Thinking of buying/building machine for PCB milling

[njhussey]

The best thing to do is to make your own, it will knock the spots off most things commercially available for the same price and in the process you'll get a better understanding of how they work and therefore (I can only guess as I'm just starting to build mine) will be better equipped to fix them when things go wrong/break.

If you look at, for example, the MD ones and the cheap China ones and then look at the build threads you will see the difference between a "home made" one and a "commercial" one. Making your own will enable you to do things like use supported rails, make the gantry much stronger (and normally a bit lower) use better quality materials for the same price you would have paid for a ready made one. There are enough guys on the forum here who will cast a critical eye over your design and once it's complete do any machining for you if you need it allowing you to assemble it with simple hand tools/workshop tools.

In my opinion a ready bought machine is only really good for scratching the surface at playing with CNC machines, in more ways than one! You will undoubtedly, no almost certainly from my reading various forums, need to update components to get the performance and finish you require. I did consider going down the ready made route before I started but soon dismissed it and now I will end up with a machine that will be capable of cutting Aluminium even though I started out wanting one that would cut Balsa and ply.

As an example of costs the MD A4 ball screw machine starts at £3,555.00 For that you'll get a machine that has a machining Area of 300mm x 220mm x Z Axis 155mm, my machine (when I get my finger out and start making it) should come in well below that and will have a cutting area of 1000 x 500 x 200. As far as I'm concerned it's a no brainer! If size is an issue, i.e. you don't need a machine that big then you'll be able to make a much stronger/accurate machine for half that cost.

[njh]

Would love to build my own CNC router one day and from reading and learning, it doesn't look as if it is as difficult as I first imagined.

But at the moment I have decided that for now I want to just get up and running and make some PCBs...

[njhussey]

Did you not see the 3040 for sale with everything you need in the for sale area? If you upgraded the trapezoidal lead screws to ball screws then you'd almost be there?

[m_c]

In other news, I asked why the UK postage costs were so huge (£200) on this 3040T-DQ Router and quickly received the response:

It's a way of avoiding some of the extorionate ebay fees, as you don't get charged fees on postage/carriage, so they include the customs charges in it.

[Jonathan]

It's a way of avoiding some of the extorionate ebay fees, as you don't get charged fees on postage/carriage, so they include the customs charges in it.

Not any more - they very recently changed the rules so sellers are now charged the final value fee on postage in addition to the final price.

[m_c]

You do realise fees vary depending on what country you're registered in, and to what country you're selling?

We get stung pretty hard for fees due to ebay's near monopoly in the UK. Other countries where ebay don't have as big a market share get far better rates. It does annoy me selling stuff on ebay in this country, that ebay end up with near 14% of everything by the time you factor in that they also own paypal.

[njh]

Discovered the 'CNCDudez New 2012 Desktop Mini Muscle CNC Machine' last night:
CNC Design Limited - CNC Desktop Router

Unlike other suppliers Sean actually replied to my email!

Has anyone else here bought one?
I suspect it surpasses my requirements for milling PCBs.

[Jonathan]

It would be ok for cutting most PCBs, but I wouldn't do much more than that with it. I've already posted my thoughts on that machine here, so I wont repeat it.

Edit 2: Link stopped working again, but you can find it in google cache here, post #19 is mine:

http://webcache.googleusercontent.co...&ct=clnk&gl=uk

Other pages:

http://webcache.googleusercontent.co...&ct=clnk&gl=uk

http://webcache.googleusercontent.co...&ct=clnk&gl=uk

[njh]

It would be ok for cutting most PCBs, but I wouldn't do much more than that with it. I've already posted my thoughts on that machine here, so I wont repeat it.

I don't have permission to view that page.

[Jonathan]

I don't have permission to view that page.

Should work now...

Edit, since the link stopped working again, I've copied my post about the CNC dudez 'mini muscle' machine below for good measure:

I agree with what Jazz has just said - frankly I'm amazed it has taken until post #17 for someone to mention any of the limitations of this machine. It could so easily be improved for little additional cost. Here's some ideas for you:

Gantry sides are weak and not braced - you have just single pieces of extrusion relying solely on joint at end. This is easily improved by adding diagonal pieces from the top of each to the base frame and a couple of horizontal pieces, or a plate, between them. You'll then end up with a trapezium which is far stronger. It looks like 1/8" thick aluminium angle has been used to mount the bearing blocks on the Y-axis. This is rather thin considering how important it is to make the joint between each axis strong.

Rails are only 12mm, which if you're intending to cut metals regularly is pushing it. This is compounded by only having one bearing block on each X-rail. The fact it's a double length one doesn't compensate for only having one instead of using two with a reasonable spacing. This makes the table more susceptible to racking when cutting near the extremities, especially with just one ballscrew. It's fair enough to have one ballscrew on X for a small machine like this, but with 2 bearing blocks not 4 it's again far from ideal. It looks like you could have two bearings on each rail and increase the spacing without loosing travel and if not for the sake of increasing the lengths of a couple of bits of extrusion to gain back the travel from increasing the spacing, it's well worth it for the rigidity gain.

Z-axis motor mount is flimsy - thin aluminium with small area in contact held only by two bolts. This is easily extended to for instance a U-shape which would offer much better support - 3"x1", 1/4" thick aluminium angle is good stuff for cheap motor mounts. Axial alignment of the Z-axis pulleys is quite a long way off so the belt is constantly rubbing against the pulley flanges which will cause premature wear. Not the end of the world, but still I'd expect better.

In the video the bed is surfaced at a pretty low speed. The machine should be rigid enough to surface it at something like 7-8m/min. Sholud be getting chips not dust...

The axes are running a lot more slowly than they could do with those motors. This is no doubt due to only running on 36V which is half of what you could and really should be using with these motors. The rest of the video shows that the acceleration is set quite low too. It's illogical to compromise the feedrates so much by using a low voltage when the whole idea of making a rigid machine is that it enables you to cut quickly. Also it's interesting to note that you may be able to get better performance with some good quality lower torque motors, since the corner speed is higher for a smaller Nema23 motor so with such a small mass to move you can end up operating past the point where the larger motor runs out of torque. If you add timing belt drives to the X and Y axes the acceleration or feedrate could be improved, in addition to reducing resonance. This may or may not be worth the increased cost, but certainly worth experimenting. I'd go for pulleys anyway since it's clear from past experience with people here that those flexible aluminium couplings are prone to shearing.

The Kress spindle is not a good choice if you're intending to cut aluminium regularly since its bearings will wear out quickly and it's somewhat lacking in power, although given the rigidity of the rest of the machine the latter is not a big deal.

In the video it takes 6 passes to cut through the 5mm aluminium, so 0.8mm per pass (perhaps slightly more since the last pass seems thinner) with a 6mm tool. If you're cutting aluminium regularly that's not very good. A machine this size with a strong frame should easily be capable of more. Also stop plunging with the cutter, especially in aluminium - it's hard on the tool and machine and there's just no need when you can use spiral toolpaths or ramping.

This machine is currently capable of cutting aluminium since having the 4 rails on Y/Z has helped compensate for the lack of strength in other areas.
A good design with 2 rails would easily perform as well, if not better (and by that I mean achieve a higher material removal rate in aluminium) with two rails. That's not to say having 4 is a bad idea, in some cases it's the only way to make a machine rigid enough, but here you might be able to save money by using two (supported rails) and investing more in other components, such as the frame.

For a machine this size it's hard to come up with a design that wont cut aluminium so long as you follow some basic guidelines. The reason is as the machine size increases it rapidly gets hard to maintain the rigidity. For a simply supported beam with a load, the maximum deflection is proportional to the length squared - so for example if you make the machine twice as large with the same material cross section then the deflection will be 2^3 = 8 times greater. Hence, if you keep the machine small which is the case here it's not difficult to obtain sufficient rigidity. There is no clearly defined limit for when a machine fails to cut aluminium and some sellers have been exploiting this for ages to make ridiculous claims, which is of course why it's good to have the video. I can cut aluminium with a screwdriver, but that doesn't mean it's a good tool for the job or that it will last very long.

The bottom two Y-axis rails should be flipped over so that the bottom rail is the other way up. The force rating for these supported linear bearings is much lower in the direction trying to pull the bearing off the rails, so it's best to have the rails mounted opposing to balance out the force rating. Currently the rails are all the wrong way round to counteract the force when you drill or plunge with the cutter, which is partly why you get the horrible noise every time that happens. If you swap them you'll always have 4 in the optimal direction for forces parallel to the Z axis.

If you want to test how a machine will perform in terms of how fast it can remove material and how good a finish it will get then the least subjective way to do it is to measure the deflection for a given force on each axis, then divide the force by the deflection to get a stiffness value in N/mm and compare this to other machines. Anything else is speculation. Even if we define a standard test using the same cutter and material, then measure the result, you can't accurately compare machines since there are so many variables. For example you can push the machine hard to get a better depth of cut, but the surface finish will deteriorate so you now need to have some measure of that, plus if the cutter only lasts a few minutes at that speed it's not an honest test. Again there are a lot of variables to get a good finish - even a weak machine will get a good finish (and hence good accuracy) with a very light cut, so the machine that gets the good finish with a large cut is the better machine. This will be the machine with the highest stiffness. That's why you measure and compare stiffness...
Could also measure backlash, but that's generally unimportant for a router when you have ballscrews, and so is to an extent measuring the size of a cut part or a centre distance with the calliper since, again, there are a number of factors that affect this - on any reasonable machine it'll be within a few 10's of micrometers at the distance most callipers measure, so things like the actual tool diameter, calliper tolerance and cutter wear become a factor. Given a couple of tries I could make a video of cutting an aluminium part with my router, measure it and get it spot on according to the (0.01mm) calliper - all you need to do is make the part once, measure the error and compensate for it in the drawing, then cut it again. The machine would have to be very poor for this strategy to work.

I was expecting "under £1000" to include the electronics and assembly since having worked out how much it would cost to build there's still room for profit in that price assuming you're sourcing the components from China. £1600 is not much less than it cost me initially to make my machine (not including labour of course), with steel frame, and that's 53 times the working volume of this yet capable of cutting aluminium much faster (although I still wouldn't describe my machine as very rigid) for a long long time. For that sort of money, if you want to cut metals, I'd advise buying a milling machine and converting it to CNC unless you really need the additional Y travel. I wouldn't be surprised if you still sell plenty since the UK CNC router market is currently exceptionally limited, so anything better than the CNC3040 is bound to be a hit!