Many people have been attracted to the low price and big dreams of the CNC3018 desktop CNC router. If you’re quick, you can pick one up on the usual second-hand sales sites with little wear and tear for a steal. They’re not perfect machines by any stretch of the imagination, but they can be improved upon, and undoubtedly useful so long as you keep your expectations realistic.
[ForOurGood] has set about such an improvement process and documented their journey in a whopping eight-part (so far!) video series. The video linked below is the most recent in the series and is dedicated to creating a brushless spindle motor on a budget.
As you would expect from such a machine, you get exactly what you pay for. The low cost translates to thinner than ideal metal plates, aluminium where steel would be better, lower-duty linear rails, and wimpy lead screws. The spindle also suffers from cost-cutting, as does the size of the stepper motors. But for the price, all is forgiven. The fact that they can even turn a profit on these machines shows the manufacturing prowess of the Chinese factories.
We covered the CNC 3018 a while back, and the comments of that post are a true gold mine for those wanting to try desktop CNC. Warning, though: It’s a fair bit harder to master than 3D printing!
Thanks to [Thomas] for the tip!
I’ve got the slightly smaller brother of this, and the biggest fundamental issue is getting the thing true. It lacks any mechanism to make accurate adjustments on the frame.
After that, sure the frame would be better steel and the spindle is underpowered, but the biggest issue it turns out is the quality of the milling bits, both those that come with it and those you’ll pick up online.
A friend who works with big industrial machines gave me some old cutting tools from his place. They probably cost more new than my CNC machine did. But it’ll cut steel with them, as I discovered when I accidentally set my safe height too low and cut through a bolt that was holding down the workpiece.
We (rightly) are concerned about stuff like rigidity and so on, but that’s not necessarily the weak point in cheap machines. When the tool cuts like a hot knife through butter, the small servos work fine.
I agree with you. I have spent a lot of time trying to true my cnc3018. There is zero adjustability for several elements. The rigidity on my machine is ok, at least for cutting slowly in softer materials.
The cutting bits on Amazon are not good. The v carving bits are particularly lacking.
I also picked up a flycutter to flatten a spoil board. Because the spindle is not at a right angle to the spoil board, it leaves steps on the spoil board. I have improved that slightly by adjusting the position of the steel rods on the spindle x axis. The lack of adjustment is the major issue.
The spindle is anemic. The controller is also frustrating to work with. It is somewhat functional, but has been cost cut to a point of being slightly too small to work with. Even the cables are too short for the controller to reach out in front of the machine while cutting.
The 3018 kit is mostly interesting as an intro to CNC, where all the basic elements are present, but not great for much beyond cutting foam.
Not sure if you’re exaggerating a bit, or if mine is a bit more rigid – my one (can’t remember what it was called… 2818? All made from 2020 struts, no flat panels for the Z-axis) very happily mills wood, which is what I got it for. It’s certainly more than a toy, it paid for itself in a few months doing custom engravings.
I’ve also used is successfully on plastics/resin, aluminium, and with a diamond bit engraving glass.
It’s got no controller, but I’ve hooked it to a pi running cnsjs with an old gamepad as a controller.
To help true it, the closest I’ve managed was a couple of metal wedges between the clamps on the axis, with a bolt through them, so tightening it forces them apart slightly, which provides some attempt at fine adjustment, but it’s not very good.
Running a gauge over the aluminium bed showed the bed wasn’t flat either, but I couldn’t lap it effectively myself. I’ve not got round to flycutting it flat, as I generally work on fairly small pieces which keeps the errors manageable.
For those of us without fortunately placed friends, can you recommend a UK seller of good milling bits with easy domestic shipping to UK addresses.
Sorotec has some decent quality but not insanely expensive bits.
For me (with CNC3020) the poor spindle has been the biggest limitation. I keep meaning to replace it with something better, but good 50 mm diameter spindles are rare and I don’t have a lathe to DIY a proper one.
Read that as “Scrotec”, and had to blink a few times. Thanks for that.
