[RustedOut]

Had a job to do on the mill and part of it was doing some scribing. Started off ok, going about .005" deep with a single flute scribe. It was a pretty slow process since I can only run at 1600 rpm for now. There was, I think, 8 different parts to scribe with 6 different scribe areas per part. So the mill head was up and down while moving through the parts all in a row.

I have no way to measure the depth of each scribe area, but they got progressively deeper as the program wore on toward the end. This was in .031 O1 steel, not heat treated. I hazard a guess that the better half of it was .020" deep. But my dro showed a constant Z-.005". Parts are probably scrap, but the fellow they belong to hasn't seen them yet.

I suspected something going on with the head, but didn't really see anything of notice during the run of that program. I had to then flip the parts and scribe them on the opposite side. Same thing, started out good, then got deep as the mill moved through the program. Since they were now individual parts at that point I was able to keep reseting Z to account for whatever was happening. I was suspecting cutter slip in my ER32 collet at that point.

I decided to abandon the single flute scribe at one point and go with my spring loaded scribe to finish them up. So I got through them and may have to get some more material and do them over. Today, on another project, just finishing one cut to bring it within tolerance and Z starts to get wacky again. This time it's in a pretty good tool holder so I'm not thinking tool slippage. I started watching the pulley and belt and noticed it cogging down. Maybe a thou or two at a time. Again with no indication from the dro that the Z was moving. It read .0000" the whole time. I decided I could live with some reduction in the Z value, so I kept going to see where it would end. Dropped a total of .030".

Last night I was changing parts after a program ended, or was reaching for my wrench to do that. I turned back to the mill to find the head slowly lowering itself. I was using a spoil board to hold my parts, so the cutter had a semi soft place to stick and it went about 1/2" into it. Then the drive faulted. It was just like the power had been shut off to the motor and was free wheeling down the screw. I reset the drive and things were good again. Never happened again, but it shouldn't have happened even once.

I'm thinking two separate issues here. I have problems trouble shooting things unless I can duplicate them and test different fixes. I'm sure the creeping is fixable, once I find the cause. Not sure I can duplicate the other issue though. So, going through the creeping Z in my head trying to think of the possible causes. I'm not really coming up with anything concrete to even look for right now. I did try to retune the drive the night before last and was getting some strange oscillations from the motor. I never felt I had it tuned as good as I had in the past, but going beyond were it is now just seemed to make the thing skittish. Could the gain and damp of the drive cause the head to creep down? And to think, I was getting ready to start milling on my belt drive conversion. That's now on hold until this thing is fixed and thoroughly tested.

Bob

[RustedOut]

Haven't been able to get Z to creep at all today. Mills been running for over 3 hours, 2 in a static mode and the third while I machined in X and Y with Z in a fixed position cutting some parts to length. I made a line on the belt and it's in the same place it's been since I made it.

Real concerned about getting into something that could be messed up with errant Z moves. Got some drilling and pocket milling to do, so I'll see how that goes.

The thing that has me most puzzled is the encoder showed no movement on the dro while I could watch the pulleys turn and the head physically come down and witness it in the cutter removing material on the end. I'm not getting my brain around that at all. And the worst part was I could tell it to jog in Z yesterday and the dro would reflect the move. Just nothing from the dro while it was creeping before or after those jogs.

Bob

[mcarvey]

I haven't had your issues, but I have found a few tricks that help with repeatable Z axis depths. But I think you have a different issue. Is this a CNC'd machine? Is the DRO a separate DRO or the Mach DRO?

The thing that has me most puzzled is the encoder showed no movement on the dro while I could watch the pulleys turn and the head physically come down and witness it in the cutter removing material on the end.

This is a HUGE Red Flag- it absolutely should not be doing this. I would check to make sure the encoder/pulley coupling to the ballscrew is securely tightened.
Once you get that worked out, If you are trying to increase the accuracy of the Z axis, it is important that you get the machine (spindle and ballscrews) at a steady state temperature before you set your Z height. The spindle can get pretty warm, and the ballscrews also heat up slightly. If you run the numbers, you'd be surprised by how much a 0.6m ballscrew can change with only a modest change in temperature. (steel has coefficient of thermal expansion of 12um/m/C, so a 10 deg C change in ballscrew temperate can make the Z axis grow by ~2.8 mils! )

Hope this helps,

-Matt

[RustedOut]

Yep, cnc mill. dro is on the Mach screen. Been thinking it would be real handy to have a separate dro display, but no plans for that right now.

