View Full Version : BUILD LOG: small mill for machining steel. Max spend 500 including computer

Andrew Wilding
22-11-2010, 09:06 PM
Hello all

I thought that I would post my build log here of a small mill. The aims of the project are below:

max spend 500 including computer and cost of base machine - this a real budget as it is the max funds the missis' will release and it will also be an interesting excercise in how to get reasonable results on the cheap.

capable of milling mild and possibly stainless steel with <6mm carbide cutters

Accuracy of +/-0.001" over an area of 4"^3

Andrew Wilding
22-11-2010, 09:17 PM
the basis of the mill is a Centec 2 horizontal purchased a few years ago for the sum of 150. This served as my mill until replaced with the AEW show in the pictures below:


This machine is very tired with lots of bed wear and backlash so the first task is to rectify this. It is also a knee mill operated with a rack and pinion system and stops. A rubbish threaded rod lead screw system has been added later to make the knee more controllable. I will have to come up with a better method to raise and lower the mill.

Andrew Wilding
22-11-2010, 09:32 PM
Ok I admit that I have already done a lot of the work, but I will go through it like a chronological build log.

Scraping the vertical slideway on the knee (before showing uneven wear):


after showing spots per inch (note rack):


Milling the knee:


The knee was aligned to a ground rod that had already been trammed. Wear was beyond scraping on horizontal part of knee.


Andrew Wilding
24-11-2010, 04:00 PM
the x axis was also machined with a dovetail cutter on the mill



Andrew Wilding
24-11-2010, 04:06 PM
The ways were built up with PTFE. This was sheet purchased from E bay and had an acid etched back to allow it to be bonded to the ways using epoxy resin.

The PTFE was then machined an scraped to alignment


Andrew Wilding
24-11-2010, 04:58 PM
Ball screws are the obvious choice for a cnc mill but alas the budget did not extend this far. I opted to make the best out of conventional screws. I selected 16mm dia by 4mm pitch trap leadscrew stock. This has a max pitch error of 50 microns per 300mm (2 thou per 12") which should be sufficiently accurate to meet my accuracy target. I may experiment with lead screw mapping later on. The 16mm stock is sufficiently stiff over the lengths I am using to only require supporting at one end. At the supported end I am using a preloaded pair of angular contact bearings to reduce support backlash as much as possible.

I am using wear compensating antiback lash nuts from delrin. There are made as a main nut and slave nut biased together using a torsional spring made from piano wire.

Below shows the nuts being made on the small lathe. the thread was roughed in the lathe to ensure parellelsim with the nut. These were finished using a tap made from some of the stock.


The slave nuts were castellated together on the mill. Note the DIY dividing head. This was made when I only had a lathe and uses change gears to provide most of the divisions required in my shop.


This shot shows the three lead screws, bearings and nuts assembled on the bench


Total cost of leadscrew assemblies about 25 as opposed to at least 200 for similar pitch accuracy ball screws.

24-11-2010, 05:46 PM
Looking good.
I'm interested in your nut design - do you have a drawing that you could post here?

24-11-2010, 07:02 PM
Andrew you have discounted the efficiency of ball screws.

The use of a knee mill is good but the downside is the amount if force needed to move the Z axis.

You could calculate the amount required using a spring balance, have you done this ?

Guys who do Bridgeports use the quill for the Z axis

I do have a Bridgeport myself and did a bit of checking.

The Y axis has an 83mm offset and took 2Kg of force to turn it

The Z axis has an offset of 185mm and took 4Kg to turn it

1000 / 83 = 12.048
1000 / 185 = 5.405

2Kg / 12.048 = .166Nm Required
4Kg / 5.405 = .74Nm Required

.74 / .166 = 4.458 So if the norm was to use a 4Nm motor for the Y axis then amount of force to drive the Z compared to the Y axis would be a 17.8Nm axis motor "4 * 4.458"


Andrew Wilding
25-11-2010, 07:13 PM
Thanks for you experiments. I realise that ball screws are more efficient. I have chosen larger steppers than I would have with ball screws. The teflon coated slides also help increase the overall efficiency of the system a bit.

As you will see when I post the detail of how I did the knee, it is counter balanced. This obviously improves the force balance on the system but the knee is still a fairly heavy part to accelerate. I tried to take the easy route out with the design and replace the threaded rod with the screw stock but I was not happy with it and had to go back to the drawing board! more on that later.

