create an account to join the conversation.

23 Jul 2021 10:34 - 23 Jul 2021 10:55 #215661 by dm17ry
what's the full name of the "Model E" ATC servo amp?
edit: oh, ok. it's FCUA-MP10. never heard of it before. it's sitting on the same lower-speed "MC Link B" bus together with remote IO units...  can't control it at the moment. however, the HA-xx motor can be driven by e.g. a MDS-A-SVJ servo drive...
Last edit: 23 Jul 2021 10:55 by dm17ry.

Please Log in or Create an account to join the conversation.

23 Jul 2021 19:54 #215710 by tropostudio
dm17ry -

X, Y, Z, and A (jig changer) are currently driven from a pair of Mitsubishi MDS-V2-0505 units. Compatible with your YMDS2P.

Spindle is driven from an MDS-C1-SPH-110. Can this be controlled by your YMDS2P?

ATC is driven with a FCUA-MP10-06S1. You are thinking it can be replaced with an MDS-A-SVJ-06 and use your YMDS2P to be controlled via linuxcnc?

Magazine is driven with a FCUA-MP10-03. You are thinking it can be replaced with an MDS-A-SVJ-03 and use your YMDS2P to be controlled via linuxcnc?

Is it 1 YMDS2P adapter card per driven servo motor? What kind of latency is experienced running the drives through the adapter cards vs. having a direct connection between control and drives?

Please Log in or Create an account to join the conversation.

23 Jul 2021 21:54 #215714 by dm17ry
if you tell what motors ATC and Magazine use, i can name compatible MDS drives.

yes, the idea is to put all drives on the common bus, together with X Y Z A and the spindle. there can be up to 7 drives on the bus. one YMDS2P card per bus.

yes, the MDS-C1-SPH is supported.

the card/driver do not introduce any additional latency - it's no more "indirect" than the original control. meldas servo cycle is 3.55ms, i run linuxcnc at 0.88ms, passing 4 setpoints to drives in each cycle. the feedback is also updated every 4th linuxcnc's servo period.
the drives however always have some non-zero following error (droop pulses) - proportional to speed. that should be taken in account if any external async event are present.

position feedback loop(s) (including load-side encoders or linear scales, if in use) are closed by drives themselves - that's how the protocol designed...

Please Log in or Create an account to join the conversation.

24 Jul 2021 02:38 #215727 by tropostudio
Magazine servo motor is an HA-FH13CY-S5 (100w). The current FCUA drive is a 300w rated drive.

ATC servo motor is an HA-F23 (200w). The current FCUA drive for that is a 600w rated drive.

That's why I was thinking of like-rated MDS-drives. I'd appreciate your recommendations. All encoders are absolute, multi-turn, with battery back-up. I haven't scoped the output, but it'd be great if they could be saved. Beautifully integrated, and everything has original, high-end cannon fittings. All encoders are dog-less, set up with relative machine zeros.

Please Log in or Create an account to join the conversation.

24 Jul 2021 16:48 #215792 by dm17ry
the HA-FH has an absolute, 13-bit serial interface encoder.
HA-F23 - i don't know such thing. haven't you missed a letter?

they are old motors... MDS-A-SVJ do know them. not sure about MDS-A/B-V - probably not...

Please Log in or Create an account to join the conversation.

Moderators: piasdom
Time to create page: 0.095 seconds
Powered by Kunena Forum