Once the issues started yesterday, the mill had been running for a few hours. I did some measuring a little while ago on a few of the early z features and they were within a thousand or two. The other day, the problems started on the second running of one other program. So yesterday it was run from the mdi and the other was from a program from my cam.

Just finished my pocketing for some cap screws after drilling all the holes first. I spent the whole time watching the line I had marked on the belt. The z was dead on every time the mark came into alignment with the center of the z motor spindle at a z-0.175 in my program. Measured the pockets depths and they repeated within .0005" or less. These were from programs done in the wizard. That's what I'm used to seeing with this mill, not the funny stuff I've been seeing the past few days.

And now, after spending the day watching the belt and pulleys, I'm going to estimate the cogging yesterday was around .006" to .008" each time it did it, always going down.

Bob

[Runner4404spd]

couple things to look for...

1. do you have a capacitor at the encoder for the z axis? its not included in the kit but does a real good job filtering noise that maybe coming in on the wire bundle.

2. check to see that your head stock is bolted tightly to the guide plate. i had a problem with z repeating as the head stock was loose, as long as you didn't go up and down your were ok, but the minute you tried doing repetive depth cuts the z would lose position and creep down. a few thou on the first pocket, a few more on the second, a lot overall on the third once the sticktion on the ways was overcome.

if your dro is not moving than that means the computer is not commanding the axis. this is an inherent problem with mach 3 and the fact that it has zero direct feed back from the motors. that being said, positionally, mach 3 should work fine and its works great on my machine, just need to watch and work within its limitations.

[RustedOut]

couple things to look for...

1. do you have a capacitor at the encoder for the z axis? its not included in the kit but does a real good job filtering noise that maybe coming in on the wire bundle.

Yep, been there since the start of my build. Picked that info up from Maris Freimanis off of the Gecko site. Looked that end over on Saturday morning before I fired up the mill. All looked ok to me, but nothing tested.

2. check to see that your head stock is bolted tightly to the guide plate. i had a problem with z repeating as the head stock was loose, as long as you didn't go up and down your were ok, but the minute you tried doing repetive depth cuts the z would lose position and creep down. a few thou on the first pocket, a few more on the second, a lot overall on the third once the sticktion on the ways was overcome.

The head's not loose. I've seen and corrected that in the past and as well as a loose nut mount. And I thought about the sticktion issue on Friday. Turned the mill off with the head midway up. If my head isn't all the way to the top when power off, the head will slowly slide down. That's exactly what it did to test the gibs and lube when I turned it off.

I just trammed the mill on Tuesday last week. I checked the head for looseness at that time with the indicator zeroed on the table. Nothing scary there, the needle moved and returned right back when I quit pulling on the head. Now that all took place before I seen the motor pulley rotating on it's own while running what should have been a fixed z operation. I ran some small pockets on Saturday with great results, so this is an intermittent thing going on.

if your dro is not moving than that means the computer is not commanding the axis. this is an inherent problem with mach 3 and the fact that it has zero direct feed back from the motors. that being said, positionally, mach 3 should work fine and its works great on my machine, just need to watch and work within its limitations.

Yep, I get that and it's a scary thought right now. It was commanding the x and y at that time though. A closed loop would have caught it right away. Mach has always worked good for me as well, up till last week. Not sure it's a problem with Mach though and I'm focused on the drive right now. Just a gut feeling I have.

Bob

[Cruiser]

I had a similar loss yrs ago, turned out that I had tuned my gecko's too tight ! and they were constantly clicking. I backed off the tune till they were quiet and the lost motion stopped too, forever ! And it is only on the Z drive because of the tension needed on it was so high at the gibs that the motor is making major demands on the gecko. That is also when I started to rub moly on the slides too.

