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  
Craig Smith 313/05/2019 11:20:27
7 forum posts

Hello John,

I added a comment to your last entry on the "How to build a Mill with Epoxy" on Hackaday forum regarding an alternate supply of epoxy here locally in Oz.

But now can I ask you this.

With the experience you have gleaned from this E/G building process, if you were to do it over, is there anything you would do differently?


Craig Smith

John McNamara13/05/2019 16:47:19
1314 forum posts
113 photos

Hi Craig

That's a difficult question for me being a bit of a perfectionist. There are many small details that I could improve on and that can go on infinitum! However There is nothing that I really regard as a gotcha,

A better choice of aggregate would have made the process a little easier, The mix was hard to tamp down, As I explained casting over a thick granite surface plate made vibration of the whole mold impossible It would require a massive vibration table, And even if I could have vibrated it I would not like my surface plate to be subjected to that sort of abuse. So I simply tamped and tamped! I still got some surface defects I just filled them after the castings were de-molded. For the next build I Will used a finer aggregate that flows better.

I chose Megapoxy because It is well priced, It is fairly thin when mixed and and it does not contain solvents, I works well for me and the company that sells it is very helpful. I know West system's products are popular with the boat building fraternity in particular. I am sure it would work well also. I have used Araldite (The retail 24 hour product) To Make a cast in place bearing. Google "Epoxy bearing material and method", It also worked well in that application, It is a lot thicker than Megapoxy. And much more expensive. Fine for small quantity use.

I guess if you work with composites all the time you would make a high quality mold apply release agent and then cast it. My choice was to apply self adhesive film to all the laser steel panels before they were assembled is a departure from standard practice. No allowance for draft was made. Angling the sides of a mold as you would do for say a boat to allow easy release from the mold was needed as the panels were all removed one by one. I wanted 90 degree corners. The plastic sheet did its job well. Yes, I ground off all the sharp corners with a small angle grinder and a flap wheel, then fine coated with polyester body filler, sanded and applied paint. I got a good finish For a "one off" a lot quicker process than preparing a perfect multi part mold. Remember no draft.

Previously I tried using several coats of release wax. It was an melamine MDF mold. The wax got scratched by the aggregate while tamping. The mold had to be pried off in pieces. What waste of time.

Maybe my biggest error is in not starting my project sooner! Too much research and overthinking and not enough building. The Heavily steel reinforced machine frame I have made has surprised me. Its turned out far better than I imagined.

I am starting to feel guilty for not posting here for a held up at the moment waiting on the linear bearings and ball screws to arrive so I can get cracking.


Craig Smith 314/05/2019 22:19:51
7 forum posts

Hi John,

Hey thank you so much for taking the time to answer. I too am a person who enjoys excellence in everything I do, and it’s easy to recognise that trait in you. The quality of work you have shown us all is just fantastic, and I know to achieve that takes full engagement and a lot of focus. Goodness knows how you have made the time to then record it for us all to see. Just amazing.

John, it’s been a great learning experience following your build and reading your posts. Over time I have had need to make many different kinds of moulds and patterns, and like you say, some with draft for easier release and others using a similar technique to the ones you made where the sides are square to the main face and so they are disassembled after curing.

I wonder if there may be some members that could glean a little from my experience with mould making and epoxy resin and apply it to an Epoxy Granit part? I will take the chance and share some of my experience and if anyone finds it of interest, I could elaborate if asked.

I learned a lot about making these kinds of moulds from an old friend who has a large pattern making business. They would box up around the pattern, which normally would create hard sharp corners all round, then after the mould release step was completed, would apply an epoxy gelcoat against the pattern and up the sides of the boxing. Next would come an epoxy paste or “tie” layer. An intermediate layer which was applied with a brush, added as a way of tying the rougher inner core to the outer surface finish. This is really just a mixture of epoxy resin and a thickening agent of some kind. Most likely Fumed silica, which in Australia is sold as Cabosil, in Europe and elsewhere called Aerosil. Not the kind of stuff you would ever want to breath in. Mask is mandatory. I think adding this layer is relevant as an option for Epoxy Granit construction.

