CNC - Temperature Distortion

For discussions related to designing and making ski/snowboard-building equipment, such as presses, core profilers, edge benders, etc.

Moderators: Head Monkey, kelvin, bigKam, skidesmond, chrismp

Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

CNC - Temperature Distortion

Post by Richuk »

Does anyone have any advice or information on how changes to workshop temperature will effect accuracy and or distort a CNC machine? Assuming dimensions circa 8' x 3' (shopbot style build). Does this partly explain the use of large heavy frames? The suggestion I have read todate suggests a 40f is sufficient to cause an issue.

I have been reading statements suggest that extrusions can be filled to reduce noise and vibration. Does anyone have any suggestions? I was thinking expandable foam (this puts wiring and any coolant pipes outside the frame).

Has anyone used VDS within their CNC build ... vibration dampening :D
IslandRider
Posts: 80
Joined: Wed Dec 19, 2012 11:00 am
Location: Kelowna BC

Post by IslandRider »

Heavy frames absorb vibrations better, and generally make the machine more stable. For ultra precision machines bigger heavier parts will allow you to hold ultra tight tolerances better, all other things being equal. I worked at a place that ground carbide to .0005" tolerances, and they'd run the wheels with coolant cycling for 45 minutes or so in the morning to get it all up to temp before touching parts. That said, an 8'x3' router type CNC isn't going to be able to hold tolerances anywhere near what temperature drift will affect. Shopbots may be good for +-.005" or so in soft material like wood and foam realistically, and temperature drift won't affect that. If you're trying to hold tolerances tighter than that it isn't in wood or foam, and an 8'x3' machine isn't the right tool. A heavier gantry or frame will get you a better surface finish with aluminum, but only assuming the bearings and everything else can handle it, and even then you can make lighter parts just as stiff. Filling extrusions might dampen out vibrations and possibly affect surface finish on aluminum or really hard woods, but seems like a pain in the ass for the effects you'll actually see. And foam fill won't do it, you need weight, and all the negatives that implies for motor sizing and running speed.

If you're buying a machine I wouldn't bother filling anything, just figure out what it's capable of and work to that. If you're building a machine just design it to be rigid enough in the first place and it will be fine. Then put the whole thing on some rubber feet to separate it from the floor for noise. Until you turn the spindle on and the cutter touches something, which is what causes 99.999% of the noise.
Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

Post by Richuk »

Thanks for the insights IslandRider.

It's a CNC selfie, limited to cutting wood. From what you have said, starting the build this winter isn't a big issue, as accuracy is derived from using the correct critical parts.

Cheers!
Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

Post by Richuk »

I'd appreciate anyones opinions on the following assumptions - they are broad brush and I will try and add links back to research material later.

Loss of accuracy on the Z axis is more critical than in other areas, hence we see a lot of ball screws being used. Rack and pinion is possible, but would require a high holding torque stepper motor and greater attention to the gearing of the motor.

Flex within the gantry is another issue affecting accuracy on the Z, so limiting the travel of the gantry to what is actually required is a limitation yielding a positive result. The accuracy of the carriage and rail system is more important here, on the Z-axis, than elsewhere.

The torsional rigidity of the extrusion used to create the Y-axis plays a critical role. Bolting together two extrusions to create a beam is a common solution, but it may not be necessary. It may depend further on the design of the rail e.g. bolting cold rolled plate to the extrusion. V-groove rails and bearing perform well enough, but the interface between the Z and Y axis with be enhanced by the introduction of a third rail.

Ideally the X axis performs best when it behaves as a simple rail. This assumes the Y-axis is not affecting the performance of the gantry carriage assembly i.e. twisting. Whether the X-axis is able to perform as a rail depends largely on the accuracy of the rails and the ability of the carriage to accomodate any inaccuracy. It is common to see a carriage with four or six wheeled carriage, although three wheels appears to be an option when using a casting. I wonder whether three wheels, (the third being used to remove slop) and two spring tensioned spindles (one free wheeling) would be sufficient.

Vibration ... it comes from the act of cutting and can come from within the motors themselves. I wont have the option to leverage the mass of the frame, so I was thinking I would build the gantry carriage assembly using an aluminium sheet/wood/aluminium sheet sandwich. I think this should create a very stiff, light and damp structural support, isolating the X-axis from the Y and Z assembly.

