Indeed, you are very, very right, Doug.
Even more than I thought. So far I didn't thread lock or pin anything on the carriage, not even the studs as instructed in the manual, perhaps for several reasons:
- Curiousity: just to see what would happen and what would come loose first.
- Maintainability: I just wanted to be able to take things apart again and fix/replace parts withouth having to go to the oven.
- Lazyness: I could have drilled at least the component holes for pins, but I didn't.
Anyways, all that caught up very quickly with me, so here is the progress so far.
I cut and screwed down the t-rail sections across the entire length of the machine:
WP_20131018_004 by mkloberg, on Flickr
In terms of chipload and cutting forces, MDF seems to be easy to work with, so I ended up cutting all the tiles for the spoilboard for the machine, on the machine itself!
While doing that, I found out that there seems to be a backlash on the x-axis of about 0.008". I don't know if that makes sense yet, but I have a theory where that comes from:
The teeth in the x pulley are a tiny bit wider than the XL belt teeth really are.
I did several tests involving the machining camera coming from both directions and when working with backlash feature in Mach3, the settings that worked the sweetest were "Backlash distance in units (X): 0.008, Backlash Speed of Max: 100%".
I'm curious to see what everyone's opinion is on that, because most folks don't use the backlash feature of Mach3 I hear. For me so far, it completely eliminated the problem.
Anyways, after these tests, I felt confident enough to load a sheet of MDF and cut the remaining spoilboard tiles.
That went really well, including the profile, the rabbets and the pocket holes on the other side, all cut by the machine itself.
WP_20131020_001 by mkloberg, on Flickr
The only thing left at this point was to distribute the spoilboard tiles along the base and drill the holes in the centers of the pocket holes to mount the tiles to the spoilboard base plate.
I did that with the machine itself also, using two different toolpaths. One to drill a core hole through each tile and into the base. Then a second path to open up the hole in each tile to allow the screw to go through clear.
Getting a little bit worried that I might pull up one of these tiles while drilling, I made a set of holddown bars with a hole at each end that would fit into the outer t-tracks.
While cutting these, I ended up calling them 'monkey bars' because they are a shortcut to clamp something like this properly for drilling.
Having the monkey bars holding down all tiles in the bed, the machine drilled the core holes and the clearance holes in two passes.
WP_20131020_005 by mkloberg, on Flickr
This went great, and I was able to bolt all the tiles down without a problem. Throughout the three sections, everything seemed to be reasonably flat, adhering to the epoxy base level.
Finally, a good looking machining surface to work with / to clamp stuff down on, yay! :-)
Eager to try out something more complicated, I found an example on the net that was a small plate with a chinese dragon and a weaved border around it, so I decided to make that a test cut.
Roughing with a .25 Endmill and finish with a 3mm ballnose.
WP_20131021_011 by mkloberg, on Flickr
Keep in mind that this is still with fully extended Z-axis, very high up. Having the spoilboard assembly in there now brought the base up a little higher, but there's still a lot of z-flex lever in play.
During the cut, especially on the right side, the machine started to stutter, indicating that there is something more than flex going on at the beginning stages of a problem I would find out about pretty soon...
For now, this very first piece off the CNC was more than I expected - with the Z still hanging that far down.
WP_20131021_016 by mkloberg, on Flickr
The next day, my GF asked me about doing some curved flat boards for a trial product they had in mind. In woodworking, this is when you cut shapes out of 3/4 boards and assemble+glue them sideways.
This little project was about 6 of those shapes that had holes in them to make a small wavy board for seeing if the machine could do it.
While making the toolpaths, I kept thinking about the still hovering task to pin and threadlock the parts on the gantry, because things could come apart!
And they did... I guess these wavy shapes and the addition of the hardness of real wood instead of just MDF triggered a chain reaction in the carriage for things to give away.
Half way throught the second part, everything started to shake so violently that I hit the e-stop button to prevent further damage:
WP_20131023_004 by mkloberg, on Flickr
With the endmill still stuck in the board, I started to clean up and get ready to take the gantry apart for pinning and thread locking (long overdue).
This is what that first part looked like by the way, it's giving some clues on what gave a way first and what went on during the cut.
WP_20131023_005 by mkloberg, on Flickr
After looking at this and checking all the bearings for tightness, I found that the lower front bearings on the y-axis were completely loose.
Everyting else was still tight.
Looking at the front bearing blocks (part #14) revealed that these two blocks have shifted under the load.
I'll have to pin these in as a first fix, but there isn't much real estate to drive the pins through on this part.
Next was a complete strip-down of the carriage, fortunately I was able to take it off the y without taking that apart too, utilizing the hole in the base - taking out the z-axis first.
All that took less than 20 minutes, yay :-)
WP_20131024_001 by mkloberg, on Flickr
Looking at the problem some more in Sketchup and where to place the pin holes into part #14 to prevent that from happening, I found some beefy places for the pins:
WP_20131024_002 by mkloberg, on Flickr
Everything is in pieces right now, probably a good time to thread lock the studs I didn't do yet too and examine some more spots where pins can help for stability.
WP_20131024_005 by mkloberg, on Flickr
--
Mac