View Full Version : BUILD LOG: Sweet dreams CNC - more than one step beyond...

28-07-2014, 01:51 AM
Having exhausted the 80's music metaphors, (ok, you might spot a few more) I thought I'd get around to posting some details and images of my CNC build "Sweet Dreams" now its finished, (does it ever finish??). Probably easier to start at the end and work backwards to the beginning...

[edit]Change of plan... I have a large box of papers, drawings, research and notes, (there were over 900 revisions to the main CAD!) as I come across the info or the images I'll post them. You may see some conflicting reference to past, present and future tense, but eventually it will all be in past tense! Apologies.. this might get a little random! lol [edit]

I'll start with the design specification, but here's a couple of teaser pics until I get the customary video uploaded.:sneakiness:


In front of the Rose window at the ally Pally.
If you were at the London Model Engineering Exhibition at Alexander Palace back in January then you've already seen and heard it working! :-) Great to talk to so many knowledgable people and thanks for all the kind words.

You may notice two Echains feeding the same location on the Z axis? One has data, the other power. This principal is applied to the whole machine, data down one side, power the other. (Echain is cheaper than finding interference!)


Table, height adjustable and can be swapped out completely for 4th axis. 40mm box steel, weighs about 70kg without sand.

some boring soldering... more to come on that... :)

The electrics section.

Blue chosen as it matches my eyes! lol. The electrical "box" is er, unusual! The rear of the machine, has a base board across its entire size with hinges at the bottom (to the right in the above image), the view above is in the open position. Laid out horizontally with all the components easily accessible. In the closed position, airflow moves from the bottom of the box and out through the top. (right to left in the pic). aided by 12VDC 3 inch fans positioned by the transformers.

The base board is 22mm Buffalo board with 3mm alloy sheet bonded and screwed to it. This alloy sheet acts as the ground plane for all circuits and helps simplify the prevention of the dreaded earth loop. The hinges are positioned such that in the closed position, the base board, with all the components bolted to it is slightly over centre so gravity holds it closed. The alloy base board ground plane is electrically connected to the main frame.

From top left heading downwards;
(just out of picture) mains in, massive mains filter, 4xMCB, mains distribution rail in 150mm conduit.
(in picture)

Transformer cooling fan low voltage distribution
2 x 5VDC transformers
12VDC transformer
24VAC transformer for relay cluster and high voltage distribution
toroid transformer, rectifier and smoothing caps.
Moving right in the image
central conduit (75mm)
right of centre conduit
Power to machine front (round conduit)
secondary distribution board
data feed front of machine (round conduit)
data feed to Z Axis (round conduit)
Relay cluster including hot swap spares, (to Omcrom micro switches, Estop and height gauge)
power to Z axis (round conduit)

8A drivers far right.

All cables are numbered for easy FF. Thankfully, apart from a microswitch wired to a spare relay, no issues with cabling.

That brought a smile (mainly of relief!)

Warning; >Digression alert<
The evening came when it was all wired... circuits tested and retested, but no power yet applied. A quick phone call to a sparks friend saw the next morning arrive with a body who knew nothing of the build and the wiring. His job was to act as a blind test and verify all was as the wiring diagram said it should be. A couple of hours later and all the circuits are given a clean bill of health. It's time to turn the power on... or we could check it again... lol. It gets plugged in (all switches off) as the first switch is turned on (12v feed to pump, fans, misting system, work lights) fire extinguisher in hand, there is a deathly silence while we search for glowing wires, the faint whiff of the aroma of the magic smoke escaping... after what seemed like hours, we turned it off and went to the pub to calm down, walking like an egyptian perhaps?

The cable runs use Igus echain but some, eg linking the rear base board to the 5 screens at the front are run through the frame. Prior to assembly, the frame members were machined to accept lengths of 22mm copper pipe. These are electrically bonded to the frame and act as conduits for the cables and keep the sand off the wiring.

28-07-2014, 04:06 PM
Ah ah Sweetdream I like that mate. . . Lol

28-07-2014, 11:49 PM
Design spec;

Seems like a long time ago I asked the question "what do I want to cut with it?" "Don't know yet" was the unhelpful conclusion! But after much thought about what I needed, opposed to what I wanted. I came up with a basic spec; a very strong, fast CNC that could work anything I threw at it including metals and portable enough to fit through a standard door.

