Thomson rail length under tension?

[erikjgreen]

So, I'm working on a design for a router table, and I've been able to source thomson linear bearings for the three axes cheaply ($5 per each double length bearing). They're in 3/4" size, and the pillow blocks are closed (the rod has to be unsupported to clear the block).

I've also got a different type of bearing with rails for the longest axis. The long axis will have the thomson rails supported by aluminum supports from thomson.. no problem there.

But I'm planning to buy two lengths of 3/4" thomson rail for the X (cross table) axis. This axis would be about 36" across, with the rod maybe 46" and supported on the ends only. I'm going to use two pillow blocks on these rails, for a total of four bearings on this axis, spread over 6 inches linear distance per rail.

My gantry is probably going to weigh about 50 lbs when I'm done, which would make the X axis center flex be something around .1" when done, and that's static, not dynamic load.

So I'm thinking about doubling the end supports, and possibly using the outer support to hold the thomson rod in tension to prevent sag. Something like mounting the outer supports on a sliding plate and using a small jack bolt to place the rod in tension.

Any thoughts on whether/how this would work?

Erik

[wildwestpat]

Hi Erik

The rails are just too thin for the intended application. This is why the makers of inear brearings offer continuously supported rails. Whilst it may appear logical to try and tension the rails you will have to apply a very large force and that implies a very substantial frame so it would be better and a lot cheaper to bin the rails and bearings and get the correct ones.

Linear ball races come in two forms closed and split. The Split ones are used with the appropriate blocks to alow the rail ro be supported throughout its length.

The static deflection is only the begining of your problems.

Hope you have not ordered the rails yet.

Regards

Pat

[erikjgreen]

Nope, I haven't ordered the rails.

I'm trying to do this as cheaply as I can since it's a first machine and since I'm strapped for cash anyway. I suppose I could just wait for better rails with supports to become available at my local surplus place, or I do have a combination rail/ballscrew from THK that's nearly the right size to use for that axis. I wanted to have similar linear rails on all axes, but oh well. It seems a shame to not use those closed pillow blocks, but I guess I could double up on the vertical axis for extreme stiffness.

I could also be more conservative and make a smaller machine, but where's the fun in that?

I am going to put together a welded steel base for this thing, not quite a table since it's made to have the router attach to it with bolts, and has no table top, only cross supports.

Erik

[wildwestpat]

Hi Erik

Before you go much further can I suggest that you set out the basic spec for the machine you have in mind - you may have done this. A CAD drawing is not necessary just a sketch on a scrap of paper. Having determined what the machine is for i.e cutting foam or heavy duty steel then decide on the accuracy with which the parts are to be made. Accuracy has two main facets in CNC. First is true dimensional and the second is repeatability from part to part.

Having got the peliminary design spec you can then set about trying various design ideas still at the sketch on scraps of paper. I guess you have done this but then got seduced by some nice looking bits which were too good to pass on. (My good wife says "do you really need that" every time I make an impulse buy!)

What you might have done was then to have looked at the requirements for the rails (or any other mechanism) and their support structure for their ability to support the load with the required accuracy. Unfortunately the load has two components the static loading due to the weight of the parts to be carried and the impulse load caused by the reaction betweent he cutter and the work. Rather than calculate the actual cutting reaction the torque of the motor at the maximum tool diameter is a safe approximation. The tool reaction can take place at any angle and as seen by the 'rails' has a plus an minus value and a direction.

This leads directly into the consideration of static deflection - you have calculated this at least for the vertical dead load. The real situation is that under this vertical load there will also be an implied cutter reaction force at right angles - assuming a vertical mill machine - the cutter reaction force will try and alternately push and pull on the rails. This will set up a vibration which leads to chatter marks on the work. The aim should be for maximum ridgidity i.e. no vibrations. What I am trying to convey is that machine ridgidity (the rails being but one component) is related to the material being cut and the spindle power. Many people are going to try for the maximum machine speed in terms of stock to swarf and this means that machine stiffnes / ridgidity is one of the deciding factors on the quality of the cut. To be absolutely fair it is normal practice to rough out with a high feed rate and then to take one or more light cuts to finish off.

There are other rail methods used that are less expensive. Suggest you might look to use scate bearings running on ground stock supported by a light section RSJ which is part of the frame. Here in the UK house walls are most often double skin brick and the wall above the window opening is suported by a rolled sheet steel section that might make a very good cheap bed for a wood working machine. I mention this as building trade outlets can turn up some nice bits if you have the eye to see out side the box.

Take hope in the fact that everyone who has built a machine from scratch has inevitably bought and collected bits that have not been used on the current build. As a result they have the makings of another smaller - bigger and better machine as well as a well stocked 'junk shelf'.

Onwards my friend.