Thanks for the post. I was at Notts 6 or 7 years ago and studied mech eng (ahh the smell of the Coates building!). I now work for a gas turbine manufacturer but still live in Nottinghamshire. I did do an autocad sketch of the nut but it was only to see if it looked right. I will try to post this or possibly just a better picture, although there was not much science behind it. The slave nut is castellated to allow the spring load to be adjusted and so there was a bit of experimentation involved. The two halves mate with a 60 deg chamfer so an increased locking torque is experienced between the two parts when compared with just plain faces for a fixed spring rate. This should increase the thrust required to separate the two halves of the nut with the same spring rate. It seems to work ok. The lash in each axis is about a thou but I an now adjusting everything to see if that can be reduced. Without the slave nuts the lash is 6 or 7 thou (I deliberately made this a loose fit so that the only friction is on the useful faces)

26-11-2010, 03:07 PM
Andrew not easy to counterbalance the knee, but good to see you have it worked out.

Good luck with the rest of the build which i will keep an eye on.


Andrew Wilding
27-11-2010, 08:50 PM
This is a view of the mill with all the remachining and scraping done as well as the start of a vertical spindle thing

This is a shot of one of the bearing blocks being centred. I find the easiest way to centre a dot punced block is to use a more taper centre with a spring cntre as a wiggler. This does rely on accurate centre puching thought. I have recently invested in an optical centre punch which makes life a bit easier. this is on the larger harrison lathe.


Andrew Wilding
27-11-2010, 09:06 PM
the next few shots show the first knee raising method. this was replaced later as I was not happy with it. It also shows a spindle that was later replaced with one with less run out and larger bearing centres. With this knee raising method the force is off centre and so the knee crabbed slightly and required a lot more force to raise the knee.


you can see the original handle at the back of the mill that was used to raise and lower the knee. If you look at some of the previous posts you can see the rack on the knee. The handle is connected to a pinion that meshes with the rack. I replaced the handle with a bike chain wheel and a bike chain with weight attached is used to counterbalance the knee. I have just realised that I do not have any pics of this so I will ahve to go and take some.

Andrew Wilding
27-11-2010, 09:58 PM
brrr its cold out there! Here are some pics of the counter balance assembly.


a weight is fitted to the end of the chain

Andrew Wilding
27-11-2010, 10:11 PM
here are some close ups of the y axis nut

28-11-2010, 12:27 AM
Andy keep up the good work and nice to see old being brought into the modern age.

Andrew Wilding
28-11-2010, 06:42 PM
cheers 2e0poz (callsign?)

I replaced the side acting z screw with a screw that acted below the approx c of g of the knee. This resulted in a much smoother action. The stepper is geared 2:1 and as stated before the knee is counterbalanced using the original rack and pinion system using a chain wheel, chain and mass (the wiring is just a lash up to test it)

the stepper turns the nut rather than the screw. The nut is supported using two largish AC bearings. No anti backlash nut is required as long as the counterbalanced force is less than the weight of the knee. This improves the raising action significantly and it is smooth with no evidence of the crabbing action seen with the first method.
The only issue is that the screw protrudes below the base of the mill when the knee is lowered so the bench has to have a hole in it! Not too much of a comprimise for a simpler system without the mass of the stepper on the knee.

The electronics are also shown in the box next to the machine. I may add a post about this later.

Andrew Wilding
28-11-2010, 06:50 PM
2e0poz I am m1cws (I think, it is a long time since I used my radio!)

28-11-2010, 08:36 PM
QSL Andy seems half off 'G' land is into CNC these days and it's been over 15 months since i used mine.

As an idea could you not mount the nut into the base of the knee and use a belt to turn the screw in the configuration you have now? a bearing block could be placed under the plate you have the motor mounted to. a couple of pneumatic rams like used on hatchbacks would help with the weight if you use them upside down and a light spring over the ram section?

29-11-2010, 09:32 AM
As an idea could you not mount the nut into the base of the knee and use a belt to turn the screw in the configuration you have now? a bearing block could be placed under the plate you have the motor mounted to. a couple of pneumatic rams like used on hatchbacks would help with the weight if you use them upside down and a light spring over the ram section?

I think Andrew has the best solution now as maybe he maybe would need a hole in the table to do as you suggest.

I went through 4 different Z builds on my mill so i know the frustration after a lot of effort it is all in vain.

Like the amp meters in the control box, i would like to see an Estop placed on the machine when the wiring is finished, in case the worst happens.


Andrew Wilding
30-11-2010, 09:48 PM
on with the mechanical build:

new gibs machined, teflon glued on then remachined. I prefer using the shaper rather than the manual mill.


