[RomanLini]

What I have done is clamp a flat glass plate to the front of the gantry (where the bearing rails will go) so the glass plate is vertical and comes down to touch the table.

Then your carpenter square stood up will test the vertical square of the gantry front plate. And the carpenter square laid flat on the table will test the XY square of gantry to table.

[JerryBurks]

Thanks for the ideas. I will play with that when I get to the Y-rails.

I had a panic attack today when I slid off the gantry to machine hard-to-get-to areas and tried to put it back on. A bit careless and the linear bearings spilled 3 of the small balls

15 minutes on my knees and I found them all, pushed them back into their channels, made sure not more were missing and apparently it all works again. Whew, I think I will not take the gantry off again if I can avoid it. It is getting a bit heavy anyway, even for 2 people to do some precision fitting.

[ger21]

Get some wooden dowels the size of the shafts and slide them into the bearings as you slide them off the shafts.

[JerryBurks]

I started playing with the lead screws. I can not machine proper bearing seats and threads at home and did not find a local shop that was interested in such a one-off job. So I cheated, drilling out precision 3/4" shoulder bolts to 15mm on the lathe (I can do that at least after annealing) and affixing them to the lead screws with cross dowels (and J-B Weld ). Looks a little blacksmithy but the runout is actually quite good. For the bearings I picked 3/4" tapered roller bearings (a.k.a. trailer bearings) and I hope with proper pre-load using belleville spring washers it will be accurate enough. The plan is to support the bearings only axially with the aluminum plate and pretty much let them float radially (except for the pre-load friction).

[RomanLini]

Hi again JerryBurks. Is that the opposite end of the screw to your motors? I thought you were going to use the motors as the axial (thrust) bearings?

If you fix the bearings axially at the far end of the screw then when the screw expands in length from heat it will bow and give you a heap of whip and other issues.

Or maybe I misunderstood your description and these bearings support the screw radially but NOT axially? ie; the screw can move lengthways through the support?

[JerryBurks]

Hi again JerryBurks. Is that the opposite end of the screw to your motors? I thought you were going to use the motors as the axial (thrust) bearings?

Yes this is the opposite end of the motors and it is axially fixed. I had originally planned to use the motor of the Z-axis only as a thrust bearing but with all the concerns in the feedback here I chickened out and will now have a dedicated set of thrust bearing for all lead screws.

If you fix the bearings axially at the far end of the screw then when the screw expands in length from heat it will bow and give you a heap of whip and other issues.

I am still planning to use all motors as support bearing (no thrust) for the other side, that is if I manage to fabricate or buy rigid couplings with excellent run-out. I understand that will require the motors to have some small axial leeway to avoid the heat expansion bowing (0.2mm should be enough) while being firmly mounted radially. I expect the flex of the large motor mounting plates to take care of this screw expansion but it will probably need some experimentation.

[RomanLini]

Ahah! Thanks for clarifying.

[JerryBurks]

Another week passed....I believe I sunk already some 200 hours into this project (don't ask how much money). But it is FUN!

Anyway, here the latest progress:

Completed the gantry torsion box and varnished the gantry. Bolted the lower Y-rail in place and tomorrow I will attach the upper rail.


Also, I fabricated new motor couplings with clamp screws all on one side to minimize runout (thanks for the tip, Roman). I did a quick test with a torque wrench and they started slipping at 30 ft-lbs (the steppers will be set to 5 ft-lbs max). I think not too bad for a wood lathe job.

[RomanLini]

Excellent work on the couplings! I really like the way you are methodical and test and measure as you go along. Lots of people doing builds should follow your lead with that methodology I think.

Just a couple of observations, the peak roational torque of the stepper shaft can be many times the max torque of the stepper motor. This is because of the situation where the stepper armature mass and/or screw mass accelerates very rapidly (even if just for a fraction of a degree rotation) then when the leadnut backlash is taken up it "whacks" hard into the load (the gantry mass). A stepper can generate enormous acceleration over a very small rotation and that can give very high rotational peak force there on the shaft.

It's fairly trivial to drill a couple more holes and tap threads in your couplings and add a grub screw onto the flat on the stepper shaft and leadscrew. Those shaft flats are there for a reason!

How is the parallel alignment of your two gantry bearing rails? Have you been able to measure the rails are an exact distance apart? I'm curious how you get the rails parallel.

[JerryBurks]

Excellent work on the couplings! I really like the way you are methodical and test and measure as you go along. Lots of people doing builds should follow your lead with that methodology I think.

