View Full Version : BUILD LOG: Small workshop CNC router, steel base, planning on ATC

29-12-2016, 09:38 PM

This is not so much a build thread as the machine is mostly complete and operational, but i wanted to share and document any further development. There are so many people over here with much more experience than me, so i'm open to criticism, suggestions, anything.

A bit of backstory, i started with a chinese 3040 router frame, which was mostly horrible. I started upgrading and reinforcing it here and there but soon realized that it was a lost cause, so i started building a machine from scratch. Goal was to have a machine rigid enough to mill aluminium somewhat reliably, but since i'm a student currently without a job, it had to be as budget minded as possible.

First thought was to make the entire frame out of 80x80 aluminium extrusions, but that thought quickly went away due to cost and the fact that i somehow got it in my head that it wouldn't be rigid enough. So i made a frame from 10mm mild steel plates with a 12mm mild steel top, with holes for mounting a vice or an aluminium t-slot table. The base ended up weighing about 50kg alone. Complete machine comes about 75kg.

Then i adapted the gantry i had made for my previous machine. Double ball screws for the x-axis, 16mm ball screws and 16mm supported shafts all round, coupled with 270oz stepper motors. I'd love to have gone with hiwin rails or something like that, but it was more than double the cost and i really couldn't afford it. I really regret it now though. Spindle is a generic 2.2kw chinese spindle. Travel ended up being 550mm x 250mm x 125mm.

Now at that point i intended to get the steel frame stress relieved and the top skimmed but the machine shop wouldn't be available for a few weeks, so i couldn't hold myself from assembling and testing it. First impressions are really satisfactory. There is a bit of ringing from the steel frame, but much less than expected. I'll probably be filling the base with epoxy granite, or something to that effect. I've made a couple of samples (like the carburetor mounts you see in the pics), and i'm quite pleased with the result. These were made with a 2 flute 8mm carbide end mill on ~12k rpms, with a 6.8mm depth of cut and 0.7mm width of cut and 1600mm/min feed rate with no chatter. That's not the limit, it's just the first combination i happened to use randomly, and worked good.

Next step is to disassemble the whole thing, get the base relieved and skimmed, put it back together and square everything up for good. I'll then fabricate some sort of enclosure, because i intend to use both mist cooling and flood cooling. I haven't made up my mind if it'lll just be a plain polycarbonate enclosure, so that the flood cooling won't be making a mess, or if it will be a substantially larger double-walled wooden soundproof cabinet. Also, i played around a bit with the VFD's torque boost function, which does make a difference and will definitely be of use, but the spindle gets really hot really quickly, so a proper water cooling system is in order, instead of just a bucket with a recirculating pump.

I'd love to hear your thoughts! Excuse the crappy quality pics, i'll be posting better ones.


30-12-2016, 10:53 AM
Hi Steve,

Nothing much to comment on really other than well done looks nice strong machine. Good start for first build make it earn and you'll soon have enough money for Mk2 which can have few more bells and whistles like profiled rails.

Only comment regarding spindle cooling is that if bucket of water isn't cooling it then probably pushing settings little too hard. I've run my spindle for BIG hours with 20ltr bucket of water cutting aluminium and never gets much above warm.
I've experimented with lots of setups and found that for short cycle times in 2-3hr range then just PC rad and 1ltr of water works fine. On machines I build fit aluminium tank on the back with pump inside which holds aprox 3-4ltr water which is good for 99% of users needs.(Those in UK) If need more cooling then PC Rad & Fan can be fitted on outside but never needed to fit one yet.! (see pic)
With you being in Hot country then maybe worth running it thru small radiator with Fan on. Probably need something little larger than PC rad but small Scooter rad would work.


30-12-2016, 01:01 PM

Thanks for your comments. If there ever will be a mk2 it will definitely be a vertical mill type machine. Maybe I didn't make myself clear on the cooling. Under normal conditions, it barely even gets warmer than room temp, like you say. It only got much hotter when I decided to test the VFD's torque boost function, which effectively alters the V/F curve. I had it as low as 800rpm , and torque boost seemed to make a difference, but I didn't try cutting anything yet.