[RustedOut]

I had a similar loss yrs ago, turned out that I had tuned my gecko's too tight ! and they were constantly clicking. I backed off the tune till they were quiet and the lost motion stopped too, forever ! And it is only on the Z drive because of the tension needed on it was so high at the gibs that the motor is making major demands on the gecko. That is also when I started to rub moly on the slides too.

This is good to hear, Don. I mulled everything over a couple days ago and decided it had to be one component causing the problem. I didn't think it was an encoder, as it didn't seem like it was lost. The most likely I figured was the drive. But to be honest, I wasn't seeing how the gain and dampening would cause it. And I still can't figure out why it would change after being so smooth for so long. Z has been my smoothest drive motor combo. But X and Y are up there as well.

So, thinking back, the motor was strange for a few days last week. Seemed to be under some tension. Not continuously growling or oscillating, not even warm really. But rapids ended with some jerkiness and belt slapping and a short growl upon stopping, especially downward rapids. Just not smooth like it has been.

I believe I started re-tuning the directly drive after the head lowered itself to the table. Memory is a bit fuzzy today on the exact time line, but if I'm recalling correctly I had no clue what was happening to cause the cogging problem but I knew there was a drive issue when the head seemed unpowered that one time. So, that would put the tuning prior to Saturday and that was the day I tested it all day long without issue.

I got a couple 1" thick steel parts to mill soon as my coated roughing cutter shows up today, so that'll give me many hours of time to watch the mill. Lots of up and down motion as there are many holes to mill then the outside profile, but they are all through holes, no critical Z features at all. Good parts for further testing. Maybe I have it fixed and just don't know it.

Thanks for the insight!
Bob

[veggiedog]

My gecko err/res pins were wired directly to enc+, and consequently the controllers were ignoring faults. The result in my case was that the Z-axis controller was faulting when the head was on the way up (I adjusted my gib too tight), and then when the head was on the way down on the next down command, Mach thought the head was higher than it actually was. My head was slowly losing vertical height relative to Mach's indication.

Could it be that your Z-axis controller is faulting (and your controller is also wired to ignore faults), and Mach is then losing track of the head's position during the fault?

I suspect that the head could drop a little due to gravity if the servo isn't holding it in place during a controller fault.

Just speculating...

[RustedOut]

No, I changed that way back last spring. I installed a momentary switch for the reset of all three drives and three relays with led indicators.

But you're right, they shouldn't be set to automatically reset. If there's a fault on any one of the axis, I'd think a guy would want to know it happened.

I've not heard back from Gecko after asking them for some advise. It did the funny movements under power to the Gecko's and the drive had been tuned for so long I forgot when I tuned them. After the problem reared it's head, I had to retune it and went through some testing to get back to having the Z motor smooth again.

What really puzzles me is how did it change so much. Wish I had or could borrow a scope, but then I'd have to find someone who knew how to use it.

Bob

[RustedOut]

I've been using the mill for small stuff without anything more than tweaking the gain and damp switches for the past month or more with good results. Wednesday I ran it right at 16 hours continuous in a variety of ways and my Z never missed a beat. I was nervous going in, but the work needed to get done. Results were just great and the parts fantastic. And I watched the Z pulleys and the dro the whole time.

Still wish I knew how an electronic adjustment that was really good went to hell so quickly. It's a switch after all and it should hold it's setting, at least that's the way I see it. Or does that need to be part of my periodic maintenance too from now on? I guess I'll call it fixed at this point and make a mental note to watch for issues there.

Bob

[RustedOut]

How's that saying go, famous last words or something to that effect.

The day after I wrote the post on the 16th, I couldn't get the mill to stop oscillating. Mostly on rapids, but not exclusively. And mostly going down, but not exclusive to that either. A few more days of that and I lost all faith in the mill and stopped using it. I didn't stop working on it though. Funny thing (but I wasn't laughing) was I could mess with the X or Y drive and retune them to run smoothly in no time. Not so with Z.

I got so disgusted with it I didn't touch it for a whole week. I had other things needing done and it was keeping me from them. The past week I've worked on it every day and now have it calmed back down a great deal. I even used it last night cutting a steel block. The Z motor did rattle right at the start on a downward rapid move. I backed off the gain a bit and it calmed down.

