[eric14779]

We have a 1987 Sharnoa SDC-80D with Fanuc Motors and drives. The Z axis is having an issue.

The Z axis Drive is a A06B-6047-H003 Velocity Control Unit

The Z axis Motor is a DC 20M Servo # A06B-0652-B005

This machine has a proprietary (Sharnoa) control with a PC motion card that receives the encoder feedback and directs the drives to move. The motors have feedback to the drives only.

The issue we are having is that the Z axis physical position is drifting from indicated position, and we have to re reference to bring it back into position.

I do not understand why, if the ball-screw encoders feed back to the motion card, how this can happen. I would think that the motion card would either have a fault indicated or require the drive to continue moving till it is in position.

Any ideas as to where to start ??

[eric14779]

Ok I started looking at the encoder input "chain" and here is what I have found:

The Z axis encoder is Heidenhan ROD type. It is connected to a terminal strip at the encoder and then a 4 pair cable brings the signals down into the control cabinet.

The cable has shielding but it is not being used.

The signal is then ran into a box that has a line driver (digital amp) for each axis, and also brings in the 5VDC from the Computer power supply for the encoder(s) and the digital amp circuits.

The signal then comes out of the box and is fed into the motion card in the PC, again with unshielded cable (I believe that it is twisted pair (4) wire)

I do not know how old this setup is but, I am wondering if the problem could be with the Digital amp on the Z axis feed ???

I also do not like that there is no signal shielding, I would like to fix that as well.

Anybody think I may be on the right track ??

[mike_Kilroy]

yes, u could be on the right track. but did it used to work for 25 years and now does not? thats a clue - what changed? what was last thing that changed from when it used to work and now does not work? I doubt the shielded wire shield dissappeared, so what did change? that is more likely where to look. nothing changed you can think of? wires were always unshileded and it used to work? hmmm prob not shielding then - unless u just installed a new welder next to the mahcine and it makes lots more noisel. so what changed from when it worked for years to now?

I think, assuming it used for work for years, that some changed. how about a loose coupling on the encoder?

[Al_The_Man]

You say Fanuc motors and drives, but mention a Motion card?
What type of control is this?
Is it all Fanuc or a composite system?
Al.

[eric14779]

Al,

Yes this is a composite system Sharnoa as you may already know was way ahead of the rest of CNC technology with their controls... They were PC based and had all kinds of powerful abilities....

The problem is now that there is no factory support in the USA and only 2 companies that "service" them here. apparently on a part time basis...

So yes the motors and drives are Fanuc the control is a industrial computer backplane system with a CPU card, I/O Card, and a motion control card wired to the encoders and to the motor drives.

The Z axis Drive is a A06B-6047-H003 Velocity Control Unit

The Z axis Motor is a DC 20M Servo # A06B-0652-B005


the encoder for the Z axis is a fairly new looking Hiedenhan ROD 426. Newer than the X and Y axis encoders... Must have been replaced at some point.

We purchased this machine used and I can see that there has been mods and "upgrades" done to it.

Yes we have added new equipment in and around in the shop, Also had some incoming power issues just before this drift problem was discovered. Coincidence ??? I do not know... This machine is NOT used everyday, it can sit for weeks or months before it is used again.. But now this issue happens every time it is ran.

The control software does NOT show any errors or faults, the Z axis just slowly drifts out of the indicated position and requires the operator to notice and re reference to bring it back in position... then it repeats.... but not always and exactly at the same rate...

So the first place I know to look is with the encoder signals..... Be sure that they are robust and accurate.... And see what happens...

Don't you agree ??

[mike_Kilroy]

so it is a new machine with little use on it so it could be shielding issue. yes, noise can creep in as you know and move position like that. so can a loose coupling on any shaft between the encoder and the final output position so don't discount that possibility also.

[eric14779]

Mike,

again this is a 1987 machine.... that we bought used some years ago, It has had many issues that were dealt with as they have appeared, this is the latest.

The encoder/coupling/ballscrew connections are tight and solid.

I believe that the home operation is to a limit switch then locate the reference marker pulse from the encoder, and offset from that for machine Zero. If the encoder was moving on the ballscrew I would expect the machine zero to continue to move. Don't you agree ??

that is not the case, after homing everything is back as it should be.... then it drifts off again as Z moves up and down, cycle after cycle.

[ROBBY 68]

Question
Does this machine has a scale on Z AXIS?
If so, position feed back is given by scale which is more accurate of encoder and could be causing drifting
regards

[eric14779]

ROBBY 68

It used to but has been removed.... before we bought it... rotary encoders ONLY now...