And now, onto the ATC portion of the thread. It really bugs me to have to change tools manually, it's time consuming, it's a pain to zero every tool, even with the auto tool zero function, and most of the times doing all that takes up more time than the actual operation itself. I've looked into ATC spindles (since the generic chinese ones can't be retrofitted with any ATC system), but first of all the cost is way high(1.5-3k ), but most importantly there is no real reports from people actually using them, so I'm not going to take that risk.

I then had an idea based on the people that convert the X2 machines to cnc and ATC. The idea is very simple, so simple in fact that there must be something massively obviously wrong with it and I still can't see it. Basically I want to make a new head into which i'll plant an R8 spindle (http://littlemachineshop.com/products/product_view.php?ProductID=1407&category=6), seated on a couple of angular contact bearings. The R8 spindle will be mated to a two step pulley that will be driven by the 2.2kw chinese spindle. My intention is not to exceed 12k rpm (or thereabouts) on the chinese spindle, the noise is tolerable on that rpm level and it puts out decent power. Pulley ratios will be 1:1.66 on the high gearing and 1:3.75 on the low gearing, so I can get about 8k rpm on the high gearing and 3k rpm on the low gearing. Belt will probably be a poly-v 4 rib belt, since I can easily make the pulleys myself on the mini-lathe and belt speed will be around 20m/s max, which is well within what they can take. Here is a quick mockup of that


One consideration is weight, I've no idea how much weight the gantry could support without deflecting. I've been thinking to make it out of 7075 aluminium (head with spindle will weigh about 5.5kg). People tend not to prefer using aluminium, partly because it will expand a bit more than the bearing in a given temperature rise (I seem to recall reading somewhere it is 1 thou more over 40 degree temp rise). That could very easily be rectified with watercooling for the head, since there is already a system in place for the spindle.

I mentioned deflection earlier, I did a simple measurement. I put a dial gauge upside down on the outer lower face of the spindle body. Then I rested a 5kg weight on top of the stepper motor. Gauge moved down 0.01mm. I then hang the same weight in front of the spindle and gauge moved down 0.07mm. Open ended bearings have inherent weaknesses, and I think most of that slack comes from the Z-axis. I need more travel in the z-axis anyway, so i'll be replacing the rails on the Z with profiled ones. Oh, and the stepper motor (nema23 270oz) didn't even flinch with the extra weight.

About the actual auto tool change, if the head is a success the rest is a piece of cake. R8 spindle can take a Tormach R8 collet (http://www.littlemachineshop.com/products/product_view.php?ProductID=2454&category=209730719), which can take Tormach TTS toolholders (or indeed homemade ones). Couple that with a drawbar preloaded with 3 pairs of belleville washers and a pneumatic actuated lever and you're there. LMS used to offer it as a conversion kit for the X2 mill called the Z-bot, and a lot of people have done a similar conversion like the one below from Hossmachines.


If you can poke holes in my plan please do, it seems to good and simple to be true. The total cost will be around 500, which is a risk i'm willing to take.

30-12-2016, 02:38 PM
Nothing wrong with that setup and thought about doing something very similair but would allow 2.2kw to still be used.
Would be side by side twin head with 2.2Kw upside down when low speed cutting. Slide out and turn around and it's back usable has high speed spindle without needing high speed bearings in milling head.

30-12-2016, 02:51 PM
I think i've seen something similar in a drawing somewhere, don't know if it was you or somebody else. I have no real use for high rpms, never do wood or plastics and rarely use small end mills, which have no problem running slower, so i don't need that interchangeability. If i need to do some engraving or a pcb or something like that, i can still setup my old 3040 and spare 1hp 24k rpm spindle and do it there.

I'm glad people with more experience than me don't see much wrong in that, i'll be putting it together as soon as possible.

Clive S
30-12-2016, 03:23 PM
Hi Steve and welcome to the forum. I for one will be keeping a close eye on the way you achieve this,:applouse: and well done with the machine.