I'll keep a watchful eye on it, but I'm backed up on other peoples milling projects and need to press this thing back into service. Next week it goes back into full time service.

Bob

[Cruiser]

Rusty
Did you check out the thread I did a while back about shaking ball screw ?
Back a few yrs ago I had been swapping out drivers and did so by pulling the wire blocks off the pins and in the process the little scrapers that contact the pins deform and loose contact. That was the point of the thread as well as the easy tune ear method. I will attest that it is super easy and only takes 20 seconds per to do and get it right. I was amazed at just how simple it really is !
The key is to back off both gain and damp and turn up first pot next to limit till an audible sound is noticed from servo then back off till totally quiet. Then do the same to the third pot and QUIT RIGHT THERE ! Before I couldn't quit, I had to keep at it till ! I must say that it has been very quiet and smooth since and reliable too. I was about to really tear into things till I saw the SPARKS !

Edit ! One last note, In the motor tuning I have been setting the acceleration low so that when I jogged the axis I could just detect the accel. I use about 5 on Z and 7 on X & Y. There is no reason to have any more than that, also, I have been leaving the Vel at 100 even tho I can take it up almost to 120.

[fatal-exception]

I had the same problem on my mill recently. The cap on the encoder did the trick. Dont know if they ever wear out, but it might be worth a try replacing it.

I wrote a program something like this to check it before and after the cap was installed:

G0 z1
m98 P1 Q100
M30

O1
G1 Z.5 f50
g1 Z1
M99
M30

At the end of the 100 cycles, your dial indicator should read the same as at the start. If it doesn't it's probably a noise problem, assuming everything mechanical is tight.

Hope you find it soon. I hate when the machine doesn't do what I tell it to do!

[RustedOut]

Yep Don, I read that and took a look at mine. I pulled it out of the cabinet and took a good look at it. It was plenty tight and I've not noticed any arching there. I've been down on the floor and personal with it for weeks. The creeping is fixed for as much as I've been running it and just getting the motor oscillations now. Messed with it so much I burnt up my parallel port a couple weeks ago and had to swap to a backup pc. Took me a while to figure that one out, but I apparently fat fingered a ground while probing terminals.

I used to have the Z drive tuning screws turned a lot like the other two drives, but now they are backed off ccw a lot. Backed off the gain a bit more yesterday. Maybe the thing was like you say, I didn't know when to quit with the tuning. Maybe it was on the ragged edge all along. Kind of hesitant to back them off all the way, but wonder what that would give me. I don't need the head sliding down again.

I ended up getting a nice older analog scope to help me out, but I've not been able to use it to tweak the tuning. I get a signal displayed, but I just can't see that much change in the sine wave while I turn the screws. I don't think it's the scope, it's just me most likely. The tune right now is all by ear and feeling the motors. They are dead calm at idle and pretty quiet while running. But the sine wave appears to be under damped.

I have caps on the encoders and had not considered one failing. I did wonder about the encoder when I initially got the creeping problem. I couldn't for the life of me find anything wrong though. I've been using the scope to measure the voltages in all the components. Never used one before, but I got to say it's a pretty handy delicate instrument. Other than me messing up the pc and taking the tuning of the Z drive to the opposite end of where it was, there's been no changes to speak of.

I'll be glad when I get to the point where I'm measuring repeatability again. That means this on going problem will be behind me.

Bob

[Cruiser]

