Oops! How not to undertake double sided milling of hard wood

Finally, I got around to cutting a composite block of cherry wood. It was formed using six pieces 10 x 130 x 110mm and I had glued the pieces together to create a solid block. I used standard Titebond III glue and left it to set over 24 hours. The machinability of the block was great with the standard carbide uncoated cutters (#201 & a 1mm tapered ball from SpeTool) leaving excellent finished surfaces behind them. The feeds and speeds I had used were the automatically calculated values by MeshCAM, which I had placed on trial for the last two weeks.

Not being content with trying to machine hardwood for the first time, I thought that I should also machine double-sided complex geometry to boot. The machining of the first side was great and all proceeded according to expectations. The flip side started well enough and then the workpiece split longitudinally. It was more or less along the glue line although slightly to one side. I don’t believe that the glue bond gave way but more that the forces imparted by the cutter, and possibly that large opening, had created an issue which manifested as a lower resistance to deformation/breakage.

It reminds me to ask about flip jigs. I had used a datum line from the centre of the block in both X and Y directions and lined them up with similar lines on the spoilboard. Does any reader know of a flip jig that can be used with a standard SO3 or at least have an idea where I can find one that I can make from a plan.

Lessons learned: MeshCAM suggested feeds and speeds are great for single sided jobs. I should exercise more judgement about feeds and speeds when undertaking a job which is double-sided. More attention must be paid to double-sided work with regard to the stresses that the component is being placed under during milling of the second side. Pick an appropriate material for the finished component. Had I successfully finished making my cuts in the wooden version, I would have probably left it at that.

The reality is that the component should have been made from metal which means that this job will take me a little longer but eventually be finished as an aluminium component. I need to source the aluminium billet and the grade of aluminium. I am not sure about whether to choose 5083 or 6082 and I am leaning towards the T6 6082 because it is apparently much better for machining than 5083.

I have already purchased Niagara cutters that are designed for cutting Aluminium and Brass and the wood test demonstrated that my machine is cutting accurately. I will take care to start with very conservative speeds and feeds and will attempt to use my copy of Millalyzer static to inform my decisions.

Overall, this was an interesting start to understanding what Sophie is capable of doing when combined with MeshCAM, decent cutters and a carefully considered plan. As always, the image is worth 1000 words. All comments would be appreciated. Thanks to @WillAdams for drawing up the component and providing it in the STL format for MeshCAM. If @Vince.Fab has anything to add about aluminium material choices for the component (the top will mirror the curved bottom and there will be a 5mm difference at the peak of the curvature) that would also be very much appreciated.

Disaster image follows:


The image as an STL file:


Your crack looks like it must have been a defect in the wood itself. Unless the project moved and got cross ways with the bit there is no reason for the wood to split that except it had a crack in it from the start.

I do wood turning and when you start turning wood you release all kinds of internal stress. Thinks like bark inclusions or foreign objects embedded in the tree. Some trees grow on hill sides so one side of the tree has more stress in it than another piece of wood. You never know what you will find in side of wood. If a piece of wood explodes on the lathe it can go 360 degrees but it always seems to find the few degrees where the operator is standing. Wood does act strangely when you start whittling away at its structure and does unexpected things.


Thanks Guy. I had not considered that the grain would potentially form something similar to cleavage lines that are found in rocks. That curving line does look as if it followed the grain. The line of unbroken grain near the cutout at the top of the image does seem to have followed the broken line at the bottom of the image. My own stupid fault for not considering all that needs to be understood before starting up the machine. Enjoy the day.

Dont beat your self up. It is impossible to tell internal defects in wood. Hind sight is always 20/20.

1 Like

Bad luck @jepho … but it’s really just one more successful failure.

The model in Meshcam has an inner curve and an outer curve, and it looks like there’s a fair amount of material in the moon-shaped cross section.

However, the cut piece has a rectangular channel cut out of the back of it and is much thinner. Is there some other shape you cut out of the back after making the piece?