30-12-2016, 03:45 PM
Thanks! I think the most difficult part will be the one I can't do myself, squaring and boring the aluminium block to take the spindle, and i'm not yet really clear on how to preload the bearings. The lower bearing is supposed to be a press fit and the upper one a slide fit. I'll have to go to a machine shop for the boring, and I dread to think how much they'll charge me. Anything out of their daily routine gets treated (and charged) like aircraft machining. Somebody quoted me "at least 400€" to skim the surface of the bed of the machine, which I find outrageous. Other than that, everything else doesn't seem that difficult to do, honestly. I'm also curious to see if I can turn toolholders from O1 or A2 steel on my mini-lathe.

01-01-2017, 12:43 PM
Nice ideas, but i have serious doubts about using the 2.2kw spindle as a power source for this type of setup. The bearings are poor and won't take the constant side-load of a belt, torque at lower speeds is terrible as well. I looked into ATC stuff recently on my Bridgeport conversion and gave up as it was beyond difficult - I need an ATC that can handle 24k + but at those rpm's everything needs proper balancing which is beyond the home shop.

I would go for a BT30 spindle, definitely not R8 in any case.

01-01-2017, 01:04 PM
What would be a bigger load, the load of driving the belt or the radial load of a 7mm doc/1mm woc cut? The torque at 12k rpm is enough for the cut itself, wouldn't it be enough when underdriven via a set of pulleys?

It is quickly becoming apparent that the most demanding aspect of the design is the part I overlooked the most, the bearing arrangement. I spent hours and hours reading up on it and it's still kind of a mystery to me. I've narrowed it down to two possible designs.

The one is a couple of back-to-back universally matchable angular contact bearings at the bottom and a common deep grove up top. The inner races of the bottom set will be held together by a KM nut (top portion of the 35mm dia part of the R8 shaft (http://littlemachineshop.com/products/drawings/1407SpindleR8.pdf) will need to be threaded), and the outer races by a flange on the bottom of the housing, which will incorporate a radial oil seal as well. How much will the bearings will be preloaded and how the preload can be measured is still unknown to me.

The other design is sort of a clone of the Tormach spinle (https://www.tormach.com/uploads/1001/SB0047_PCNC1100_SpindleRebuild_0316A_WEB-pdf.html), ie a couple of angular contact bearings at each end of the shaft. The tormach design incorporates preload washers at each end for the bearings, but weather they are used to have the best possible contact on the inner races or as a seal (or indeed anything else) is unclear to me. Again, how much the bearings need to be preloaded is a bit of a mystery to me.

If anyone has any ideas, please share. I'll try to post a drawing of each design soon.

01-01-2017, 01:13 PM
The difference to me is that the cutter load is random in position and loading as well as intermittent, the belt is constant in position and load, it may well work out but don't forget these spindles are seen as disposable - they are pretty much dirt cheap and a set of decent quality matched bearings will be around 2 x the cost of the spindle whole - there is no free lunch.

Preload is a subject in itself, it should be documented somewhere on the bearing manuf. sites.

01-01-2017, 01:44 PM
You're right in that, I guess i'll have to try.

SKF has published a huge pdf with info about bearings, setups etc, I spent a lot of time going through it. Some things are still a bit unclear to me though, and quite outside the spectrum of a hobby/home machinist in my opinion. I'll do my best to figure it out.

01-01-2017, 06:26 PM
Hi spresv

Nice first machine, well done. I've sketched, deleted, pondered and wondered about doing an ATC spindle driven by my chinese spindle, along the lines you describe. Things I thought about:
Running the spindle upside down (Like Deans says) - but didn't think about turning it the other way for direct machining!

Limit to 12,000rpm to make the cassette bearing spec manageable (plus limit centrifugal forces in the the ATC mechanism)

Tormach TTS system based on R8 spindle (but hadn't seen the LMS one, was thinking of making my own before you pointed that one out, thanks)
In case you missed it they added the drawings - and they are metric:

Gear down with at least 1 pulley set, maybe 2, to get the speed way down whilst multiplying the torque (hadn't decided on lower speed but maybe to 2000rpm ?)

I also thought about the constant side loading but I was prepared to give it a try. Could rotate the spindle body around from time to time I suppose to load a different part of the outer race?

