What's new
What's new

Help, machine zero wanders

sander96

Plastic
Joined
Oct 1, 2022
Hi, I am currently working on a 1992 Okuma Howa act4 lathe fanuc 15tf. I have this problem where the machine zero ref. point seems to wander +-5mm between power offs. This happens for both x and z axis. Machine decelerates and finds the zero point every time without error, its just in a different place than before. I checked the zero proximity switches and cleaned them, but no luck. It has incremental encoders, so it shouldn`t have any axis batteries right? ( well I couldn`t find any). It has no problem with accuracy, so I think the encoders are fine. Maybe the zero return prox. switches are just getting old and need to be replaced? Maybe this is caused by some software offset or parameter? I really don’t know what should I try next. Machine is usable as it is, but its PIA to offset both axis before every use.
 
How about a little history of you and this machine. Semi new to you? Had it lost it's parameters and you've since mostly restored them? Have you swapped out any servos or drives or connections of such things?

The only time I've seen this is on an OMC 4th install I did were unknowingly I bought an Absolute fitted servo and installed it on a machine that's running Incremental. And wired it in like an incremental. Like you it would home just fine but in different positions. I'd have to correct it with a small EXT. A axis work shift. During use it would work just fine.

You might try holding down P and Cancel during startup one time (to clear previous offsets) and see if that does anything. Semi doubtful but...

BTW - Do not try any other key combinations unless you want trouble, or know exactly what you're up to in doing so.
 
I would scope the DC power supplies and the encoder outputs. If the DC power has a lot of AC ripple that could be part of it. If the encoder signals are weak or dirty you will get stuff like this, but happening for both X and Z at the same time is an unlikely deal. Both encoders and/or cables don't fail at the same time.
 
Sorry for not giving enough backround : ). I am a 26yo mechanical engineer in Estonia. I have a small company where I design and build belt grinders. For a while I was looking for a cnc lathe and came a across a good deal on this old okuma howa lathe. Machine has been sitting power off since 2015. Luckily it was in warm room and the batteries didn't run empty. All memory and parameters survived. So far I have ironed out most of the problems and machine is starting to make parts again :)
 
I would scope the DC power supplies and the encoder outputs. If the DC power has a lot of AC ripple that could be part of it. If the encoder signals are weak or dirty you will get stuff like this, but happening for both X and Z at the same time is an unlikely deal. Both encoders and/or cables don't fail at the same time.
Thats a good idea where to start looking. I think I had an anomaly once where i could measure 13v ac in 24v dc circuit when testing hydraulic valves. I dont know are these events related. Could be a faulty reading.
 
Thats a good idea where to start looking. I think I had an anomaly once where i could measure 13v ac in 24v dc circuit when testing hydraulic valves. I dont know are these events related. Could be a faulty reading.

You're not looking at voltage here. You're looking at the waveforms. This is oscilloscope stuff. Any cheap 2 channel oscope will do this.
 
I had a lathe that would do that sometimes. Turned out to be the distance between the home switch closing and the encoder Index was so short that sometimes it would go an extra turn of the encoder to index. The fix was to move the switch ramp about 1/2 the distance of the screw lead, 2.5 mm. Had to reset all the Z offsets but no problems since.

Ed.
 
I'm with the post above for 50%.
I think that you can watch this somehow in DIAG, but I'm not that fluent.
But you may need to move either the dog or the switch on the axis, or - if neither of those are easily done, then maybe loosen the motor coupling and slip the motor shaft 1/2 turn.

The other 50% is that I would really look at those switched good - if you don't replace.
Doo they click/click solidly?
What readings does your meter show in either case? (continuity of switch)
If nothing has been wrenched on recently (and you really don't know what happened to it soon before it was taken from service) then I would expect switches just need replaced.


-----------------------

Think Snow Eh!
Ox
 
I'm with the post above for 50%.
I think that you can watch this somehow in DIAG, but I'm not that fluent.
But you may need to move either the dog or the switch on the axis, or - if neither of those are easily done, then maybe loosen the motor coupling and slip the motor shaft 1/2 turn.

The other 50% is that I would really look at those switched good - if you don't replace.
Doo they click/click solidly?
What readings does your meter show in either case? (continuity of switch)
If nothing has been wrenched on recently (and you really don't know what happened to it soon before it was taken from service) then I would expect switches just need replaced.


-----------------------

Think Snow Eh!
Ox
Ill try to move the switch or dog and see what happens. Thanks!
 
You're not looking at voltage here. You're looking at the waveforms. This is oscilloscope stuff. Any cheap 2 channel oscope will do this.
I dont have any oscilloscopes on hand. But I do have an extra 10A 24V power supply. Maybe if I bypass the machine transformer and rectifier with a known good power supply and test?
 
Hi,
machine was sitting in cold storage throughout winter and now, I finally resumed to this issue.
Today I was able to fix it. I checked ball screw pitch, encoder pulse values and reducer ratios and concluded that parameters 1816 and 1820 were wrong.
These parameters are about Command Multiplying Ratio, Detect Multiplying Ratio and Reference Counter Capacity. With these parameters machine calculates its positioning. I am amazed how the machine functioned this far, with these core positioning values being completely off???
I inserted factory parameter values for 1816, 1820 and now machine zeroes accurately.

For future reference, if anybody has similar issue.
 
Loose encoders? Loose screws/thrust bearings?
One time, during IMTS, I had a brand new Doosan nearly wallop the sub spindle. Couldn't figure out why. Called the factory tech over and he found the wrong fasteners were used in installing the Z axis. It was wandering all over.
 








 
Back
Top