By continuing to use this site, you agree to our use of cookies. Find out more
Forum sponsored by:
Forum sponsored by Allendale Jan 24th

DIY Epoxy Frame based CNC MILL

A heavy bench top V mill

All Topics | Latest Posts

Search for:  in Thread Title in  
John McNamara26/11/2018 10:21:38
1314 forum posts
113 photos

Hi Martin


The Swiss company Fritz Studier Held one of the early patents. Now lapsed.
They used "polymer concrete" With their proprietary name Granitan using Ciba Geigy resin.

Polymer concrete is the description used by the Europeans, Epoxy Granite is used more on the other size of the Atlantic by the US

The reason for the break from Cast iron was not cost it was about the considerably higher vibration damping properties. The almost, and some say zero shrinkage of the cast after curing. and the advantage of being able to place steel inserts, pipework and other fittings within the almost cold mold. Casting accuracy of a thou .001" is not uncommon. Compare that to cast iron and the considerable amount of metal removal required to finish machine it.

On the con side it has a much lower tensile strength. You have to use thicker sections. and possibly for me anyway you have to rethink a project model to gain the benefits it offers.

For the home builder polymer concrete opens many doors. Form work can be just about anything; plywood, metal, chip board, plastic etc. Simple molds can be made for once only use or more durable materials for reuse.

And the rest of the world? See link below.



Edited By John McNamara on 26/11/2018 10:24:08

David Taylor26/11/2018 23:12:27
133 forum posts
39 photos

This is an amazing project John. The forethought in the design is something else.

John McNamara02/12/2018 14:02:49
1314 forum posts
113 photos

Hi All

The first casting was made today!

Image below
This is the last time I will see all the components I made. as you can see there are no complicated parts.
Mainly a lot of simple lathe work, tapping holes and preparing threaded rod.
A few laser cut 2.5mm sheet parts. deliberately rough ground on one side for adhesion.
Also 2 16mm x 70 rail supports. I lapped one side flat (A lot of work)

Before starting the cast all the parts have to be cleaned with solvent I used methylated spirits.

The surface plate was very carefully cleaned then checked with the palm of the (clean) hand, this will detect tiny specks that you cannot see.

The surface plate was then covered with a new sheet of 80 gsm plan paper (with 2 large slots cut out 25mm all around wider than the bearing support rails) The bearing supports rails sit on waxed baking paper not plan paper, (When I measured it I found it has very uniform thickness) This is to protect the granite surface plate. The waxed baking paper was sprayed around the edges with silicone.

The 5mm thick mold bottom plate was then placed on the surface plate.Note all the long nuts screwed in from the bottom of the plate in countersunk holes all carefully checked that no CSK bolt head sits proud of the bottom of the plate.

The top of the plate is covered with self adhesive plastic to stop the epoxy sticking.

Laser cut mold parts ready for assembly. These will be reused.

I am sorry I don't have any pictures of the molding process I was kept very busy placing and tamping the 4 mixes of about 5 litres of Epoxy aggregate This was topped of by a thin layer of sand epoxy as seen above.
I will try to get a helper to do the photographs on the next pour, There are 4 pours to go!
As you can see the finish is so so, I kept the mix ratio to 14% epoxy by volume, this makes a fairly stiff mix.
The 8mm Aggregate contained a few larger stones, For the next pour I will sieve the larger stones out. they made the tamping more difficult. I did it layer by layer. so the layer on the bottom would have been unworkable (but still very sticky) by the time I was doing the top layer.
The process took about 1.5 hours The weather was cool so there was no problem with the mix going off before it was placed

Exotherm? The mixing bucket got slightly warm only. When placed in the mold I could not detect any heat with my hand.

The mix was one part aggregate and about and 0,45 parts washed fine sand.
This was determined by filling a container with aggregate then slowly pouring the sand over the aggregate until no more sand could be poured in. I used a mechanical vibration. Note when this process was finished the volume is only one part. The sand simply filled the voids between the aggregate particles.
The final step is to sieve the aggregate to recover the sand which was then measured to determine its volume.

I used a cheap disposable bucket to mix with an 2500 Watt electric drill, and I needed those watts it was only just enough. The agitator was a low cost spiral mixing unit purchased from the local hardware for about 10 dollars.
Power mixing is essential.

