Author Archives: tumblebeer

The man in black fled across the desert, and the gunslinger followed.

You know, I was trying to find a good quote or reference to use as a title for this blog post. Since this post is supposed to be the “I’m finished!”-post, I needed something to signify that. I thought about going the Borat route, Great Success!, or perhaps a Douglas Adams quote (you know which one I mean). Both of those would relay the success of the project, and the fact that it’s finished.

But that’s not how these things work. Anyone that’s ever truly dedicated themselves to build something knows that.

Instead I went for this one. I’m not going to explain it, since that would ruin one of the best book series ever written, but should you have read them all, you’d know. You’d understand.


Anyways, here’s the video I’ve been promising! It’s a bit of a mashup, showing a quick fly-by of the machine, some 3d ballnose profiling and some straight up end mill slotting action. Pure porno.When you’re done watching, go read all the blog posts if you haven’t already. They’re fun. Everyone that’s read them all will agree with you. (Only I’ve done it that I know of so far, and you can’t disagree with me until you’re at least nr 2).

You can find all of the blog posts linked here.


That’s with two D’s for a double dose of my pimping.

Lately I’ve been upgrading the CNC, something also referred to as “pimping”. As of this moment, the machine is actually… wait for it… Finished! That’s right, by adding the things I’ve called “final pieces of the pussle”, I’ve done everything I initially set out to build. I’m going to do a small video where I showcase all of the features, and perhaps even a milling job. I’ve already milled quite a lot of aluminium, with great results, but nothing that I’ve managed to film. But for now, I’m just going to try and describe the things I’ve built since the last post.

I’ll try to remember all the things I’ve built and/or added to the machine, but it’s been quite a few of them. Let’s see, I’ve built:

1. A dust boot.

2. Another dust boot.

3. A third dust boot. Not really, but I’ve modeled one that I’m going to build eventually.

4. Started working on a vacuum table.

5. Added limit and homing switches.

6. Added an air blast and mist cooling setup. With a quick regulator for coolant flow.

7. A crosshair laser.

8. Chip guards.

9. X-box MPG.

10. Auto tool height setter touch plate.

11. Touch probe.

12. Clamps

13. Final pieces of the puzzle.


1. A dust boot

I keep my cnc in the living room. Actually, I keep my cnc in pretty much the only room we have. This means that both me and (more importantly) my significant otter have to live with everything that gets thrown out of it. Without a dust boot, that is a lot. Dust, chips, burnt rubber (not intentional), various cutting fluids and other unspeakable horrors. At least they feel like unspeakable horrors when you find them between your buttcheeks when waking up in the morning. As such, I’ve spent a lot of time trying to build the perfect dust boot. I haven’t succeeded yet, but I’m getting closer. This first attempt(first working one at least) does its job superbly  at containing dust and wood chips. It’s a floating dust shoe, which means it doesn’t move up and down with the z-axis. Instead it rides on the bristles, which avoids them getting crushed on plunges. It’s held in place with linear bearings riding on aluminum rods.

Dust boot in action


2. Another dust boot

This one isn’t nearly as pretty. It does however have two very important features. I can see the workpiece while cutting and I can using cooling. It works, but not much more. My next iteration will incorporate the features from both versions to form a perfect bust boot. Like using a hobo and a rabbit to make a hobbit.


4. Vacuum table

I haven’t really gotten around to this yet, but I have a bunch of vacuum ejectors (venturi effect type thingys) that I’ve been playing around with.


5. Limit/homing switches

This was a pretty major piece that I needed to fix. Limits are good to have to avoid crashing the machinea, but what I really needed was the homing to autosquare the gantry at every startup. The switches are inductive proximity switches (I’ve tested them, and 9 times out of 10 they’re dead on according to the dro, 1/10 they’re 3 thousands of a millimeter off. That’s impressive.). I had some trouble finding good mounting spots, but that was nothing compared to the trouble of fitting even more cables in my e-chains. I had to use flat UTP cable. Noise isn’t an issue, even though they’re not shielded, since I use 15V and run them straight in to an optocoupler. To be safe, I also generated the voltage with a traco switchmode converter, to keep it from having a potential against anything else in the system. The homing is dead on accurate like a said, mainly due to a “double tap” macro I wrote and added to the mach3 homing function. I’m not going to put a separate picture here, it’s up top.