This tie layer of epoxy paste is allowed to reach its gel stage where it takes on a comparable hardness to modelling clay, but still remains a sticky, then the bulking mix is added. The mix is usually a ratio of one-part epoxy to four or five parts washed sand. Even with such a high epoxy resin content, and a fine even aggregate, if the tie past wasn’t used, there would be surface imperfections and air bubbles just beneath the gelcoat waiting to cause issues once demoulded. So the dense and smooth tie layer acts as a buffer between the surface finish and the rougher aggregate as well as giving a good resilient interface for the bulk fill to bed into and bond well.

If you would like to get a nice surface finish without filling voids after demoulding, this technique allows for a great finish straight out of the mould, and done well creates a surface that is ready to use as is, or at worst only needs an undercoat after a good sand back. I have made moulds this way and, in an attempt to achieve a nice surface finish out of the mould as well as save on resin, have applied a sand/epoxy layer directly after the tie past, and then mixed a coarser aggregate with a far less percentage of epoxy to fill the larger interior volume. This once again minimises the chance of air voids close to the surface and gives a stronger outer shell less likely to crack when the inevitable knock happens.

To be cont.

Craig Smith 314/05/2019 22:20:23
7 forum posts


Instead of using an epoxy Gelcoat, it’s even possible to spray an epoxy undercoat like Dulux Luxepoxy or a two-pack polyurethane directly onto the mould surface. Then add the tie past layer which protects the surface from aggregate attack.

After the epoxy moulding has cured and different epoxies require differing curing processes, I apply a cosmetic fill to the rough B surface as John described using a car filler. Personally I use a mix of a faster curing epoxy resin and glass spheres, which is normally sold under the trade name Q-Cells. I will throw in a bit of Cabosil or talcum powder to toughen it up as the Q-Cell mix on its own can be a bit light and fluffy and when spread can have trapped air bubbles. By adding a bit talc or Cabosil it smooths it out which helps it to spread better.

The advantage of using this mix is that it sands very easily without clogging the sandpaper, and you can even plane it with a hand plane. It doesn’t have the shrink back of a large mass of car fill, which normally uses a polyester resin as its base.

Not using a tie layer and just going with the main mix against the mould is certainly quicker and in all uses less epoxy resin, but it does require more finishing if that is important to you. Also the larger aggregate mix is more likely to crack if knocked than a more resin-rich mix with the finer sand. On smaller parts I will even substitute the sand with silica flour. This is ground up silica (basically sand) looks just like talc, but feels a lot more abrasive to the touch and is a lot cheaper to buy. It can be mixed in with epoxy at a ratio of five parts flour to one-part epoxy and makes a much denser and stronger mix. Another thing I have done to toughen up and strengthen the initial sand/epoxy layer is to add some chopped glass fibres to the mix. These fibres are about 6mm long and make the resulting mix incredibly strong and a lot more resistant to cracking if bumped hard.

Okay, this is quite long. If anyone is interested, I can elaborate a bit more on mould release systems and different epoxies and ways to strengthen, or stiffen (two different things) aggregate parts. If not, I hope this has been of interest to someone that is not that familiar with the process.



Chris Atack28/06/2019 09:01:34
1 forum posts

Hi John.

What are you going to use for the table?


John McNamara29/06/2019 02:35:15
1314 forum posts
113 photos

Hi Chris

And to the patient souls that want me to get cracking on this project. I have been away from the workshop, very frustrating....

The base design is uses laser cut 10mm plates spaced by shouldered steel inserts that are threaded and will form the points of attachment for work pieces. there are also inserts threaded to attach the bearing blocks. Rather than have the epoxy visible on the sides I designed side plates that also locate all thread reinforcement bars. the completed assembly is filled with epoxy from the underside. through a multiplicity of holes.

Once assembled and mineral cast the top and bottom will need to be made flat and parallel. this is a job i plan to farm out to a surface grinding contractor, I could do it by hand scraping around all those holes! but it would take a lot of time, it is also above my skill level. I could also mill it on my vertical mill, but not to the level of accuracy obtained by grinding.

The table is probably the most difficult of all the components of the machine to produce as a fabrication rather than a conventional cast iron casting. I did consider using a single thick steel plate, that would also work fine. as would a cast iron casting. With t-slots. or tapped holes, it will depend on what machinery is available to the builder.

The images below show the design I am using.
The 25 x 25mm Aluminium angles are part of the swarf guarding

Neil Wyatt29/06/2019 12:37:05
18558 forum posts
723 photos
78 articles

Nice to see progress John!


