Kb58's granite CNC router build

[kb58]

Eight hours on a big mill resulted in a more-or-less leveled gantry. It not being my machine, I chose not to correct its tram - meaning that the spindle-to-table angle was not 90 degrees. For some work that's not a big deal, but with a large diameter cutter, the error was rather huge, with one side contacting the work, and the other side about 2mm up in the air! The combination of my previous mess, the tram error, and the low cutter speed, resulting in a less than optimum finish. By keeping the cutter on the center line of the linear rails, however, the error ends up as only a slight symmetric dish, so the linear rail will still sit level. As huge as the mill bed was, it still wasn't quite long enough, necessitating moving the work several times, but thankfully, the machine was rigid enough that alignment was maintained, more or less.

 

[kb58]

This pic shows why surfacing rectangular cold-rolled steel tubing isn't really an option - it's not all that "rectangular." With the gantry now surfaced, the rest of the holes were drilled and tapped, then it was painted. This didn't turn out so great because cutting oil got under the raised stock during machining and kept leaking out during paint. I used acetone to get rid of as much as possible, but I give the paint a "C" rating and don't expect it to last. The right way is to either sandblast it or have it powder coated, but if that's going to be done, there's a bunch of pieces that may as well be done at the same time due to set-up charges. Holding off for now until the machine has proven itself.

Anyway, with that mini drama aside, the next puzzle is how to attach the gantry beam to its end plates, which in turn bolt to the carriage plates. I thought about welding plates onto the end of the gantry but didn't. There's bolting the gantry on via flanges, but where they go and how many to have, that'll take some more staring and pondering. This relates somewhat to my issue with over-constrained assemblies in CNC machines, and such is the case here. It's a gantry design, which means there's a servo and ball screw at each end, moving in-step. The question becomes:
1. Do you build a rigid gantry and end supports, where if one servo gets out of step, the whole affair binds up or back-drives, causing all sorts of stress in the ball screw assemblies?
or
2. Do you let the gantry "float", free to pivot at each end, and leave alignment solely up to the two servos? If one gets out of step, the gantry pivots slightly instead of binding up. Granted, alignment goes off, but it would also have gone off with a rigid setup above.

In both cases above, getting "out of step" is mentioned. These stepper/servos have active feedback, so they move to where they're told, know where they are, and report any inability to move as commanded, so theoretically, if one gets out of step, the entire machine shuts down to prevent binding. Regardless, this will require some more thought.

 

[kb58]

Sat the gantry in place - the thing is growing! Took a few measurements to confirm, then cut the gantry end plates. Usually I leave the aluminum with the end mill cuts as character. For the end plates, I thought they'd look better properly finished, and used the surfacing cutter to skim them smooth. They turned out pretty nice. Both still need ports cut into them for the ball screw shaft and motor adapter.

Regarding how to attach the gantry beam, the more I think about it, the more it seems like welded-in plates in the ends, which are then threaded, are most logical.

 

[kb58]

Welded in the gantry mounts, drilled end plate holes to match, and it's in - sort of. On the not too distant horizon is the iceberg-infested sea known as "Alignment Nightmare Ocean." Piotre, whose design I used as guidance, spent days checking/adjusting/rechecking/readjusting nearly everything. and it about drove him nuts. Of course, he was aiming at micro levels of precision - I'd be happy with 10 microns (0.0004"). There aren't many situations that I'd need better than that. We'll see how close it comes to that lofty goal. Right now, things are just bolted together to create the overall assembly, then taken partially apart and rebuilt with precision.

While things were apart, everything that moves with or is part the gantry was weighed: 122 kg (270 lbs). By the time it's done, the entire machine will probably be near 1,000 lbs.

 

[kb58]

Installed the X rails and ball screw, only to remove the ball screw again because it would be in the way during alignment. It took several hours to learn how a rail acts as it's bolted down - they're kinda ornery. Tighten one bolt to set alignment just right, but tighten the next bolt and it changes everything slightly, so there's a lot of back and forth. For a while I didn't think 0.0004" (10 micron) was doable but by the end of the day, half the rail is dialed in to 0.0002", or 5 micron. Of course tomorrow is another day, and aligning the other half of the rail will no doubt mess with what's already been accomplished, but so it goes. Oh, and because this is going together for the first time, I left out some bolts, knowing things would be coming back apart. Unfortunately, I forgot to install all the bolts holding down the plates that the gantry sits on. We'll see how much things shift when they're all installed.

 

[kb58]

Toward the end of the day I had enough of adjusting rails in what seemed like circles, so for something more upbeat, most of gantry assembly was bolted up just to see what it looks like. It's a bit pretentious since all the rails aren't yet aligned, but it's still nice to see it coming together. All the parts are on-hand for the control box, so that'll be a separate project... when I'm tried of endless aligning.

 

[kb58]

Worked on the control box. Tomorrow the bottom holes will be cut. I'm following Clough42's layout and in his system, he has two sets of cables, one running from the internal boards to bulkhead connectors, and another set running from the bulkhead to the servos and so on. That means that every signal passes through four connections. Since I won't be frequently disconnecting the control box, having one cable run through a port instead of a bulkhead fitting, saves two connectors and to me, the simplicity is worth the small overhead of opening the control box to disconnect it.

After that, the X servo-to-ball screw assembly will be modified. It's apparently flexing the coupler somewhat (it squeaks) and its assembly is more complicated than it needs to be. It's one of those things that's already bugging me about fixing it. If there's time after that, wiring of the control box will be started.

 

[kb58]

Marked out, cut, and installed the items on what will be the bottom of the control box. With that messy and noisy job done, attention turned to the wiring. I have the AC power-side done and even (carefully) plugged it in. No sparks or flames, so there's that

 

[kb58]

Though looking much the same as last time, good progress was made wiring the control box. It'll probably be installed into the enclosure tomorrow or the next day.

In other news, I ordered a set of 0.0001" collets. What was received says 0.0002" right on the box. I've contacted the seller and hope that this resolves in a straightforward manner. My concern is that they might claim "oh don't worry, they're the high precision parts, we just share boxes." How could I tell what was received, as measuring to that level could be challenging.

 

[kb58]

Regarding the collet set, it went about as expected. Paraphrasing their reply: "In our description, we state that they're 0.0001 - 0.0002, so you have the correct parts." Well, sort of, because the parts are advertised as 0.0001.... period. The wider spec is only mentioned in the full description further down the page, so it's misleading. They'll probably be fine for my needs. Probably.

On a somewhat related note, I measured how much the spindle deflects under side force: around 60 microns off-axis. There are a bunch of variables though: I don't know how hard I was pushing, maybe 5 kg. I have no idea how much force there'll be on the cutter during actual milling, and that varies with cutter sharpness, speed, depth of cut, material, etc, etc. I'm not going to worry about it, but couldn't resist measuring it anyway.

The control box is basically done, at least to the point that it can go into the enclosure. There'll be additional changes but those can be made in-place. I recently realized is that the VFD/spindle circuitry probably requires a separate box, and in a way becomes a separate project. I'm reading on various CNC forums about how people build these systems, and some of them recommend a contactor that's controlled by the emergency stop switch ("EStop"). Their thinking is that in an emergency, the EStop should kill power not only to the spindle, but also the servo power supplies. At first this made sense, but after thinking it over, it seems like there's a middle ground. That is, instead of removing spindle power, just stop it. For the servos, same thing; instead of removing power, just stop them. Still thinking it over.