The main thing stopping me doing the detailed work was that I should really finish my current machine (which is direct drive) - . . . although designing it in now before making the Y and Z parts would be better than a retro fit . . . . hum . . .:distant:

01-01-2017, 08:08 PM
Thank you!

12k rpm is asking much, Chris @ LMS says it's good for about 10k rpm. Angular bearings are rated at 12k rpm, if you double up they drop to about 9600rpm and if you decide to fit a radial shaft seal their limit is a surface speed of 18m/s. At 40mm diameter (the lower part of the LMS spindle) that's about 8400rpm.

Still, my main concern is how much and by what means to preload the bearings. I hope I get to the bottom of this.

02-01-2017, 04:43 PM
Ok, I think I concluded to a design I'm comfortable enough with testing. It requires threading the 35mm part of the R8 spindle, so I can use a set of 7207BEGAP angular contact bearings. They are universally matchable, so the inner races are factory ground for a light preload, so fitting a KM7 nut will secure and preload them. Top bearing will be a standard deep groove 6206. The lower gap of the pair will have a flange that will secure the outer races and house a 40x62x8 radial shaft seal, series CRW1 or CRW1A.


Also, (assuming the rest of the design works) the more I look at it, the more I don't like the design of the lever operated drawbar. The benefit is that you can use a small and cheap pneumatic cylinder, but the end result will look a bit bulky and amateurish to me, although fully functional. If I can spare the expense I might change the design to a 100mm bore 10mm stroke air cylinder. I think I'll need about 6mm of travel for 3 pairs of washers to release the collet, but I'll test as soon as i get them.

02-01-2017, 05:07 PM
Try and get those lower bearings as close to the end of the spindle nose as possible, it will reduce deflection.

I think either operating method is good, both have merits, the lever and small cylinder will possibly be lighter.

02-01-2017, 09:26 PM
Yes, i've read that recommendation elsewhere, and it really makes sense to me, but it poses 2 problems. By far the bigger one is that I have no machinery to turn it down and grind it. The second problem is a design flaw that I couldn't predict. The end of the spindle head will be flush with the bottom of the Z-plate which is about 135mm from the table. That means that at the end of the z-travel, assuming a toolholder with an end mill with 20mm stickout, it will still need another 40mm or so to reach the table. If I get the bearings 20mm closer to the spindle nose, that becomes 60mm. I could maybe build longer toolholders, or use ones with er collets, or mount the head below flush from the z-plate, but wouldn't that be a problem?

02-01-2017, 09:31 PM
ER collet holders are very good, i have never liked the ones with a screw in the side to clamp the tool.

02-01-2017, 09:41 PM
They do have a cost advantage though, since you could build them yourself. If the rest goes to plan I'll get a tormach one, build some of my own and measure runout on both (including er ones) and report back.

27-01-2017, 12:31 AM
Hello guys

Exams and a streak of bad luck kept me away. A couple of weeks ago I was building a set of carb mounts. 4 tools needed per side, and I can only mount one at a time, so a lot of tool changes. ER nut started to feel a bit notchy, no visible debris in the threads. I gave it an air blast, still notchy. Grabbed a pointy sharp tool and had a go at the nut threads, something did come out, so I did that again meticulously on both nut and shaft. Squeaky clean and all was well, no hint of blemish or damage on either threads. Last part, last side, roughing operation finishes, I go to unscrew the nut to change tool, and it just seized solid. I've never seen anything like it before, no excessive mechanical stress, no thermal stress, no nothing, and it just seized. When I say seized I mean I held the nut on a vice, I grabbed a 21mm spanner and a 1m long lever and it STILL wouldn't budge. Eventually with a lot of patience and even more swearing I managed to get it out. Bottom threads were absolutely destroyed, nut too, but the worst thing is that trying to get it free I must have applied force in a way I shouldn't have and I damaged the bearings, shaft wasn't turning true anymore.

