[YorkCNCWorks]

I have a 1986 vintage Daewoo Puma lathe (I have been told it's a Puma 4, the model number is worn off and all documentation has long since been lost) with a Fanuc 0T control. I have a problem where sometimes the machine goes to different locations than commanded. For instance I have a section of a part that gets turned to 1.220" dia, I can run one just fine, and then load another part, hit start without changing anything in the program and it will cut it to 1.300" dia. It will just keep doing this, doing nothing but loading new parts and hitting start will result in some good parts and some bad ones seemingly at random. It has also happened on the Z axis, it will face a part to the correct length, then running again with the same part left in the spindle it will sometimes come back and cut about 1/16th off the end. I have been seriously working with this machine for a few weeks now and this problem has been getting more and more common, it used to only happen maybe once every few dozen parts, now it's happening on the majority of them and I've had to give up actually running parts until this is sorted out.

What I have done so far is clean and check all the limit switches and dogs, I also pulled the cover off one of the servos to check if the encoder was dirty or something, inside was spotless as far as I could see. They are fanuc branded servos with yellow caps if that helps. I have also run the program in question a bunch of times over the same part to see if I could find a pattern to the behavior and nothing jumped out at me.

If there is any other information that would be useful please let me know. I'm mostly a manual machinist with some backround in manual g code programming and some basic electrical concepts, but this is my first foray into actually repairing a CNC machine tool.

Any suggestions about what could be causing this would be greatly appreciated.

Thank you for your time

[dhardt]

Does your Fanuc motor have a coupling that attaches it to the ballscrew?

Is it loose?

[YorkCNCWorks]

Does your Fanuc motor have a coupling that attaches it to the ballscrew?

Is it loose?

It has couplings with shear pins between the motors and screws. I just checked one by turning the screw manually and looking for any relative movement in the coupling and didn't see anything. I'm going to pull the motor off to get a closer look at it.

[YorkCNCWorks]

I pulled the X axis servo off and checked the couplings, they are on the shafts good and tight with no play that I can discern, and the shear pin fits snugly in the slots on the couplings. However I did notice that it has ball bearings that fit between the couplers, they fit into a hole on the ball screw end deep enough for about 1.5 balls and a shallow tapered recess on the motor end (it looks about like what you get when you do a standard spot drilling operation). Now my concern is that one of the holes had 2 balls in it, and the other only has 1, which doesn't even come up above the face of the coupler to engage with anything. I was hoping someone familiar with this kind of servo could tell me if this is normal or if someone who had this apart before just lost a ball. I'm going to try and attach some pictures I have of the couplings in question.

[dhardt]

Yup, that is old!

It seems to be a torque coupling. But if the balls (possibly backed by stiff springs) are not jutting up against the two divots of the other surface, it might not function accurately.

Are there springs behind the balls?

What is the information from the machine's placard?

Make, model, serial#...?

[underthetire]

Those balls should be sticking out and very stiff to push back in. Normally it's just 1 ball per side with a stiff die spring behind them, but being that old...
I'll see if our old Daewoo prints are around, think it's a 4 from the late 80's, very early 90's.

[dhardt]

The Puma 4 (vintage 1986) used a two piece Motor/Ballscrew coupling that had 4 steel balls
Item Part# Description availability
9 L04430033 Servo motor coupling not in stock
14 L04430043 Ball Screw (X) coupling in stock
13 S6250180 Steel Ball in stock

[YorkCNCWorks]

I pulled the Z servo off and it has the same setup, except all 4 balls are present. Neither of them has a spring in the coupling, the balls seat against the hard bottom of the bores they sit in. I'm not so sure about this whole thing though, it sure seems that as long as the shear pins are intact and secured properly (which they were) the balls shouldn't be required to transmit any torque, but I could be wrong.

I found the machine serial # (450015) on a placard that lists the machine power requirements, but I can't find anything with the model information stamped on it. I checked the whole exterior of the machine, as well as inside both cabinets. All I have is the painted plate on the front of the machine on which you can just make out the words "daewoo" and "puma" but the model number has been worn off. A repair guy who was here working on another machine told us he thought it was a Puma 4, so that's what I've been going with. The small chuck and travel of the machine (6in chuck, not much more than a foot between the chuck face and turret) makes 4 sound about right to me, I can't imagine the series going any smaller.

