Kb58's granite CNC router build

[kb58]

The 19 mm aluminum tool plate is here. By the time the drill program was about to be run, I set something wrong and it was just "cutting air" way above the part. That's the scary thing about this stuff, kinda like flying a plane. It's fun and all, but one tiny screw-up, one overlooked factor, or a brief lack of attention, can cause things to go very wrong in a hurry. Granted that there's a feed override control - and an Emergency Stop, but I've found when I start pushing to get something done toward the end of the day is when I make mistakes, so everything was shut down and I'll try again tomorrow.

Once it's bolted to the granite, next is deciding what bolt pattern to use for attaching work to this plate. I saw one setup where they used 1/2-13 thread, which seemed huge, until seeing that fit hold-down clamps from a standard milling machine set - that's clever. Another video showed using shoulder bolts, specifically the type with no head (I'm calling them "shoulder pins"). The plate might be threaded 2/3 of the way through, and the upper third reamed for a snug fit on the shoulder. This would provide very accurate reference points to slide parts up against. A combination of the two patterns could be even better. The more I think about this the more I like it.

Regarding the above holes: blind or through-hole? Threaded through holes are easier to deal with during fabrication... except that once done, if any coolant or lubricant is applied, it can get in the holes and sit there forever between the aluminum and granite, probably doing bad things. To coin a phrase I've been hearing lately, "this is not optimal." Blind holes, on the other hand, keep fluids and chips from getting stuck between the plate and granite, which seems like a good thing. They'll both still pile up at the bottom of the holes, but at least there's a chance they can be removed with compressed air or a vacuum. Just have to decide on the size of shoulder bolts to use. Not sure there's any advantage to using small one, so maybe 12 mm shoulder... pins, but this is subject to change at any time.

 

[kb58]

Most of the holes are drilled. It was nerve-racking watching the drill approach the work with the granite right there, but it went fine. The last pic shows the plate shifted back so the spindle can reach the front row of holes. The same will be done in the other direction to reach the back row. The side holes... I'll probably do by hand because they're non-critical and awkward to have the router do them. Keep in mind I'm at the interim stage of having no proper way to hold down this panel while it's being drilled. Once these holes are done, the panel can be bolted to the granite, then the higher-precision holes machined without risk of the panel shifting.

Threaded inserts are being used, the type that get hammered in and expand to anchor in the hole. For fun I tried pulling one out with a claw hammer to see how well anchored it was. It was rather shocking how easily it came out, so epoxy will be added insurance. Maybe I didn't install it correctly (not wanting to beat on the granite base and risk cracking it) but regardless, that's not acceptable.

The next project after the fixture plate will be a dust/chip shoe. Because of the higher (than a mill) spindle speed, it was impressive how far the chips were flying, and this is just drilling. Once side cutting starts, it's going to be all over the garage. The shop vac really helped, catching perhaps 98% of the chips, so that's a strong clue about what works. At this time I don't want to do a full enclosure.

After that will be a mist coolant setup. I can't justify flood cooling for what I do - though it can always be added later. The nice thing about the mist coolant is that it actually sprays very little coolant, delivered right to the cutting edge through a pressurized tube. In addition to being a lubricant, it blows chips out of holes, keeps the chips from getting welded to the cutter (often causes them to break), and keeps the cutter cooler.

 

[kb58]

As an aside, I've always questioned how efficient - time-wise - a CNC machine is to make just one part, typical of what's done in a hobby workshop. "Time-wise" means from beginning to end, starting with creating the drawing, to having the finished part in-hand.

Watching it drill the holes gave me some insight. If curves or many repeated features are present, the CNC-produced part will likely be faster. As far as being more accurate, yes... but since only one part is being made, there is no room for error. On the other hand, once more than one part is produced, the CNC will win every time. Creating the initial drawing is a wash since it's needed for both an operator of a manual mill and to feed CAM software. Time spent generating the G code and speed and feed settings is a burden only on the CNC side. So whether making one part in a garage setting via CNC or by hand depends on the number of curves, needed precision, number of holes, required consistency, and repeated features. A whole bunch of variables that make a straight answer nearly impossible! Adding in the enjoyment factor can bias the preference in either direction.

 

[kb58]

Well this kicked my butt. Remember how I said how a diamond core drill can drift due to not having a point? Well it sure does, quite a bit as it turns out. The inner rows and columns aren't an issue because the spindle forces it to stay put. The outside rows and columns... not so much. They were drilled by hand, and just how messed up they were didn't become apparent until the fixture plate was put into position. Some of the holes were off by a little, and some by a lot, up to as much as 5 mm! Having no choice, out came the file, and hours later, it's "good enough." Not pretty, but functional. Partially because of this, the important holes will only be placed where the router can reach, which will leave open areas around the outside, but oh well. That area will be reserved for large 1/2-13 threaded holes for hold-down clamps, and they don't have to be precisely placed, so it works out.

