Cause for jumping G54 Y coor?

[Shotout]

Hello all,

One of our shop's TM-1 mills (2001 model) jumps the Y-cooridinate on start up. Specifically the fixed jaw of the vise is G54 Y-2.8912, the next day after starting up the mill, homing the axises with Power Up Restart and edge finding the same jaw it will be say Y-2.7412, and after the next day's start up back to Y-2.8912. Would the Y axis limit switch failing cause this?

I do get an occasionally failure on the Y axis return on that machine. As far as part accuracy I have noticed a small (+.0015) concentricy issue on milling circlular bosses using both I & J as well as R. when using CW and CCW interpulation. This is with slower feedrates such as finishing 316 SS or with faster feed rates in 1018 CRS and can not be deflection. Also the machine inspection sheet shows +.0002 for such operations when it wa new so to me this is a significant difference.

I wanted the front office to issue a PO for a service call but was asked if I could work around it, which admittably I can so they said not now, but I'm concerned about it being a symptom of a serious problem developing. Not to meantion when I need that machine to run a part with a tighter tolerance I might not be able to. Any ideas suggestions would be appreciated.

[HuFlungDung]

You might need to check for cleanliness around and on the limit switches for that axis. There could be a stray chip causing a premature trigger of the machine homing routine on that axis. I suppose that when the machine senses the limit (or the chip) via proximity sensor, that it then searches for an encoder index, which would explain why the error has two discrete values, rather than random values, because the encoder index is detected once per revolution of the ballscrew.

[Shotout]

I had assumed that to be the problem origionally, had that problem with the z return on the other machine. After cleaning the machine as well as I know how failed to resolve it I began to notice a pattern to the shift. It is the same jump amount daily and always +Y every other startup returning to the origional value the next day, in other words day one Y, day 2 Y+, day 3 Y, day 4 Y+ and so on. I experimented a bit by powering down and then powering up again and edge finding the fixed jaw. Apperantly the machine has to be powered down for a certian amount of time for the shift to take effect. You can imagine the headache of continuing a program from day to day on an incomplete part. Honestly it can be worked around, but it adds set up time plus increases the chance for error when shifting cooridinates etc. I appreciate the info.

[HuFlungDung]

Have you talked with Haas support about this problem?

Another factor could possibly be that when the machine homes on the limit, that it finds it is already sitting on the encoder index. So, sometimes, it immediately finds the index and registers the current position as home, other times, it decides to search one whole turn for the encoder index. There is most likely a tiny bit of variation to the detection of the limit switch, so this is why you find the machine has two states of home in that axis.

This could be one of those uncontrolled factors from machine to machine, because the encoder index position is uncontrolled as everything is put together.

Now I am a bit more read up on Mitsubishi controllers, but there are options in Mits setup to define a "grid mask", which I believe, is a trick to automatically force the motor to rotate a few degrees at the moment that the dog (limit switch) trips. The purpose of this is to force the encoder off the index position, after which, the motor then turns in a certain direction to establish machine home. This eliminates the ambiguity factor of the machine finding the encoder index at the same instant as the limit switch trips.

[gar]

070225-1111 EST USA

Shotout:

The difference in your two readings is 0.1500". Is this the pitch of your leadscrew? I doubt it.

HAAS sets up the limit switch position to be approximately 180 deg from the once per revolution on the encoder. I believe thay also do a test that if the limit switch position is too close to the once per revolution mark that an error is generated.

If 0.1500" is not the leadscrew pitch, then you re looking for some other problem than the relation between the limit switch actuation and the once per revolution mark.

What does G54 have to do with your measurement? For the test you are doing you should use the machine absolute position.

If you are measuring the vise position relative to G54, then check the content of G52 Y, and G92 Y.

Also HAAS may do some temperature compensation modification of an axis and you have no idea what it is because it is hidden and I do not believe there is anywhere to view its value. Normally you would not expect its value to be this large. It also modifies machine absolute position and you have no way of knowing of its existence. This comment is based on experiments on our HAAS lathe.

If the absolute position measurement shows this error and the leadscrew pitch is not 0.1500", then it is likely an electronic problem. However, 1500 is not a simple 1 bit error in binary. For our VF series machines I believe the encoder increment is about 30 millionths, and the encoder arround 8000, so our pitch is about 0.25". I do not have an easy reference to verify this at the moment.

.

[Shotout]

HuFlung

I had noticed that the axis slows after hitting the limit switch and then finds the machine home. I had assumed it was something along the lines of what you stated. While the Haas manual doesn't elxplain it the manual for our Mazak QT-250 does and it staes that it homes exactly as you said. I assumed the Haas did it's own version of this as well. What you said makes sense if the limit switch is failing, which since on some occasions, mostly cold mornings or after the weekend it has failed to home at all instead tripping the servo from trying to overtravel. If the limit switch isn't always signalling at the same position, or in that case at all, so that could very well explain the shift since it has to rotate a differing amount to find the home on the encoder. Since it has gotten worse last week I am going to call Haas monday, the only problem is since I'm the only guy in the shop by the time they open I'm pretty busy and forget to call until it is to late in the day. Luckily I finished out everything I needed to get done Thur (4 10hr shifts) so I should have some breating room monday, unless the salesman slaps it on me again with short due dates again.

Gar
The difference is .15xx, but I can't remember the last two decimal place values. It is always the same amount however. The G54 relates as such. I edge find and on the offset screen highlight the approp axis, in this case it is Y that differs, the X axis is consistant to .0002 which I assume is me not the machine, and select part zero set. The reading at the bottom of the screen is the machine absolute value for which ever axis is selected which is where I am getting the Y-2.xxxx from. G54 0.0 is Y-2.xxxx in machine coordinates in other words so I guess I didn't state myself clearly in the origional post.

Thanks for the help

[gar]

070225-1549 EST USA

Shotout:

Check your manual for Alarm 165-8. This describes the limit switch relationship with the once per revolution encoder mark.

On our VF and SL machines the encoders have 32768 steps per revolution for the 2000 year units. The 1993 VF2 appears to have a 2000 count encoder and reference is made to a 6 MM screw. However, no direct statement that the leadscrew pitch is 6 MM. Note 6 MM = 0.23622 IN.

It does appear that you are looking at the machine absolute position.

Try to find out what is your leadscrew pitch. 0.15" seems too small to me. If the leadscrew pitch is not close to your 0.15", then I do not think it is detecting the wrong zero position.

When your table is homed measure from the back edge of the table to the column for the two different conditions. If this distance remains constant, then it is not a table zeroing problem. If it shifts by the 0.15" value, then it is a zeroing problem. If there is free space on your table at the home position, then you could measure from the spindle center to the table edge.

.

[Shotout]

Gar,
The tape did show a difference. I didn't find the error listed that you stated but Haas provided a solution. The tech said that the limit switch was sticking, most likely from dried coolant after he had me describe how I had cleaned it. He suggested flooding it with WD-40 and letting it soak. So I placed a couple rags uderneath it and did so, I only thought I had cleaned it well. It eliminated the problem. Thanks to you and HuFlung for the suggestions and help.

Scott