6. Coolant

I added an air blast/mist coolant system to lubricate, cool and blast away chips when cutting alu. I use 99.8% isopropanol alcohol when cutting. It’s great for alu, leaves me with a mirror finish on the climb side. There’ no cleaning, since it evaporates without leaving any residue. Since I bought a real cheap mist thing from china, I also had to add a high tech regulator for the coolant flow, see pictures. I’m going to upgrade this to an IV bag next week. That’ll be fun.


7. Crosshair laser

This is a great addition. It’s a small laser mounted to the spindle-mount. By clicking one of the shoulder buttons on my x-box controller, I turn the laser on. I can then place the cross wherever I want my zero, then click the other shoulder button to execute a macro that moves the bit to that position and zeros the dro. Since reflections can make the exact center a bit hard to see, I also added a pot to set the brightness of the laser.


8. Chip guards

Even with a dust boot, sometimes a few chips fly out and land on my rails. I didn’t like that, so I added a couple of removable walls made out of cardboard. They’ll be replaced by acrylic later on.


9. X-box 360 controller MPG

This one is great. There’s a plugin for mach3 to use a controller as an MPG, and it’s perfect. Custom macros on every button and fully analog speed adjustment. Wireless as well, so no hassle with stuff getting caught.


10. Touch plate

A piece of pcb with a crocodile clamp that plugs in with an audio jack. Great for setting tool height fast and simple.


11. Touch probe

This one is also great. When engraving, I need really flat surfaces. That’s rarely the case, so I built this probe to autolevel. (I use g-code ripper for autolevelling). The switch is a standard micro-switch, the repeatability is good enough.


12. Clamps



13. The final piece

This piece and the one on the other side were the last two things I needed to put the machine for it to be completely finished. They’re milled and engraved by the machine. The result is just perfect, couldn’t be happier. It cut the alu like butter, left a mirror finish and the accuracy of the parts is better than my calipers can measure.

This means that the machine is completely done. I’m just going to make a final video to show the machine and all of its features, then I’m off to my next project. Hopefully that will be my thesis work, still haven’t managed to get in contact with anyone that wants me. And yeah, I made this video to attract some viewers. Because you know, bitches love Arduinos.

Hello old friend, we meet again

Ok, this time the reference was a bit obscure, I admit it. I am im fact (incorrectly) quoting the movie Ink. Watch this to get an idea.

It’s been too long since I updated, and lots of things have been happening since last time. I’m actually nearing completion. Though I’m already planning upgrades, namely a vacuum table, mist cooling and a 3d digitizing touch probe. And of course homing switches.

Anyway, the pictures should be more or less self explanatory. I’ve attached the table, and milled it flat. The finish came out ok, but I wanted more than that. A bit silly perhaps, but I wet sanded it and polished it with some autosol. It’s not quite a mirror finish, but it’s as good as it gets without a buffing wheel.

I then spent tons of time basically stripping it down to parts again to straighten it out and fix the angles. A few shims were needed, and it came out decent. The spindle could perhaps still need some tramming, it’s about 0.05 degrees tilted now. But I’m going to call that good enough for now. I also checked the backlash, it’s 0.01mm in the y axis and 0.006mm in the x axis. The runout of the spindle appears to be 0.02mm, though I’m not sure I dialed it in the correct place. So far, it’s within what I was hoping for looking at rigidity (about 2500N/mm) and accuracy. Don’t know about repeatability yet. I’m sure I’ll get more info once I start cutting things for real.

I also rebuilt my electronics enclosure, now featuring fans and blue blinkies and stuff. It didn’t turn out quite as nice as I had hoped, primarily due to the fact that I drew all of the cuts 10mm wrong in all dimensions. It’s good enough that I don’t care to redo it though.

Next post might be the “I’m finished, for now” post, but we’ll see.

