Lowcost capable CNC - Make vs Buy?

[Andrewg]

Ok, you can guess the answer, because I wouldn't be writing here if I hadn’t concluded ‘make’ remained a compelling option (and wanted your help!). But I hope laying out some facts and reasoning will help others who come here with this question. And those who know better can correct/encourage me.

Since the arrival of cheap motion control systems gave birth to Home CNC, huge ingenuity has gone in to replicate the key expensive mechanical functions- linear motion and positioning. The standard solutions of profile rails and ballscrews were too expensive hence draw slides, bike chains, skate wheels, timing belts and V-slot. These have added to the part count and complexity of design but have (for some) found a sweet spot of performance and price that has attracted a lot more makers .

So in the maker ‘design around the problem’ corner we have people like Open Builds and Carbide 3D who offer a kit of parts based on skate bearing on V slot extrusion with timing belts or leadscrews. Today a Workbee or Shapeoko is available as a kit for around £1,400

To me this is a base line - For £1,400 I can be up and running in a fairly short time with plenty of support and a community to help me through the learning process. Granted the machine will have significant weaknesses, but at least I am cutting and learning.

However the continued superiority of the profile rail is evident in the multiple upgrades of these machines, resulting in the slightly bizarre solution of V-slot extrusion used as a base to bolt linear rails. Carbide 3D now offers a linear rails and lead screw or ballscrew upgrade for the critical Z axis. QED.

Meanwhile the increase demand for cnc systems has reduced the costs for key components like Hiwin profile rails and spawned cheaper copies. Add sites like Aliexpress and Banggood that allow access to the Chinese supply chain and the DIY builder can now use the correct parts, just of lower quality.

For some time it bothered me that few manufacturers seemed to be exploiting this route. After all the Chinese 3040 and 6090 machine that flood ebay are still mainly using unsupported round rails(Good enough for the market they target I guess?) The better Chinese makers like Omni CNC do offer linear rails and ball screws in the 3040, 6090 sizes but don’t have a UK distributor. Another Chinese maker, Quick CNC supplies larger machines (6090 and bigger) which are available locally. Their smallest 6090 is marketed as the Laguna IQ in the States, and is available in the UK as the Itech K6090T for £5,400. The suspicion is that ti would be mechanically good but electrically more suspect.

More recently I2R launched a new range of budget machines with their smallest 600x600 cnc available in the UK for less than £4k inc through Axiom in the US and StoneyCNC in the UK and sports linear rails and ballscrews. Felder, a quality woodworking machines brand have recently entered the budget end of the market with the badged Hammer 47.82 (around £6-7k I believe), though interestingly with leadscrew not ballscrews.So the market is getting more competitive and crowded.

So to buy a ‘semi serious’ home machine, I am looking at £4k minimum and probably £5-6k. This remains a big jump from the £1,400 Workbee. It seems therefore there remains a sweet spot for DIY solutions if by smart sourcing and support from people on forums like this, I can design and build a machine in a reasonably short time for around £2k that exactly meets my needs (and works!).

I should of course acknowledge that for some the journey is the point, and I do feel the pull to test myself by building my own, but I am trying to make a more rational calculation. Many of the builds described here are epic and must have left the builder forever changed (and better for it one hopes) but a cold calculation of time spent and opportunity missed (assuming you want the machine to make things) would put the true cost of the machine far in excess of the £10k for a decent industrial no compromise machine!

I started writing this to convince me to stop reading this forum and dreaming of building...but it seams to have backfired. So I will now try and distill what I have learnt here and elsewhere into a plan of action.

[JAZZCNC]

It's simple really.!! . . . If you need a machine for business and have the work for it then buy one or have one made by someone like me as you are up and cutting quicker, so earning quicker. I've built machines for people costing £5k and the machine as paid for its self within the first few months or even weeks in some cases.

However, if your DIY and it's just for pleasure and have the time, patience along with the tools then just get on with it start building forget how long it takes, it doesn't matter. What matters is how you do it.
If you try to cut corners and do it on the cheap then at best you'll have an under-performing machine and at worst you'll have a pile of expensive scrap. Cheap solutions like using Mdf, linear draw slides, chains and even belts if not done correctly are a waste of time if you want a serious machine.
The only way to have a machine that's worth the effort is to do it correctly from the start and buy components that are fit for purpose. This doesn't need to be expensive either, many secondhand industrial components can be found that even if 10+ yrs old will still be far far superior to any of the cheap alternatives used.

However, you don't even need to do this if you are prepared to buy from china and look around. With £2k if you shop around and buy wisely you can build a very decent machine that will match and in some cases greatly outperform many of the machines you mentioned. You just have to wait and get your hands dirty, which is half the fun if your into DIY.

