Kb58's granite CNC router build

So... the dust shoe. The wish list includes:
1. Transparent top, to see what's going on.
2. Independent of spindle position
3. Have two attachments, a close-clearance air gap (no brush) for flat stock, and a brush for "lumpy stuff"
4. Have the vacuum hose attach behind the gantry for better visibility
5. Easy tool changes
6. Connects to a 2.5" hose and shop vac
7. LED light ring would be nice

I saw one design that uses a stepper motor to adjust the height. Seems like an extravagance for the coolness factor, without being significantly more useful. The rolls of tape represent a possible height for the spindle surround. The issue is that with taller cutters (think: drill bits), the Z axis has to be higher above the table. If it's too tall, it may rise above the top of the dust shoe surround, creating a big air leak. A factor contributing to this is that the spindle is currently placed so that when at cutting height, the Z carriages are centered between the X rails, minimizing deflection. A compromise is to lower the spindle just enough to increase the amount it can move without leaving the confines of the dust shoe surround.
 
Two things showed up today that I put off buying due to cost. The first is the wireless pendant that allows controlling spindle position when you're setting something up. At first it seemed like an expensive and redundant toy, but after seeing videos of them being used and how handy they are, I reluctantly ordered one. The other arrival is a granite tri-square/triangle. It was another reluctant purchase because I kept thinking how rarely I'd use it - and I still do. Thing is, there's no other precise way to determine whether two axes are square to each other. I already tried other methods and wasn't satisfied with the accuracy - looking at a metal rule and estimating what fraction of a millimeter was shown. I bought the smaller one as a balance between cost and use, and dang it, it works really well. The X and Y axis are now within 0.02 mm of square over 300 mm, and that's good enough for now. This needed to be dealt with before making any precision aluminum fixture plates, since if axes squareness is off, everything ever made with the fixtures will be off as well.
 
Well the fussy part of the dust shoe assembly is done, the bit that hangs from the Z back plate. Two bars slide up and down, and can be locked in place to hold the shoe right above the work. I reserve the right to redefine "fussy", as the shoe itself appears to be a never ending story based upon how many YT videos there are. Now need to finalize a candidate design.

1721875354586.png
 
Designing the dust shoe means using the router - and the 3D printer - to make parts. The problem is that I haven't used the 3D printer in so long that I've forgotten the workflow. Its job is printing the vacuum hose adapter that bolts to the shoe. It's tempting to skip the 3D printer and just adapt some plumbing parts to get it working, but that just perpetuates the problem. Whlle that would work, the 3D printer really needs to start earning its keep as a tool and not a project. If I can't get it working this coming week, Black Friday is right there... beckoning with sale prices. Bambulab has an extremely impressive unit that everyone says "just works", https://store.bambulab.com/collections/ ... er?skr=yes. Hmm.
 
Ugh, me being me, I tend to flail around using new stuff, get all frustrated, calm down, try again, and usually with more success. Yup, right on track. I'm using functions in Alibre I haven't used before and get stuck, a lot. If not stuck outright, it'll refuse to do what I want, usually with either a really cryptic message, or worse, nothing at all. I'm very sure I'd be whining about this regardless of CAD package, so there's that. Also, at the back of my mind is "You know, if you would have spent today actually making the dust shoe hose adapter, you'd be done by now." Yes, I would, and I still wouldn't know how to use CAD or the 3D printer, and that problem's going to keep confronting me until I push through it.
 
Regarding the 3D printer, I have my answer:

Figured out my issues with Alibre and the dust shoe top is done. As mentioned, the 3D printer needs to be useful, so as a test, the file was sent to it instead of the router, the idea being to create a prototype, test the printer's abilities, and delay dealing with the intimidating router, hah. Anyway, a common 3D printer issue is first-layer adhesion, and sure enough, the first layer of PLA developed what almost looks like patches of peeling tree bark. Varying the nozzle and bed temperature sometimes seemed to cure it, but then it would randomly return. About 2/3 of the way through the first layer, it started behaving, a problem that goes away for no obvious reason isn't good. Cancelled that print and flipped the build plate over to a fresh new surface, thinking that was it. Same problem, the first layer peeled up in small patches. Oddly, it only occurred in long runs, but never short ones - what's with that? What's especially annoying is that it didn't used to do this, though admittedly I never printed long objects.

As the print built up, it started lifting off the bed in several place, some by a little, and nearly 5 mm in one corner! I know that these are mostly settings issues, but this goes back to wanting a tool and not a project. I learned a ton building the printer, but now I want to actually use it and not keep messing with it. What settled the question was when the lifted corner caught on the extruder, broke off the fan duct and dislodged the nearly complete part from the bed. If this was just for fun, oh well, but four hours are gone instead of having something to work with.

So, I guess that's my answer, or the above is just a big excuse to buy something shiny and new...
 
So as I dismantled my 3D printer, I reflected on what was learned, the money spent, the time invested, and wondered if it was worthwhile, given that I eventually concluded that it wasn't good enough. This isn't the first time this has happened - the last time was building a rotating-drum fish pond filter, where it worked, but I ended up admitting that it wasn't good enough, cut it up, and bought a commercially-manufactured unit. In hindsight I could see what "should have been done" - but only after looking over the commercial unit. And dang it, that thing just keeps running and running, a sign that R&D is a real thing and costs money - who knew!