Spindle started. Spindle rough machined and drilled


taper set on the top slide (female centre drilled)


taper set using existing more 2 centre


taper in the end of the spindle was finished with a mores 2 reamer. TIR was less than 0.0002.

Andrew Wilding
30-11-2010, 09:51 PM
rough machined spindle and housing


pulley key machined


spindle exploded assy


Andrew Wilding
21-02-2011, 09:19 PM
I thought I would post an update on the mill progress. I have been a bit busy fitting a kitchen, drawing tractors for my son and somwhere along the way had time to make anther sprog - due soon!

As you can expect progress has not been great on the mill front, however I have the spindle operating and have made some cuts in delrin then steel . The machine performed ok but I am now making improvements. The next few post will bring you up to latest progress

Andrew Wilding
21-02-2011, 09:36 PM
These shots show the spindle drive pulley arrangement. Tensioning of the high speed belt is achieved by rotating a block in which the counter shaft is mounted (eccentrically). The low speed (v belt) is tensioned by rotating the pulley assembly.


note that no guards have been added yet but are on the list before the mill is complete.

Andrew Wilding
21-02-2011, 09:45 PM
The anti backlsh seemed to work ok but I had an idea how to improve them. I modified the design so that the two nut halves are radially fixed to each other but axially free. this was acheived using 4 off 3mm rods in close fitting holes in both nut halves. The slave nut is threaded with a fine thread and an outer collar threaded on. this is biased with a tprsional spiring to take up any backlash. hopefuly this greasy sketch might help - it was used when I was machining the parts so it is not the clearest:

drilling the holes for the rods in the nut halves


Andrew Wilding
21-02-2011, 09:49 PM
machining castelations in the collar

some surgery to the bed to fit the larger diameter nuts in - I love machining cast iron!


X axis nut in position but biasing spring not fitted over collar (lash up wiring shown!)


Andrew Wilding
21-02-2011, 09:56 PM
one more picture showing x axis nut

21-02-2011, 09:57 PM
Nice build :)


22-02-2011, 11:26 AM
thanks for sharing your build keep up the good work.

Andrew Wilding
22-02-2011, 12:56 PM
Cheers all

BillTodd I looked up you roller nut for your mill conversion when I was wondering what would be a good low cost ballscrew alternative. Are you still using this and if so how is it going? I would be interested to see how the nut has worn. I decided against it as the chosen solution for a couple of reasons:

mainly space - it is tight enough with the current arrangement and i am not sure that I could have got a roller screw to fit without making it more complecated
Time - it seemed a time consuming (but rewarding) method

but could see the efficiency advantage.

22-02-2011, 07:23 PM
BillTodd I looked up you roller nut for your mill conversion when I was wondering what would be a good low cost ballscrew alternative. Are you still using this and if so how is it going?

They're working really well, well enough for me to consider replacing the fixed nuts on my Haighton (http://www.mycncuk.com/forums/showthread.php/1231-Haighton-Major)as well.

As you say, the space requirement is the main drawback. I've been eyeing Mike Everman's quad bearing design (http://www.cnczone.com/forums/linear_rotary_motion/13593-different_better_kind_screw-66.html) and wondering if I could use it with some compact bearings instead of my twin roller type.


Andrew Wilding
22-02-2011, 08:12 PM
looks like an interesting solution. Looks like it was well covered in the thread so I will not regurgitate that here, but how would that work with large pitch (high lead) screw like the one fitted to your mill. You may only end up with one or two line contacts per bearing? I guess it is not an issue with the type of duties our mills see. Obviously you will have to post results when you get onto manufacture!

Nice job on the refurb by the way.

Andrew Wilding
25-02-2011, 01:01 PM
I tried the new x axis nut out last night. Backlash was very similiar to the previous nut at about 0.0007" and did not reduce even if I cranked the nut up tight so I suspect that I need to look at the preload on the AC bearings or check for movement of the bearing in its housing.

Rapids have gone up from about 30" min to 150" min (+ acceleration increase) without lost steps so nut drag is much reduced.

I am happy with the solution. Now to make the y axis (Z does not have a fancy nut as gravity is working in the same direction ad the cutting force so eliminating lash naturaly)

Andrew Wilding
13-04-2011, 01:20 PM
I have had a busy week as my second child was born on Friday! All is well and Lydia is starting to become a bit more aware of her surroundings. It does mean that I have some paternity leave and so in between changing nappies and winding I have had a bit of shed time.