This CNC was not going to be an exercise in what was theoretically possible, just a no frills, very strong, quick and easy to build structure that a tank could drive over! It had to be built without recourse to another CNC or exotic measuring equipment, just the equipment I had, could borrow or was prepared to buy. I did have some milling done on the frame, (making the 26x 22mm holes for the conduits, we had a jig on the mill and just knocked them out but that could as easily have been done with a pillar drill but would have taken a lot longer. Apart from that milling job, it was built with power tools and a pillar drill.

A few additional criteria added during the build, the spec stayed the same and was met. It fits through a fairly standard door frame too!

For the frame, I went for a PVC thick wall tubing with copper plumbing fittings. Skate bearings and ACME thread and I found a couple of old plastic pallets to use as the bed too.. LOL, only joking, just checking you're still awake. I realised fairly early on that for the performance I was wanting, a hybrid of bolted and welded steel construction and profiled rail were the right option. So large quantities of 80mm thick wall box section steel were calculated. The Hiwin catalog and price lists took a beating too.

Other staples were the standard 2.4kw water cooled spindle, NEMA 34, 8amp drivers, Omcron switches, nothing unusual except the later addition of a 250 rpm single phase spindle to deal with any low speed spindle requirements, mounted parallel but alongside the main spindle.

Although I had a good idea of what I wanted to achieve and some ideas I'd not seen used before, when I started I did not know all the questions that I'd need to answer before the build was finished. For the benefit of anyone reading this before building their own CNC thats a small but important point. You may not know all the questions you'll need to answer. :-)

But of course, the devil is in the detail and how you put it together.

So first build a workshop! lol..


or convert an old shed. After repairing the roof [picture coming] putting in a concrete floor, [find pics] some storage arrives courtesy of hearing about them being thrown out!

The wooden carcasses are solid but the plastic trays look a little fragile. Might swap them out for wooden ones as and when they break.

I've ended up with the back wall of the shop, floor to ceiling with these bins and a bench cantilevered out between them at a suitable height for standing at the bench. I had enough left to put them on one wall from head high to the ceiling giving a total of about 200 bays, 17inches deep, with either 3 inch or 6 inch high trays. Some I put shelves in as seen in the Hammer store below.

The storage starts filing up...

All workshops should have a good selection of tools!


In case of working late, some refreshments may be required! :-) [update] the beer has long gone! There's half a bottle of gin, 5 litres of cider and some fruit based cordials in case of visitors.. :-)


29-07-2014, 08:18 AM
Good way to start, prioritize refreshments. Well done sweettooth!!!- I mean sweetdream. G.

29-07-2014, 09:11 AM
Liking the cut of your jib...looking forward to seeing details of the build and the machine...

29-07-2014, 10:46 AM
I think any man with a beer fridge in his workshop is going to fit in well around here :highly_amused:

Although chilling Spitfire is not an approach I would advocate ;)

29-07-2014, 05:51 PM
Good way to start, prioritize refreshments. Well done sweettooth!!!- I mean sweetdream. G.

haha, closer than you think! the ice crusher attachment for the CNC is almost ready to test!

29-07-2014, 05:53 PM
Liking the cut of your jib...looking forward to seeing details of the build and the machine...

I'll get there eventually! I had a go pro running in the workshop during the build, taking a shot every 5 secs. should make an interesting time lapse video if I ever get the thousands (and thousands) of shots stitched together!

29-07-2014, 06:42 PM
Ah ah Sweetdream I like that mate. . . Lol

Thank you. :-)

29-07-2014, 09:31 PM
I think any man with a beer fridge in his workshop is going to fit in well around here :highly_amused:

Although chilling Spitfire is not an approach I would advocate ;)

I thought the fridge was a requirement for joining this illustrious gathering??

There's got to be a joke about spitfire and cold in the spirit of the other ads.. just can't think of it..

29-07-2014, 11:00 PM
Spindle cooling;
There's lots of ways to do this, here's my version.

Design spec was for a compact autonomous cooling system. I figured that when it was running, I'd have enough to worry about so anything that could be automated, I'd automate.