Forget any Cheap hack routes they don't work well or last longterm.

It really is that simple. If for DIY Build it good with decent bits. If a business just Dig deep buy it and earn it back quicker.

[Andrewg]

Indeed Jazzcnc,
So given I don't have a business case, but want to explore what the capability would bring to my workshop....the choice comes down for me at least to buy a Workbee or build my own for a budget of £1.5 to 2k. This forum shows that this second route is viable and is the route I am keen to take.

So what have I learnt a 'correct approach' looks like?

I should say I want a machine to cut hardwood in my furniture making workshop. 600x600 work area because I don't have the space and have plenty of processes for which this is big enough. Also helps with accuracy keeping small I believe. Want to cut joinery so 0.1mm repeatability. It will be a workshop machine that won't run all the time but when used needs to be reliable and quick (5m/min cutting?). A build I can complete in two weeks full time ideally..(planning, dreaming excluded)

Frame
Stiffness directly effects performance so important. Unlike manufacturers, DIYer does not worry about shipping, so a steel frame, welded ideally. Epoxy leveling process allows imprecision to be addressed. However I would like to go Ali extrusion and shimming if possible alternative to avoid epoxy and steel working plus flexibility to extend/change frame in future.

Mechanics
Linear rails, ballscrews and correct support bearings sourced from good chinese supplier as set. Leadscrews seem possible compromise that many designs accept, attractive as their lower profile nuts allow leadscrews to more easily fit between bearing rails and keep design simole. Guess wear and lower efficiency not a killer for home machine that operates fewer hours or am I wrong?

Dual drive Y axis seems to be preferred for simplicity now electronics can cope with squaring Gantry especially if drivers have stall detection.This especially of value to me as I want to be able to machines end of pieces held vertically through table.

Principle design challenge are the plates required to mount the components and join it all together. Aluminium can be machined with wood tools. Aluminium tooling plate is already flat, so either they can be cut and drilled myself or waterjet cut by contractor. Adjustability is principle to adopt rather than precision.

Gantry
Make it L shaped and use stiff large box section or extrusions. Extrusions make fitting everything much easier. Rails on front or top&front. One thing that eludes me is how best to join two extrusions to make L, what fitting works that way?

Electrics
Nema23 steppers are big enough for small/medium machine (exact force and inductance to be determined), matched with proper drivers (digital, stalll detection?) and serious power supply (toroidal). Homing limit switches, inductive for accuracy and reliability. Get shielding and earthing of wiring right. Think through positioning and routing of everything.

Electronics
Mach3 and printer ports are old hat. Ethernet and UNCNC or Acorn seem to get thumbs up, though there are also dedicated controllers and open source solutions based on grbl are getting better.

Build challenges
Machine accuracy is determined by build accuracy, so need a dial indicator and know how to use it!

A good design should make build process one of bolting together and aligning ideally, plus lots of wiring.

What have I overlooked?

[JAZZCNC]

Indeed Jazzcnc,
So given I don't have a business case, but want to explore what the capability would bring to my workshop....the choice comes down for me at least to buy a Workbee or build my own for a budget of £1.5 to 2k. This forum shows that this second route is viable and is the route I am keen to take.

Forget the WorkBee or any of those you will be disappointed I promise you that.!
The Budget needs to be 2K to do it right if building DIY.

I should say I want a machine to cut hardwood in my furniture making workshop. 600x600 work area because I don't have the space and have plenty of processes for which this is big enough. Also helps with accuracy keeping small I believe. Want to cut joinery so 0.1mm repeatability. It will be a workshop machine that won't run all the time but when used needs to be reliable and quick (5m/min cutting?). A build I can complete in two weeks full time ideally..(planning, dreaming excluded)

600 x 600 is the perfect size to start your first DIY build. 5Mtr/min cutting no problem, if done correctly you'll double that with a bit to spare.
0.1mm repeatability, again no problem and easily able to achieve better if done correctly.
Building within 2wks could be a stretch for your first build and would require everything to go perfectly and put in some long hours. And obviosuly you have to have everything needed to hand.
For reality's sake, I'd allow your self a full month for fully built and working machines and work shorter hours.
Don't underestimate how long some of this stuff takes, just the wiring alone could take a full week if you are not used to doing it.

Frame
Stiffness directly effects performance so important. Unlike manufacturers, DIYer does not worry about shipping, so a steel frame, welded ideally. Epoxy leveling process allows imprecision to be addressed. However I would like to go Ali extrusion and shimming if possible alternative to avoid epoxy and steel working plus flexibility to extend/change frame in future.

Again Ali extrusion is perfect for first machine as it's easily worked with and accurate. However, you must use the correct type, size and not go with the light duty stuff often found on machines like the Workbee's etc.
It's also very very important how it's all fastened together and plates etc. Done correctly it's very strong and perfect for wood routers.

Mechanics
Linear rails, ballscrews and correct support bearings sourced from good chinese supplier as set. Leadscrews seem possible compromise that many designs accept, attractive as their lower profile nuts allow leadscrews to more easily fit between bearing rails and keep design simole. Guess wear and lower efficiency not a killer for home machine that operates fewer hours or am I wrong?

Don't compromise go with ball screws every time. A quality lead screw and nut that can get close to ball screw efficiency/accuracy will be just as expensive as Chinese ballscrew and won't be massively better in quality.

Dual drive Y axis seems to be preferred for simplicity now electronics can cope with squaring Gantry especially if drivers have stall detection.This especially of value to me as I want to be able to machines end of pieces held vertically through table.

Not strictly required on a machine this size if just cutting woods. However, it is better if you want to cut aggressively or want to cut very hard materials or aluminum, brass, etc.
Not all Drives have stall detection, in fact, very few do, even then they only work in a limited way. A much better solution is to use Closed-loop Steppers which have come down in price massively. Far better than the standard stepper system.

But again going back to usage and Budget, using a single screw setup is much cheaper and if you don't plan on cutting aggressively or harder materials then it could be wasted.
The other thing is that if you design the machine and plan the control box correctly it's not a very difficult upgrade at a later date to just stick another ball screw on and fit another drive in the cabinet.
I've built many machines for people like your self who were on tight Budgets and didn't exactly know what their needs were going to be so this approach was taken (all my small machines are electrically designed to do this). Very few actually found they needed to upgrade and those that did then it was a simple few hours to upgrade which they did them selfs.

Principle design challenge are the plates required to mount the components and join it all together. Aluminium can be machined with wood tools. Aluminium tooling plate is already flat, so either they can be cut and drilled myself or waterjet cut by contractor. Adjustability is principle to adopt rather than precision.

If you cannot make them your self by any of the methods you mention then I suggest you get someone to machine the parts rather than cut them on a waterjet or laser. unless they are thin parts. Both water and laser leave a taper on the edges of thick parts which can cause you more grief than it would to cut them your self with a hand router.

Gantry
Make it L shaped and use stiff large box section or extrusions. Extrusions make fitting everything much easier. Rails on front or top&front. One thing that eludes me is how best to join two extrusions to make L, what fitting works that way?

The best method for the "L" gantry is to bolt into gantry end plates with the ends. On a narrow gantry like what you want you'll get away with just one Bolt in the center which goes from the underside into a T-nut on the upper profile.(or can drill and tap the profile directly) However, it's important that you have a piece of material which is a nice tight fit that goes into the slots between upper and lower profile. This stops movement from sliding and the single-bolt clamps the profiles together. More bolts could be added but on short length like this it's not required.

Electrics
Nema23 steppers are big enough for small/medium machine (exact force and inductance to be determined), matched with proper drivers (digital, stalll detection?) and serious power supply (toroidal). Homing limit switches, inductive for accuracy and reliability. Get shielding and earthing of wiring right. Think through positioning and routing of everything.

Electrics are the KEY to a great machine. Take your time, don't cut corners and if the budget allows use closed-loop steppers, in fact, save up a little, wait a little longer if needs be, it's worth the wait.

Electronics
Mach3 and printer ports are old hat. Ethernet and UNCNC or Acorn seem to get thumbs up, though there are also dedicated controllers and open source solutions based on grbl are getting better.

Again KEY don't cut corners. Forget Grbl or open-source, yes it's getting better but it's still miles away.
If you want quick and good at a reasonable price then go with UCCNC and one of there controllers like the AXBB, there is nothing better for a wood router for the money.
Or if your like a challenge then take a look at Linux CNC and Mesa but get ready for a big learning curve.!

Build challenges
Machine accuracy is determined by build accuracy, so need a dial indicator and know how to use it!

A good design should make build process one of bolting together and aligning ideally, plus lots of wiring.

What have I overlooked?

Design is VERY VERY important and if done with a forward-thinking attitude then you can have a great machine for not a lot of money.

Lastly, let me just say don't get caught up in the "Steel is better than Profile" or " Must use Steel if want a strong machine" because it's often said by those who haven't built a machine using it.
I build machines using both or Mixture of both as is often the case and I can tell you now profile is more than good enough for any router which is to be used in a DIY environment or even a Small/Med business environment where speed and high material movement rates are not required. Accuracy and repeatability can still be achieved with a profile just like it can with Steel, that's down to how it's built the design and components used.

Good luck if you take this route and if I can help just ask.

[Andrewg]

Ah Kitwin, the siren call of ‘just make it a bit bigger’! But I take your point, extra Y length does not change the design or performance significantly.

And thanks for the appreciation Brian, I had hoped this would be of value to others. However others have already steered you away from gantry routers for steel

So where has Jazz left me:

Budget and spec is realistic (Hurrah) but time under significantly! I can live with the greater time commitment, though I expect I will be tempted into stretching the budget when it comes to the motors.and electronics. Definitely worth press on then:

Frame
Wood and plastic cutting router, so aluminum extrusion frame keeps things simpler for me.(avoids welding and steel work) Plates I will definitely be looking to cut/drill my own so will want to keep design simple.

It's also very very important how it's all fastened together and plates etc. Done correctly it's very strong and perfect for wood routers.
.

Sure I will need to come back to what joining methods for extrusions are best once I have a first stab at the design.

Mechanics
Ballscrews all round, but only 3
Single Y ballscrew will save a bit of budget, however I do need to cut ends of boards set vertically. With an under-table ballscrew and cross brace, I will needs to have the spindle sitting forward of the bearings enough to do this. I only need 65mm of cut distance beyond table. Is that a sensible design option?

Gantry

However, it's important that you have a piece of material which is a nice tight fit that goes into the slots between upper and lower profile.

I get the material in the slots to lock the two extrusions together, but see that 10mm aluminium bar is not a standard size so what to use? Options seem to be either machine a ½” aluminum bar down to 10mm slot size on a router table or forget the slot filler and have multiple bolts. By creating a gantry length ‘washer’ with regular clearance holes along it, I could use it as a guide to drill the access holes in the lower extrusion, then install bolts in it to mate with slot nuts in the upper extrusion, so they are bolted every 50mm or so. Avoids having to machine material to tolerance?

Electrics
Tempted by closed loop steppers - looks like for 2nm nema23 steppers it is about £40 stepper plus £20 driver vs £80 for closed loop motor and driver, so maybe extra £20 per axis well spent. Where will I see the benefit besides getting error rather than inaccuracy due to missed steps?


Electronics
There are enough challenged already, UNCNC it is.

Design is VERY VERY important and if done with a forward-thinking attitude then you can have a great machine for not a lot of money.

But still all down to the design, so guess I need to fire up Fusion360 and start designing. Plan to start from the Z axis and work out axis by axis. The Z seems a simple matter of sandwiching 15mm linear rails and 1605 ballscrew between two plates and getting the heights and clearances right...we shall see.

[JAZZCNC]

Mechanics
Ballscrews all round, but only 3
Single Y ballscrew will save a bit of budget, however I do need to cut ends of boards set vertically. With an under-table ballscrew and cross brace, I will needs to have the spindle sitting forward of the bearings enough to do this. I only need 65mm of cut distance beyond table. Is that a sensible design option?

Single or double ball screws it doesn't make any difference to if you can overhang or not. This is down to gantry design.
Many will sweep the gantry beam back from the centre of the bearing plates to recover lost tablespace. However, in your case this can't be done because you need the spindle to project past the end of the bed so your better with the gantry beam inline with the bearing plates.

There are other design methods, like having the long axis frame rails extend past the end of the bed. In this way you can have both methods and have any overhang amount you like(within reason). The cost is it takes up more physical room and slightly longer rails which cost more money.
I've used this design on several machines so that a 4th axis can be used on the end for turning and fluting etc. The same can be done with extending the gantry at one side and using the length of machine for longer work. (See pics)

Gantry

I get the material in the slots to lock the two extrusions together, but see that 10mm aluminium bar is not a standard size so what to use? Options seem to be either machine a ½” aluminum bar down to 10mm slot size on a router table or forget the slot filler and have multiple bolts. By creating a gantry length ‘washer’ with regular clearance holes along it, I could use it as a guide to drill the access holes in the lower extrusion, then install bolts in it to mate with slot nuts in the upper extrusion, so they are bolted every 50mm or so. Avoids having to machine material to tolerance?

Your way way over thinking this.!! . . . For machine this size it's not required. To be honest at this size you would get away with NO bolts. If not using the spacer method then 2 x M10 bolts spaced evenly along length and tapped into the upper profile will be more than enough.