Thanks...I am an office guy for a living but that engineering education has to pay off sometimes I believe I spent about as much time measuring and building measuring jigs as I did cutting parts and assembling.

....It's fairly trivial to drill a couple more holes and tap threads in your couplings and add a grub screw onto the flat on the stepper shaft and leadscrew. Those shaft flats are there for a reason! ...

I have to watch that. So far I tried to stay away from setscrews (or grub screws) at least on the lead screw side of the coupling because I don't want to mess up the ends of my pretty C5 grade precision lead screws where I may have to take off the ballnut sometimes (remember the fixed bearing is on the other end). If really necessary I may use setscrews with brass tips like McMaster is selling.

How is the parallel alignment of your two gantry bearing rails? Have you been able to measure the rails are an exact distance apart? I'm curious how you get the rails parallel.

I am quite satisfied with the distance parallelity, I got it to better than 2/1000", that is within the uncertainty of my dial gage jig. I measured using a 3" wide strip of plywood that I can rest against the sides of both rails with another square wood strip attached that is riding on top of the lower rail. The dial indicator is bolted to the plywood strip and measures against the bottom side of the upper rail (you can see this contraption in the picture on the table) and I can watch the indicator change while sliding along the rails.

I am not entirely happy with the twist. I hope it is good enough but the top rail ended about 10/1000" out of plane across the length. Not sure how I screwed that up but if I end up with tramming problems due to this twist I will have to go back and move the top rail.

That is actually how I got precision alignment: I built the gantry from bottom to top starting at the X-rails, measuring parallel and perpendicular alignment over and over with every new piece being attached and sanding or planing until it fits exactly. Basically one side of the gantry box every day. Since everything is glued and screwed together there is no room for error. Once the glue sets I can only start over. I made especially sure the rail mounting surfaces were already parallel, perpendicular to the table and straight.

The rails were then set in epoxy (actually J-B Weld because it does not easily drip or run) for leveling. The fine tuning of the parallel adjustment was done with moderate force by 4 or 5 clamps. Once the epoxy had cured, I drilled the bolt holes through the existing holes in the rails, countersunk and fastened the bolts.

I noticed building such a rigid structure comes at the price of serious accuracy requirements. A flexible machine is much more forgiving to assembly errors.

[louieatienza]

Thanks...I am an office guy for a living but that engineering education has to pay off sometimes I believe I spent about as much time measuring and building measuring jigs as I did cutting parts and assembling.


I have to watch that. So far I tried to stay away from setscrews (or grub screws) at least on the lead screw side of the coupling because I don't want to mess up the ends of my pretty C5 grade precision lead screws where I may have to take off the ballnut sometimes (remember the fixed bearing is on the other end). If really necessary I may use setscrews with brass tips like McMaster is selling.



I am quite satisfied with the distance parallelity, I got it to better than 2/1000", that is within the uncertainty of my dial gage jig. I measured using a 3" wide strip of plywood that I can rest against the sides of both rails with another square wood strip attached that is riding on top of the lower rail. The dial indicator is bolted to the plywood strip and measures against the bottom side of the upper rail (you can see this contraption in the picture on the table) and I can watch the indicator change while sliding along the rails.

I am not entirely happy with the twist. I hope it is good enough but the top rail ended about 10/1000" out of plane across the length. Not sure how I screwed that up but if I end up with tramming problems due to this twist I will have to go back and move the top rail.

That is actually how I got precision alignment: I built the gantry from bottom to top starting at the X-rails, measuring parallel and perpendicular alignment over and over with every new piece being attached and sanding or planing until it fits exactly. Basically one side of the gantry box every day. Since everything is glued and screwed together there is no room for error. Once the glue sets I can only start over. I made especially sure the rail mounting surfaces were already parallel, perpendicular to the table and straight.

The rails were then set in epoxy (actually J-B Weld because it does not easily drip or run) for leveling. The fine tuning of the parallel adjustment was done with moderate force by 4 or 5 clamps. Once the epoxy had cured, I drilled the bolt holes through the existing holes in the rails, countersunk and fastened the bolts.

I noticed building such a rigid structure comes at the price of serious accuracy requirements. A flexible machine is much more forgiving to assembly errors.

I have the same ballscrew assemblies as you, but decided to have the ends machined down. In hindsight, this is the more cost-efficient method; so I'm very interested to see how having the "fixed" end on the opposite side works...

.001" of "twist" is pretty darn good...

