[rsfairman]

I recently acquired a FANUC OM knee mill for my home shop, and something is not right with the x-axis. First, some background.

The machine is from about 1991, and came from a local technical college, at which I was a student. So I know the machine somewhat, but it was never any of the students' favorite. I'm not sure of the model; we all called it "the Johnford," even though it doesn't say "Johnford" anywhere on it. The manual says it came from Machinery Group USA, in Huntington Beach CA, and is their model 375CNC. The most similar model I've seen by searching the internet is a YCM40 Supermax. The machine sat idle for about two years before I purchased it -- it may have been turned on a few times, but was not used much, if at all. The batteries were changed however.

After moving it to my place, the fun began. The first problem I noticed was that the x-axis would get hung up. When jogging in the X- direction, the table would simply stop moving. If you then jogged a little in the X+ direction, it would then allow you to move in the X- direction again, but would get hung up again a few inches later. With use, that problem seems to have gone way.

Next problem was over-travel errors on the x-axis. These seemed to be due to some soft limits (stroke check) that had been set in parameters 700 and 704. Fixed that. Incidentally, I am friends with an instructor from the school (since retired) who recalls that there was a student in the year before this machine was taken out of commission and that this student may have sabotaged some machines in minor ways. That may explain where this error came from.

Next: alarms 401 and 424. What does VRDY stand for anyway? My instructor friend vaguely recalls that they always had to keep the feed override set to 30% on this machine, and doing so does avoid this error, but you're guaranteed to get it if you bump the feed rate above 50%.

Once the machine was working (more or less), it seemed like a good idea to exercise it to get it lubed back up again. I homed the machine (G28 X0Y0Z0), then executed a program that moved each axis repeatedly between the extremes. That is, it moved between roughly X-32.0 and X-1.0 repeatedly, and similarly for the other axes. After running for some time (45 minutes?), it hit the limit switch on the negative x-axis side. I thought maybe it's just working out some kinks, so I ran it again, and the same thing happened. So I reduced the travel in the negative X direction a bit and ran the program again. After a few iterations, it hit the negative X limit switch again. Each time I ran the program, and hit the limit switch I would home it (G28), then jog over manually to hit the other limit switch and note the amount of travel shown on the CRT. I then reduced the amount of travel in the program and ran it again until I got another over-travel alarm. What I found is that the amount of travel from home to the other limit switch went from -32.30 to -31.96, then -31.72, -31.64, -31.49. It seems like the servo/encoder was losing steps (not a stepper motor, I know), but with use it was getting more accurate and NOT losing these steps.

I got out a dial indicator and ran some tests. First, I found that by moving from near the X+ limit switch to the X- limit switch I was losing from 5 to 20 thousandths each time. That is, I put something on the table at the X- end, slowly jogged over near the X+ limit switch, then slowly jogged back to the same place (dial indicator against the object on the table) as determined by the position shown on the CRT. Each time the indicator showed from 0.005 to 0.020 more travel than the time before. While I had the indicator on there, I tried something else. I brought the table to the X+ end, then used MPG and the handwheel set at the highest step rate to move back to the X- end. If I moved back to the X- end at an even steady rate, then the table moved about 0.6 inches LESS than the position on the CRT indicated. If I moved the handwheel in a jerky way -- fast spin for a revolution, stop, another fast spin, stop, etc. -- then the table moved about 3 inches (!!) less than the position on the CRT.

A couple of stats on the machine. The FANUC version is 0469-21, and the x-axis servo says it's Model 0S, type A06B-0313-B002#7000. However, I can't read the actual B002 number on the plate. Someone scratched out whatever was there and used a magic marker to say it's a "2," so the original plate could read B00?, where '?' is anything at all.

I'm thinking that it's a bad encoder, but any advice would be appreciated.