Day Something – The Tinkering

Since my last update, reality kinda caught up with me. Not in the horrible I-woke-up-to-remember-that-I-was-actually-born-a-pygmy-shrew (It’s fucked, basically), but more in the I-can’t-spend-17-hour-shifts-building-anymore. This means that I’ve spent quite a few days with a little tinkering here and there. Don’t be fooled though, these things take a tremendous amount of time. For example; yesterday I planned out what I was going to do. A reasonable estimate was about 1 hour. But to be safe, I said to my significant otter that it would probably take all day. I didn’t quite manage to finish half of what I had planned. All in all, above expectations.


Now, what I have done since last update is first of all to get it moving. I was so happy I actually peed my pants a little. The first axis to get going was the y-axis. I had expected to be able to get around 2000mm/min out of it, plus whatever the better driver I had on that axis could give me. It turns out I could push it up to 14000mm/min. Holy smokes, Batman! The other axes could go up to 5m/min, which is both disappointing and incredibly impressive. Going from my calculated 2m/min to 5m/min is all thanks to the PLCM-E4, which is awesome. However, this means I really, really want to replace all my shitty old chopper drives with the AM882 digital driver, doing that would also allow me to increase the drive voltage to 70-80V, further upping the speed with like, a lot.

What I’ve done apart from the actual getting shit to move is to sort out the wiring, cramming them into the e-chains, cabletie-madness and so on. Cable management is a bitch. I then built a temporary electronics enclosure. This one will later on be fitted with walls and a front panel, and several cooling fans. Then I fixed the cooling for the spindle. It’s a 75W pump in an Ikea box, sealed with silicone. The coolant fluid is demineralized water with some anti-freeze (glycol) in it. The glycol isn’t in the photos (it’s a fancy red color).


There’s still quite a bit of work left to do with it, so keep on coming. Hopefully the next post will have the first job in it. While you wait, here’s a video!


Day 5 – Pictures!

I feel pretty much like a woman post partum. I’m tired, I’m sweaty, I’ve probably shat myself, I hurt all over, I’m a bloody mess. But most of all, I’ve created something. It sure as hell ain’t finished, it might not even work properly, but  oh dayum it’s looking good so far.

Today I’ve been assembling the parts. A few things had to be changed, some holes countersunk a bit more etc, but mostly things have worked out well above my expectations. I still have a lot to do though. A few more parts need to be screwed on, like the table and the ballscrews on the x-axis. Then I need to mount the motors, do the wiring, set up the electronics, fine-tune everything with the dial indicator and so on.

Unfortunately, these five days of intense building was all the time I had for now. Next up is a little bit of school, packing and going on a vacation, perhaps spend a little time with the missus. Might be a couple of weeks before I get time to build again. I had hoped to be able to do more in these five days, but from what it looked like a couple of days ago, I never thought I’d get this far.

Until next time.

Day 4 – Getting real tired of this shit

Ever worked as a professional shoe shiner? Well, me niether. But I do imagine this is about how I would feel after a 16 hour shift (with a 15 minute food break) if were to have tried shoe shining as a career. And if the shoes had to be wet sanded with 7 different grits of paper before polishing. And the shoes were made out of aluminum.

The polishing didn’t turn out great unfortunately. The smaller pieces were ok (and by ok I mean mirror finish), but I couldn’t get an even finish on the larger ones. Probably a case of shitty tools again.

On a more positive note, I’ve started a bit of assembly. I haven’t got any pics today, but tomorrow should be an exciting day. Stay tuned.

Day 3 – Things are looking good

No pictures today. Sorry.

I’ve managed to drill and tap all of the holes. Not counting the countless holes I’ve missed, and certainly not counting the endless hordes of bad holes that need to be redone/thrown out the window.

Anyway, it doesn’t sound that impressive when I put it to paper like this, but I’m happy with what I’ve accomplished today.

Jens out.

Day 2 – We have liftoff

Day 2 has been a lot better than day 1. I’ve managed to drill and cut 6 pieces today, and I would have been able to do even more if not for some tap-breakage-issues. Note to self, always buy quality taps (and tools in general). I managed to get the first tap I broke out, but the second one is lodged in like a piece of popcorn between my teeth. Luckily, the hole that was ruined was more or less cosmetic and shouldn’t matter that much. Still, wasted a lot of time trying to fix it. Also I need more taps.