18DS20 sensors at the input to the spindle and output of the radiator send the temp data via a one wire system to the Arduino Mega that is looking after the cooling. That looks at the difference in temperature and either turns the radiator fans on or off and feeds the current status to a small LCD screen.


phone pic, sorry. :(
Pump on, fans on, temp in 18c temp out 17c. There's not a lot of heat generated by the spindle, haven't seen those numbers move much.

I had done a bench test with the radiator in free air and hot water in it, turn on the fans and watch the temperature drop over a set period of time. I then repeated the test but with the radiator bolted 5mm away from a sheet of 10mm alloy. The temperature of the alloy went up by one degree, but surprisingly there was not a difference that I could see over the average of 5 of each configuration in the speed the water was cooled. Reversing the direction of the airflow made no difference.

The radiator is bolted to the back of the Z axis with rubber spacers and the fans pull cool air through the 5mm gap and venting to the rear of the Z. Foam could be used as an air filter between the radiator and the back of the Z if required but there is, as yet, no sign of this being an issue despite routing some ply down from 22mm to 6mm.



image above is a the reverse of holder for a remote monitor and e-stop, allowing me to keep an eye on operations from the comfort of an easy chair. ;-)

The pump and reservoir are mounted above the Z axis, reinforced hose connects it all together.

The arduino also rotates the spindle work lights LED sequence (like a circular white cylon type pattern?) while the spindle and axis are still, so it can be seen at a glance that the pump is working ok. The work light is 3.3vdc, and minimal amps. I've got it through a relay but this will be revisited and modified at some point. My initial concerns that zeroing the tool on the workpiece without a stationary light were unfounded, its actually easier to see the tool tip with the light source rotating around the tool.

The cooling setup is a modified Thermaltake PC water-cooling unit, works really well.


The 3 fasteners above remind me that anywhere I had a horizontal fasteners I'd use 3 of them but vertical fasteners were used in 4's. The Z was fairly tight on space, as per the design spec, so this strategy kept them apart.

I'm posting this as I come across it but if theres' anything specific just ask. :-)

29-07-2014, 11:59 PM
I think any man with a beer fridge in his workshop is going to fit in well around here

Does chocolate in the 'fridge' at -70C count? :onthego:

30-07-2014, 09:16 PM
Does chocolate in the 'fridge' at -70C count? :onthego:

..chocolate counts only if you can eat it!

31-07-2014, 05:28 PM
As mentioned the frame was initially going to be 80mm box section, measuring the steppers showed that by careful arrangement, the frame could act as the stepper motor mounts, saving the costs of dedicated mounts [pic] this worked rather well on the X and Y axis but as there was no box section on the Z, it couldn't be done there. The Z axis had its own design spec, small, as compact as physically possible and strong, very strong.. The Z stepper mount needed to be adjustable in one axis (same as the X as it happened) but solid in every other. 4 pieces of 10mm alloy sorted that.

Getting the spindle as close to the gantry as possible, mainly to reduce the leverage effect from the spindle under load, was another area I spent some time on. It's ended up 140mm from the centre of the spindle to the centre of the Y axis and the Z axis is 144mm wide! This minimises the amount of table space lost to the spindle. The only parts on the table, profiled rail, and ball screws that have to be there but cannot be used by the Spindle when the Z axis is 1mm off the stop switches is 146mm on the Y Axis, 73mm each side.

It should also be remembered that smaller chunks of 20mm ECO plate are cheaper than Large chunks of 20mm ECO plate! :-)


The height [?] is where the space was found to mount the cooling system pump / reservoir and the Z axis stepper pulley arrangements. A couple of junction boxes take care of the wiring and act as an anchor for the twin runs of echain.

Very happy with the result, the Z axis stays within the confines of the 4 Y axis trucks wherever it moves, the Horns do their job and the 5mm thick steel triangulated plates that were shown not to be necessary on the simulation, but for the sake of 30mins welding, "over engineering" was preferred to "always wondering". lol.

02-08-2014, 11:02 PM
I mentioned earlier about the 'boring soldering'! One of the things I anticipated happening when I started running a new untested machine, at least for the first few times, was that you needed eyes on a swivel! Even by phasing the testing to only specific circuits, there comes a point where you have more new bits running than you have eyes to watch them! The automation of voltage and temperature checking etc. meant that I could focus on stopping the spindle diving into the sacrificial bed or any other undesired behaviours.