 

[kb58]

Well that was a ton of work, but it's done. I could have had the router do more of the work, but had no way to hold down the plate - a Catch-22. Also, due to being larger than the work envelope means that the holes outside of it had to be placed, drilled, and tapped by hand. I was getting urged to single-point tap the holes. Again, that can't happen because the plate being machined sits on the granite, and is intended as the base for all work, not to be worked on itself! So, I got to tap all those holes by hand - not a lot of fun.

Oh, and I kind of outsmarted myself. I wanted the plate-to-granite holes to be evenly distributed, and ended up with a 100 x 110 mm spacing. That's all fine, but it means that the plate can't be turned 90 degrees and offset to machine the holes outside the work envelop. It also means that all the various threaded holes are not at a 45 degree angle, though I can't think of a reason why that matters, so proceeded. We'll see how smart that was.

So what's next? Probably an MDF spoil board for now. Later, an aluminum fixture place is likely. I think the first real project will be a dust shoe, then after that, a cutter mist system.

 

[kb58]

There's been some drama behind the scenes after finding that the small 14" touchscreen I'm using apparently isn't well-suited for this application. Due to its software (or the controller software - which the mfg denies), there seems like there is a race condition where if you briefly tap an icon, the touchscreen sees the tap and highlights it - BUT, will only perform the function about 2/3 of the time. You have to deliberately hold your finger down for a minimum of maybe 1/3 of a second for it to always work. The mfg said, "that's because you have a sucky capacitive touchscreen; you need a single-point resistive touchscreen." I said, "and how would I have known to buy that, it's not listed as required anywhere on your site." "Oh we've discussed it (in threads impossible to find)."

Anyway, the project: The touchscreen that's on the way is a larger 17" unit, so where it lives needs to be dealt with now instead of leaving it sitting around risking damage, along with the keyboard and very small PC. I decided I want a swivel table at the front left corner of the tool cabinet, one that adjusts vertically so I can work on it sitting or standing. It'll swivel around to the left side to stay out of the way, or so that I can run the router outside the line of fire (of chips). Leftover metal from previous projects will get a second shot at becoming something after not quite making it into Midlana or Kimini.

But yes, a dust shoe is pretty much Project #1.

 

[Eddyde]

Wow what an incredible build! Thanks for posting such detailed coverage of your journey. Tons of great info, I'm sure it will be helpful to many.

 

[kb58]

More progress on the keyboard/monitor swivel table, and pretty happy with how it's turning out. The metal frame will be painted black, but unsure how to finish the wood. I like working with wood except for the sawdust and don't like dealing with paint/stain/whatever - no fun. I think I'll see if I can use some sort of clear finish and leave it at that. The wood surface isn't yet trimmed but will extend out enough to offer hand support while typing. The last pic shows where the tiny PC will live, out of the way but still accessible.

I almost added a cup holder but thought that might be a bit much

 

[kb58]

Related to spoil board and fixture plate fabrication is gantry alignment. At first, I thought I could use the sides of a small granite surface plate as a 90-degree reference. Turns out that - no you can't. Surface plates, at least ones that I can afford, are only guaranteed to have a flat top. It may not even be level, but it will be flat. Worse, there is no guaranty that the sides will be 90-deg to each other, so it makes for a poor reference. The last attempt was to mark a large triangle with the spindle and measuring the hypotenuse. Again, it seems rather "meh" for accuracy.

This is brought up because if an aluminum fixture plate is fabricated with the gantry not square, nothing made on it ever will be. Searching ebay turned up a few precise granite tri-squares, but even used, they're very expensive, never mind shipping. Also, since they're a single-use reference tool, I couldn't bring myself to spend the money. Instead, I bought a smaller more affordable one, 10 x 6 x 1", which will have to be "good enough." An indicator off that should get it pretty close.

 

[kb58]

Started on the first real job for the router - an MDF spoil board. I'm only using drills for now to avoid side-loading the work with router bits until proper hold-downs are available. Even though this is the first job, it's already proving to be a lesson in how there's a million ways to do things. To sink the flat head screws into the MDF, a big countersink could be used, or a smaller one used in a circular pocket operation, or first drill the holes to depth so that the countersink has less work to do, and on and on. I started by drilling 1/2" holes for the bolts, then the idea was to use a countersink, but since the flathead bolts are nearly 1" in diameter, it's a lot of material to remove with a cutter not intended for side-cutting. I'm going to go back and use a 1" drill to remove material down to where the countersink starts, minimizing the amount of work it has to do. Given that it's a 3/4" countersink, it doesn't have to remove too much additional material.