Pat

[ger21]

Originally Posted by erikjgreen I'm trying to do this as cheaply as I can since it's a first machine and since I'm strapped for cash anyway.

Doing it twice is a lot more expensive than doing it right the first time. Just about anyone who's tried to use unsupported rails like you're talking about has been very unhappy. It just won't work.

[erikjgreen]

Thanks for the feedback, guys.

Here's what I have in mind for this machine, I've got a rough plan drawn up on paper.

0) Gantry type mill using a router for a spindle

1) Will cut mostly wood, MDF, and plastic

2) Repeatable accuracy to about .01"

3) Table size about 4'x5' if possible, smaller if absolutely necessary

4) Materials: metal, steel table frame and maybe bed with MDF sacrificial top, steel or aluminum (I have access to cheap 80/20) router bed frame and gantry.

5) Rails: I was hoping to use linear rails for smooth movement and ease of fabrication, but I could use something different. I have rails and trucks for the Z and Y axes (short and supported respectively) but nothing for X. I have a couple of leadscrews or a short ballscrew that I could use for the Z axis, but for X/Y I was considering roller chain drive or cheap ballscrews.

6) Electronics: I have a HobbyCNC 3 axis board built, plus power supply parts, plus 6-7 surplus stepper motors of varying sizes. I have two large slo-syn motors for X and Y with about 350 and 500 oz-in of holding torque and a couple motors with about 120 oz-in, one of which might work for the other axis. I plan on using Mach3 for control, not sure on design/CAD software yet. The power supply will by about 34 volts DC, 20 amps (I got a big transformer cheap).

I'm starting with designing and welding the table up from 2x2 square tubing. I plan to bolt the router to it for stiffness. The router bed will be basically a rectangle with short feet bolted to the table frame. Cross braces for stiffness across the bed, one each foot or so, with elevator bolts in them to level the MDF top.

Pretty standard gantry design using round thomson rails for the vertical axis with a counterweight for the cutting head and a geared down 120 oz-in stepper for motion. I expect about 4-6 inches of motion for the vertical axis, about 3.5 feet for the cross, and about 5.5 for the long axis.


This is a first project for me, to get my feet wet and to make a woodworking router I can use to make carvings, shape wood panels, etc to make some things for myself and friends. Mostly to learn and make use of the HobbyCNC board and some surplus parts I picked up. If it's not terribly accurate or fast, that's no problem, I just want to make something that works.

Long term I want to build a second project, a metal milling CNC machine capable of about a 30x20x20 work envelope, probably linear rails and ballscrews, with servos driven from gecko controllers, made out of mic6 plate or fortal with a VFD driven 3 phase head. Very heavy and stiff for medium precision machining. Alternatively I wouldn't mind CNCing a knee mill if I could find a used one cheap.

Erik

PS: I have available in my shop a band saw, TIG welder, 9x20 lathe, and a mini-mill to fabricate parts

[wildwestpat]

Hi Erik

Comments inserted as follows:-

0) Gantry type mill using a router for a spindle

Good versatile configuration.

1) Will cut mostly wood, MDF, and plastic

This reduces the demand as a metal cutting gantry for steel would be a tall order if thick material cutting was to be undertaken.

2) Repeatable accuracy to about .01"

The accuracy in terms of repeatability can be achieved but you also need to think about absolute positional accuracy over the whole of the machining envelope. This becomes important if you are going to make other parts that have to fit. Ball-screws good but a chain or better still a toothed belt would be better given the amount of dust MDF generates! I fear that MDF may become the new asbestos so I would build into the gantry an efficient scavaging system for swarf. The link I have given you has a neat apron idea shown in one of their videos. See my response to your statement number 5 for the link.

3) Table size about 4'x5' if possible, smaller if absolutely necessary

Suggest you consider the short dimension as the 'X' travel and the five foot as the width as this will let you machine a full sheet of material in two goes.

4) Materials: metal, steel table frame and maybe bed with MDF sacrificial top, steel or aluminum (I have access to cheap 80/20) router bed frame and gantry.

Like the idea of starting out to design in the sacrificial base and suggest that a sub-base of one inch or better MDF be used as this will have to be levelled to the 'X' axis rails and you are going to have to cross brace the frame anyway. Also consider how you are going to hold the work down to the sub-base. The hold down method will depend on what you are making. For sheet materials a vacuum system is nice an quick if there is enough sheet left!

5) Rails: I was hoping to use linear rails for smooth movement and ease of fabrication, but I could use something different. I have rails and trucks for the Z and Y axes (short and supported respectively) but nothing for X. I have a couple of leadscrews or a short ballscrew that I could use for the Z axis, but for X/Y I was considering roller chain drive or cheap ballscrews.

