I found some 20mm linear shaft and bearings that are 60 inches long on vxb.com. Would these be strong enough at this length to hold a gantry for the x axis motion? I'm not familiar with how string these are. I would use one 20mm linear shaft on each side and a leadscrew down the middle under the table to move the gantry for its X axis.

I know most people would prefer rails instead - I'm just interested in what people say about the shaft design.

Thanks!

Warren

At 60" long, I would imagine that you would have intolerable deflection and vibration during use if the shafts were unsupported.

Scott

I was going to support them on each end with brackets like this unit in the picture below. This machine below uses up to 36 inch linear shafts on each side the web site stated. I really wanted to stretch it to 40 inches for the X axis instead of using the whole 60 inches. What do you (or anyone else) think about that?

Thanks

http://www.3drouters.com/specs2.htm

http://www.iambigbrother.com/router2.gif

Scott's right. Too much deflection with a 20mm shaft and trying to span 40 inches with any load.

A few things to note;

The link you've given describes sizes and accuracy for a machine with Bishop Wise Carver bearings and that is not what is in the picture - different machine.

From the machine in the picture and guesstimating from the size of the router (3.5" dia) the X axis is ~36" but the shafts are over an inch dia, say 1 1/2". Even then, if the gantry/router weighs 50lbs and with 1 1/2" shafts you're still going to get between 0.002" and 0.006" deflection due to weight alone.

IF you try to span 20mm shaft for 40" in the same design then the 50lb gantry will cause deflection of the X axis of between 0.03" and 0.12" Yes that's almost an eigth of an inch.

Also is this 40" for length over all or X axis travel? - if it's travel its going to be a lot worse.

The range of deflection is the difference between a perfectly fixed end point and a simply supported end. Actual deflection figure will be somewhere in the middle with the design linked.

Andrew

Thanks for the info. I needed to know that before I ordered the shafts and bearings. I suppose the best solution for the X axis is rails huh? If the rails are mounted down on something that will not move any then they should be fine I imagine. I keep looking on all these sites at rails and cannot find ones long enough or find the pricing. Maybe I'll make a seperate post asking for a direct source to find this. I looked on ebay a bunch but only find shorter ones. I want to move the gantry 4 feet for the X and Y axis. Of course, I could always load up on a bunch of roller blade bearings! :)

Thanks!

Warren

Just now I found this setup on this site. He is using two shafts per axis for X and Y. What size do you think those shafts are? I emailed him but may not hear back since he has not posted again since last year. I would guess thats about 3 feet on those shafts?

http://www.iambigbrother.com/router3.gif

There's a freeware application called 'BeamBoy' sorry I can't remember where I downloaded it from but it allows you to calculate the deflection on loaded beams, both hollow and tube of differing materials, if anybody wants this I can trawl back through my backups.

CNC machines are a bit like boats. Add a foot in length and double the price...

Rails are good but not just because they are rails but mostly because they are supported along their length. The problem transfers from support to alignment as you have to align the entire rail *and* support rather than just the ends with a shaft.

The last design you linked is probably trying to half the deflection by spreading the load across two shafts on the X, and this should work. However but the end fixing of the shafts isn't doing much to stiffen or locate the ends of each shaft and you've doubled your alignment problem.

There are ways to reduce deflection of a shaft; Physically support it along some portion of it's length, that would then require open bearings of course, but it becomes something like a rail arrangement.

Alternatively just use a larger diameter shaft and bearing but that costs more.

Finally you can support the ends of the shaft properly. the longer the span or the larger the load the stronger and more complex the end support needs to be. For example Shafting with two clamp supports at either end spaced, say 4" apart with significantly reduce the deflection. Thats pretty much the defelection calculated as the lowest deflection in my post above. It will still deflect, just a lot less than one support. Four end supports stiffens the shaft considerably by forcing the deflection to bend the shaft in a recurve - or three bends, one each end downwards and one at the point of load upwards. The worst thing to do is just drill a hole through a piece of 1/2" mdf or AL. and either just bolt through the ends or just clamp it in place. This means the deflection only has to bend one curve and the ends are effectively hinged - this is the largest deflection case in the note above.

I've attached a very rough spreadsheet I did a while back to calculate deflections in this stuff. Simple supports are like the second linked machine. Fixed supports are mathematically similar to the first machine, just with two clamps at each end of each shaft, or approaching a rail where it's supported along its length where you can add the shaft *and* the support into the bending calculation.

Whatever, a 4 foot travel with small deflections will be expensive however you do it.

Thanks for your info fyffe555! Its funny you mentioned about supporting it along the shaft and using open end bearings. I found a shaft (only 16mm though) that is 41" long but it has 9 tapped holes it in on one side and includes 2 open end bearings. So the shaft could be mounted down all along and include supports on each end. This would work just like a rail also. They were $89 each which I don't think is bad including 2 bearings per shaft.

They are all the way at the end of this web page.

http://store.yahoo.com/cnclinx/linearslide.html

Good write-ups, fyffe. I like your style.

Kudos.

Scott

Warren,

A fixed shaft would be better and the smaller diameter would be ok as long as its load capacity met your needs.

I forgot something which Scott ( thanks Scott!) picked up first time round - vibration. You're way better off with fixed rails or shafts rather than end supported shafts when you consider the vibration caused by actually moving and cutting.

When working out the deflections under the weight of the axis its easy to forget the real loads are the forces of moving the cutter through the piece. A 1/2" bit in a 3hp router can produce *way* more than 50lbs of load and it will do so in all directions. If a Shaft is just end supported then it can and will be deflected under cutting side to side as well as just up and down.

Have you looked at hardened shafting at McMaster or MSC? their prices for the real stuff are not much more than drill rod and I think cheaper than the site you linked to. You could get the 20mm open cars from vxd and shafting from McMaster? For a 48" travel you need some pretty hefty rails..

Misumi has some good and fairly economical stuff too.

Andrew

Thanks, I'm satisfied that I either need to use rails or shafts that are tapped and can be mounted all along its length to prevent movement and vibration.

Can I take a drilling rod and tap threads into it myself along the bottom to mount down? I can get a 12 foot drill rod from ToolAndDie.com for $75 and cut it to length.

Or cnclinx.com has one already tapped every 9 inched but its not long enough for my needs. Wonder if I cut them and mount them evenly together to make a longer rod if that will work?

Warren

Drill Rod isn't hard enough. The bearings will wear grooves in it. You need to use hardened shaft.

Thanks about the drill rod. I'm not doubting you but on the web site it states the following about them being hardened. They offer this drill rod as water hardened or oil hardened. Do you still think they will get grooves? Seems like that would be alot of wear to put grooves into them. Just my thoughts though... :)

Oilcrat Oil-Hardening Drillrod, 12ft. Made of AISI/SAE type O-1 steel, annealed for easier machining. These 12 foot long, nominal diameter sizes are made to exacting standards for tolerance, size, straightness, surface finish, and uniformity. Each piece is centerless ground free of decarburization and then polished to final size. Ready for packaging in a sturdy box for prompt delivery and easy storage.

I'm using 20mm VXB bearings + rails on a 36" x axis and there is at least a .015 or more droop in the center with no extra force and I can defect it alot with hand pressure. I'd say that about 20" is about the max span that you could have acceptable results from 20mm on a light machine.

42-45mm is more like what I need for a 36" axis for unsupported rails.

Now THAT is some real world, no BS info right there. Thanks for the insight, anoel! Good info!

Scott

Thanks about the drill rod. I'm not doubting you but on the web site it states the following about them being hardened. They offer this drill rod as water hardened or oil hardened. Do you still think they will get grooves? Seems like that would be alot of wear to put grooves into them. Just my thoughts though... :)

Oilcrat Oil-Hardening Drillrod, 12ft. Made of AISI/SAE type O-1 steel, annealed for easier machining. These 12 foot long, nominal diameter sizes are made to exacting standards for tolerance, size, straightness, surface finish, and uniformity. Each piece is centerless ground free of decarburization and then polished to final size. Ready for packaging in a sturdy box for prompt delivery and easy storage.

Gerry's right. Note the word 'hardening' not 'hardened' and 'annealed for easier machining' . These rods are not yet hardened and its not feasible to harden this stuff at home, and even if you could they wouldn't stay straight or round.

The Ball bearings have point loads on the shaft. At any time you've only got a few dozen balls in contact with the rail and they've got to support the entire weight/load. They will and do cut up drill rod easily and quickly.

Nice lateral thinking though..

Andrew

I was going to support them on each end with brackets like this unit in the picture below. This machine below uses up to 36 inch linear shafts on each side the web site stated. I really wanted to stretch it to 40 inches for the X axis instead of using the whole 60 inches. What do you (or anyone else) think about that?

Thanks

http://www.3drouters.com/specs2.htm

http://www.iambigbrother.com/router2.gif

Where do you get alum extrusions this big? I have looked and looked and cannot find it.

Mike

Mike,

I think those are just 3x6" extrusions. 80/20, T-slots et all. have them.

http://www.8020.net/T-Slot-5.asp

look up 40-8016

Carlo

Where do you get alum extrusions this big? I have looked and looked and cannot find it.

Mike

Hello Mike

I own one of these machines the aluminum rails are the 3060 - series 15 from 80/20. You can find them at there e-bay garage sale store, here's the link.

80/20 Garage Sale Store (http://stores.ebay.com/8020-Inc-Garage-Sale_W0QQsspagenameZL2QQtZkm)

Later, Steve

More 'real world' info..

Check out my gallery for my home made router.. I'm using 20mm shafting supported by alu extrusion on both x/y axis.

gallery (http://www.cnczone.com/gallery/showphoto.php/photo/310/cat/500/ppuser/1399)

A dial indicator put on the spindle will show about 0.2mm deflection on the table if I get up and stand on the gantry (75kg) with the table centred.

