Mori Seiki TV-30 MSC-802M (M50) interpolation problems

[Custom Billet]

Hey Guys,
I just bought this machine and I have been having a few problems with it. Straight side cuts are perfect, but when you go into any angle or circular interpolations, it leaves a stair stepped or jagged finish.

I found a bad X axis ball screw. It had .004" backlash and it was very noisy. I had it rebuilt and I reinstalled it. I figured that had to be the problem.

The backlash is fixed but the problem remains. I tuned up the gain on both axis, and it helped, but only a little. It got to the point that it started to vibrate the machine. So I backed off the gain a little, but it is still not an acceptable finish. Any ides?
Thanks!!

[Harri J]

It's late here so I'll just throw in a couple of random ideas/questions..

1. did you replace thrust bearings? Often fail before ballscrews
2. is the finish perfect cutting in x/y only, or is there a difference in either axis?
3. if finish is perfect in x & y, sounds like servo mismatch or bad settings
4. can you feel/hear roughness when operating x/y by handwheel or feed?
5. interpolate a circle, are the steps the same everywhere or is there a difference, could you post a pic?
6. check your axis parameters, they should be almost the same for x&y
7. set your gain low to test, too high a setting will cause judder. There are lots of params to adjust, hope you have manuals. If not, you can download them at meau.com
8. while you are at it, make a full backup of all your parameters!!!! NOW! You'll be incredibly sorry if you loose them or if your backup batteries fail. After backup change the batteries as described in manual. I've personally tested this one and I can guarantee it's not something to look forward to...

I have the same machine and it will cut very smooth, so your's certainly has something wrong. If you need any help, just let me know.


Regards,
Harri
Finland

[Custom Billet]

Originally Posted by Harri J It's late here so I'll just throw in a couple of random ideas/questions.. 1. did you replace thrust bearings? Often fail before ballscrews 2. is the finish perfect cutting in x/y only, or is there a difference in either axis? 3. if finish is perfect in x & y, sounds like servo mismatch or bad settings 4. can you feel/hear roughness when operating x/y by handwheel or feed? 5. interpolate a circle, are the steps the same everywhere or is there a difference, could you post a pic? 6. check your axis parameters, they should be almost the same for x&y 7. set your gain low to test, too high a setting will cause judder. There are lots of params to adjust, hope you have manuals. If not, you can download them at meau.com 8. while you are at it, make a full backup of all your parameters!!!! NOW! You'll be incredibly sorry if you loose them or if your backup batteries fail. After backup change the batteries as described in manual. I've personally tested this one and I can guarantee it's not something to look forward to... I have the same machine and it will cut very smooth, so your's certainly has something wrong. If you need any help, just let me know. Regards, Harri Finland

Man it is so good to hear from you! I was about ready to take the gas pipe! Mori and Mitsubishi told me that this machine was not designed to machine. Just drill and tap. They got me thinking I just bought a very expensive, really cool drill press.

1. No, I did not replace the angular contact bearings. They were clean, smooth and I had less than .0001 endplay. Plus they are $200 each!

2. X and Y finish is absolutely perfect.

3. First off, I got no paperwork with this machine. Yeah, I didn't understand that either. How do you loose manuals??!! I did however con Mori into sending me a set of parameters off of the same machine with a different serial number. All parameters matched to mine.

I did buy the operation manuals from Mori. What a rip off! $300! So if anybody needs them, I got them, but they are about as good as comic book for a 9 year old. I am going to make copies and give them to people that need them.

4. There is no noise at all since I replaced the ball screw. The grease line broke going to the X axis at one point in time. It looked like it was right after the machine was made because it had the original color grease on the ball screw side of the break and the blue grease on the other. So that ball screw never got any fresh grease.

5.On a circle, the steps are small on the y axis far and near sides, and it increases the step in size as it rolls around to the X axis on an outside interpolation. Then it leaves a tiny triangle protrusion (about .001") on each side of the X axis. Backlash huh, ......nope! The backlash is only .0001". So it is slightly bumping out and then coming back in to the right place. Now, here is the other weird thing....The bump-outs are not at 90 degrees! They are about 5 deg out of 90. Where the backlash would be is a nice transition.

6. The parameters are the same...double, triple, quadruple checked.

7. The lower I set the gain, the worse it gets. The higher the gain, the better until it starts to vibrate. I think it is parameter 2205. I can't get it up high enough for it to be acceptable.

I wanted to get the parameters set and working right before I saved them, but your right, I need to do that.

I will try to post pics tonight.

Thanks for all of the help! Any more info you might have I would greatly appreciate it!