The wiring shown is for 5 LCD screens mounted into a console at the front of the CNC. The screens show various information about the machine without have to go look for it or fret about it. This is in addition to the touch screen mounted onto the frame, but facing away form the swarf! :-)

The Blue LED section in the wiring pic (above) alerts you by turning red if anything goes outside the parameters you set. For the startup tests, this was set somewhat conservatively! :-) During the tuning phase these parameters were relaxed somewhat as the machine was dialled in.

The screens are augmented by override buttons, for instance, capable of turning fans on if they are off, or off if they are on. They are fed from 2 CaT5 cables, one from the Z one from the rear wiring.


03-08-2014, 10:58 AM

04-08-2014, 01:06 AM
The suicide test;

So the machine is running without smoke pouring forth, all the bells are ringing and the whistles whistling.. (why don't bells belling?) spindle has been powered up and a couple of small tests run. so still air cutting at this point. ...Commissioning continues.

Time to up the ante.

Bearing in mind that body parts may well be found in the same vicinity as gantries and other moving parts, cue the suicide test.

The suicide test
Set gantry at one end of the X axis.
Set speed in excess of 30 metres per minute.
Hold down the button and drive the gantry at the highest speed you can get into the stops at the other end of the X axis!
Wait for the very loud bang!

Repair as required

I hit the switch and there was a whirr and a blur as the gantry shot from one end to the other in what seemed like well under a second. There wasn't time to take my finger off the button before the limit switches did their job. I don't have an exact figure of the speed it reached but I can tell you that the gantry was still accelerating when it crashed into the the Omcron microswitches at the far end of the profiled rail. lol.

The switches have 3mm of movement, .1mm between open and closed. Assuming 4mm of total travel, the speed of the gantry when it hit the switch can be calculated from the time taken between the switch being activated and the distance traveled before stopping. The speed of light and the price of mars bars being unequal, the gantry was travelling in excess of a "greased lightning" when it got to where it was going!

Confidence went much higher after that test. The limit switches worked great, even in the worst case scenario of a runaway gantry, the worst that I'd expect to see was a crushed switch as it didn't get as far as hitting the metal stops.

If you are building a machine, fit the limits and estops before you wire the motors. Watching a gantry move at high speed is really not something you want to see from the wrong place. If your gantry moves as quick as this one did, you'll never make it to the power switch in time.

04-08-2014, 11:34 PM
The CNC is primarily steel box section,welded AND bolted! There are some new thoughts on the design, assembly and set up. I've not followed the classic route for a gantry CNC, I believe that there are major improvements that can be made to the 'normal' design which improve strength and accuracy both on the Y and Z axis. The X axis is probably the easiest one to strengthen.
On the Y and Z axis, the ball screws sit between the profiled rails, with the trucks outside the rails of the adjacent axis.

The Z axis is 20 mm ECO plate aluminium, built into a small but very strong arrangement. We had a large 4x4 parked on it at one point without any effect.

19-08-2014, 01:49 AM
The cooling system is mounted onto the Z axis and all contained on it. Standard arrangement to the spindle but one side fed from the 12v pump which gets the coolant from the clear reservoir. the return side heads to the twin fan cooled radiator and the radiator feeds the coolant back into the reservoir via a bubble trap. a couple of chunks of styrofoam give a visual that the coolant is moving as it should. The radiator and fans are mounted on the back of the Z axis, making it a very compact solution. total length of the pipe work is less than 4 feet.

Other bits about the Z, the trucks are mounted top and bottom, side by side, giving maximum load carrying in each direction that the forces will be applied. Ballnut is through the centre and in-between the line of the trucks.

19-08-2014, 08:19 AM
This thread definitely needs more pictures. :whistle:

16-10-2014, 12:25 AM
Phew... been a busy couple of months there..

If anyone is around the Lemington Spa area, the beast (must be 400kgs... F.heavy!) is on display and making large bits smaller at the model engineering exhibition from Thursday through Sunday, 16th to 19th October.

If you are going, come and say hello.. Full tours available! :-)

Forgot to mention there is at least one other CNC there, CNC lathe and I understand some 3d printers and laser cutters too... During a brief wander around earlier I also spotted some reasonably priced end mills and nice chunks of alloy I shall be revisiting. :-)

16-10-2014, 07:50 PM
Hi sweetdream, thanks for mentioning exhibition at LS. I have just booked tickets on line for tomorrow, so will probably see you there. G.