[cjfisher]

I would think that the control would be throwing codes if the encoder were missing... Are the mechanical connections between the servo motor and ball screw tight? A loose coupler could easily be the cause of what you describe. I know on some smaller mills they used timing belts to couple the motor and ball screw. May be worth a through look at the machine's mechanical state - Especially if sabotage is suspected.

Charles

[underthetire]

I would think that the control would be throwing codes if the encoder were missing... Are the mechanical connections between the servo motor and ball screw tight? A loose coupler could easily be the cause of what you describe. I know on some smaller mills they used timing belts to couple the motor and ball screw. May be worth a through look at the machine's mechanical state - Especially if sabotage is suspected.

Charles

I agree 100%

[rsfairman]

Gentlemen, mechanical slippage is certainly part of the problem. There is definitely slippage on the shaft of the motor. There is a timing belt type setup, and the pulley on the motor is slipping on the shaft. I have yet to disassemble this to see exactly what the problem is, but you've pointed me in the right direction. Incidentally, I forgot to include what may be an important fact. When the machine was originally delivered, 20+ years ago, it came with a metric ballscrew when the customer ordered an inch ballscrew. So the x-axis ballscrew was swapped out after the machine was delivered. All of this predates my friend's association with the school and is only what he recalls being told, but maybe the swap wasn't done as well as it should have been. But then 20 years is a long time. I keep you posted on what I find.

Thanks much,

Randall

[underthetire]

Gentlemen, mechanical slippage is certainly part of the problem. There is definitely slippage on the shaft of the motor. There is a timing belt type setup, and the pulley on the motor is slipping on the shaft. I have yet to disassemble this to see exactly what the problem is, but you've pointed me in the right direction. Incidentally, I forgot to include what may be an important fact. When the machine was originally delivered, 20+ years ago, it came with a metric ballscrew when the customer ordered an inch ballscrew. So the x-axis ballscrew was swapped out after the machine was delivered. All of this predates my friend's association with the school and is only what he recalls being told, but maybe the swap wasn't done as well as it should have been. But then 20 years is a long time. I keep you posted on what I find.

Thanks much,

Randall

Ballscrew pitch won't matter as long as the parameters are set correctly. Metric is just easier to calculate.

[rsfairman]

The problem is resolved -- I hope! There was mechanical slippage on the x-axis. Here's what I found.

This machine uses a timing belt kind of setup, and the pulley on the shaft of the x-axis servo was slipping. Once I removed the pulley from the shaft, I could see why. The pulley was on there with only a friction fit (no key), and the taper was garbage. The narrow end of the taper in the pulley fit just well enough that the machine sort of worked most of the time, but the wide end of the taper was a good 0.075 too large for the shaft. Once the belt tension was removed, I could tilt the pulley around on the shaft at will. The y and z-axes also have a friction fit, but not as badly done. In fact, the z-axis is pretty good.

As background, I have information from a person who was around when this machine was purchased. It did come with metric ballscrews when inch ones were ordered, but they did not swap in the correct ballscrews. Instead they sent the servos somewhere (back to the manufacturer?) and they "did something" to the servos to compensate for the wrong ballscrews. I can't imagine what was done since this seems like a simple software switch, with no need to mess with the servos, but that's what he recalls. It also explains why the plate on the servos was changed with magic marker. In any case, whoever made this modification did a crap job. A friction fit seems like a very poor choice here and they didn't even do that very well.

The plan is to fix up the tapers on the pulleys, then cut proper keyways. Assuming this is the only problem, it will be a very solid machine.

Again, thanks for the advice. You were spot on.

Randall

[cjfisher]

Randall,

Glad to hear you have found the problem. Don't knock a taper friction fit... done right it will transmit a hefty amount of torque. My big mill uses RingFeder lock rings on a straight shaft (no key) and the smaller mill (aka the rebuild project) uses motors (similar to yours) with a tapered end and the pulleys are well attached. I'm curious to know if you solved the 401 and 424 alarms. I believe the VRDY is a voltage ready signal that indices if the drives are fully powered and ready to go.

Charles