Distantly related, I'm selling the big telescope I built, though for different reasons. In that case, it certainly is "good enough", but I don't use it enough to justify the money tied up in it, never mind being a large cumbersome thing to move and store. Light pollution only gets worse, and most recently, a next-door knucklehead neighbor set up a security light that detects us in our backyard, plus any cars coming down the street. It's always going off, yet the neighbor says it's warning people. Worse, he adjusted it nearly horizontal, shining straight into our eyes. I haven't said anything to him, as I'm just so impressed that someone can be either so utterly clueless, or just doesn't care. Part of me wants to set up a floodlight that triggers off the light from his, shining it straight into his living room. But I know how such childish things can escalate. The place to use such a scope is in the desert, about 1.5 hours from here. That sounds good but didn't work out. It means heading out, setting it up, staying up all night, sleeping some portion of the following day before heading home. My wife wouldn't enjoy it, and our elderly dog wouldn't like the change in routine.

So I reflect on the money spent on things I've built that ended up being dismantled or sold and have mixed feelings. I learn a ton doing projects, but I do wonder sometimes if the gained knowledge is worth the money spent. I guess it comes down to defining "worth." Did I greatly enjoy building them, yes. Did I gain knowledge about new things, yes. I still have thoughts that someday I'll combine this diverse knowledge into something cool. Only time will tell.

Lastly, gloomy thoughts aside, I'm really pleased with how the router has turned out, though admittedly it's yet to run a job other that drilling some holes. While waiting on the 3D printer, it might be worth running a test cut on a piece of plywood before with the polycarbonate. Lexan is being used because there's some left over from the windows on Kimini. It's really tough stuff, flexible yet highly resistant to breaking - I folded a piece over on itself and it didn't shatter or even crack. That's good in case a cutter breaks at 20,000 rpm and shoots out sideways.
 
Decided it was time for the router to make something for real - a plywood version of what will eventually be Lexan, but the job failed on two counts:
1. I apparently don't have the CAM software set up right; there's a groove around the outside of the part but it was almost completely ignored during the job.

2. The big problem happened about 2/3 of the way through the job, when one of the two gantry servos stopped driving, twisting the gantry out of alignment. The two servos run in parallel, with identical signals going to both. Because of this, it seems likely it's an intermittent cable or defective servo. The router was rebooted and the gantry manually commanded to move back and forth. It moved fine but after a couple minutes, it happened again, and this time I saw which servo stopped. At this point, I still didn't know whether the problem is the servo or a cable, though after rebooting the router makes me think that a cable issue is less likely. One power supply powers both, so it's not suspect, though it could still be a power cable issue. If it is a power issue though, I'd expect the servo to reboot itself once power returned, but it didn't.

The servos were swapped side-to-side to help isolate the issue - and of course now it won't fail. That's very frustrating because I can't trust the machine until this is resolved. I think I'll rerun the job, but with no cutter and leaving the spindle turned off. The problem is that even if it fails again, it doesn't tell me anything unless I know which direction the gantry was moving. Without knowing that, it could be either one. I'm afraid I may have to just sit there through the entire run and watch it. Good times.

Oh, and the 3D printer is due to arrive tomorrow, so at least there's that.

In the pic below, the cutter is shown driven into the work short of where it was supposed to be. The machine properly shut down, preventing some crazy new cuts and potentially breaking the cutter.

1721876150367.png
 
Spent all day working on it and the "good" news is that it's fairly reproducible . The bad news is that every time it does, it really does a number on the poor gantry, but there's no point in aligning it until this is resolved. It "seems" to be the servo originally installed on the left hand side being bad - it or its cables. When disconnecting the gantry yesterday from the servos, it was a bear disconnecting the right-hand side, and it jumped a good half inch(!) toward the rear when it finally sprang free. Because the gantry had been driving that direction, it suggests that the left-hand servo was still working. As mentioned, the servos were switched left-to-right, and when it's failing now, it appears that the left side is failing. I'll have to abuse it some more this weekend to confirm, so we'll see.

Oh, and the Bambulab 3D printer arrived. I ran the obligatory "benchy" model boat and it turned out better than my printer ever produced... sigh I used the same questionable PLA filament just as an experiment and it printed fine. I'm not going to bother test printing the dust shoe part because it has its own issues that need resolving first. In the case of the dust shoe though, I think it's important that it's transparent so the cutter can be monitors, and the contents of the shoe monitored for blockages.
 
Mechanically disconnected one servo and let the suspect side drive the entire gantry assembly. Moving at 4000 mm/min, after about 10 minutes of back and forth, I started noticing it sounding different toward the back of its travel... that's not good. It eventually shut down with the same error code as before. Curious, I touched the ball screw and was surprised that it was warm, and the ball nut was even warmer... so, mechanical binding... that's not good.

The servo manufacture's app was used to monitor torque, and the gantry motion repeated, and sure enough, torque increased as the gantry moved aft. Within a few minutes though, it started increasing even more, eventually getting so high that it shut down again. Pretty sure it's due to heating of the ball screw shaft and nut, causing binding. So the question is, why. Misalignment, obviously, but where? I checked horizontal parallelism and it's good, and ran out of time to check vertical parallelism, that'll happen tomorrow. It's a sure thing that it's off in some direction, because backing off the bolts between the ball nut and gantry completely eliminates the increase in torque. So for the moment, I no longer suspect a servo issue, but we shall see where the tea leaves lead.
 
You must log in or register to reply here.
Back
Top