I believe it’s “Sorotec”
On the mill & lathe I use cheap no-brand cutting tools as disposables for learning (AKA messing up & breaking stuff) or for general rough / “good enough” work and then have a few better quality ones bought from reputable suppliers for “best”.
You can get decent quality tooling for sensible money at the low end of the industrial catalogues – single flute cutters for aluminium are great, and a few decent carbide inserts for the lathe make a big difference too.
For what it’s worth, depending on the type of milling you do(!), “getting the thing true” is near trivial if done in the software that generates the gcode for it. Granted, I wrote my own mods to FlatCam for XY trapezoidal correction but it’s literally just a simple linear equation which allowed my DIY machine to mill accurate PCBs (the further X moves, the more Y offset needs to be added, basically). It’s also way easier to calibrate than moving an axis physically. For something as cheap as this, a software solution appears to be much more appropriate.
Motor + power supply + ESC + bearings + other materials is already more expensive then a complete set with BLDC motor, but with DIY you have the option to make a better quality spindle, and that alone may be worth it. But to make a decent spindle, more is needed than what is shown on this video. Simply using decent angular contact bearings would have been an improvement. The bearings also go in way to easy, which means the shafts are too thin and the holes to big, and that way you can never have a good runout.
For the BLDC spindle motor sets, there are a bunch of options, With ER11 the collet holder is often directly on the motor shaft, and runout will be good, but the collet holder tends to have a lot of stickout and that is bad for rigidity. With an ER16 collet holder, it’s clamped (with a simple setscrew) on a thin motor shaft, and this is also far from optimal.
One modification I’m curious about myself is to use the BLDC set with ER16 collet holder, and then:
1. Take out the lower bearing.
2. Fix the collet holder permanently to the shaft (For example tapered bushing & loctite)
3. Grind the outside of the collet holder down a bit (Use the inside with a brash bushing as bearing)
4. Put a bigger bearing on the outside of the collet holder.
5. Make a new bearing block as interface between the motor body and the bearing.
This is a modification that is relatively easy to do and it solves a few problems.
In step 3, the runout will be very low because the taper of the collet holder is used as the actual guide.
The big bearing on the collet holder makes the stickout shorter, and thus a lot stiffer.
The problem with the thin motor shaft is resolved.
Also, I would never ever buy a router with those round unsupported axles. As soon as yo try to do anything more then light engraving, then lack of stiffness becomes a huge problem.
If you search Aliexpress for “mini CNC 3020” you find a relatively cheap (around EUR500) machine that is built with square rails and PVC sheet material. It’s not a great machine (those cost EUR2000 or more) but it’s a decent start, and there are lots of opportunities to improve it.
Another very interesting machine is the “low cost EUR500 mill” as shown on instructables. It is a combination of square rails, aluminum profiles and wood sheets. It is a nice compromise to have something decent, while still keeping the costs low. The Aluminum profiles have easy straight cuts, and the wood can be made to size with “manual methods”. Unfortunately the author does not sell wood kits anymore. Also note that this machine is pretty much a copy of the Sorotec Compact line. (But those are fully aluminum and cost several thousand EUR.)
My description was not very clear. The Instructables I meant is the one below from Vulcaman
https://www.instructables.com/LOW-COST-DIY-500-CNC-MILL/
Note: I have no affiliation with him in any way. It’s just one of the best designs for a DIY router I know on the ‘net.
While I do think that design has some merit for a not too dissimilar price you can get a manual mini-mill and some stepper motors etc. Which I think would be for many folks the better choice at this scale. As even cheap crappy casting are likely stiffer and smooth running enough with a decent spindle for the lateral loads. You are still going to have plenty of work to do turning the probably very sloppily finished and put together mill into something more useful while adding the electronics, but the end result is a machine that should be able to take pretty meaty cuts even in harder materials and retain useful accuracy without too much chatter.
NB I did say not too dissimilar price – I’d still expect the mini mill version to end up a little more expensive, but probably not much. So if you get as lucky as some folks have with their mini-mills in initial quality its a massive step up in performance for almost no effort (and if you don’t the castings etc are hopefully just needing a little work).