[ROBBY 68]

If there is NO alarm, it means encoder is NOT "seeing" machine is moving.
Normally power is applied to motor to keep it ready and stand still waiting movement command, encoder will advice control if any displacement is detected.
so I would check coupling between encoder and motor, it may be loose.

[eric14779]

ROBBY 68

The servo is a Brush type DC and it is driven by a DC Velocity control

Fanuc:

A06B-6047-H003 Velocity Control Unit

20M Servo # A06B-0652-B005

The encoder is mounted to the end of the Ballscrew NOT on the motor. The motor has a Tach on its shaft... again 1987 technology.... 24 years ago..

[mike_Kilroy]

ok so good you checked for tightness on encoder coupling. You ask obvious question: if rehoming brings it back from the drift then doesn't this prove the encoder is not moving?

Obvious answer is yes, but real answer is not necessarily! Since you home to a SWITCH then go to marker (are you sure it does? A lot of machines just home to the sw and ignore the marker) the switch gets you very close and then the marker is only another possible +/- some portion of the ballscrew turn. So it may seem to rehome each time since this partial motion after home sw to marker may be too small for your operator to notice....

Being 1987 technology, the enc prob has a glass disk. This disk is glued in so unless you know for sure the rehome brings you back to the SAME marker location each time I would still have slipping encoder in my list of possible culprits. If you are sure the marker is used, and that the rehome goes to exactly the same spot, then I would discount the encoder now.

I am lazy so not going back to study initial posts, but are you sure the encoder uses DIFFERENTIAL outputs rather than single ended? Of course single ended often has this creep issue.

[Al_The_Man]

AFAIK Delta-Tau was the only PC based card that integrated Fanuc drives to the card, if so, this is a very sophisticated card, in spite of its age.
Like all the PC based motion cards, the encoder closes the loop back to the card so if the encoder signal is valid and getting back to the card, then either the card corrects it if it can or goes into error.
All the motion cards I know of use the switch to indicate the seek for the encoder marker pulse, much more accurate.
If it homes OK, what is the position displaying when it drifts, unlike most of the commercial systems that display the Anticipated position, motion cards use the encoder feedback as the displayed position.
So if the display also shows a drift, then the encoder signal is getting back and it is most likely in tuning or parameter, I would have expected a following error to be displayed.
Al.

[eric14779]

Mike,

the encoder is a Heidenhain ROD 426

it looks newer than the other two on X & Y

Al,

The display on the control (again Tiger 6 Sharnoa) indicates the "programed" position.

the actual Physical position is off, after re-referencing it is brought back so the "programed" position and actual are equal.....

then it starts to drift again more and more ( about 0.100" to 0.150" over 4 to 6 hrs ) until the operator re-references again.

[mike_Kilroy]

I have not looked up rod 426 but that is irrelevant: it may have differential outputs (a,a-, b, b-) and your control may only have single ended (a, b) wired up. You need to verify you have both a- & b- wired up. I would say this is troubleshooting step NUMBER 1. Cuz if not using the differential outputs then you WILL have 90% probability of noise issues as you describe.

[Al_The_Man]

When you say 'drift' do you mean the axis appears to move when it shouldn't or is the commanded position off after a few moves, both are different, I cannot believe there would be an error between the encoder detected position and the commanded position without the control doing something about it or going in to error?
BTW. ROD 426 is a differential output.
Al.

[eric14779]

When I say drift I mean the following:

If the axis is not commanded to move it stays in position indefinitely.

As the axis is moved + and - over time it loses position more and more, as it is moved.

when at rest after moving it stays out of position directly proportional to the amount of moving it had done.

Control indicates that it is in commanded position at all times.


yes the ROD 426 encoder is differential... there is a converter box inline with all 3 encoders.

The compliment signals are all hooked to V0 and the a b z signals are ran thru a differential line driver circuit to rebuild the compliment signals, then all 6 signals/axis are ran into the motion card.

[Al_The_Man]

OK, that does not sound like my definition of drift, but mechanical position error, the first is when an axis is moving and it is supposed to be stationary, position error is when the control commands a move and thinks it is in position via the encoder but mechanical wear or other factors such as loose encoder coupling have physically positioned it at a different location, this error can gradually increment as moves increase.
Although a loose encoder coupling will change your actual home position up to one revolution of the encoder.
There is a simple test using the H.W. to find out if your controller displays actual position or anticipated position, it will also tell you the amount of following error that is allowed by the controller.
Al.

[eric14779]

AL,

there is no hand wheel see pics

I do not know if anybody knows how to display following error either.

[Al_The_Man]

AL,

there is no hand wheel see pics

A Mill and no handwheel.!?

Is there a slow jog?
Al.