In the pictures you can see my spindle mount. It was a pretty novel idea, you’ve hopefully already read about it some posts ago, but I’m happy to report that the fit is absolutely perfect. My cut isn’t very pretty, but the mount as a whole is a lot better than the standard chinese mounts.

In the pictures you can also see my workplace, I’ve managed to keep it a lot more tidy today. That felt good.

Overall, I’m still about 2 weeks behind schedule (after two days), but I’m alot happier than I was yesterday. My hands are pretty cut up though, some of the cuts are looking pretty nasty. There’s a lot of aluminum lodged in the wounds that I can’t get out. My kitchen floor is covered with a blood-cuttingfluid-aluminumchips slurry. It’s pretty disgusting.

First day of building

There we have it. I made it through the first day. Honestly, I had expected to work all through the night, but I’m about at tired as I’ve ever been in my entire life. Going to get something to eat now, and we’ll see if I continue a bit after that.

Anyway, to summarize this day in one word: utter apeshit. That’s right, I used two words. Fuck you.

Where do I begin? Everything went wrong. First of all, I had expected to use my old cnc to mark the locations for the holes. That took about half the day to realise that my old one was alot worse than I thought, missed steps or backlash, I don’t know. It sure couldn’t place holes accurately at least. That means I had to mark all the holes by hand, which took hours and hours. For 3 pieces. The other 20+ are still on my to-do-list.

Then to the drilling. That was a lot harder than I thought, and a lot messier. You can see that even my vacuum separator gave up on life. The drilling was a nightmare, and that was before I broke a drill almost all the way through a piece (I broke like 5 other drills, but they weren’t that much of a hassle). Took me about 2 hours and a messed up hole to get it out.

I tried to polish one of the pieces to get it to look a bit nicer. It didn’t turn out as nice as I had thought, but I only spent about an hour doing it, I guess I should expect three times as much to go through grits 180, 240, 500, 800, 1200, 1500, 2000 and then the polishing wheel.

Oh well, food now and a new try tomorrow.

And so it begins…

Today, the final piece of the puzzle fell in to place. Or rather, the final parts arrived. The actual falling in to place will probably be more of a sweaty, frustrating, angry microadjustment-with-sledgehammer-redo-everything-piece-of-fucking-shit thing over the course of a few weeks. I can hardly wait to get started :).

The plan is this: I’m moving back to Uppsala this weekend, and hopefully by monday I’ll be able to start building. I’ll have a few days to build, before uni starts and I’m going to Gran Canaria. Because fuck uni, that’s why. That means that this is the last boring post with no acual building to show. I’ll use this time to go over the parts in a bit more detail (but still a very watered down version, do you have any idea how many books I could write on the subject of hobby cnc by now?) and post a more or less complete BOM (Bill Of Materials).


The frame

cnc v3 1

The frame is built completely out of aluminum, with a variety of alloys. The machine will weigh around 120 kilos, which isn’t very much. I could fill the profiles with dry sand later on if I wanted to increase the weight and get rid of some vibrations. Then again, I wont be able to move it as it is. Haven’t really figured out where to put it in my tiny apartment yet…

Anyway. I’ve ordered a grand total of 23 pieces of aluminum plates from and The plates are 20mm, 15mm, 5mm and 4mm. The majority is of course 20mm. Most of the plates are 5083, a few 6060-T6. 5083 is almost as rigid as the very common 6061, but should be less susceptible to vibrations. Also great if I’m using the machine in seawater.
The gantry crosspieces are 80×80 alu profile, the rest 60×60. The table is made up of 4 pieces of 160×16 profile that is surprisingly massive. You should be able to see some alu angles that connect the profiles. Those are bought from china, and I deliberately made sure that no real stresses would be put on them. Turns out they were excellent quality and rigidity, I could have used them alot more than I did. Pleasant surprises all ’round (in fact, I’ve bought a lot of items from china, mainly using, and every single piece has far exceeded my expectations).
The parts are fastened by means of drilling, tapping and applying screws. Very high tech. Oh yeah, you know those hammer-head specialty nuts that you use for profiles? Also bought those from china at practically no cost. Excellent quality. Insane prices here in Europe.

The beauty of the frame isn’t really in the parts though, it’s in the design. Not going to talk about that here, but one of the main considerations I had when designing was to avoid dependecy on the accuracy of the cuts when I ordered the parts. Having measured them now, motedis claimed a 0.2mm deviation from the ordered lengths, and that criterion was met. However, the width and thickness of the plate parts were consistently 0.3-0.4mm too thick and wide. Doesn’t matter in the slightest for me, but something to consider should you buy from them. Apart from that, I’m very pleased with their cuts and packaging, I’d definitely recommend them to anyone needing alu.


The linear units

The linear units were bought from fy bearing on aliexpress. Wendy was a great seller, easy work with and great engrish. My order was completely custom made, and most of the things I ordered wasn’t even in his store. You can see from the pictures up top that the packaging was slightly damaged from shipping, but none of the items were harmed. They were pretty dirty from the cuts, so the first thing I did was to clean them up and oil everything up with some machine oil. The rails are, as far as I can tell, actually genuine HIWIN parts. That’s pretty fucking impressive. Next up we have ballscrews. C7 quality 1605 ballscrews, they appeared to be perfectly straight, or at least as near as I could tell, the machining was looking good as well. I’ll just throw in the entire parts list here:

2 x HIWIN HGR20R 1000mm linear rail
2 x HIWIN HGR20R 700mm linear rail
2 x HIWIN HGR20R 300mm linear rail
12 x HIWIN HGH20CA block (He actually sent 2 used blocks, which I didn’t notice until after I’d voided my purchase protection. When I told him, he didn’t believe me, but would have a talk with his employees. 1 day later, he sent 3 new ones free of charge.)
2 x SFU/RM 1605-1000mm ballscrew
1 x SFU/RM 1605-700mm ballscrew
1 x SFU/RM 1605-300mm ballscrew
4 x 1605 ballnut
4 x ballnut holder/housing DSG16H
4 x BK12 ballscrew supports
4 x BF12 ballscrew supports
4 x 8mmx10mm flexible ballscrew/motor couplings.

Can’t think of anything more that wouldn’t be too detailed at the moment. But you know, there’s a comment field if you would like to know more.


The spindle

There is a literal fuckton of chinese spindles scattered over fleabay and aliexpress that look virtually the same. Protip: They aren’t. Your garden variety chinese spindle probably boasts something like “german import ceramic bearing, more longer life very good”. Inside you’ll find 3 cheap bearings that might of might not be angular contact bearings. The connector will have 3 or 4 pins, the fourth will probably not be connected to the housing. It’s a gamble if you get a good one.
I’ve spent a lot of time finding a good supplier, and I finally did. What you really want is 4 bearings, which will increase the service life alot. An earthed housing is pretty much essential as well. My spindle appears to be perfect, though I haven’t yet put it to the test in the machine. In fact, the clamping nut for the collets even has two drill holes that indicate that the spindle has been balanced.

The actual spindle is a 2.2kW water cooled three phase asynchronous motor, 0-400Hz. The water cooled motors are a bit more work than the air cooled ones, but it does pay off in the sound level and the ability to maintain low rpms without overheating. I can tell you, when running my motor at full speed (24000 rpm) I can hardly tell if it’s even on. With my wood router, I needed double ear protection and it still hurt to run it at those speeds. To run this kind of motor, you need a VFD. I went with the Huanyang VFD, though a lot of people recommend getting a quality vfd; Hitachi, ABB or similar. I got a set of 14 collets for the spindle, ER20, meaning I can use up to 13mm bits.

I also got a 75W water pump that I intend to use with distilled water to cool the spindle. Since it’s an induction motor, it doesn’t produce a lot of heat until you put some load on it. I could probably mill circuit boards without using any cooling at all.

It’s very important to use shielded cable for the spindle power, since it’s really dirty and will mess up your limit switches and steppers. I used 4 core 1.5mm² CY cable. Got it from CY cable is practically impossible to get your hands on in Sweden. I was going to buy the cable when I was in London, went to a CEF but they were out of stock. Bummer.