I did find a Fanuc warranty card in the electrical cabinet, but all the labels are in Japanese. It did reveal that the control is specifically an 0T-A. The other numbers on it are "Y 24146","A02B-0083-B501", and "10127" if those mean anything to someone.

[dhardt]

I recommend getting 4 steel balls and replacing them all. I'm willing to bet that it will solve the diameter issues.

[YorkCNCWorks]

When I first measured one of the balls with calipers I got 9.5mm (which comes out to 0.374") and assumed that was the size, however I just pulled out and cleaned all 7 that I have and measured them with a vernier micrometer. 6 of them measure 0.375" +/- a tenth or so, and the last one measures 0.3747". My question here is have these been replaced with a (slightly) wrong size, or did it come from the factory with 3/8in balls (which would be odd seeing as basically everything else on the machine is metric). I will try to track down some replacements, but it would surprise me if those balls being so slightly wrong could result in such a large error (40thou radius on X and around 60thou on Z).

[YorkCNCWorks]

I just had a breakthrough and figured out what the balls actually do. I was pushing on parts of the coupling on the ballscrew side and found out that it has an outer sleeve which can slide along the axis of the coupling given enough force. The bottom of the bore that the balls fit in is actually the face of this sleeve. If a shear pin breaks the balls will cam out of the tapered spots on the motor side coupling which in turn push on this sleeve and slides it out of range of a sensor (hall effect or something like that) and triggers the shear pin alarm.

So good news is I finally know what those balls are for, bad news is that they shouldn't really have anything to do with my actual positioning problem.

[dhardt]

the part number for the ball:
S6250180 Steel Ball in stock

Attached are drawings of the coupling, showing the ball (3/8") in place

I believe that the missing ball is the reason for your positioning issue. I believe that it is allowing the drive force to rotationally fluctuate from pivoting on center of the coupling to shifting the torque center to flexing off of the shear-pin.

Think of it as the positioning equivalent of "ringing-out a wash cloth." The motor is turning, but the loading-up part of the motor's rotation is starting to "flex" an improper coupling, prior to transferring that force into the ballscrew.

[YorkCNCWorks]

I put the machine back together (including replacing the missing ball) and so far things seem to be back to normal. I ran a simple test program 30 times and didn't see any large inconsistencies like I was getting before. I'm not sure what exactly fixed it, it could have been the minimal cleaning I did of switches and connectors, could have been a simple case of something mysteriously working after disassembly and reassembly for no apparent reason (which happens to me probably more often than it should), or it could be the result of a different approach to programming, which I'll elaborate on below for anyone interested.

I'm going to put the covers back on the machine and try and run some more parts, keeping a close eye on it. I'll update again after I finish the job or if something goes wrong.

As for the programming thing, here's the deal. I was not using the work coordinate system and tool offsets in the machine because I didn't know how they worked, instead I was using G50 position register (same as G92 on some other control systems). Now the deal with position register is that you kind of need to be at machine zero each time you call it (after each tool change) for practical reasons. I contacted a machine tool repair guy we have worked with before about my issue and he said the problem was likely to be with switches or sensors in the zero return mechanism being inconsistent. This made sense to me, because if it were true it would cause the zero point being referenced by my position register command to be shifted, and subsequently my tool positions when cutting. What I was told to do for testing was use a program that doesn't do zero returns between tools, to hopefully isolate the problem. That meant I had to buckle down and figure out the coordinate system and tool offsets for real. Now that I am using that system, my tools can no longer be affected by machine zero issues as long as the initial home on powerup is good. So it's entirely possible that I still have some issue with inconsistent zero returns but I'm just bypassing it. This is workable for now, but I'd like to make sure it's fixed for real at some point when I get the time.

[dhardt]

Your understanding of how a machine finds reference is in error.
Moving the switch, or faulty action on the switch would offset the reference position by at least one full rotation of the ballscrew...either forward or back.
Your issue was likely due to the missing ball bearing allowing flex of the coupling.

[Superman]

I have a section of a part that gets turned to 1.220" dia, I can run one just fine, and then load another part, hit start without changing anything in the program and it will cut it to 1.300" dia.

0.080" error approx.= 2mm ( or 1/3 rev of the metric leadscrew pitch )...... looks like a pulse "skip" on the encoders when homing. I think there are 3 pluse positions on an encoder, hence a 2mm change on a 6mm leadscrew

Did you HOME an axis before the size change ??
- mark a line on slide & base when homed, repeat homing.....do the marks always line up ?