I thought I might re-iterate a bit on what I meant about "backing off". Let me get away with changing the names of pots to numbers, please.
1st pot being power, After backing off pot three and two full ccw, reduce pot 1 till it has just enough power to drive servo. Then without driving servo at all just turn pot 2 up till servo makes some noise, then very slowly back it off till all is quiet. Now increase pot 3 till it sounds off a bit and back it off till quiet. Now turn up power to full for Z and about 80% for X & Y.
This is where you leave them well enough alone.
Back when Aaron was still active, He made special note that things change with the season and need periodic clean, adjust, tune. The way that tuning was described in the old days, It led me to trying to "Get The Most" out of the tuning process and I would keep trying to get things stable with turning the pots up and up till servo complained then nudging back till it was tolerable. This is NOT proper. According to Gecko instruction, the pot 3 is backed off fully and pot 2 is backed off fully before starting to re-tune ! Then tune as described above. I only just recently stopped kicking myself in the ass for being such an ass !
The only other two things I can think off that causes problems like this are, One, getting a little coolant inside the encoder housing which can cause condensation to form on the encoder, since the micro lines are printed with "BUG SPIT" and won't tolerate "ANYTHING MOIST" this can create a dead spot that will not go away. I lost a servo on the Y axis because of this issue. And that brings up the second item, damaged servo ! Where for what ever reason it burns out one winding and gets a nasty attitude.
One thing that might be worth looking at is the housing for your Z encoder. My issue with mine on Y was that the shaft end was turned too small and it did not have that sealing interference with that little O-ring and it allowed coolant to wick in on the shaft, then the condensation started and encoder was trashed.
The funny thing is that with the trashed encoder, momentum would carry through the bad spot at times, with an audible jerk, and show up with the lost motion. This can be seen because the spot on encoder will come up at the same location with each turn of the shaft. At some point I took the encoder out and looked at it with a loop and seeing the spot got the great idea of cleaning it ! and clean it I did, there were no sign of those pesky lines left on the area I cleaned !
This is when I made another big mistake, when I put it all back together, I used some old locktite that now makes getting the wheel off of shaft impossible by conventional means ! We really need a different method of holding the wheels in place on shafts for maintenance and repairs. I DON'T LIKE THE LOCKTITE METHOD AT ALL !
I really really need to come up with a solid way to get a grab on that ball screw in order to re-adjust the preloads and get rid of the back lash mechanically. The software just doesn't do it proper.

GOT ANY IDEAS FOR HOLDING THAT DAMNED THING ?

sorry for such a long post, I guess I'm bored !

[RustedOut]

I thought I might re-iterate a bit on what I meant about "backing off". Let me get away with changing the names of pots to numbers, please.
1st pot being power, After backing off pot three and two full ccw, reduce pot 1 till it has just enough power to drive servo. Then without driving servo at all just turn pot 2 up till servo makes some noise, then very slowly back it off till all is quiet. Now increase pot 3 till it sounds off a bit and back it off till quiet. Now turn up power to full for Z and about 80% for X & Y.
This is where you leave them well enough alone.

I kind of thought that's the way you were describing the pots. I was messing with X yesterday, having the scope attached, trying to affect the sine wave. But I was also listening to the servo motor. Seemed that nothing I did affected the signal, but I could follow the settings with the sound from the servo.

I've never found anything concrete on the starting positions for the pots prior to tuning other than the bench work prior to installing them in the machine. As my problem is with the Z, I've been hesitant to lower the gain (pot 3) too much. Not too sure of the outcome. However, yesterday I did just that and the head came down on it's own under gravity. Couldn't get the drive to take control again until it came to rest on the vise. That sucked! So I won't be doing that again.

Back when Aaron was still active, He made special note that things change with the season and need periodic clean, adjust, tune. The way that tuning was described in the old days, It led me to trying to "Get The Most" out of the tuning process and I would keep trying to get things stable with turning the pots up and up till servo complained then nudging back till it was tolerable. This is NOT proper. According to Gecko instruction, the pot 3 is backed off fully and pot 2 is backed off fully before starting to re-tune ! Then tune as described above. I only just recently stopped kicking myself in the ass for being such an ass !

I had the Z servo working pretty good for a couple days, but it got the rattles on a rapid down movement and I got back into tuning it again. One thing is it seemed soft to me, not like X and Y at all. Maybe that's OK and maybe I need to leave it there, that seems to be what you're saying too. That is if I get back to that condition today.

The only other two things I can think off that causes problems like this are, One, getting a little coolant inside the encoder housing which can cause condensation to form on the encoder, since the micro lines are printed with "BUG SPIT" and won't tolerate "ANYTHING MOIST" this can create a dead spot that will not go away. I lost a servo on the Y axis because of this issue. And that brings up the second item, damaged servo ! Where for what ever reason it burns out one winding and gets a nasty attitude.
One thing that might be worth looking at is the housing for your Z encoder. My issue with mine on Y was that the shaft end was turned too small and it did not have that sealing interference with that little O-ring and it allowed coolant to wick in on the shaft, then the condensation started and encoder was trashed.
The funny thing is that with the trashed encoder, momentum would carry through the bad spot at times, with an audible jerk, and show up with the lost motion. This can be seen because the spot on encoder will come up at the same location with each turn of the shaft. At some point I took the encoder out and looked at it with a loop and seeing the spot got the great idea of cleaning it ! and clean it I did, there were no sign of those pesky lines left on the area I cleaned !