John McNamara03/08/2019 14:35:25
1314 forum posts
113 photos

Hi All

At last I have had some time to work on the mill. I have installed the rails for X,Y and Z. The images below only show the X and Z rails. I have removed the cross-member sitting on the Base casting and it covers the Y Rails, it is very heavy. Also note that the cross-member is rotated 90 degrees from its installed position. to allow drilling and tapping.

The first step was to drill the rail support bars using the Jig I built, I showed this jig a few pages back in this post but thought I should attach a couple of photos showing a test set up here. As you can see I utilized part of an inexpensive (I found it in a bin!) hand drill stand, it was made by AEG. The column attached to the removed base was 25mm mild steel. I simply replaced it with a piece of 25mm shafting I had to hand. I turned a boss and fitted the column to a piece of mild steel plate. this plate was also drilled to fit one of the linear bearings that was later to be used on the machine Also note the second boss at the rounded end of the plate, this was precision bored to fit the 25mm drill bush carrier.

The drilling bush is hardened and ground with a short taper that engages the rail when the bush carrier is pressed down perfectly centering the drill. see the second image below. At the top of the bush carrier is a second hardened bush, the long series 5mm drill is fully restrained.

Type 25 linear rails as used here are drilled 7mm and counter-bored for for M6 cap screws. The difference between the 7mm hole and the 6mm screw allows the rail to be accurately positioned after drilling and tapping. This sounds like a lot however it relies on very accurate placement of the mounting holes. both with regard to spacing in this case 60mm, not just for one rail but two! Both rails must be parallel.

Any errors above 1mm (+-0.5mm )and THE RAILS WILL NOT BE ADJUSTABLE

I thought on this for some time an the method I have designed is the result. It has worked very well. It relies on accurately positioning the rails before drilling using the cams on each side of the rail and a known accuracy ground straight edge. A stretched piano wire and a cheap USB microscope would also provide an excellent reference, particularly over several metres where straight edges become very heavy hard to find and expensive. I have done this before, it takes more work but the result will be the same.

Anyway in this case I have a good 1200mm straight edge and I used it. As you can see I used only one edge.

The first step in the process is to accurately position the straight edge parallel to the rail supports, The cams have limited travel and an error here will create problems later. It does not have to be in the center as long as you have access to it for a dial indicator. I used a 0.001mm dial Indicator.

Once the straight edge was clamped in position (It must not be moved until the job is complete), the rails then can be positioned with the cams using the straight edge and dial indicator mounted on a linear bearing to align the rails to straight edge. The cams are drilled and countersunk for a 30mm M8 screw, by rotating the cams clockwise until they bind then then tightening this screw the cams press against the side of the rail foot. Once all the cams are tightened you will find the rail is clamped quite tightly. There is a bit of a learning process to get this procedure right but after a few trials I was able to reduce the error on the indicator to + - one 0.001mm division, as the rails are at this point not screwed down the error will be greater than this but good enough for drilling and tapping.

I used a G clamp to hold down each end of the rails, They were bowed down slightly. Don't assume that the brand new rails you receive from the factory (In this case Bosch Rexroth) will be straight. They have to be bolted down to a flat surface.

Below are 4 photos of the setup used to drill and position the rails using a straight edge.

I will describe attaching the saddle in more detail later, here it is temporarily bolted to the cross-member rail bearings. using the straight edge still installed below it runs within 0.001mm following the rails as expected.
Also note the vertically attached bearings ready for the Z axis carriage.

I temporarily bolted on the Z axis carriage. centering the bearing rails to 165mm exactly. 

The Z axis spindle; This is a big question?
As seen below I have placed a fabricated spindle I started making a while ago. It was designed to have a BT40 taper, Not done at this time.
Once the machine is finished I want to use it for steel cutting, It has the mass and rigidity required to do this.
However I Also have a ubiquitous 2.2kw Chinese spindle given to me by a friend, see photo, I will use this first.

Next is reinstall the cross member

John McNamara04/08/2019 12:02:00
1314 forum posts
113 photos

Hi All

Before reinstalling the cross member the ball screw mounting plates need to be set up. These plates are located in pockets in the casting attached by 4 M4 countersink screws. There is a space under the plate to allow its height to be adjusted to the correct mounting height for the ball screw assembly. once the final height is established the void behind the plate will be filled with epoxy by injecting it through the central hole in each plate.