Clive S
16-10-2014, 09:52 PM
+ one ..Clive

17-10-2014, 01:08 AM
hi guys, it was heaving in there today... you couldn't get near some of the stuff on show. They were 3 deep most of the day!

I did meet a couple of lurkers from here, (you know who you are! lol...) They'd read this thread and were going anyway. Thanks for stopping by and saying hello, really nice to chat and thank you for the kind words gents.

As always, there's lots of interesting people who want to talk CNC stop by and one or two luddites who don't get it. One of the more eclectic conversations this morning was with an old boy who recounted that when he was an apprentice, his boss thought that no apprentice should be allowed to use "powered" machinery until they had done 6 months on a treadle powered one! LOL, brilliant....

Clive, Geoffrey, look forward to meeting you.

22-10-2014, 01:40 AM
I had intended to add something to this log after each day at the show, but lack of sleep and a duff Mac software update killed that idea! (don't load the 10.9.5 Mavericks update, it killed my machine sufficiently to provoke reinstalling OSX!)

Back to the show;
A HUGE thank you to Graeme, an extremely talented and helpful guy who was kind enough to spend a lot of his time helping me on Saturday, (I should have that item and the plans in the post to you on Wednesday).

Also to Tim, who got collared into manning the stand with me and hung around till the end of Sunday to help me load the beast back in the car while explaining the vulgarities of making casement windows on the CNC! lol. Big, big thanks to you both!

Met up with Clive and Geoffrey and got nothing done while we chatted. lol. Geoffrey that is an incredible bit of work you did, you should put the pics up! Clive, thanks for passing on the message. Great to meet you both and glad you enjoyed the day.

I was surprised at how many Lurkers came to say hello, there must have been 30 or so. That's only the ones who mentioned seeing this build log! I heard about some great ideas and some amazing gear but was surprised to hear the reasons why they were lurking rather than posting. There was a surprising number of team builds too, where a couple of people got together to build a CNC.

Rob and Rob... thanks for the kind words, yes, it is a beast of a design. :-) I like the principal of building it stronger than the hardest material you want to cut. Yours sounds fine for doing wood but throw a pic or two up here and get the "illuminati" to take a look if you still have concerns.

There were a couple of commercial CNC designers I spoke to who were also building their own machines, but even if its your first machine, there's lots of ways you can help or inspire others to build theirs.

Sweet Dreams seemed to act as the focal point for anyone with an interest in CNC at the show and what I found nice was seeing groups of people by the stand chatting to each other about their own CNC or their plans for one. Thankfully(?) no one came up with any improvements to the design of Sweet Dreams, so I don't have to build a second one! lol.. Thanks to everyone who took the time to come and say hello and chat about their CNC or ask about Sweet dreams. I had a great time at the show, learned from some generous and amazing people and had conversations with experts about subjects I knew little about!

Going to one of these shows is highly recommended if you want to talk engineering and see some incredible work. Very inspiring.

Here's some images of workmanship to drool over;

Image 1: musical box!
Image 2: is the modified 3020 from Society of mechanical and Experimental Engineers, running on a Raspberry pi!
Image 3:scratch build radial aero engine...

22-10-2014, 09:45 AM
I heard about some great ideas and some amazing gear but was surprised to hear the reasons why they were lurking rather than posting.

Come on then enlighten us to why.?

Sorry couldn't make it like I hoped to but just too busy and after speaking with Clive about the journey(4hrs to get home) that just doomed any chance.!!
Sounds like the space ship took off well again and we will meet up one day.! Don't know where, don't when but I know we'll meet again.!! . . .Opp's went War time on ya then. . Lol
Coming down your way in not too distant future so will drop you a line and maybe meet up then.?

22-10-2014, 10:54 AM
I heard about some great ideas and some amazing gear but was surprised to hear the reasons why they were lurking rather than posting.

That's a real shame. I'm sure the complete novices on here, myself included, would value more feedback on our posts. So come on you lurkers, I'm almost certain if you all became active members the forum would be a much richer place for it. Don't be shy now. :whistle:

22-10-2014, 12:17 PM
Hi Sweetdreams,

I'd missed this post so was surprised to come across your machine at the Engineering Show. Your colleague gave me an overview, plus mentioned that it was out of action following a suspected break out board failure. Hope you get that sorted soon.