Electrics
Tempted by closed loop steppers - looks like for 2nm nema23 steppers it is about £40 stepper plus £20 driver vs £80 for closed loop motor and driver, so maybe extra £20 per axis well spent. Where will I see the benefit besides getting error rather than inaccuracy due to missed steps?

Depending on which you get then the main differences will be smoothness and speed/torque.

The smaller closed-loop motors (<=3nm) all tend to be 3 phase and the motors use 0.9Deg step amount or 1.2 rather than 1.8deg found on typical hybrid steppers. This gives a smoother action but requires more pulses from the controller to get the same speed. But controllers like AXBB or UC300 etc have more than enough frequency so this isn't an issue.

Being 3phase they provide more torque and allow higher rpm's. Typical RPM where torque starts to drops away is 1000-1100rpm for hybrid stepper motor. With closed loop it's around 1500Rpm, but this does depend on other factors like quality and voltage etc. But in genral they are faster and stronger.

Larger (=> 4Nm) Closed-loop motors can be found either in 3phase or two phase versions.

But still all down to the design, so guess I need to fire up Fusion360 and start designing. Plan to start from the Z-axis and work out axis by axis. The Z seems a simple matter of sandwiching 15mm linear rails and 1605 ball screw between two plates and getting the heights and clearances right...we shall see.

Don't use 15mm for the Z-axis as it makes more difficult to build regards clearances and they are fiddly. 20mm cost very little extra and make for a much stronger and easier to build Z-axis.
Whatever you do don't cut corners on the Z-axis, it's THE one area you don't want to get wrong because it holds the tools so if it flexes or vibrates this will be seen in your work.!


Below are 2 machine designs, both can be single or twin screws and placed under the bed for protection. The larger one will extend past the end of the bed and the side, it also adjustable length 4th axis.

Edit: also a larger version with extended side rails for more front extension.

[Andrewg]

Your way way over thinking this.!! . . .

That is what I need - Keep me focused on the real issues


Yes this is what I aspire to:Neat, clean, capable looking machine!
From your previous posts I am assuming I am looking at
- 45x90 extrusions on Y axis with two for the L gantry.
- 12mm Plate for ends and gantry sides
- 20mm linear rails all axis
- 1610 ballscrew, 1605 on Z driven via belt and pulleys so steppers can be tucked out of sight, 1: 1 gearing? is it worth added parts?

A little worried about the X/Z carriage getting complicated. If rails on front face and ballscrew on top, then couldn't it be a single plate again?

I am assuming this design cannot cut beyond the table, but by removing the gantry sweep back and maybe mounting the rails on the face of the gantry (and removing the control panel) I could create sufficient over run.

The smaller closed-loop motors (<=3nm) all tend to be 3 phase

The ones I was looking at for price were like these appear to be just steppers with encoder and therefore not 3phase though these look similar but are described as 3phase and are £40 more so guess your advantages of 3phase closed loop will cost a little more. Know little about steppers besides frame size (NEMA) and torque (Nm) plus a vague idea that inductance needs to be worried about.

Don't use 15mm for the Z-axis as it makes more difficult to build regards clearances and they are fiddly. 20mm cost very little extra and make for a much stronger and easier to build Z-axis.

Again feedback from real experience appreciated. How much should I be worried about keeping Z axis weight down? I was assuming that on a small machine, I want to keep the weight being flung about to a minimum to ensure good acceleration and performance whilst allowing modest (lower cost) motors and drivers to be spec-ed.

Working on a Z axis, need to upgrade the rails to 20mm!

[Kitwn]

Andrew,
Sounds like you've done your research (Reading MYCNCUK with a pack of HobNobs and a large pot of tea?), I'm looking forward to reading the build log. Jazz will love you for not having actually bought any bits before asking for advice
My only suggestion would be to make the bed a bit longer if you have the room for it. My own machine can cut a little over 600 x 900. It doesn't affect the gantry in any way and I've never had a project that needs all that area at once yet, but a bit of extra length has proved necessary on occasions. My woodwork projects seem to regularly require parts that are narrow but a bit longer than a 600mm square machine could cut in one. Mostly accessories for my wife's weaving hobby.

[BDH]

Thanks Andrew & Jazz for the very useful questions and answers, they have helped me to understand DIY CNC a lot better although my 'need' is for a smaller machine but one that will tackle steel.
Brian

[JAZZCNC]

Thanks Andrew & Jazz for the very useful questions and answers, they have helped me to understand DIY CNC a lot better although my 'need' is for a smaller machine but one that will tackle steel.
Brian

If you want to cut steel and the parts will fit then buy a Milling machine is my advice. Building a machine to cut steel is at a different level to one for wood/aluminum.

The principles of the building process are the same but machine design, materials, and components are very different.