Any thoughts at this stage would be appreciated, I'm trying to get to the point whereby I understand enough to make informed choices.
twizzstyle
Posts: 2204
Joined: Tue Mar 07, 2006 8:25 pm
Location: Kenmore, Wa USA

Post by twizzstyle »

I built my entire thing from steel. It was overkill, but its heavy and stiff. With the machine running, even doing heavy cutting, I've felt all over the machine and can feel virtually no vibrations anywhere.

My x and y axes use 20mm rails with linear bearings. This was the easiest solution for me, and I got all of the parts fairly cheap on ebay (no issues so far). The linear bearings I use have virtually no lateral play in them at all, but they do have a tiny amount of off-axis play. On the x-axis I use two of these bearings per side to remove this play. On the y-axis, which has two rails, I have four bearings.

I actually use ballscrews on both my y and z axis. I have rack and pinion on my x-axis, but it has built up some backlash over time as my gear alignment isn't perfect, and the gears have worn slightly. Replacing the gears should get me back to where i started.

The steel has been great for rigidity/vibration, but I was taking a bit of a gamble since I welded everything together. Everything turned out solid and square, but if my welding skills weren't up to par I could have ended up with a very heavy, very crooked, frame that was unusable. With the driving gears on the x-axis disconnected, my gantry (that must weigh about 75lb) slides with almost no effort.
IslandRider
Posts: 80
Joined: Wed Dec 19, 2012 11:00 am
Location: Kelowna BC

Post by IslandRider »

The machine I built sounds very similar to Twizz'. 20mm linear rails on x with two bearings per side, and 16mm on y with 2 rails and 4 bearings total. Main frame is steel, but I used aluminum in the gantry cross pieces to keep the weight down a bit. No issues at all so far, and linear motion stuff is dirt cheap on ebay. I looked at v-groove bearings, but decided it wasn't worth the hassle when proper linear bearing rails are so cheap. Once you add up all the bearing parts, and buying a bunch of cold rolled flat those v-groove setups end up getting pretty expensive quickly. I spent ~$550 on all the rails, bearings and ballscrews, with custom machined end sections on the ballscrews to suit. Including freight from China. Hard to beat really. Regardless of how you do it, as long as there is a way to adjust tension on whatever bearings you choose, you shouldn't see much slop in the bearings themselves, more on the drive side of things.

Z-axis accuracy up and down (like the one you're concerned about profiling a core) is the easy part, and like you mention a ballscrew is the simplest way to do it. The part of z accuracy you're considering with all bearing details and stiffness actually results in inaccuracy in x and y from cutting loads. First step is to make an estimate of what your actual cutting load is going to be, and size parts not to deflect based on that. I personally wouldn't worry too much about damping, wood is very forgiving to work with, and vibration issues would have to be pretty severe for you to notice it cutting soft materials. Metals are a different story, but you're almost designing a different machine if you mainly want to cut metals. For a first pass design, just come up with an x, y, z rail arrangement that can put the tool where you want it, without worrying about 'performance' details. You will quickly find that there are about a million other details that become more important than exactly how well the bearings are performing.

I posted a bunch of pictures of my build on CNCzone a few days ago. Total cost to build was less than the partial kits cncrouterpart.com sells. Link is here if you're interested:

http://www.cnczone.com/forums/diy_cnc_r ... hread.html
Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

Post by Richuk »

Thanks Twizz, from what you have said, I can tone down some of my concerns. I think I have read somewhere that the gears are usually made from slightly softer material for the purpose of preserving the rack. Its good to see a consenus on the design of your two machines. I can see myself spending a few hour drawing something up in the near future!