This is for Neil Wyatt. Yes the Limit switch wiring is routed through the casting, they will go straight into the control box shown earlier this forum thread.

it's been a long journey, at last it will be possible to test how accurately the bearing support rails are aligned. Hopefully they will not have moved during the cure and I will not have to re-lap them.
The next step is to strip the mold, I will wait 48 hours at least to allow the material to cure.
This will be followed by attending to any small voids with body filler grinding off any rough spots, and rounding sharp edges followed by epoxy paint.


Neil Wyatt02/12/2018 14:22:27
18558 forum posts
723 photos
78 articles

Excellent progress, John.


John McNamara11/12/2018 15:01:21
1314 forum posts
113 photos

Hi all

I have cleaned up the casting, rounding edges and filling blemishes with body filler, there were more blemishes than I would have liked, the part has been given a quick undercoat, once the whole machine is complete it will be detailed and properly painted. The part is very massive too heavy for me to lift. Tapping it with a hammer yields a dull click, even on the metal parts. due to the damping properties of mineral cast epoxy.

The next step of this project is possibly the most important. the accuracy of the machine depends on this stage being a success. The accuracy depends on the rail mounting points being both straight to fine limits and co planar. As the XY and Z axis are completed they must also be correctly located in relationship to each other. At this point we are dealing with the X crossmember axis only.

As stated earlier in this forum thread a surface plate was used as used to align the steel rail supports. the surface plate partly formed the bottom of the mold. there were holes in the mold and the steel rail support bars projected trough these holes in direct contact with the surface plate.

To test the accuracy of the rail supports the part was placed on shop made jacks. They bear against a cast in steel face not seen, the columns will mate against that face.

The part was aligned to the surface plate to within .0001" using a dial indicator, one div = .0005", mounted on a very rigid Mercer base. Once set up measurements were repeatable to within .0001" or better. The one knob type indicator mag bases being sold widely are simply not up to it.

The rail support bars 70 x 16mm in cross section were ground and hand lap/ground on one side prior to casting, this was not a complete success, the bars were too flexible and I found it quite difficult to get repeatable markings from he bluing. as the results show below there was some error after casting. the upper rail in the picture with two rows of tape had a maximum error of .0008" bowing op in the middle the second row of tape shows the error reduced by lapping to .0004" not a huge job a bit over an hour so far, there was not a lot of metal to remove. More to do, I want it as close as I can get it within the limits of my tools.

The other rail has an error of .0018" almost two thou, mainly caused by a dip at one end
It will take a little longer to adjust.

Simple steel laps made from 25mm x 25mm mild steel. The abrasive is 40 micron mylar on the longer lap 120 grit aluminium oxide on the shorter lap for roughing. not glued on just held with fastenings at the ends. Every workshop should have these, mine never leave the bench

When lapping I sometimes make a small light scribe line to help me gauge progress between indicator checks that take time due to cleaning, not too heavy on the scriber ! see below.

The surface plate It must be kept perfectly clean and free of grinding particles, you cant just drag the indicator around.

The surface plate itself may not be perfect I purchased it from a toolmaker he said it was grade A. as a double check I am also checking the flatness with a precision straight edge, a recent find brand new still in the wax paper. note the date! Both edges give the same readings indicating it has not moved in 73 years.

I must admit I am excited by these results. It is clear that rethink of the rail mounting support design is in order, I would like to improve the preparation prior to molding. They will have to be stiffer and my lapping skills will have to get better.

On the other hand the existing method has worked very w

John McNamara13/12/2018 14:31:02
1314 forum posts
113 photos

Hi All I was asked these questions at another place.

In answer to your questions A and B.

A. “To get ground stock (can be bought ground, but suspect $$$), predrill (most) holes and cast against surface plate as you have done”

I considered precision ground stock and phoned around to locate a source, it was available at a cost, quite a lot actually; however the supplier would not guarantee flatness on longer lengths. They would provide a tolerance for constant thickness but not flatness, however for this project only one side has to be flat the other side is buried in the casting.

Laser cutting precision ground stock which is normally a water or oil hardening carbon steel is likely to make any hole or laser cut edge very hard. Also my laser cutter charges extra for precise positioning small pieces of material, they prefer to work with large plates. This hardening also occurs to a small extent with mild steel, making it hard going for high speed steel tools however I did not have a problem with HSS on the steel supplied by the laser cutter.

