An Enhanced X2

This is the Z-axis axle with the drive gear. The gear sits in a push fit pocket to help keep everything concentric. The gear is held with 3 5mm screws that clamp the ball nut, and gear to the axle. Three additional 5mm screws hold the nut to the gear. I am waiting on the screw. It has been in Chicago for a week now. I have received empty boxes in the past when they have been held up this long, so I hope I don’t have to wait for another one to be shipped. I can fabricate the head attachment for the screw once it gets here. Then it’s just a matter of putting it all together. That should finish the mechanical part of the upgrade. It will still need some electronic part installation, but its completion is within sight.
Z Bearing Stack.jpg
 
The ball screw for the Z-axis finally arrived. Getting things from asia is always interesting. It arrived in Chicago about when I expected, but then it sat the for 10 days. I was getting fearful that it had fallen off the belt or torn apart by some machine. It finally went to customs. The next day it was on my door step. No trasnfwer from customs, no arrival at the destination post office, no out for delivery. I try not to order international stuff because you never know when or if you will get it.

I drilled and tapped the needed 5/16 hole to attach it to the head. No particular gotcha when I put it all together. It goes up and dow at 60ipm, even without any counter balance. I will probably install something to take most of the weight off the screw. I might even get some more speed out of it.
Z Axis Done.jpg

The last major mechanical work was to put the motor back on. Minor problem. The left support riser was too long. The motor control box wouldn't fit. I happened to have a face mill on the G759, so it only took a few minutes to mill off .120". Motor and control boxes are on.
Motor Done.jpg

The spindle spins and everything moves. Hot darn it really works!

I still need to install some limit switches and put the wiring in flexible conduit before calibration.

Larry
 
The electrical work is done, I think. Home switches for all 3 axis. There are connectors for all the cables. The motor and home switch for the X-axis needs some sort of flexible conduit. I got a chunk of Igus link carrier, but it doesn't work right.

The original incarnation of the machine used micro switches for home and limit. Limit switches proved to be of little real value. The micro switches were prone to failure and collected swarf. The switches for this version are mostly sealed and have over travel so nothing gets bent. I originally used intercom cable for motor drive. Solid conductor cable doesn't do well on things that move. I am currently using 4-conductor trailer wire. Over kill as far as gauge goes, but low resistance is always a good thing.

The Z drive doesn't move as fast as the others, but seems reliable. I did not re-install the spring counter because it limited the Z travel. I don't think air springs will work any better. I will set up a counter weight if it has any problems with stalling.

IMG_1444[1].JPG

IMG_1445[1].JPG

IMG_1446[1].JPG
 
Before I embarked on calibration of the axis, I needed to get the column square and trammed. I couldn't just tram the column to get it square to the table. The rotating head queered that. Instead I mounted a 1-2-3 block on the table. I mounted a .0001 resolution dial indicator to the head so that it bore on the block. I then ran the head up and down whilst tapping the column into alignment. I managed to get it down to .00015 in 2.5", or .0006 over 6". I have no way to check the accuracy of the block, it is probably good enough for anything I would make. I then mounted my tramming device in the head and adjusted the rotation so that the spindle is square to the table.

I had intended to invite Ray over and have some fun calibrating the steppers, but I discovered that I really didn't know how to do the calibration. The original hacked up machine was run with Mach3. Once that machine had passed the smoke test and hadn't fallen apart after a few hours of running, I needed to make a decision on legal software to run it. Mach3 is nice, but I still don't think the few times I will probably make things with it justify the cost. The version 1 machine finished life running with linuxCNC. The ancient PC that I am running the machine with also likes Ubuntu better than the recent versions of windows. So, the ini file to run linuxCNC was converted from the original Mach3 one. That method didn't work now that I had different screws moving the axis. After some experimenting I found the ini parameters that needed to be changed. Only when I was about done with the calibration did I figure out how to use the built in calibration routines. The axis are now accurate to about .0003. I still need to check backlash, but I am putting that of until I solve the next ugly problem.