Jim

[Harri J]

Hi Jim,

Mori's official opinion is that it's a drill & tap machine. They'll also try to convince you that machining is going to muck up the toolchange, however this is not true. Ok, the design is prone to the occasional chip but this machine has been performing flawlessly except for one bad bearing on the spindle motor.

It's a very robust and well built little machine with lots of power and good rigidity. Where it lacks is the control, or to be more specific, documentation and ease of use. I also think that the reason Mori doesn't recommend it for machining is the way it's tuned. It was set up to tap and drill at blinding speeds (for the time at least). However, to get really good 3D performance you'll need to tune the axis parameters a little. One example, there are a whole bunch of params dedicated to quadrant protrusion, depending on interpolation speed and direction, which have just been ignored at the factory.

Ok, back to your problems..
I'm sorry I haven't got an immediate fix, but maybe these thoughts will spark some ideas with you or someone else reading this.

Did I get it correctly, looking from above the steps are smallest at 6 and 12 o'clock, worst at 3 and 9? If this is the case the problem is where X is at it's slowest and Y is moving at max. You mentioned a broken grease line: are the linear bearings ok on X&Y? You probably checked this while the leadscrew was being rebuilt, if not, put the machine in estop and push and pull. Everything should move easily and sound/feel smooth.

Another question, do you get the same results all over the table or have you just tested in one spot?

The error you see when crossing quadrants is just lack of adjustment, and it will typically show at about 5 degrees so we can forget about that for now.

You mentioned getting manuals: are they Mori TV-30 or Mitsu Meldas? To properly adjust the axis and servo parameters I'd recommend the Mitsu manuals (green color), the Mori ones are useless and the two of course conflict happily wherever possible. Anyway, the tuning and quadrant crossover stuff is only in the Meldas manual, I've got one if you can't get the info. Also, in case you didn't know, most of this stuff is set by hex strings. Something that takes a little figuring out if you don't have the info.

Ok, one final thought. Just have the machine interpolate a circle without cutting anything and with spindle off. Listen, feel and look at the movement. Often you can get a good idea of what's wrong, you can hear the servos fidgeting, kicking, growling or whatever. You can also check the droop and see if that stays more or less constant or fluctuates. You may also have a feedback problem, on the encoder or cables, which can easily cause all kinds of bad behaviour. I'll be around, for at least 3 hours at the shop and later at home. Hope you get it sorted out!

Harri

[Custom Billet]

Hey Harri,
OK, the linear bearings feel good. After the new ball screw, it is tighter (harder to push) than before, but I attribute that to a nice snug ball screw.

It really does not matter where on the table it is.

I have downloaded all of the manuals I could find from meau.com on the servos and drives. The drives are MDS-A-SVJ10. I forgot to write down what the servos are when I had it apart. Are yours the same?

If you could email me a PDF of your manual, that would be great! At least the part about the tuning, I would be forever grateful!

I noticed a few things that were HEX, that didn't make any sense. Most of it that pertained to the drive setup, was not. The one I was looking at was only 4 digits long (set to 4400) and it said in the manual to change the 5 bit to a 1. Where the heck it the 5th bit? So I got a little lost there and decided I didn't want to mess with that one too much. I did go through the setting procedure in the manual and it really didn't help. It said, if you are getting a bad finish on 45 deg angles, to turn up the gain until it vibrates and then back it off. It didn't work.

You mentioned feedback. A buddy of mine was wondering if my power source was causing some issues. I have 240V 3 phase with a wild leg. Mori told me that is not a problem, just put the high leg on L2. None of my other machines have a problem with it, so it shouldn't hurt, but what do you think?

One more thing, is there a way I could get you to jot down your 2200 series parameters, just for comparison? Did you have to tune yours, or did it just run out of the box?

Again, thank you so much for your input!

Jim

[Harri J]

Quick reply..

I have SVJ6, SVJ10 and SVJ20 (APC and high power spindle).

I'll scan the manual and send relevant pages asap.

HEX: here's the catch: take the hex string (4400 in this case) and convert it to BIN. Now you'll have 0100010000000000, this string corresponds to the chart with things to adjust. Just change any bit that needs changing, convert back to hex and punch in. Incredibly dumb system, but there it is.

I think you can rule out the power source issue, doesn't fit the problem.

Mine ran fine the way I got it but it took a lot of adjusting to get good quadrant performance. When I get the RS232 issue sorted I'll print out my parameters and send them to you.