If I was able to locate precision ground stock with a guaranteed flatness it would have to be machined to size and the 16mm holes for the cam pins would have to be drilled. These holes do not have to be located to high accuracy so marking out and centre punching in the normal way would be fine. The cams will compensate for any small errors.

The 16mm thick standard mild steel plate supplied by the laser cutter has a reasonable surface finish free of the normal flaky scale you get on black bar stock, there is a black oxide coating and some mill marks. After cutting it is not flat there was heat distortion, bowing of the 16mm axis in particular. I mechanically straightened the pieces and then sent them to a firm with a surface grinder. This was not a success as they did not completely flatten then pieces. I suspect their operator was not up to the job. The big enemy is the magnetic chuck. Grinding long pieces requires packing the work so the chuck does not pull any bowing down before grinding a side then, then turning it over and repeating the process for the other side. This process may have to be repeated several times until any stresses are relieved and the piece will lay flat on the surface table. Even better would be to have the part stress relieved professionally another cost.

In the end I hand lapped the errors left by the surface grinder. As the post above shows there were some residual errors.
I have been working on them with the lap and now the maximum error is .001” over 800mm testing the part on the surface plate with an indicator. I am confident I can get that down to a tenth or so, however just turning the part 90 degrees would probably change that, gravity is a universal force. I guess I could set the part up on its airy point’s that would possibly reduce the measurement error however temperature is also a variable. My workshop has a skylight. We happened to have a hot day yesterday a high of 33 today the high was 20. As expected my error measurements changed with the sun shining on the part; the top face was heated and the face bowed up. Digging the depths of high accuracy is an interesting challenge.

Question B. “Use mild steel bar stock, cast in standard mold, not requiring surface plate, and pay someone with a big cnc mill to quickly machine the steel and drill the holes?”

That will work no problem, apart from finding a contractor that has a CNC machine and equipment that can handle large heavy parts and cost.

It was tempting me to take that option, were it not for the desire to to find out what is possible using a surface plate, small lathe and basic tools. The laser cut mould is optional it could be made of MDF for one time use. The machine does use laser cut parts, they could also be made by hand a large time cost. If I manage to make a workable machine and demonstrate a method that worked I hope others will maybe create their own designs and contribute improvements to the methods I have used for this machine.

Edited By John McNamara on 13/12/2018 14:32:21

larry phelan 115/12/2018 16:51:56
975 forum posts
14 photos

As you said John, it is a long Post ! I,m afraid you lost me a long way back !

To put it mildly,this is WAY ABOVE my pay grade ! Never heard of machine bases being made like this,but then,I still have a lot to learn.

Funny that you mention about threaded rod being cheap over your way,I bought some here recently and I dont know how they can produce it for the price.. It,s one of the most useful things I buy,[I use it for everything ],and still dirt cheap.

Marcus Bowman15/12/2018 17:52:53
169 forum posts

All the same, this is fascinating stuff, and just shows what can be done with oodles of skill and good planning. I look forward to the next instalment.

As for threaded rod; yes; I agree about the price. However; although I buy the occasional length in B&Q, I have a friend who uses it in quantity for a commercial product. Think of the cheapest price you have ever seen, then divide by 3. That's what he pays. It beggars belief that at that price it can still be manufactured at a profit.


John McNamara16/12/2018 13:40:15
1314 forum posts
113 photos

Hi Larry and Marcus

Thanks for the words of support. Glad to see the threaded rod is low cost for you too. I am not sure the galvanized type would stand up well in severe conditions it is very easy to remove the coating with a wire wheel. I suspect it is applied by a process other than hot dipping. for this project it does not matter, raw steel would have been fine, I just used what was in stock from the local Bunnings store All i want is "reo" with threaded ends for fittings.

Hi All

This weekend I continued the lapping of the rail mounts, its been slow going to get them down to about +- 0.0001" one division on the indicator is .0005" I Interpolated between the divisions, however this overstates the overall accuracy obtained. Temperature and measurement errors have an effect on the results as well. However I am very pleased with the results so far.