Suggest you look at comparable commercial machines. One that has been flagged recently on the zone is the 1800 Legacy machine. (see http://legacywoodworking.com/products.cfm?product=6)

This will give you some ideas on the 'X' axis support and more. I am tempted to try my have at a similar one as I am interested in making clocks.

6.) ) Electronics: I have a HobbyCNC 3 axis board built, plus power supply parts, plus 6-7 surplus stepper motors of varying sizes. I have two large slo-syn motors for X and Y with about 350 and 500 oz-in of holding torque and a couple motors with about 120 oz-in, one of which might work for the other axis. I plan on using Mach3 for control, not sure on design/CAD software yet. The power supply will by about 34 volts DC, 20 amps (I got a big transformer cheap).

I always struggle with motor sizing. I hope others will give you some advice as my gut feeling is that you might need more power if you want a high speed action. Geared down enough the motors would do but the question is how much.

The 2 inch square tube sounds Ok if it is thich wall. Don't forget welding reduces the strength so use cross braces and the odd diagona strut or use flat sheet guset plates. Look at the 'Legacy 1800' general construction as this will give you some good pointers on stiffness - those side plates serve several puroposes as well as stiffening the structure - guards the operator from the router - helps with dust and chip control etc.

Hope this helps and we look forward to some pictures as the build progresses.


Regards

Pat

[erikjgreen]

My replies in bold.

The accuracy in terms of repeatability can be achieved but you also need to think about absolute positional accuracy over the whole of the machining envelope. This becomes important if you are going to make other parts that have to fit. Ball-screws good but a chain or better still a toothed belt would be better given the amount of dust MDF generates! I fear that MDF may become the new asbestos so I would build into the gantry an efficient scavaging system for swarf. The link I have given you has a neat apron idea shown in one of their videos. See my response to your statement number 5 for the link.

I'm planning on integrating a small shop vac controlled manually into the design.. I have a spare little unit that should keep things clean nicely. I'm inclined to use chain for motion since it's easy to get.. timing belt would be ok, but I'd have to order that just like the ball screws.



3) Table size about 4'x5' if possible, smaller if absolutely necessary

Suggest you consider the short dimension as the 'X' travel and the five foot as the width as this will let you machine a full sheet of material in two goes.

Excellent idea.


4) Materials: metal, steel table frame and maybe bed with MDF sacrificial top, steel or aluminum (I have access to cheap 80/20) router bed frame and gantry.

Like the idea of starting out to design in the sacrificial base and suggest that a sub-base of one inch or better MDF be used as this will have to be levelled to the 'X' axis rails and you are going to have to cross brace the frame anyway. Also consider how you are going to hold the work down to the sub-base. The hold down method will depend on what you are making. For sheet materials a vacuum system is nice an quick if there is enough sheet left!

I'm thinking instead of a sub base I'd use elevator bolts, which I'd level to the X axis before placing the sacrificial MDF and surfacing it. But if I use a vacuum hold-down system (simple, and I have a vacuum pump) then I'd need some kind of hollow base below the sacrificial sheet that I could seal for the vacuum.

5) Rails: I was hoping to use linear rails for smooth movement and ease of fabrication, but I could use something different. I have rails and trucks for the Z and Y axes (short and supported respectively) but nothing for X. I have a couple of leadscrews or a short ballscrew that I could use for the Z axis, but for X/Y I was considering roller chain drive or cheap ballscrews.

Suggest you look at comparable commercial machines. One that has been flagged recently on the zone is the 1800 Legacy machine. (see http://legacywoodworking.com/products.cfm?product=6)

This will give you some ideas on the 'X' axis support and more. I am tempted to try my have at a similar one as I am interested in making clocks.

I've been looking at this machine: http://buildyourcnc.com/blackToe4x8.aspx He seems to have pretty good success with the chain drive.

6.) ) Electronics: I have a HobbyCNC 3 axis board built, plus power supply parts, plus 6-7 surplus stepper motors of varying sizes. I have two large slo-syn motors for X and Y with about 350 and 500 oz-in of holding torque and a couple motors with about 120 oz-in, one of which might work for the other axis. I plan on using Mach3 for control, not sure on design/CAD software yet. The power supply will by about 34 volts DC, 20 amps (I got a big transformer cheap).

I always struggle with motor sizing. I hope others will give you some advice as my gut feeling is that you might need more power if you want a high speed action. Geared down enough the motors would do but the question is how much.

Well, I was thinking the bigger ones would need just a small step down, maybe 2:1 if that. The 120 would probably need a 4:1 or so. High speed is not important... if it takes several hours to cut something that's ok as long as the whole thing doesn't need a lot of manual tweaking.