The steppers and drivers

When buying steppers, what you’re looking for is primarily the form factor (usually NEMA), the winding inductance and the holding torque. Larger motors are alot stronger, but comes with a higher inductance. NEMA 23’s are usually a good compromise. Preferably you’d want an inductance at near 3 or lower. The reason for this is that steppers are really strong when standing still, but loose torque at higher speeds. This effect can be mitigated by running them at a higher voltage. Gecko recommends sqrt(inductance)*32 iirc. Most of my motors are 3.8mH, which would mean 62 volts as an optimum level. Now, I’m going to use the steppers and drivers from my old cnc. These steppers are NEMA 23, 3Nm, 3.8mH. The drivers are DM542A, analog drivers with a maximum voltage of 50V. My power supply is at 36V. This means that my machine will be far from the speeds it could be with good digital drivers and a proper voltage.
Since I’ll be using a fourth motor this time, I’m taking the chance to try something better. My new motor is a NEMA 23, 4Nm, 3.0mH ( I’m driving this with a Leadshine AM882 (aliexpree), which is a really good digital drive, capable of up to 80V. This could produce perhaps triple speeds, with less noise and stall detection. Should I fall in love with this combo, I’ll probably replace the other drivers and the power supply eventually.
The cable is a 4 core 0.75mm² CY cable. Shielding is important, yo. Didn’t use shielded cable in my last machine, and I couldn’t use limit switches despite RC filters and such.


Motion controller

Now this is an exciting chapter. Should probably make an entire blog post on this subject alone. Maby later.

First of all, stepper drivers take 2 input signals to make shit move. Step and Dir. Dir sets direction and one pulse on Step means the motors move one step (or microstep). One step is usually 1/200 of a revolution, or 1.8° (a microstep can be almost impossibly small, the AM882 supports 512 microstepping. That’s 100 000 steps per revolution or 49 nanometer per step with a 5mm rise ballscrew). Usually, you have mach3 or linuxcnc generate those step pulses and send them through an old fashioned parallell port to a breakoutboard that sends the signals to the drivers. The thing about mach3 is that it runs on Windows. Windows is a great operating system when it comes to most things. Realtime operations is not one of those. Not being a realtime OS means that Windows doesn’t really care all that much about when things happen, only that they happen fast. Things like microcontrollers and FPGAs aren’t that fast (compared to a modern processor at least), but they are great at doing things at precisely the right time.
A motion controller will talk to mach3, in its own time. A little bit of information here (pun intended), a little there. Usually by USB or ethernet. The motion controller then makes its own step pulses, with blackjack and hookers. Properly timed pulses means a happy stepper motor. A happy stepper motor can be run alot faster without stalling. A cnc operator likes that. Everyone is happy with a motion controller.

Actually chosing a motion controller is another story. There are quite a few. The smoothstepper is perhaps the best known, but other contestants like planetCNC, CSMIO/IP-S, PLCM-E3 and EdingCNC are also avaliable. The thing about all those (except the CSMIO, but it’s insanely expensive) is that they all need a separate breakout board. A breakout board can be had on the cheap from china. A cheap board will mess up your signals though, which kinda defeats the purpose, ye? I was looking around, trying to find a good board with integrated breakout of the signals. In my desperation I stumbled on to and went google translate on their asses. Lo and behold, I found this thing called PLCM-E4, a motion controlled that is reasonably priced with intergrated breakout. I contacted them and managed to buy one, with some persuasion. I have it in my hands now, currently untested (a bit nervous I have to admit). I think I’m probably the only one west of the iron curtain that has this particular board. Pretty mindblowing, but it might also mean support might be a bit hard to get. Then again, I’ve heard that purelogic has great support, we’ll see if I need it. When I get it running, I’ll make a review of the board and perhaps introduce it to the rest of the world.


Other stuff

I’ve got a shitload on my mind, and there’s probably tons of items that I’ve bought, or planned to buy, or will discover that I need when I start building. We’ll see.

I’ve bought 5m cable carrier, e-chain, w/e. Should look real nice to get the cables sorted in them. A pretty small size though, we’ll see if I can fit the stepper cables, spindle cable, water pipes and limit switch cables in it. I’ve made some calculations, and it’s tight as a 5-year old.