I ground the 25 x 25mm mild steel bar face flat for this stage it was not perfect and was affecting the results.
As shown below I used various grades of paper from very rough 60 grit for ripping down 120 grit aluminium oxide finishing off with 40 micron Mylar backed strip. I found the Aluminium oxide paper to be better than emery. As you can see by the number of used strips the paper looses its sharpness and looses cutting power on the large lap fresh paper cuts fast, it will be fine for other work

I will stop the lapping process lapping here until the linear rails arrive. it will be interesting to see how accurate they are when attached and aligned by the cams. Any further adjustments will be made then.

It is time to start the two column castings.

John McNamara31/12/2018 01:15:56
1314 forum posts
113 photos

Hi All

The workshop has been a little quiet over Christmas, cant wait to get back to working on the project.

I posted a note on limit switches here: **LINK**

I think I will stick to the plan at this stage however the ability to home the machine to within a few microns is a possible upgrade.


Edited By John McNamara on 31/12/2018 01:16:30

John McNamara06/01/2019 08:20:45
1314 forum posts
113 photos

Hi All

The first Gantry column has been cast and preparation for undercoating started.
The other column is next. The same mold will be used, By turning the base plate upside down and rotating all the mold side plates 180 deg you get a mirror image of the first mold. A bit like making a pair of shoes. The left column does not have to provide for a motor so those parts will not be used.

The steel molds are amazing, every part just fits together, there is no draft. I did not use any mold release. The self adhesive plastic sheeting works really well, a light tap and the mold parts just fall apart. there is a bit of mess I did not protect every face. A quick clean on a wire wheel and the mess is gone.

I have been working on the Epoxy / crushed granite mix mixing about 5 litres of material at a time then placing it in layers, Due to the complexity of the molds and the many inserted steel components they contain I think I will have to increase the percentage of epoxy above 15% to get a wetter more flowing mix. More experimentation is needed. Less epoxy is better (stiffer).The next cast will be the left column.

Mold prepared inserts all attached.

Mold filled with epoxy/granite, the top layer is 5mm of epoxy/sand for a smoother finish.

Body filler applied and sanded in preparation for painting

Motor pocket metal form work partially removed. See removable cast in position motor plate at bottom of pocket.The form work was wrapped in duct tape prior to casting this provides clearance making it easy to remove.The M8 threaded rods attach the mold base and locate the motor plate and form work in exact position in the mold.

Removable motor plate cast in position. It is wrapped in adhesive plastic film, this provides clearance for removal and epoxy won't stick to it.

View showing the cast in position mild steel 10mm mild steel shoe and 2.5mm mounting faces for the gantry attached to the column

Michael Gilligan06/01/2019 08:43:38
17658 forum posts
811 photos

Great progress, John

... a really fascinating build.


Marcus Bowman06/01/2019 09:44:32
169 forum posts


Looking at the first photo in your post dated

06/01/2019 08:20:45

I would suggest that when you have removed the second side, you clean the mould carefully then pour clear resin into it; polish the finished article, and mount it for display in your workshop. 'Tis a thing of beauty.


SteveI06/01/2019 10:15:55
247 forum posts
22 photos


Thanks for the updates, a most interesting build.

I am looking forward to learning about your linear rails and how they mount with particular reference to your need to lap the steel to the tolerances you quote. Which brings up the question what are you going to machine on this tool and what kind of tolerances are you aiming for?



Marcus Bowman06/01/2019 10:46:32
169 forum posts


Talking of tolerances, what type of leadscrews will you be fitting, and what type of belts to drive the axes (unless they will be driven directly)? I assume ballscrews, but wonder what class/tolerance and arrangements for controlling backlash. The same is true of belts, which vary in the extent to which they introduce errors in positioning.

I know on my own mill, belt design has advanced in the last 10 years, and modern profiles are much better at transferring motion with reduced error.


John McNamara06/01/2019 12:29:08
1314 forum posts
113 photos

Thanks Fellas

Yes clear resin mold would be a great conversation piece.

I would be tempted to experiment with polymer concrete if the shrinkage and cracking can be controlled instead of epoxy but that will be after the Epoxy/aggregate mill is built.