Sure, with those cast iron machines you can have a stiffer frame, but not for a similar price. Cost of the smallest bare machine is around EUR800 (more money is bigger and stiffer frame. EUR 1500 is closer to a “decent” machine), and those prices are without stepper motors, power supply and other electronics.
You also have to deal with the extra friction of the dovetails and acme thread (and backlash), and therefore you will need bigger stepper motors. As a result, you would need a minimum budget of around EUR1300 for such a machine (but closer to EUR2000 for a decent sized machine). Then it would work acceptable for metal, but it’s too small to be very usable for sheet material. It’s a different type of machine for different sort of work.
You can also get a gantry type machine that is stiff enough for decent steel milling (instead of very slow / shallow metal removal) They can be either based on Epoxy granite, UHPC, cast iron or a steel frame. But those are in a different price class.
(Also, you need some budget, for mills, clamping, enclosure, dust removal, etc. and those costs all add up).
At the scale of these tiny gantry ones “suitable for sheet work” is never really true for either IMO, or equally true if you feel generous – at this price point the working areas always just end up too small to be much good for sheet work as a rule. Though the DIY router kits never have the Z travel to do much else or have the flexility in tooling they can use either…
I’ve seen mini-mills substantially under your 800 EUR mark, though not new at those prices its still a good condition piles of castings and a solid spindle often barely used – seems lots of folks try a mini-mill and upgrade or buy one either specifically for a project or as a project so there is usually a good supply of great condition used ones around (that may even come with a pile of the required tooling further improving the value).
Which is why I suggest at this sort of scale and price point the mini-mill is probably the better option. You have flexibility in the tools you can use to a much greater degree, more stiffness, usually dramatically more Z travel with pretty comparable X-Y for a very comparable price. You have to get a bit bigger for the gantry to really start to separate itself from the conventional mill in part because these mini mills are such mass produced casting they are stupidly cheap for what they are, but that doesn’t hold so much as you get bigger.
I wonder if clamping/bolting/epoxying one of the 3018s (or similar) to a granite countertop offcut would be of any use. I have an 18″x24″ no-name beat to hell granite surface plate of unknown accuracy that I paid $10 for at an auction, that I’m not using for anything, that I might experiment with.
Hello!
I applied a dc spindle to a Proxxon MF 70, but with bad results. The spindles shown in the link have already been converted and made usable.
https://forum.hobbycnc.hu/topic.php?topic=181
“The usual second hand sites”: do tell, where are the cool kids shopping these days?
Well, I saw one on Facebook marketplace half an hour drive from my house, in barely used condition for £60. I missed it by an hour or two, or i’d have one here to learn on. I’ll just have to use the bigger machine at the makerspace in the meantime.
So you need a lathe with which to make the adapter for a BLDC… might be better just going for a corded “palm router” tool, to my understanding some come with a “chuck” which takes a proper ER-something series collet, rather than being able to take only tools with a 1/4 inch shank. They’re usually around 900 watts, which gives you more power than a small BLDC. Then with cheap ballscrews available, including ballnut and ends bearings, for about £35, 3 of them, one for each axis wouldn’t be a bad idea.
I bought one of these as it had a laser. Got that working without too much hassle.
The CNC software just doesn’t work for cutting though so I’ve never been impressed.
That’s the weak spot for any CNC “engraver” or “router” – the software. But I’ve found that using FreeCAD and Universal Gcode Sender, I can actually do things with my 3018. I do most of my design in OpenSCAD, but one of FreeCAD’s advantages is that it can accept models made in practically any CAD system, so you can use your favorite, and just use FreeCAD as a way of converting each operation into a gcode file, then UGS to drive the machine. FreeCAD may even have a gcode module, but I’m happy with my workflow.
I’ve been tempted to buy a laser module for it, but when I’ve looked, the power available was sufficient only for burning the surface of wood, or cutting paper. So it would be a poor laser cutter as well as a poor mill.
I have found FreeCAD quite difficult to use with STL models, though it can import most OpenSCAD designs directly.
For STL, Kiri:Moto is nice.
Given the purchase price plus upgrades plus time spent dicking around with it, is there a version / variant of these that is good out of the box for a bit more money?