While messing about looking for documentation for the HAL parameters, I discovered that my version of linuxCNC is 2.5 and no longer supported. The current version is 2.7.7, but you can't upgrade to it unless you do an upgrade to 2.6 first. I don't have networking on the controller PC. Upgrading once is a pain, but upgading twice to get to the current version is not going to happen. I saved all of my custom HAL settings. I downloaded the latest version of linuxCNC and burned it to DVD. Tomorrow I will receive a wifi dongle that is supposed to work with ubuntu. tomorrow afternoon I will lock myself in the shop and reinstall linuxCNC. Worst that could happen is that I have to retreat to 2.5. If I'm really lucky and hold my tongue to the roof of my mouth, while tapping my left foot and singling Dixie, I may end up with version 2.77 and wifi access.
 
Scraping in dovetails is far easier than upgrading ancient software. I got my wifi dongle. It said it was linux compatible. Well, sorta. It is compatible with the newer linux versions, not my Ubuntu 10.04.

I disconnected the cables and hauled the computer into the office so I could upgrade it. After dozens of re-installs and a new CD drive, I discovered that version 2.7 and 2.6 detect my wifi, but crash starting up linuxCNC. Versions 2.5 and 2.4 don't detect wifi, but linuxCNC does function. In an effort to get wifi, linuxCNC, and a supported version, I bought a refurbished computer and had it shipped over night.

I'm not sure what constitutes 'refurbished' The computer I received beeped like a calliope when I turned it on. The system battery was dead and the hard drive fan was unplugged. And the left side cover latches were broken so the side fell off when I moved the box. It went back. I gave in. I will use linuxCNC 2.5 with no network connection until I get rich and buy a new desktop.
 
I have the machine running and cut a circular test pocket. It looked round so the backlash compensation is working. It was a 1.250" pocket cut with a 3/8 end mill. It was too small by a few thousandths as measured by a dial caliper, electronic caliper, and a micrometer over a telescoping gauge. I adjusted the pulses per inch and the pocket became the right size.
IMG_1449[1].JPG
The ugly problem is that manual moves with the MDI interface are now off by the amount I adjusted the pocket for. Something is wrong with the cam or the post, or 'I don't know'. I snuck away from T-day celebration to do the MDI measurements. I haven't had a chance to look into it. Anyone else have this kind of discrepancy?
 
Steps/unit are better set using a test indicator and a piece of known length (123 block or similar). Mach3 will even do the hard work for you

measure your end mill. I have noticed that cheaper end mills are typically undersized. Carbide end mills seem to be a very common offender for this. New or otherwise I always measure my end mills before they go into my tool library.

If that's not an issue, post up the Gc0de. should be easy to find out if your cam is leaving extra stock.
 
I didn't see in your posts above if you are using the parallel port or and external motion controller. Correct me if I'm wrong but somewhere I read that axis calibration only works if you are using the parallel port. Doesn't work with USB or ethernet motion controllers.

Tom S.
 
I will correct you :D I have used it on USB and ethernet smoothsteppers and it worked like a champ.

So that said, I just now remembered you're running LunixCNC...So that's of no use any ways, haha.
 
Sorry about not doing an update sooner. This is a busy time of the year.

I redid the manual calibration. First I did all 3 axis with a digital indicator over a 0.45" distance. I had intended to do 0.5", but the plunger travel was barely that much and I kept hitting the stop. Next I used a mechanical indicator over 1". I am satisfied that the calibration is accurate and repeatable to less than 0.0005". Since backlash had already been done, I ran the test program again. The pocket was still too small. It acted like the backlash compensation wasn't working.

I then measured the backlash. I still had backlash. So I opened the 'ini' file and added 0.002" to the x-axis. Mach3 is so much easier for setup. LinuxCNC gets its operating parameters from an 'ini' file. While a script populates some of the possible commands, backlash is not one of them. The backlash must be manually entered with a text editor and the program re-started. The command for joint 0 (x-axis) is:
backlash = 0.002​
I rechecked the backlash and it was slightly less than 0.002. So I made the backlash value 0.006 and it was still 0.002". After muttering to myself for a while I realized what the problem was. See the command above? Well it is wrong. It should be:
BACKLASH = 0.002​
I have written operating programs for real machines doing real testing on real ICs. I would never, ever have made a machines ini file case sensitive. Some day someone would inevitably do exactly what I did.

Anyway -- I measured my 1/4" end mill and it measures 0.0002 over size on 2 calipers and a micrometer.

I still need to find time to get out to the shop to run the test program.
 
Last edited:
Back
Top