So, new bearings or new spindle? Two pairs of nachi bearings from bearingboys in the appropriate size were almost as much as a new spindle. So new spindle it is, it took almost 10 days to get here from Germany, ffs, but yesterday I got it. I mount it and straight away I noticed it was MUCH stiffer in rotation. Give it a spin by hand and it barely does one rotation. Hang on... Trusty Mitutoyo indicator doesn't lie, 0.025mm runout. Advertised as LESS than 0.005mm, that's at least 5 times more than advertised. Send the seller a message, hasn't come back to me yet, but I feel it's not going to end well.

In other news, I decided to make a small fly cutter, I wanted to see if it could be usable in that spindle. I had some 7075 bar (don't freak out, it's just a proof of concept), so I turned a 12mm shaft with a 30mm body and a 8mm slot on a 10 degree angle to mount a turning tool with a polished carbide insert in. Now, I would swear I had a 95 degree left hand tool somewhere, turns out it was right hand... The one pictured on the fly cutter is the only thing I had that would mount and make a cut, even though it's the wrong tool to use.


Turned out quite nice though. I tried slow cutting air, to get a feel for the speed. Around 2200 rpm vibration starts being noticable, 3000 and it's too much. I tried it on the lathe at about the same rpm and it only had a really slight sort of harmonic to it. My guess is that whatever imbalance it has is greatly accentuated by the spindle runout. Anyway, 6082 plate, 0.1mm DOC, ~45mm WOC, 2300rpm, 300mm feed et voila. Not bad at all, spindle could handle it, I guess if it was runout-free it would give me much more confidence. Not that I would go deeper, I don't care about material removal, just surface finish.

Lastly, I decided to put the ATC project on hold for a while. I need to invest in other things first, an enclosure, a new cabinet for all the electronics, a half-decent hand-controller or panel etc. Most importantly I need a bigger, more reliable and robust method of workholding, so I could mount fixtures, jigs etc to be able to make multiple parts at once and minimize the frequency of tool changing. I got my eye on this, anyone tried it? http://www.arceurotrade.co.uk/Catalogue/Workholding/Machine-Vices/ARC-Versatile-SG-Iron-Milling-Vices

Long post, that's all for now, cheers!

27-01-2017, 07:18 AM
Its a crap-shoot with these spindles, never know what you get, I have two and both totally different in feeling.

Fly cutters are murder on the spindle, be careful and if possible, counter balance it a bit.

27-01-2017, 11:02 AM
Nah, I'm not gonna use it much, I just liked it as an experiment.

Seller came back to me with this

sorry friend,we cant open the link you send.
Friend,would you mind keep the machine?
We will give you 5GBP as compensation.
Because it is the Chinese New year now,our technician is on holiday so that we cant let them confirmation in time.
They will go to work on 6th on February.
Thanks for your understanding.

Well, this is borderline insulting. I've spent around 400 with them and they'll give me 5 quid compensation for a downright faulty machine? Do I have grounds for a paypal dispute, or this is used when items don't arrive at all etc?

27-01-2017, 11:18 AM
You should have support under the distance selling regs - DSR but not sure what best course of action is.

FWIW there are UK sellers (if you're in the UK of course) - i would use one of them just in case, prices are similar but returns much better.

27-01-2017, 11:22 AM
I'm not in the UK but I prefer to shop from UK sellers. They were claiming to be in the UK, package shipped from Germany and they are obviously Chinese. I replied kindly requesting either a full refund or a replacement spindle.

07-02-2017, 02:50 PM
Hey guys

Fed up with the chinese crap really, so i'm looking for alternatives to get rid of it completely. I'm thinking of replacing it with a 1hp 3phase 2 pole induction motor (only ~2kg penalty over the chinese spindle) and have it drive either the spindle i designed earlier, or a ready-made one. I've been given the choice of a 4 bearing precision 8k rpm capable spindle with either an ER32 collet system or an MT2 taper with a drawbar, both at a similar price, good enough to steer me away from risking of building my own. I kind of like the idea of the MT2 taper because it means i could have preset height tools and toolholders and not touch every tool off before each operation. I do fear though that it's going to get constantly stuck, and that i won't be able to find low-runout tools/toolholders in that taper.

What's your take on this?

07-02-2017, 03:28 PM
I would not use MT tapers - they are self locking and can be a bugger to release :)