Its pretty clear that the total accuracy of the machine will be a lot less than the accuracy I was able to apparently achieve with the steel rail mounts, I said apparently because do not really know if the surface plate is truly flat to inspection grade i suspect it is within about 3 tenths or for that matter if the indicators I used are in perfect order although I think they are.

Alexander J Slocomb at MIT mentions an "error budget" in his book "precision machine design". also in this paper,

"Understanding and modeling errors in machines"
W >**LINK**
orth reading, it confirms common sense that overall the accuracy of a machine will be the sum of all the separate error generating parts. We must assume that each part will generate some error, nothing is perfect.

The mil I am building is not going to be a micron level super machine, hopefully it will be a lot stiffer and accurate than many of the small import mills being offered and it is designed for CNC at a budget. I wanted the design to be simple enough to allow a skilled person with a small lathe, a drill press, access to surface plate, hand and measuring tools and access to laser cutting for the mold and some of the machine parts. Optionally the mold could be made from MDF by hand.

So far the machine has rewarded me greatly, every day it explains very clearly what works and what can be improved. I the end it will reflect the sum of the parts.

What will I make with it? more machines!

John McNamara06/01/2019 12:55:22
1314 forum posts
113 photos

Hi Marcus

I allowed for double nut Asian rolled ball screws this should reduce the backlash somewhat while not breaking the budget. There will be more lead error than from precision ground ball screws. The same applies to the linear rails. however if higher spec rails are needed they should just bolt straight in as long as the hole spacing is 60mm and the overall height from the bottom of rail to to top of the slider block. Many manufacturers offer this as standard.


For the timing belts GT2 belts appear to offer the best accuracy, more expensive than AT style belts.

Edited By John McNamara on 06/01/2019 12:56:48

John McNamara13/01/2019 12:38:50
1314 forum posts
113 photos

Hi all

And then there were two....

The columns that support the gantry cross member are now prepared and undercoated.

I increased the amount of epoxy to 18% by volume on the first and last layer (I placed and tamped the epoxy/aggregate in 5 layers) This gave a better finish.

The epoxy was allowed to cure then the side was covered in automotive body filler while still in the mold. This was allowed to cure then sanded flat using the mold as an edge guide. The other sides were also prepared after removing the casting after a 24 hour cure. As you can see from the painted images below the result is very crisp lines. I used a vibrating half sheet power sander and a half sheet plasterers joint sander by hand. All the non steel edges were rounded carefully by hand to about a 1.5mm radius. Finishing as usual took a fair amount of time, Finally a coat of etch primer was applied, There are still a few blemishes they will be fixed before the top coat which will be done after the final assembly of the frame is completed.

Comparing the two columns visually the steel mating surfaces are in perfect alignment. I will measure the actual error later when the next mold is removed from the surface plate. The design allows for a maximum misalignment of 2mm. I am guessing the error is under 0.25mm. The column on the left has masking tape over the metal face.

Next is the base, when this is poured I will be able to test assemble the machine and check for alignment, there should not be any problems so far all the parts are within spec. The columns are going to be screwed straight down on the base without any adjustment, hence the importance of their accuracy. When the gantry cross member is placed on the step and bolted in position and the built in jacks used to align the linear bearings it carries to the base ways there will be a small gap left that will be filled with epoxy before final assembly.

The following paper is worth reading. it covers building a precision machine, it uses the same method to assemble the components.



SteveI13/01/2019 16:04:10
247 forum posts
22 photos

Hi John,

Wonderful job. Very interesting and I am working my way through the paper you linked to.



Neil Wyatt13/01/2019 16:18:04
18558 forum posts
723 photos
78 articles

More remarkable stuff, John.


All Topics | Latest Posts

Please login to post a reply.

Magazine Locator

Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!

Find Model Engineer & Model Engineers' Workshop

Latest Forum Posts
Support Our Partners
JD Metals
Eccentric July 5 2018
Eccentric Engineering
Subscription Offer

Latest "For Sale" Ads
Latest "Wanted" Ads
Get In Touch!

Do you want to contact the Model Engineer and Model Engineers' Workshop team?

You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.

Click THIS LINK for full contact details.

For subscription issues please see THIS LINK.

Digital Back Issues

Social Media online

'Like' us on Facebook
Follow us on Facebook

Follow us on Twitter
 Twitter Logo

Pin us on Pinterest