Thanks Guy. I don’t feel so bad if it is impossible. :grin:

Hi Gerry, yep, I used a chisel to cut down both sides and try to see if I could find any reason for the failure. It all looked great except for the point that Guy made about stressed areas. I did not see anything and to my untrained eyes it looks like I just did something wrong.

Hindsight is great and I will make a flip jig to contain the next piece, metal or wood. I am aware that I made the clamps very solid to ensure nothing moved because I wanted everything to match up when I flipped the workpiece. I am wondering now whether I have overdone the workpiece security. The split is along the X axis and I discovered that MeshCAM always flips along the X axis. I now think that I may have caused the issue when securing the workpiece. It is possible that I had clamped it too hard along that axis, to ensure the two milled halves mated correctly.

1 Like


Some alternatives you probably know about that don’t exert sidewise pressure on the stock, that might be worth experimenting with, are:

  • Drill holes into the stock and secure it to the holes in your spoilboard using bolts. Flipping the stock is simple too since you have the holes as locators

  • If you don’t want to apply pressure to the stock at all, try a right angle bracket mounted on the spoilboard for locating the piece when you flip it, plus blue-tape-and-super-glue for fixing. This is surprisingly firm.


Yes, I had thought about registration pins. I have several 1/4" parallel A2 stainless dowels that are 2 inches in length so they would work well. I also purchased a high quality 1/4" drill bit for a different project but it would help me to register the stock accurately.

I like the idea of keeping the stock free from stresses. I have now managed to amass a fair bit of experience with painter’s tape and CA glue and I have a lot of confidence in my usage of that method. I will buy a little angled bracket and drill it for the 6mm Cap heads I use with my Gator clamps. It would then screw directly into the spoilboard sea of holes.

I am awaiting some 6082 T6 T651 plate to arrive. I should get it by 9th April and then I can concentrate on milling it correctly. It’s weird… in October I was struggling to make my machine fit together and then it needed another 6 weeks to fettle it and make sure it was running true. It hardly seems possible that now I am about to mill aluminium in a complex shape. Very satisfying and fun.

Been watching this progress with great interest - it’s the sort of thing I like to do. Part of the diagnostic of a failed cut like this that I run through is: a) what if I had cut it the other side first - would the stresses have been better absorbed? b) what if I added end-tabs to give some more structural rigidity? c) if I re-glued the split would it be possible to complete the job in-situ and contemplate a minor dressing cut to tidy up any imperfections? …provided the integrity of the piece hasn’t been completely lost owing to the split, d) making it in something like aluminium - with the different learning curve it brings (speeds/feeds, chip load, coolant/lubricant) and to ponder whether the event that caused the split in the wood is still ‘there’ and in aluminium would instead result in a broken cutter.
All fascinating stuff!!

1 Like

Hi Andy. I had thought about which side of the workpiece to start the cuts from. I figured that the external arc would be stronger and less likely to give way when there were only the four tabs holding it in place. I thought to place the tabs midway on the body of the structure and made then 5mm wide and 2mm thick. My rationale for begin the cuts on the side of the internal arc, was that the weaker structural strength would be supported by more stock during the cut.

As MeshCAM is a new software to me, I had no idea how to place supports at the ends correctly. The software wanted to append the supports from under the edge at the end (the corner if you will) and then run the support longitudinally under the long edge. I could not encourage the supports to be moved or added in the way I had wanted to apply them.

I did not think about repairing the split. I used the opportunity to surgically remove far too much material with a chisel, in the somewhat forlorn hope of discovering a reason for the failure. In hindsight, I now wished I had waited to get a sense of what the community hive mind could bring to the problem. Instead, I trashed the piece and then threw it in the trash, when I could not make any sense of it.

The aluminium should be fun to work with. I have purchased some really nice cutters which should be able to cope without difficulties. The aluminium stock is eminently machine friendly and I will temper the fairly efficient (aggressive?) feeds and speeds from MeshCAM to align them with the recommendations of Millalyzer. Achieving a nice finish and dimensional accuracy with aluminium will be a great first step in machining complex geometry and will be another brick in the CNC wall for me.

I will use registration pins and a right-angled support (as suggested by @Gerry) to remove all externally applied forces acting on the workpiece. Once I have produced a reasonable facsimile of the MeshCam drawing in Aluminium, I will revisit the work in hard wood and see what I have been able to learn from the experience. I am really happy with the Shapeoko. Compared with my two previous machines, it is properly capable and the community help is just fantastic.

i would approach it like this
Dont center it in the stock, when you flip it you will have it resting on the finished edge rather than floating supported by the tabs.

1 Like

Hi Michael. Thanks for your suggestion. I had not considered this possibility.

My question is you can see that the crescent shape component (purple) would be resting on its thinnest aspect. In your drawing (side 2) there is sufficient stock (green) to rest on the waste (all of the green area) next to the purple crescent shape.

Removing the internal arc waste (yellow) in the side 1 illustration is trivial. Even a set of camps would not be in the way of the milling cutter. I don’t understand how to remove the remaining waste (green) in side 2 and still hold onto the component while the external arc is milled.

The work-holding solution to resolve most of this situation is likely to be registration pins. There is nothing to hang tabs from. Pins would still need to have stock which is not going to be removed, as tabs would.

The second set of illustrations makes the supports explicit and I get to see what you are considering. I need to go away and consider how each of the suggested solutions will apply to and impact milling weightier aluminium stock piece. I appreciate your giving up the time to share your thoughts for my benefit. Thank you.

When I’ve done really delicate two-sided flips my solution has been to machine a fixture which the cut half will fit into — this solves the problem of registration as well.

1 Like

Determined to make me work harder eh? :grin:

Edit: I should have said that it is a good idea, nevertheless.

1 Like

I machine aluminum almost exclusively and can say that there is nothing that beats a dedicated fixture for second operations on small, delicate or difficult to hold parts. They are a little more work to design and set up but they are worth it. Especially if you are making more than one.


Thanks for the advice Nick. :+1: I am only going to be machining the part as a one off. I liked the method proposed by @Gerry of using a piece of angle screwed to the spoilboard. I would opt for two pieces so that I can create a right angle to butt the material up against.

It will be crucial to centre the component on the work piece and all of my centred work is also placed at the rapid centre in CM. The flip along the X axis should place the billet in exactly the opposite orientation. That is to say that I expect the Y- edge to become the Y+ edge while leaving the X axis unchanged.

I will also want to use 4 x 1/4 x 2" registration pins set into the spoilboard. (belt AND braces) With the proviso that the hole centres are strictly equidistant from each other, I should be golden. The supplier of the billet of aluminium is providing it to my stated measurements and it will be cut square with smooth edges. I think there is a ±2mm condition imposed on all cut to order requests but that will only affect my workpiece placement.

As for designing a dedicated fixture, I thought about the suggestion made by @WillAdams and that, when combined with your advice, points to the sense in creating a half fixture. It may be beneficial (albeit a little slower and less efficient) to mill each side as a single job and just ensure that the registration of the billet is perfect. I say this because I suspect that I had set up MeshCAM wrongly and did not see an equal half depth of my stock being cut. The original cut should have been 14.5mm deep based upon half the stock thickness but it turned out to be 29mm deep. :roll_eyes:

1 Like

I think the other reason I like fixtures is because I hate tabs. When cutting aluminum, tabs can be problematic because of how hard it is to plunge into with these machines. End mills are not usually meant to cut with the face. It isn’t really a problem with soft materials but it can cause issues when you start cutting the shiny stuff. My process is to use oversized material in a vise for the first op and then flip and hold it in a fixture or a vise and machine away the remaining material. I do this so I never have to deal with tabs.

1 Like

An excellent point — one cool thing for designs w/ holes such as this one can machine away around the holes first, then insert a fastener (I use nylon or aluminum) which mate into threaded connectors in the fixture, then complete the machining.

1 Like