The plates will be drilled for the ball screw bearing after marking out in position.

In order provide an opposing force to the adjusting screws small pieces of rubber were placed behind the plate. A straight edge with an attached dial indicator resting on the rails was used to align the mounting plates. A simple process.

Three photos below:

Below is a photo of the tapping setup used to attach the rails. After drilling I used a modified T handle tap mounted in the drill press. The T handle has an attached shaft that slides in a bush mounted in the drill chuck. This keeps the tap vertical and also reduces the chance of tap breakage. After about fifty tapped holes in "gummy" 16mm steel laser plate I am pleased to say there was no breakage. Carefully cleaning the tap for each hole and liberal dose of Trefolex aided the process.


Michael Gilligan04/08/2019 20:11:29
17658 forum posts
811 photos

Keep it up, John

... it's looking amazing.


John McNamara07/08/2019 14:30:35
1314 forum posts
113 photos

Hi All

Now that the rails are attached I have reassembled the cross member and test aligned it. This is not going to be the final assembly, there is further steps to go that will require it being disassembled again.

As the photos below show I have test aligned the components using the linear bearing sliders reference points. I wanted to test the alignment system built into the machine. As noted earlier in this post there are built in adjustment screws to enable the cross member to be set parallel and perpendicular to the to the rails that will support the table. I had no trouble aligning the reference points to .01mm on the dial indicator if it was the final assembly I would have used a .001mm indicator. The rails are already aligned to +-.001mm to my straight edge.

As you can see I have used a Moore & Wright precision square sitting on a parallel bridging two linear guide bearings.

The linear guide bearings have a ground reference edge, this has been used to check the height using a dial indicator siting on a steel block to increase the height resting on a linear guide bearing. (The assembly was moved alternately to the left and right rail until the zero point on the indicator was the same for both sides)

The height adjustment is made using the two vertical jack screws on the top of the cross member, you can see a T handle and hex key if you study the photo. There are also 4 jack screws to set the gantry to perpendicularity to the table guide rails. see the two Allen keys I left in position. I have also marked the various points referred to with arrows.

Also as noted earlier in this post once the machine is finished I will fill the small gap between the columns and the cross member with metal filled epoxy. **LINK**
The jack screws will have done their job. I do plan to do the first cutting tests without the epoxy in place. That will be an interesting experiment. It will significantly reduce the vibration damping of the cross member. It will be interesting to see how much?

Reflecting on the design of this machine and in particular the way the rails for X,Y,and Z are all located on flat planes that are generated separately and the way the rails were easily aligned to very high accuracy makes me realize that the the biggest errors will be caused by my scraping and lapping of the bearing support plates. Before assembling the machine I will revisit the lapping and try to improve the accuracy I documented earlier in the post. I guess this is being obsessive but It will be worth the effort. Unfortunately all the geometric errors, temperature effects, deflections due to gravity and vibration etc will be compounded, Time will tell what the final result will be?



Edited By John McNamara on 07/08/2019 14:32:01

John McNamara08/08/2019 10:57:35
1314 forum posts
113 photos

Hi all

I test fitted and aligned the X axis saddle that carries the Z Axis today. Those following this post will remember I set the gantry yesterday using a 0.01mm dial indicator set against two bearing blocks. With the carriage in place I was able to use a straight edge allowing me to test the full travel of the carriage. there was a 0.005mm error in the height. No problem, I tweaked the height of the gantry casting to reduce this to less than 0.001mm along the full travel. Remarkably this was better than the accuracy I obtained setting up each rail individually, that was using 1 bearing block. The saddle is mounted on 4 bearings. Averaging must have worked in my favor. I am very pleased with the result.

Testing the X Axis travel height.

This photo shows testing the X axis travel near the tool in the horizontal plane.
Remarkably no adjustment of the alignment done yesterday was necessary.
The indicator moved less than 0.001mm

Once the table is made It will be installed and tested in a similar manner.


Vic08/08/2019 11:12:48
2750 forum posts
1 photos

It’s certainly coming on John. I’m looking forward to seeing it in action.

Derek Greenhalgh02/05/2020 20:13:36
26 forum posts
4 photos

Hi John, I just read the thread from the beginning, including some of the links. At this moment I look like a fish gasping for breath, an amazing build and will follow to the end and hope to see a video of it working.


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

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