Re: Help: Position Feedback with Mitsubishi Servos

On Jun 12, 11:13 pm, Peter Nachtwey <p...@xxxxxxxxxxxxxxxx> wrote:
On Jun 12, 6:22 am, "thrugoodmarsh...@xxxxxxxxxxx"



<thrugoodmarsh...@xxxxxxxxxxx> wrote:
Position Feedback with Mitsubishi Servos

Does anybody have experience with Mitsubishi servos and time to answer
a question?

Most parameters, except target positions, are set by GX-Configurator
QP.

All the equipment in question is from Mitsubishi.
I've got a QD75M4 motion control module and four MR-J2S servo
amplifiers. They are controlling four HC-KFS43K motors.
The PLC involved is a Q12H, not a dedicated motion control PLC.

I'm using simple single-step (end) or two step (continue, then end)
motions to send servos to PLC-specified positions.
I'm using a couple of bits to generate a "move complete" status in my
PLC:
X14 (positioning complete--the step is complete)
NOT XC (busy--the servo is executing a move)
And I'm monitoring position feedback using Md.20.

The problem is that in certain moves when the servo is mechanically
prevented from achieving its target position, I still see
X14 AND NOT XC
And my position feedback (via Md.20) shows that I have reached my
target position.

I have tried to find another method using this equipment to achieve
the goal, but I'm having no luck.

I want to send the servo to a position, get feedback indicating "at
position" when it's there, and monitor the actual position whether the
servo

reachse setpoint or not.
Has anybody successfully done this?
How might this be done with the equipment above (using the general-
purpose Q12H PLC, not a dedicated mitsubishi motion control PLC)?

A more detailed version of the above question is restated below.

Thanks for your help!

The manual I reference below is QD75M Position Module Software Manual
V2-D.pdf, identified as
Type QD75M Positioning Module
User's Manual
IB(NA)-0300062-B(0404)MEE

The servos drive ball screws with "sleds" mounted on the nut to carry
workpieces
That's not really important as the encoders are attached to the
motors, not the ball nut.

All axes are independant and all moves are simple "go-to-position"
commands.
The moves all use CTRL Method 1: ABS line 1 (Axis #1 Linear control
(ABS)) move, which controls speed to a set position.
Each move can consist of either
a 0: END (End command) segment
a 1: CONT (continue positioning control) segment followed by an 0: END
(End command) segment.
The second type of move is for high speed approach followed by creep
speed final positioning.

Two of the axes are pick-and-place, and reliably reach their commanded
position every time.

The other two axes bring two workpieces together, and due to lack of
precision in prior measurement sometimes cannot reach their final
position.

For all axes, I create a "move complete" bit in the PLC using X14
(positioning complete) and NOT(XC) BUSY per p6.3 (Fig. 6.3 ON/OFF
timing of each

signal at start of positioning).

Also, for all axes, I read current feed value via Md.20

For axes which do not achieve target position, we expect to see either
the "move complete" (generated in the PLC as above), or an error.
We also monitor the current feed value from Md.20 as above.
This is so that we can adjust the position of the next equipment in
the process.

It seems that the QD75M4 reports move complete and current feed value
Md.20 = target event though position has not been achieved.

To test this, the Engineer onsite installed an oversized workpiece so
that the ball screw could not achieve final position, then executed
the move to

go to that position.

1. We get "move complete" (X14 TRUE and XC FALSE with no faults, per
sec. 6.3 above) even though the sled doesn't make it to the specified
position.

The command in-position width is .00100"

We would expect the servo to fault on a following error if it never
made position, but it just stops with no fault and maintains force
against the

workpiece. We monitor axis faults at Md.23 (Active Error Code).

So there is only one position from the controller and that is the
target position?
This is one way to make a controller look good. When the actual
position does not reach the target position there should be an error
generated when the output builds due to integrator wind up.