[Custom Billet]

OK, got it a lot better. I would say 90% of what I would expect. There was a frequency filter set on the X axis. I reset it to 0 (it was at 205) like the Y axis and it allowed me to turn up the gain a little without getting vibration. It could still use more gain, but I don't like the way it sounds. Also, the instructions on fixing the protrusions increase the steps on angles, so the better I get the protrusions, the worse the steps on the angles get. Right now I am holding within .0002" on a square contour and on circular interp. Not too bad, but the angles are not quite up to par and there is still small protrusions. I am looking forward to seeing what your parameters are set at!
Jim

[Harri J]

Hi Jim, it's nice to hear you have some progress!
I'm still a little surprised by the steps, how much (if any) acceleration control are you using and how are angled cuts now?

I've never had stepped appearance on a cut, just sloppy quadrant changes when the machine was adjusted on the loose side.

It's a pity this control isn't like the Control Techniques I have on a second machine. Same era, but has full auto tune to get accel and vff just right. It also has the benefit of using plain English instead of cryptic hex strings for settings. So much easier to debug and tune.

Sorry I haven't had the time to dig around the Mori, I'll get there soon..

Harri

[Custom Billet]

Hey Harri,
I don't remember seeing anything about acceleration control. The angles are much better, but still not right. I can't get it any better because the servos are at boarder line vibration. It doesn't vibrate at high speed, just at some feed rates. It is not bad, but I know it is still not right. The quadrant changes are pretty good. Unless you are really looking for them, you can't see them. I am just kinda anal about appearance, so stuff like this really bugs me.

There is some sort of auto tune on this one too, but I can't get it to work. If you go to the servos page, it shows voltage, gain and auto tune, but it just gives me an alarm. Everything is set to zero.

I know what you mean by ease of adjustment. I have a Bridgeport with an Anilam 3300 and it is a breeze to dial in.

Jim

[Harri J]

Jim, it's been some time, but I dug up my notes and here is what you can do (at your own risk of course )

Raise VGN1 (#2205) speed loop gain until it just about vibrates, then back off a little
Raise VIA (#2208) speed loop advance 500 at a time, do test cuts and when it's getting good, change +-200 units until perfect

You can have the machine interpolate a circle with a nice feed, spindle stopped, listen and feel for any groan, vibration and hiss.

tweak until is as good as it gets.

You'll probably have some jittere in the 45 degree area of the circle, you can turn on adaptive filtering or jitter comp (#2227)

then it's a matter of trial and error.

You might like to decrease in position width too (#2224), it's usually set at 50 (0.05mm), but cutting it too much will produce clunky movement. Mine is at 20 and it works ok.

Please note that #2227 is set with hex!

[Custom Billet]

Nice to hear from you Harri!!

OK, I have done all of that. The circles don't look too bad. They are acceptable. The 45 degree angles however are still choppy. Acceptable, but not perfect. when I decreased the position width, it made it very sloppy. It would over shoot every time it changed direction. I will check it out today and see what it is set at.

I have been running this machine like crazy for the last week. It makes chips faster than I can get them out of there! I need to get my chip conveyor modified to fit it and a wash down pump instaled.

Also, still having a problem tapping. When you program your feedrate, it does its own feedrate. It is runs completely different from what you program.

Could I get you to send me a short program, with tapping and tool changes? I have to get this right before I type in the wrong code and crash something.

I appreciate all the help!!
Thanks!
Jim

[Harri J]

Here's a program, notice the use of F with G84. F specifies tap pitch.

%
O0000
G00 G40 G49 G80
G00 G90 G54
(M2.5 TAPDRILL 2.3)
G91 G30 Z0. M192 T22
M08
G90 G0 G43 X9.26 Y0.13 Z10. H22 S8000 M03
G00 X9.26 Y0.13 Z10.
Z2.
G01 Z-10. F400
G00 Z2.
G00 Z10.
M5
(END TOOL)
(CHAMFER)
G91 G30 Z0. M192 T1
M08
G90 G0 G43 X9.26 Y0.13 Z10. H1 S3501 M03
G00 X9.26 Y0.13 Z10.
Z2.
G01 Z-1.65 F175
G00 Z2.
G00 Z10.
M5
(END TOOL)
(M3 TAP)
G91 G30 Z0. M192 T33
M08
G90 G0 G43 X9.26 Y0.13 Z10. H33 S6000
G00 X9.26 Y0.13 Z10.
G84 Z-8. R2. F0.5
G80
G00 Z10.
M5
(END TOOL)
M09
M192
G91 G28 Z0.
G91 G30 X0. Y0. Z0
M30
%


BTW, if you do helical interpolation, you'll need to command G17 in the block with G02/G03; example:

G17 G02 X_ Y_ Z_ I_ J_...

otherwise it will first do the circular part and then plunge the z movement.