Machine looked good anyway.


22-10-2014, 03:29 PM
@ RouterCNC
Sorry I missed you! We either caught the cable by squeezing past it, (it was tight around the back of the stand), or might even have been pulling the dust covers off it but somehow managed to rip the connection socket out of the board.. works great in the workshop but shifting it in and out of cars and around at shows, it needs to be a bit more resilient to hanger rash. :-)

23-10-2014, 01:10 PM
That's a real shame. I'm sure the complete novices on here, myself included, would value more feedback on our posts. So come on you lurkers, I'm almost certain if you all became active members the forum would be a much richer place for it. Don't be shy now. :whistle:

It's an interesting point. The problem might be that there are a few very experienced machine builders who contribute, and who can talk from their experience. There are others with deep theoretical knowledge (and some practical skills as well, but that's not the point) who can offer a different insight into problems or questions. It's worth a beginner listening to all of them. Then there is a small number (fortunately very small, and possibly near zero at the moment) who claim to know a lot but it's just what they've read somewhere else. Usually not adding much to the discussion and sometimes just misleading or wrong. Then there's the other 98% of us. So, specific example - I built a CNC router to the JGRO design 2-3 years ago. Hadn't found this forum at that time, and it seemed like a good idea. Taught me a lot, mainly how not to do it, but little of that is worth repeating. I did electronic engineering at university, and can find my way around the problems of interconnecting the various bits of electronics needed to make a machine work. I understand the difference between n/o and n/c proximity switches, for example. But with my Mk1 machine, and the Mk2 currently being built. I only have experience of one particular kind of PC/BOB/driver, etc, combination. Ditto mechanical engineering - I've been a model engineer for nearly 40 years, with a decent workshop and set of machines and tools. I've worked in the IT business for most of my working life (and now retired) but have only used LinuxCNC and a limited range of CAD and CAM software. So, as happened a little while back, I was able to help someone who lives nearby configure and make work a machine he had bought (including learning Mach3 basics during our tea break!) but that's not anything like the range of experience that others have.

So, where I can speak specifically from my own certain knowledge and/or experience, I might pop in a comment, but frankly I feel that there are others better qualified in almost any topic. And that may be why there are so many lurkers. Not forgetting, of course, that any attempt to stick your head over the parapet and give an opinion could result in getting your head blown off! It's only too easy to write something that was meant to be polite and innocuous but puts someone's back up quite inadvertently. I'm sure that this post will upset someone, somewhere, even though I have no desire to do so. You need a hide like a rhinoceros to participate in online discussions...

I'm curious whether my thinking out loud mirrors any of the comments received at the exhibition referred to.

23-10-2014, 01:49 PM
thanks Neale, some interesting thoughts there. Far too rational and considered to be on a public form! :-) yes, It's not only technical skill that varies, communications skills vary too.

This topic might be of interest to posters and lurkers alike, although perhaps in its own thread rather than hidden in a build log?

23-10-2014, 02:12 PM
Well said Neale. (Even thou some of the negative could be aimed at me. . lol)

Everyone lurks to some degree. I did and still do on other none CNC forums where I deem my knowledge of little help.
Has can be seen here I'm not afraid to stick my head up and I'm a crack shot at blowing heads off if I feel the need. What stops me from posting on other forums is that unless I'm 100% sure and preferably thru experience of what I'm talking about then I don't comment as I know from experience with helping with CNC that others who don't EVER speak but at same time grab onto any info provided. This speaking without full knowledge or first hand experience can and does send people down the wrong path, even when not meant to do so, but still it does and I'm aware of it so very careful not to give diffinative answers and I make it very clear if I'm not 100% sure.
If others did the same then there be less head butting I'm sure.!! The theorist's and "know it all but never done it" types you'll never escape and IMO they deserve blasting if they comment with authority having never done it and sending people wrong.!!

Sweety pie dreams I'm still curious to why.? . . :eagerness:

23-10-2014, 03:51 PM
...Even thou some of the negative could be aimed at me...

nah, you're from yorkshire, you're excused! lol.


23-10-2014, 04:08 PM
nah, you're from yorkshire, you're excused! lol.


Ahh but it's the Irish in me that brings out the Devil. .:beer: