Router Forums banner
1 - 19 of 19 Posts

·
Registered
Joined
·
29 Posts
Discussion Starter · #1 ·
If I'm clever enough to build a CNC router with no backlash and absolute accuracy in terms of motor steps to mm of movement, my wee brain tells me that the only source of errors in the final cut will be deflection of the structure of the machine due to the cutting forces exerted on the tool. I'm assuming all errors will be such that the tool cuts short of where it should be, in other words not enough material removed rather than too much.

Sooooo, a roughing cut will take off most of the material and then a finishing cut, designed to minimise the deflection forces will result in a wonderfully accurate final cut.

Can any of our more experienced readers prove this to be too simple a theory? Am I likely to have problems with too fine a finishing cut failing to remove any more wood and simply rubbing over the surface generating heat? Is there any merit in running the same finishing cut twice or is that just wishful thinking?

The benefit of your experience will be much appreciated.

Kit
 

·
Registered
Joined
·
1,216 Posts
If your first finishing pass is pretty aggressive then it may be worth it to run it twice. If you do a clearance pass with a bit not too much larger in diameter than the ball end bit you finish with, then I wouldn't both doing the finish pass twice.
 

·
Registered
Joined
·
1,584 Posts
I'm going to be buying a cnc soon, and I can't wait to get into headaches like this! Just kidding. I'm very glad to know about potential problems, and what people have done to correct them. Thanks.
 

·
Registered
Mike
Joined
·
3,843 Posts
Kit it is important to leave some material on the project for the finish pass. What you leave will depend on the size of the bit you want to use. You don't say what software you will be using.

If you leave too much material then the finish pass might not give you good results because of over loading the bit.

If you don't leave enough material then you might not get rid of your roughing cut lines and would git a poor finish cut.

With some software you can also do REST Machining that will cut the small detail that larger bits can't get to.

Every machine is going to be a little different but with a few small 3d projects you should be able to tell where your settings need to be for your machine.

What I normally tell new owners of CNC machines is to go through the alignment procedures and get everything as close as possible. THEN use your machine and have fun, DON'T NIT PICK THE LITTLE THINGS OR YOU WON"T GET ANY PROJECTS FINISHED.
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #5 ·
Wise words Mike, you have to get a feel for the beast sooner rather than later.

I'm in the process of upgrading my machine with a more rigid gantry, stiffening for the unsupported fixed rails, larger motors and better anti-backlash thrust bearings and nut-blocks on the ACME leadscrews. I've not tended to bother with finishing cuts for the bits and pieces I've made in the past but am now looking to make more accurate cuts. What I'm really asking is 'have I missed anything?' Can I assume that errors produced by flexing of the machine's structure during a roughing cut can be effectively removed by using a finishing cut (or even two) designed to put much less stress on the framework.

Kit
 

·
Registered
Joined
·
372 Posts
Can I assume that errors produced by flexing of the machine's structure during a roughing cut can be effectively removed by using a finishing cut (or even two) designed to put much less stress on the framework.

Kit
I would not make that assumption. Any moves that cause rapid direction changes can also cause flex in a structure, even if the cutting loads are low, the inertia of the moving parts is the main culprit (blame that Newton guy”s first law :smile:). You can lower your acceleration rates and speeds, but that has a adverse impact on cutting times, and will not “fix” a non-rigid machine. It also means you are not cutting fast enough for the chips to carry away heat, causing burning, overheating and reduced bit life.

That’s the main reason I personally don’t think machines like the shapeoko, xcarve, and shark type machines are really suitable for serious woodworking, although I’m sure there are many who can’t afford better and happy to have even their capabilities.
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #7 ·
Thanks Richard, I hadn't considered the acceleration imposed forces when asking the question. There's no doubt that overall rigidity is the ultimate aim but this iteration of the machine is still using plenty of MDF in the framework so is less than perfect in that respect but not as heavy as some.

I already have some solid ideas about what my next, all new, machine will look like and it won't include any MDF, but I'd like to get some useful work out of this version before I start work on that one.

Kit
 

·
Registered
Joined
·
372 Posts
A lot can be done with wood to make a rigid machine. The wing droop on the spruce goose was/is much less than much smaller metal planes. I’d love to see pictures of what you have and what you’re planning.

One of the most knowledgeable posters on the CNCZONE DIY CNC Router forums personal machine is wood, the machine he has been designing/building (for the last 10 years!) to replace it is also wood.

If you’re making new parts, consider using torsion boxes for stiffness, and Baltic birch instead of mdf w(better screw holding).

Richard
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #9 ·
Richard,
The machine I'm upgrading is shown in the link below. It's primary weakness is the lack of stiffening in the gantry and the 1m long unsuported fixed rails which flex under load. The new version re-uses much of the hardware but replaces the belt drives on the long (Y) axis with ACME leadscrews and stiffens everything up as described above.
The next upgrade will be a completely new machine capable of taking 800 x 600mm sheets of ply (full sheet cut into 6), probably with a concrete frame and 30mm (40mm?) unsupported ground and hardened round steel shafts for the fixed rails.

https://vimeo.com/240094760

Kit
 

·
Registered
Joined
·
372 Posts
If you have not purchased the unsupported round rails for your future upgrade, please look at SBR type supported rails. They are relatively inexpensive (compared to profile rails at least). Look at the build @MikeMa is doing. Any type of unsupported rail will always have more flex than a supported one.

Looks like you’re doing good work with your existing machine! The acme screw upgrade should be a great improvement. Look at replacing the spiral type couplings with Oldham style as well. On my machine it made a big difference (I only had one on Z).

I like homebuilts!
 

·
Registered
Joined
·
1,042 Posts
If you have not purchased the unsupported round rails for your future upgrade, please look at SBR type supported rails. They are relatively inexpensive (compared to profile rails at least). Look at the build @MikeMa is doing. Any type of unsupported rail will always have more flex than a supported one.
Just to add, the primary reason for me going with the SBR20 on my build is the added rigidity that they add with their support. If you are looking for savings, both ebay and aliexpress have bundles for all 3 axis that are fairly cost effective. I got a bundle deal on ebay, with one caveat. I decided that the rails that came for the X axis weren't long enough, so I did replace those with longer ones to get a few more inches of travel.
 

·
Registered
Mike
Joined
·
3,843 Posts
If your roughing cut is too aggressive it could cause cuts bad enough to cut into the project deeper than the finish face because of deflection in the machine so the answer would be yes you can have problems.
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #13 · (Edited)
Richard,
Thanks for the positive comments on the work so far. The clock parts have been sufficiently accurate to keep the clocks running but there are measurable errors and it's not up to making parts intended to accurately mate together. Very slow feeds and speeds are needed to keep the distortions down, cutting out the clock parts took 90 minutes and I do see wisps of smoke and a blackened tool on occasions which is not a good sign! I've seen several comments on the forum regarding the merits of Oldham v spiral couplings and will take note.

Mike,
SBR rails are definitely the other option I'm thinking about for the next big upgrade. However my thinking in using unsupported rails is to avoid the need to make the underlying frame absolutely straight and parallel as is required when the rails are bolted down. I know of one other project where self-levelling epoxy resin has been used (link below) to get a perfectly flat surface for the rails to sit on but my thinking revolves around buying high quality, accurately straight round bar of a large enough diameter to avoid significant flexing. Any fixed bend in the rails can be measured and the rails oriented to have both bend directly up or down which puts the resulting errors in the Z axis which is only an issue when cutting pockets where depth is critical.This means that parallel alignment consists of shimming the end mounts without worrying about the accuracy of intermediate fixing points and the frame simplifies to something resembling an upturned table with short, rigid legs. This design idea is really at the heart of my question at the top of this thread. Unsupported rails of a given diameter are noticeably cheaper than the SBR version so this looks to be a cost-effective option at the moment. I'm only doing this for fun so the cost is a critical part of any design decision.

I'm also concerned about differential expansion of the steel rail v the aluminium support. Temperatures in my shed vary between about 10C to 50C so this may be an issue.

Timescapes - Digital Timelapse Discussion - View topic - CNC Router Build

Kit
 

·
Registered
Joined
·
372 Posts
I'm also concerned about differential expansion of the steel rail v the aluminium support. Temperatures in my shed vary between about 10C to 50C so this may be an issue.
Has not been an issue on mine.

I have 1800mm SBR rails on my X and 1500mm on Y, and my machine has seen temps from 40-100 deg F (5-38 C) with no issues. The coefficient of thermal expansion is in the 10 to the -6 power, on my machine less than a mm difference which is accommodated by the steel slipping relative to the aluminum (hard polished steel over an anodized surface is fairly slick). The parts will always essentially be at the same temp as each other, and the bolts tying them together have clearance holes on the aluminum side, they thread into the steel. The clearance holes accommodate the slight relative slipping.

Straightness of the mounting surface is much more critical when using profile rail than with SBR. My extrusions were plenty straight enough.

For the size of your machine, I think a 16 or 20mm SBR rail attached to a wood torsion box or other substantial frame would be stiffer than a 40mm unsupported steel rod. Looking what’s available to me (in the U.S.), the SBR rails/bearings seem cheaper than buying that size unsupported rails/bearings, but I know pricing can vary significantly depending on where you are.
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #15 ·
I love forums like this one! So much experience available for the asking.
I do have some Jarrah timber (Western Australian hardwood popular for use as underground railway sleepers in Victorian London) recovered from a 20 year old car-port which I assume is reasonably stable by now which could become a framework for the next machine. I have no experience that can tell me if a 40mm unsupported rail will be better than a 20mm SBR rail screwed to a length of this timber after a pass through a thicknesser which I have access to. I'm still open to suggestions as to what is the best cost/performance balance for my next build.

I have added a precision spirit level to my Christmas list (I also have a lathe to align) so I may be able to hand sand a piece of this timber to a perfectly flat profile if I put in enough effort. Something to try out as an experiment. If nothing else, the exercise will do me good:smile:

Kit
 

·
Registered
Joined
·
1,042 Posts
All my SBR is mounted on 8020 extrusions or aluminum plate. No issues with flatness. The friend who helped me design my machine, his first machine his SBR rails were mounted directly to an MDF torsion box bed, and no shimming was needed.

From a cost standpoint, for unsupported rails, you wanted want to go with thicker rails. The Shark CNCs I think use 1.5" (~38mm) rails for a comparative size machine to mine, which is using 20mm SBR.
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #17 ·
Mike,
I've had a quote for 40mm unsupported rails which isn't too painful but the linear bearings are a LOT more expensive than the smaller sizes. That's before I've considered organising freight or collection from a supplier 1300km away (the nearest!). The rails are 12kg each.
From the comments in this thread, and from looking in more detail at other designs, I'm thinking that is not the best way to go and will look at something a bit more conventional, probably using 20mm SBR on an MDF or ply box base.
I'm continuing with the current upgrade to the existing machine though I have a sneaking suspiscion it will mostly be used for cutting shapes for the more serious new build.

It's really useful having all the comments and sugestions from other forum members. Thanks to everyone who's contributed to the discussion.

Kit
 

·
Registered
Joined
·
29 Posts
Discussion Starter · #19 ·
Richard,
I hadn't though of that, they only went live a few days ago.

Just had a look and there are 2 vaguely relevant results under 'linear bearings'. Thousands under 'bearings' and plenty under 'linear' but nothing under 'SBR20' etc. I guess we'll have to wait awhile for the catalogue to fill up and the index to mature.

Kit
 
1 - 19 of 19 Posts
Top