Now that's cleaning things! LOL! Going to look at the Z encoder output today with the scope to see if I can find any problems there. Been thinking of adding the EM1, I think they're called, to my encoders that Marriss has recommended. Once I get my chip collection panels made and installed, the control cabinet will have to move making longer cables necessary. But if I got a bad encoder I might as well make that change now. Probably pull the cover too and take a look at it, although I think my problem is right there in the drive. I don't run coolant, but I've seen an oily mist on the Z wheel before. I have my spare back at Gecko getting checked out.

This is when I made another big mistake, when I put it all back together, I used some old locktite that now makes getting the wheel off of shaft impossible by conventional means ! We really need a different method of holding the wheels in place on shafts for maintenance and repairs. I DON'T LIKE THE LOCKTITE METHOD AT ALL !
I really really need to come up with a solid way to get a grab on that ball screw in order to re-adjust the preloads and get rid of the back lash mechanically. The software just doesn't do it proper.

GOT ANY IDEAS FOR HOLDING THAT DAMNED THING ?

sorry for such a long post, I guess I'm bored !

I've thought that a lock nut against the pulley would be the best approach, but there's just no room without some other changes. And I ground small a flat on the 3/8" diameter section of the ball screw for the encoder wheel set screw. That worked wonders for holding the wheel to the shaft.

Thanks for the follow up Don, I'm spending a lot of time researching and testing this thing. Wish I was able to use it, but I don't trust it right now. I have learned a bit about the old scope I bought for this purpose. Yesterday I probably learned more than ever, but still got a long way to go with it.

Bob

[Cruiser]

I purchased an inexpensive scope a couple of yrs ago, then tried to set it according to the Merris info. After spinning my wheels trying to figure out the scope, I put it away and haven't touched it since. Wasted money again ! and appropriately marked as such on my score board. I think I need a new board, this one is full up !
I hope that you get this thing figured out soon.
I have not had my Z fall in a long time and that time it was my fault. It stays put with no power as long as I don't go and fool with the hand knob and give it some momentum to amplify.

[RustedOut]

It's taken quite a bit of nudging to come to grips with buying the scope. I've made bone headed buys before that I regretted directly afterward. I was hoping this wasn't one as I didn't even know how to use it.

After many, many hours of reading, I decided the info for using the scope in the Gecko 320 manual was for checking the encoders. It's found under step 3 and talks about the voltage and time settings. Under the tuning portion, there are some connections that need to be made, but no mention of volt or time settings.

I've been fooling with all the settings on my Tektronix 475 and use .1v (due to the 10x attenuation of the 10x probes) and 1 µs (microsecond) for time instead of the 10 ms (millisecond) talked about there in the encoder testing portion. And I get a sine wave.

Now from there it starts to break down a bit for me as I can't seem to get the correct damped sine wave I'm hoping for. Close but not matching the picture in the manual. A lot of noise on the front of the signal from something. It doesn't change by adjusting the pots either.

But today I started testing the encoders and decided to go ahead and order new ones. The originals looked noisy to me even with the caps installed. I even replaced the cap on the Z encoder with no change. I looked up the price on the EM1's, then the AMT102's and went with the AMT102's from DigiKey. Just placed the order a few minutes ago.

So, I'll let it sit until I can at least change the Z encoder and take a look at the signal from that. I hope the little flats I ground on my ball screws don't give me problems with the new encoders. They don't use set screws.

Bob

[Cruiser]

How closely do they represent the OEMs from IH ? I have replaced my Y three times, all the replacements came from IH ! Those replaced were from the coolant wicking & condensation leading to my cleaning off those damned lines ! As I recall, simply pinching them with your fingers could smear those lines.