[JerryBurks]

I have the same ballscrew assemblies as you, but decided to have the ends machined down. In hindsight, this is the more cost-efficient method; so I'm very interested to see how having the "fixed" end on the opposite side works.....

I hope this picture explains the concept. Since I don't have real bearing blocks, the bearing pair is pretty much free to self-center when the Y-plate is moved close. They have about 1/10" latitude to move to either side. If the friction turns out not to be sufficient to hold them in place while operating on the other side, I can put a bead of silicone around the outer race. I know, also a bit unconventional and I don't expect all my ideas to work perfectly

[ger21]

.001" of "twist" is pretty darn good...

10/1000 = .01"

[louieatienza]

10/1000 = .01"

Yes it does... reading too fast again, could havev sworn it said 1/1000!

[JerryBurks]

Yes it does... reading too fast again, could havev sworn it said 1/1000!

yes it is annoying. If I would tram the spindle in the middle and have 5/1000 error to the left and -5/1000 to the right over 12" distance between the rails that should cause a worst case path step of 0.0001" for a 1/4" end mill. I suppose for a ballnose it does not really matter.

But I am not a machinist and have no idea if that would be visible or if that is still pretty good. Guess I will find out in a couple of weeks.

[RomanLini]

Re the grubscrews on the shafts, if you don't have a flat on the shaft that can be a problem as the screws bite in a bit and will make defects that make it hard to take the couplings off. With a flat ground on the shaft the place the screw bites in is always inside the circumference so it doesn't interfere with the shaft roundness or smoothness.

Thanks for the info on how you lined up the 2 shafts. Normally I would have used one shaft for alignment and the other one adjustable a bit, even if just the small amount of play in the mounting holes allows 0.5mm (+/-0.25mm) adjustment. Epoxying the shafts in is unusual.

When you say 5/1000 error is that 5 thou ie 0.005"? That's generally 5 divisions on the dial gauge (assuming a typical dial gauge of 0.001" divisions)?

[JerryBurks]

.... Epoxying the shafts in is unusual....

Well, this is my first home-built CNC machine, so I am just trying some stuff. Actually, while the rails are set in epoxy to create a conforming mounting surface, I had put some silicone grease on so that the epoxy does not stick and I can remove them if necessary (I hope....)

When you say 5/1000 error is that 5 thou ie 0.005"? That's generally 5 divisions on the dial gauge (assuming a typical dial gauge of 0.001" divisions)?

Yes, that is what I meant. I will probably map the x-y position error with a microscope across the table surface when the machine is complete. Then I could program a separate postprocessor for the G-Code to transform the coordinates and eliminate the repeatable errors. One of the projects I am planning to use this machine for is plastic gears that need to be pretty precise.

[RomanLini]

Nice diagram! It's looking real sturdy.

Re the z axis, have you considered lowering the bottom 4 bearing trucks down as low as possible, to near the bottom of the structure? It looks like you could make the 3 lower leadscrew bearing plates smaller and then extend the 2 side plates (that mount the trucks) fully down to the bottom. I think that would improve Z axis geometry a lot without any real costs.

[JerryBurks]

..........Re the z axis, have you considered lowering the bottom 4 bearing trucks down as low as possible, to near the bottom of the structure? It looks like you could make the 3 lower leadscrew bearing plates smaller and then extend the 2 side plates (that mount the trucks) fully down to the bottom. I think that would improve Z axis geometry a lot without any real costs.

I played with this layout for hours on the CAD. Not sure if I got it right but the reason the bearing blocks are rather high is that I want to utilize the 8" gantry clearance fully (I am planning to machine some 3"-4" tall pieces). That means to allow for the spindle and bit to protrude a few inches below the Z-Plate, the lower edge of the Z-Plate (including the rails) must be able to pull up well above the gantry lower edge. But maybe there is another clever way of doing this?

Anyway, even in the current configuration the bearings blocks are spaced a generous 10" apart vertically (since the rails are 18.5" long) and I hope this will do. I guess this is also more than the usual spacing for routers made from aluminum extrusion.

[JerryBurks]

Next step....Y-plate attached with linear bearing blocks.

One thing that worries me is the friction/stiction in the linear bearings. It takes 3.5 pounds force to get the x-axis moving (4 linear bearings) and 4.5 pounds to move the y-axis (6 linear bearings). I am wondering if that is because of stresses in the gantry or if that is just a matter of the wiper sealing lips of the bearings rubbing on the shafts.