2. The current feed value Md.20 shows that actual position = target
position, even though we can see it has not occurred.

Previously, the Engineer scribed a line on the ballscrew carrier while
the sled was at the target position.
During this test he observed the sled is around an inch short of the
line, even though the current feed value Md.20 reported target
position had been

achieved.

Are you sure there isn't another register for the actual position?


I've looked in the manual and tried the registers I could find.
That is the core issue here, though.
Md.20 and Md.21 seem to be the likely places for actual position.
If somebody knows what this "other register" might be, I'd love to
hear about it!







I would expect the feedback to report the actual position, not the
target position. If not at Md.20, then I would expect some other
parameter to

report the actual position.

To check this, the Engineer grabbed the sled and moved it back about
6" against the force of the motor. After releasing the sled, the
motor returns

it to the position against the workpiece where he grabbed it.
Current Feed value at Md.20 and Machine Feed Value Md.21 never
changes.

If the Engineer drops 240V power, but not control power to the servos,
(just opens the safety gate) current feed value from Md.20 abruptly
changes

without the sled observably moving. Reported value in Md.20 is
roughly consistent with actual observed positon.
At that point, moving the sled produces the expected change in current
feed value Md.20.

We have tried:

Monitoring Md.21 instead of Md.20 Both behave similarly, except that
Md.20 reflects the offset applied during setup via OPR.
Changing Pr.21 (Current feed value during speed control) to 1 (The
current feed value will be updated.) produces no change in the
problem. Ref 5.34,

Pr.21.

I've read multiple Mitsubishi motor, PLC, servo, and motion controller
manuals, reviewed the parameters the software GX-Configurator-QP,
looking for

some parameter I have set wrong, or the correct parameter I should be
monitoring. I'm finding nothing.

I want to send the servo to a position and get feedback to the PLC
that tells me:
1. I've reached position within tolerance.
2. The actual position of the encoder at all times

This seems like a fairly straightforward application of a servo (a
very basic one, actually), but I just don't see how to do it with
these products.

Has anybody successfully done this?
How did you do it?

Yes, but not with a Mitsubishi controller. There are other motion
controllers that can communicate with the Mitsubishi PLC via Ethernet
and can do these application easily without the PLC.

And, this is the other possible answer:
If I can document an authoritative statement that you can't do it with
a Mitsubishi Controller (MR-J2S, Q12H), then we know that the wrong
equipment was specified for the task at hand and we can move on.




There is a Mitsubishi forum onwww.mrplc.com. I think there is a
person from Mitsubishi that answer questions but that my only apply to
PLC questions. Some one there should be able to answer your question
or point you to someone who can.

You should also trywww.plcs.net

Peter Nachtwey


.

Relevant Pages

  • Re: Can an SLC 500 RIO a 1771 Rack & Cards.
    ... PLC 5, 1771, rack and I/O modules? ... What kind of home brew experimentation ... One project is to use the analog I/O for analog circuit control, ... temperature and internal temperature of the smoker. ...
    (sci.engr.control)
  • Re: A challenging 3 equations and 3 unknowns
    ... circle using three ultrasonic sensors. ... any case the OP on the plc forum said this must run on a PLC so the ... is used in industrial control systems. ...
    (sci.math.symbolic)
  • Re: PLC PAC operating system
    ... PLC or other PAC by Siemens? ... some means are required to load the control program. ... some models also executes applications originally ...
    (comp.arch.embedded)
  • Re: honeywell tdc3000
    ... Allen Bradley DH+ using the SIM module ... And there are a few neat things that I like about the Experion ... Model Predictive Control in the Controller, ... There are of course some tasks which will be more suited to a PLC, ...
    (sci.engr.control)
  • Re: how to organize several class and their instance in windows programming
    ... motion control and networking using MFC. ... Though I think the "design pattern" could solve my problem, ... I'd suggest that the mainframe is the WORST ...
    (microsoft.public.vc.mfc)
Loading