IslandRider, thank you for the link. The belt drive you have is an inspired solution. The sourcing of parts on ebay has already started (I'm gathering links)... unfortunately ebay.co.uk is about a 1/3 more expensive than .com. Import direct from China is probably the way I need to go. I'll spend a bit more time reviewing your build log.
IslandRider
Posts: 80
Joined: Wed Dec 19, 2012 11:00 am
Location: Kelowna BC

Post by IslandRider »

Richuk wrote:The sourcing of parts on ebay has already started (I'm gathering links)... unfortunately ebay.co.uk is about a 1/3 more expensive than .com. Import direct from China is probably the way I need to go.
I feel your pain, same goes for .ca vs .com. Best bet would be to contact the seller directly. Its all getting shipped direct from China anyways, I'm sure they'll be willing to work with you. None of the stuff I ended up with was straight from an auction, I sent drawings and he quoted and machined everything to my specs.
User avatar
MontuckyMadman
Posts: 2395
Joined: Fri Jun 20, 2008 9:41 pm

Post by MontuckyMadman »

pcb linear
sammer wrote: I'm still a tang on top guy.
Cornice
Posts: 73
Joined: Mon Sep 10, 2012 10:08 pm
Location: AK TOWN

Post by Cornice »

My old PR shopbot is one sloppy jalopy beast. It is probably the most pieced together CNC router being used on this site. You guys have some nice equipment!! The PR machines axis run tracks called uni-strut. Its this stuff you get from the hardware store to mount whatever on.
Anyways my point is i would be LUCKY to get a tolerance of .01" on any axis. It does not make a bit of difference in my ski making. Cores seem identical, holes are where i wanted them, ski bases are identical. What more do ya need?
knightsofnii
Posts: 1148
Joined: Tue Jan 08, 2008 6:02 am
Location: NJ USA
Contact:

Post by knightsofnii »

just get a cncrouterparts kit, or have them customize one for you.

a cnc is quite an expensive project to declare a "do-over" at any point.
Doug
Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

Post by Richuk »

Thanks man, but its getting done.
gav wa
Posts: 381
Joined: Tue Feb 18, 2014 8:58 pm
Location: Perth

Post by gav wa »

you will be surprised how much vibration and deflection you can remove with different approaches to the tool path, tool cutter angle, tooth pitch etc.

If you have a round edge cutter (like a cove bit) you create cutting forces at multiple angles and also quite a bit up verticle, which helps hold you work piece down to the table and also stop tool deflection, or in the case of small machines, slide deflection.

Practice different tool paths too, you will work out what works best too, like making sure you have a cut width bigger than 60% of the tool width, so you have the next cutting tooth enter the cut before the previous one exits, this usually makes a cut heaps smoother.

Climb milling is better for tool life but usually worse for machine rigidity, though this is sometimes worth trying both, not every job is the same.

If you travel along the length of the core making the core profile, you should only cut the profile from the tips towards the centre, from each end, this way you are using the side of the cutter not the bottom, which stops the cutting action trying to push your Z axis up while you are feeding it down. You end up actually pulling down on the Z axis and removing backlash.

I'm only new here and new to board building but have spent 15 years in some major machine shops. Even spent some time teaching CNC programming to people.
I've been out of the machine shops for a few years now but hopefully I will be able to remember enough to maybe provide some help at times. Although there have been a lot of beers between then and now. :?
User avatar
MontuckyMadman
Posts: 2395
Joined: Fri Jun 20, 2008 9:41 pm

Post by MontuckyMadman »

Hey gav wa sounds like you got some good info.
Cove bit type for milling hardwood? Linear style profiling?
Or rather parrelel finishing?
i have my longer core at 1680 x 155mm milling with a 12.7mm straight 2 flute bit at 129mm/sec engage so it takes about 18min per core with my 210mm/sec transfer rate at 16k ish rpms. Im thinkin switch to a 26mm 3 flute straight bit. And get it down to maybe 12 min per core?
My finish right now is awesome couple quick scrapes with the razor blade and its ready to press.
Would love to know your thoughts.
I see allot of use of intial horizontal roughening follwed by another parrellel finish to cut mill time but im not there combining those toolpath types yet.

Your tip on cutting tips in towrds center to prevent distortion and backlash is baddass.
I have an all steel shopbot so i dont think there is much but some maybe.
Thanks
sammer wrote: I'm still a tang on top guy.
gav wa
Posts: 381
Joined: Tue Feb 18, 2014 8:58 pm
Location: Perth

Post by gav wa »

Mate I just wrote a page long post on all sorts of stuff for you, including crappy pictures I made, and it logged me out as I pressed submit :evil: :evil:

It is late here now, so I will rewrite it all in the morning, sorry mate.
Post Reply