The 2 inch square tube sounds Ok if it is thich wall. Don't forget welding reduces the strength so use cross braces and the odd diagona strut or use flat sheet guset plates. Look at the 'Legacy 1800' general construction as this will give you some good pointers on stiffness - those side plates serve several puroposes as well as stiffening the structure - guards the operator from the router - helps with dust and chip control etc.

I'm thinking of using corner gusset plates for the table along with a couple cross braces.. I just got given some spare bed frames, so angle iron is available at the moment for extra bracing. The main tubes will be 14 gauge 2x2 steel with some 1 inch round tubing here and there.

I'm thinking more and more I'll get some 80/20 for the actual router frame... I can get it for about $1/pound at the moment.


Hope this helps and we look forward to some pictures as the build progresses.

Yes, thanks for the input. I'll start a thread in the wood router build log as things come along.



Regards

Pat[/QUOTE]

[erikjgreen]

I've been thinking about my design as I collect parts... I found a guy yesterday who does salvage around here, and he's willing to watch for specific parts for me. I bought a couple giant steppers (2200 oz-in) off him for future use (cheap) but the real reason I went to see him was encoders... I picked up three 1000 line differential encoders complete with helical couplers (they're shaft based) for $20 total.

Since I already have about 6 electro-craft E543 servos from a while back that needed encoders, I'm thinking now I'll use them for my router table. That means I can't use the hobbyCNC board, but I can still experiment with it. The Gecko servo drives are cheaper than their stepper drives, too.

So, on to my new question about linear motion.

Assuming I build skate bearing type rails for my X and Y, how would a well built skate bearing set up compare in accuracy to supported linear rails? I've reviewed a number of threads here for tables that use them and I think it's no problem for me to build some using aluminum, probably with steel pipe or cold rolled flat for the rail.

I'm guessing that the skate bearing setup would be plenty accurate for my stated goals (0.01 repeatable precision) if I do things right? And if I drive both sides of the X axis at the same time I can avoid racking?

It would extend the time needed for my build of course, but it would probably be cheaper than trying to make random linear components that I run across fit together into some sort of solid design.

On a side note I've decided that long term I want to CNC my G4000 lathe too... I expect I'll use most everything I buy for something anyway.

Erik

PS: I'm going to try to draw my design in a paint or CAD program (suggestions welcome, but I don't have money to buy one) and start a build thread.

[wildwestpat]

Hi Erik

Looks as if you have found a good source of second use bits for your CNC adventures.

1. Skate type ball bearings can be used to make very good linear motions. There are some sophisticated commercial offerings that are right up there with the best. For low cost home brews the limits are the accuracy of the track. Your options on available metal for the track-way will determine the actual construction from which you can then design the trolleys for the gantry. There are ball races that have a grove or flange on the outer part and these are useful in reducing the complexity in some circumstances. Some even have 'v' groves and are designed to run on a 90 degree edge. You might get lucky and find some of these.

I would stay away from round tube as the trolley construction will get complicated compared with square or channel sections – this is a very personal choice and available materials will rule!

The use of the slippery plastic sold for making routing jigs might make a very cheap and easily built gantry leg support bearing. I think the 'Legacy Mill' link I sent you uses this approach. The bearing surface appear to be so close together that a simple slider would appear to be used and not ball bearings. I have used one of the fancy low friction materials sold for routing jigs to mke a sledge to enable me to machine cross grain wood that kept grabbing the the router table and fence. The home made sledge used to carry the wood that was being difficult slid like a dream over the cast alloy as if it was on well oiled ball bearings.

2. The racking of the gantry can be reduced by a number of different methods. Obviously making the gantry bearing surfaces long in the direction of travel helps and also increases rigidity. Long (wide) gantry supports will have less tendency to twist and using a drive to each side that is synchronised is the most common one used. Racking can also be sorted by using a figure of eight cord that is made of non stretchable material to link the two gantry legs. I am looking for a length of Kevlar cord with which to experiment. The problem is that there would be some 15 feet of cord linking the sides. (This sort of mechanism was used on large drafting tables when drawings were done full size using pencils! The larger boards were the size of the bed of the machine you are constructing. The resulting drawings were certainly accurate to 1/64 inch.)

You might find some of these old drawing tables being scrapped in which case you might get a couple of top mounted slides that could be adapted very easily for gantry leg supports. These used bar that was ground with a chrome finish and can be 1 inch or more in diameter. The rail being supported with cast brackets or channel throughout its length. A carriage containing ball races traps the carriage and maintain accuracy throughout the 5 foot or more travel..I you mention this as I got one that I used to make a superior circular saw sledge type of carriage for a friend when I was in Ireland. He tells me it is still working well and I am not having it back.

In my book a pile of useful bits is a prerequisite for making things and fuel the imagination for the next project ad infinitum.

Now looking forward to your build.

Regards

Pat