Do you program G28 U0 W0 for the axis returns ?


The older machine have a limit switch for homing...as well as an overtravel LS for the "hard" stop
- I have used a machine that had "skipped"... have found the error could be a loose / sticky / faulty LS, even had swarf buildup around the same LS.

Homing is the only time when the machine origins are usually set, any reference returns make the machine travel back to those origins, irrespective of the G50 values....but the G50's are set relative to the current position of the slides
- so your use of G50 is not so incorrect....it's where the slides are when the G50's are executed.

[dhardt]

Superman,
Two slight corrections.
1. X-Axis ballscrew of 6mm pitch would cause a 12mm diameter. So the calculation you gave would be corrected to 1/6th of a turn of the ballscrew.
2. The home switch is at the start of the process, but does not create a miss-position of any less than a FULL rotation.

Home procedure is as follows:
(Middle of axis)
Ref. Switch = 1
Request reference.
Travel toward reference at quick speed
Trigger Ref. Switch causes...
a. Slowdown to specific speed.
b. Ref. Switch = 0
Travel until Ref. Switch = 1
Look for Z-channel mark on encoder. (1per rev)
Continue to +value in 1850 or reverse to -value
In that specific location, report that the position is whatever is in 1240

If the switch were slightly at fault, the Z-channel would be found on the same rotation. Which would not be a problem.
If the switch were severely at fault, the same Z-channel would be found, but on an entirely different rotation of the ballscrew. Causing X to be off in 6mm radius (12mm diameter ) increments only.

The problem was the damaged coupling.

[YorkCNCWorks]

Your understanding of how a machine finds reference is in error.
Moving the switch, or faulty action on the switch would offset the reference position by at least one full rotation of the ballscrew...either forward or back.
Your issue was likely due to the missing ball bearing allowing flex of the coupling.

I know how it is "supposed" to work, where it comes off the home dog and the first pulse from the once-per-rev encoder sets home. That said, I don't really have a reason to think this thing is operating 100% correctly as far as those sensor electronics are concerned. I have to assume you are used to a different type of servo where the balls are used as some kind of torque limiting coupling. I know I'm new and I'm not trying to be rude or anything but I've had these things apart and operated them with my own hands and there is no way that the balls on this machine are transmitting any significant torque between the motor and screw. They don't even have springs or anything to keep them pressed against the motor side coupling. I'll try and draw a diagram of them if I have time later. And I had the same problem on both axis despite only one of them having a missing ball.

Did you HOME an axis before the size change ??
Do you program G28 U0 W0 for the axis returns ?
The older machine have a limit switch for homing...as well as an overtravel LS for the "hard" stop
- I have used a machine that had "skipped"... have found the error could be a loose / sticky / faulty LS, even had swarf buildup around the same LS.

Yes, the machine was getting homed with G28 U0 W0 at the start of the program and before each tool change. It does have a home limit switch and an overtravel switch. That "skipping" problem is the kind of thing I am worried about may be/have been happening.

Homing is the only time when the machine origins are usually set, any reference returns make the machine travel back to those origins, irrespective of the G50 values....but the G50's are set relative to the current position of the slides
- so your use of G50 is not so incorrect....it's where the slides are when the G50's are executed.

Basically that is what I was thinking, that when I send it home before G50 it isn't quite going to the right place because of some kind of problem with switches/encoders/whatever and G50 is getting referenced from the wrong point. I guess I may have worded it poorly before.

[underthetire]

If it does start happening again, id take a goo look at the turret. May not be clamping correctly.

[dhardt]

York,
There is no need to make a drawing for me. I work as technical support for the company that built it. I have access to the drawings that allow me to build the part. I have access to the designers and engineering department that said that the coupling lacking a ball was the reason your error was occurring.
I'm glad that you are repaired. Have a great day.

[sting007]

Hi dhardt,

As i see you are Doosan service technician, so i wonder if you could help me with my Puma 6s with Fanuc 0t, the machine stopped for a long time and lost all the parameters, I've load the parameters manually as i do not have a soft copy, and now everything is working except the spindle, in MDI mode when write say S500 M03 then input then cycle start; cycle start green light but the spindle does not rotate and cycle start keep on with out even showing any alarms, the spindle only rotate if INCHING, not rotating in manual mode
I checked all the parameters but i did not find anything wrong
can you help me to solve that problem

machine serial :PM6S 0717
PMC:0T-C-Q5

sting