I have rebuilt the mill and painted it. Here are some pics of the z assembly showing rotating nut (not pressed in in the photos)





Andrew Wilding
13-04-2011, 06:14 PM
Here is a picture of the newly painted mill part assembled

Jobs I am hoping to do in the next couple of weeks:

tidy up wiring with conduit and built swarf protection for limit switches
screw and way protection
Modify spindle ratio
guards and machine mounted E stop

21-04-2011, 02:42 PM
Hi Andrew,

Looking good

How long did it take to design and build the Z axis ?


Andrew Wilding
21-04-2011, 09:42 PM
Hi Andrew,

Looking good

How long did it take to design and build the Z axis ?


Cheers. Design was carried out during daydreaming on the train and manufacture two or three saturdays of pottering in the shed plus a bit on the computer to purchase belts and bearings etc.

Andrew Wilding
24-05-2011, 09:52 PM
I had my spindle running at 3500rpm and it was seriously heating up. Tried all the usual things, preload lubricant etc but to no avail. Thread on it here:


Long story short I have rebuilt the spindle with better bearings and different seal system.

Rubber seal replaced with alu labyrinth (shaft clearance 0.0015")


I have also made a perspex guard over the pulleys. It is smoked rather than black so you can see which pulley has the belt on


without guard


I am running the bearings in and will post progress when I next get a chance:smile:

Andrew Wilding
08-06-2011, 09:50 PM
I fancied a change of tack so I have started checking and improving the geometry of the axes.

First check is the perpendicularity of the knee. This is the set up for the first method


This was the second method which I was planning on checking the first method with, however the results of this method were not very consistent so I abandoned it.


The results show an error of about 0.001" out of plane using the first method which can be improved with a bit of judicious scraping.


I has previously used the gound side surface of a parallel as my reference. This was not a very good reference so the next task is to scrape a reference surface. Oh joy - blue fingers over the weekend then! Once the horizontal surfaces are complete I will then check the other two surfaces of the dovetail are parallel.

Andrew Wilding
17-06-2011, 07:51 PM
I have made a straight edge to fully scrape in my dovetails. Before I used a piece of round plate which was not that good.

after scraping to 20-25 spots per inch

Andrew Wilding
09-09-2011, 11:16 PM
some more scraping updates:

dovetails on the knee were scraped parallel


Horizontal ways on table were scraped parallel to top of table (measured with mic)

Back way scraped to front face.


front way scraped parallel to back way using smae method as the knee.

old rack and pinion spindle bearing removed to increase travel by 1"


saddle scraped to knee by blueing knee ways.

x horizontal ways scraped parallel to y horizontal ways. Gauge block used to measure tops of undulations (indicator tip is too small and waviness was being measured)

table to knee perpendicularity checked and scraped etc. Precision pin and blocks used to provide reference to compare table way direction.


The scraping reference book that I managed to find in the library, interesting but by the time I had found this I had done most of the work! The cat did not think much of it.


nearly finished the scraping! I will be glad when I have a fully square and straight mill.:tongue:

Andrew Wilding
21-09-2011, 11:59 AM
Final alignment checks to see if I have met the Connelly prescribed 0.001" in 18" accuracy:
one side was trammed to y axis then other face of block measured to z axis.


block clamped to z ways and y axis traveresed


similiar done to x axis.

x-y 0.001" in 18
y-z 0.0015 in 18
x-z 0.0015 in 18
I am going to leave it at that as I am loosing the will to live!:tongue:

Andrew Wilding
21-09-2011, 12:09 PM
final mechanical assembly:

4622 4623 4624

4625 4626 4627 4628

Just final wiring with shielded multi core back to controller and I am done, oh and I have to learn how to use the thing!

21-09-2011, 12:16 PM
Nice job Andrew ;-)

Andrew Wilding
28-10-2011, 10:20 PM
It's alive!
Did my first tests and it seemed to go ok. Machined some pockets in some scrap mild steel.


the pocket was 30mm x 20 mm and all dimensions seemed to be within 0.01mm of nominal (although this was just a quick check with digi calipers). I will do some more serious machining and see how a bigger part measures.

Cost is a bit over budget as the electrical bits set me back more than expected but I recon the original machine, stock, hardware, motors, controller and computer cost me about 600. That does not include the extra tools I bought to do the job of course, they are an investment!

I have enjoyed the project and have learned a lot from it. I hope you have enjoyed it too.

Thanks :tongue:

20-07-2012, 03:46 PM
hi andrew, the mill is looking great

im planning to convert my centec 2 soon and i was just wondering...

how do you find the screws? have they lived up to your expectations?
what motors are you using?
is there anything that you wish you had done differently?