The only real hassle was drilling/tapping the hardened shaft to bolt it through the extrusions to the table supports. But it's still possible to do accurately with a simple wooden jig and drill press :)

So far it's been a very rigid workhorse of a machine, and built for a fraction of the cost of linear railing type slides.

Furry,

That's a very nicely engineered machine. Send me one!..

Your deflection numbers are interesting. If I understand it you can measure a 0.2mm deflection on the table - so is that the X axis deflecting only or have you got the indicator on the Y rail ? where are you measuring the deflection between?

You're using 20mm shafting and the extrusion supports are spaced so there's ~12-14" unsupported between the supports? If the axis is between the centre two supports then shaft can be considered a 12" shaft with fixed ends and a centre load at 6". With your 85 kg (nothing like mixing units!) on top of the middle span the deflection of the span should be about 0.07mm or 0.003 inches.

If the math is right then your 85kg is probably also deflecting the Square tubular frame under the rail the extra .1mm or so. IF you're going to mill with 85kg loads on the gantry then in theory you would support the centre of the frame with a further upright or triangulated and so stiffen it up and reduce the deflection by more than half. As is the thing is obviously stiff enough to maintain really good deflections.

How did you go about drilling and tapping the shafting? Have you screwed up from the extrusion into the bottom of the shaft?

It would be interesting to compare what the formulas say with your real world measure. If you're measuring the table deflection without support from the Y or Z when standing on it maybe you could try supporting the frame under the X rails further with a quick bit of 2x4 or something and jump on and measure again and see if the deflection is reduced.. ... If you decide to try it let me know..

You've done a 'proper job' as they used to say back home..

Andrew

Thanks. I think the deflection measurement is a bit innacurate.. the dial indicator does have a fair bit of play as it was a 'gimme' from a friend. The indicator was place on the router holder, so was measuring both x&y deflection combined.

Good to get some theoretical measurements..thanks for the info. I wasn't too concerned about the .2mm Z axis defection, it was more important to be tight and true in the XY axis as it mostly does 2.5d stuff.

Yep, the shafting is screwed upwards through the supports. Drilling is easy once through the hardened surface. Use an angle/die grinder in a stand and simply grind through it. Drilling and tapping into the soft core is easy from then on.

Quite a bit of trouble to save money on an extremely critical part of the machine. I opted to buy THK rails after going through rigidity issues using 3/4 inch shafting on another project. You can never have too much rigidity, particularly when you are working with metals. While it is possible to use shafting and get acceptable results, after spending the money on all the other parts of the machine, do you really want to compromise the carriage that holds the whole thing together? Swede's site www.5bears.com has great information on selecting rails for a CNC application. Not every rail is suitable for every application.

That said, its perfectly feasable to use shafting, but I'd stick with 36mm or larger if you are spanning the distances you are talking about. Linear shafting is typically chrome plated and/or surface hardened to make it last under the high contact forces created by point contact with the balls. (nuts)

Hey Furry, can you tell me where you got the open end linear bearings for the 20mm shafts? How much were they? The machine I'm building is using the roller bearing on pipe concept but for my second machine I'd like to go with your methods.

Thanks!

I'm not sure why 36mm shafting is necessary if the shaft is supported and screwed down over it's entire length. Most of the loads applied will be normal to the shaft surface and be distributed to it by 2 or 3 bearing circuits minimum. Essentially in this configuration it's similiar to railing anyway.

Not dismissing your opinion (which is very valid!) but I've seen many commercial machinines using THK railing with very poor underlying supporting structures which flex more than even some unsupported round shaft designs. Which simply goes to say, the design is often more important than the components in some instances.

After saying all that, if you can find me some cheap 6ft THK railing and guides, I'll happily convert my router tomorrow :)

Warren.. try www.vxb.com and look under linear motion.. not a bad deal on 20mm open bearings with housings.

I actually bought my roller bearings from vxb.com and saw their linear shafts there before. So its $40 per 60" shaft and $25/open pillowblock bearing. Thats $90 per side. Not too bad.

One question I have is how did you align the shafts before you mounted them down? Did you just mount one side and then with the gantry on it mount down the other side?

Thanks

Warren

If U buy hardened drill rod U won't be able to put the holes in it.
If U buy it soft, U should run nylon or something like that on it. The joint could be tricky. I've seen thks butt togather. They abrasive cut the 2 ends to be joined. "Perfectly" square.

As you guessed, the shafts were aligned by tightening them down after the finished gantry was sliding on them. Getting it all squared up afterwards only took a few hours by making test cuts in acrylic sheet.

I'll also post some pics of a smaller router I made using unsupported 20mm shafting. I'm still using it to this day as it's extremely precise... perfect for wax models, thin metal sheet and inlays where there isn't a lot of load on the router/spindle.