[peteeng]

Hi Thunza - For your application if you get straight extrusions or steel foundations they are close enough. Even plywood foundations can be lapped true enough. Epoxy has to be at least 6mm thick to self level and epoxy is effectively rubber, why mount your rail on 6mm of rubber? Peter

[Thunza]

Hi Thunza - For your application if you get straight extrusions or steel foundations they are close enough. Even plywood foundations can be lapped true enough. Epoxy has to be at least 6mm thick to self level and epoxy is effectively rubber, why mount your rail on 6mm of rubber? Peter

I did find the idea a little odd as i know epoxy isn't all that tough, maybe a little harder than rubber but I definitely see your point. I imagine perhaps with the weight being spread over the rail and 4 carriages perhaps that makes up for the epoxy not being that hard? Just a speculation for me though.

I did have an idea, maybe you could level with epoxy then router away material so you only have a very small amount of epoxy left in height. As a router would be referencing off the flat epoxy surface it would preserve the shape. Well hats my theory but I'm not entirely sure it would work. Perhaps having too little epoxy could be problematic if it gave way, deforming more than the surrounding hard metal


When you suggest lapping instead is that something that's feasible to do yourself? I've used an angle grinder and granite plate marked with blue.. stuff to flatten a bobbin sander table in the past successfully so I imagine the practice would be the same for leveling the rail supports, the only thing is getting a wide and long enough flat reference, I think other people also use engineers levels for checking coplanarness.

[joeavaerage]

Hi.
I agree with peteeng, standard extrusions and hollow sections are usually straight enough for most purposes.

Some years ago, before I started with CNC I made a 'Trammel' mill, which relied on a 1.5m length of 50 x 50 x 5 RHS steel on which ran a shuttle which
in turn carried a router. I used it for making wing cores for model aeroplanes. Clearly any deviation from straight would cause an inaccuracy in the wing cores.

I did not have much in the way of measuring equipment at that time, certainly not a granite surface plate for which the measurement almost demanded, and
the best I could measure was that the piece of steel which is just as supplied, was better than 0.05mm from straight. I concluded that the material was good enough to use
as is, without be straightened or machined or anything.

50 x 50 x 6 RHS in Christchurch is $37.00/m (NZD)....and that's retail, if I bought a full length, 8m, it would be cheaper again.

Once you start welding it then you'd really need to stress relieve or it will be significantly bent, but if you don't weld, but bolt and/or bond then you'd probably not have to straighten or
machine anything. Even if you do want to get it machined flat, then; when still in short, say 1m or less lengths and not assembled in any way you could get then milled or ground
for bugger all. The biggest surface grinder in Christchurch, at least to my knowledge, has a capacity of 1.8m.

Steel is still by far and away the cheapest and stiffest material for engineering projects. Don't make things over complicated or try to use exotic materials or techniques when a simple
and well thought out design in steel will do the job every bit as well for a small fraction of the cost. That means you've got money left over for a real good spindle, tooling, coolant/dust extraction
or what-have-you.

Craig

[joeavaerage]

Hi,
on an unrelated matter just purchased (EBay) an Atlanta Drives low lash (<1 arc min), high torque (120Nm) servo reduction box for a fifth axis for my mill. Its new old stock for
$226USD shipped to my shipping agent in Portland Oregon.

I have another one, again Atlanta Drives, slightly larger (400Nm, <2 arc min), for the fourth axis/trunnion, and have had for a while.I have just finished fitting that drive with a 750W Delta servo, and its working great.

Craig

[Thunza]

Steel box section is what I'd like to go with, I'll.

I was under the impression for mounting hiwin rails rails you need an extremely flat surface or the bearing wont run smoothly. My end product tolerance will be much lower than what many people making metal parts need, perhaps this is what you are partlysuggesting?

Thanks

[joeavaerage]

Hi,

I was under the impression for mounting hiwin rails rails you need an extremely flat surface or the bearing wont run smoothly.

There is a grain of truth to that assertion, certainly the linear components work best on a flat surface....but how flat is 'extremely flat'?
My machines have been by-and-large either milled to say 0.01mm or ground to even less, but having said that if the surface were no better that 0.05mm I suspect that would be
okay to. Were the 'humps and hollows' close together then yes you might expect increased friction in the rolling elements, but the more likely gentle bow of 0.05mm should not
be a problem.

If you plan on assembling your machine rather than welding it then it might be possible to machine the rail surfaces prior to assembly, which would then be a simple task for any halfway
decent machine shop.

I am inclined to be a 'perfectionist' and want this ground and that machined and laser aligned and this and that!!! The problem is that if you adopt that approach the project goes
well outside your budget or just simply beyond your capacity. You'll get more satisfaction from a machine accurate to 0.1mm than a machine accurate to 0.01mm that you can never
get built due to cost or complexity. One of the things that CNC as a hobby has taught me is to be realistic about what I can achieve and what I can afford. Perfect is the arch enemy of
good enough.

There is one machine characteristic that cannot be compromised ever, and that is machine rigidity. You will be very disappointed if you build a machine only to find it cannot do as you
want or intended due to lack of rigidity. For this purpose use the stiffest materials and techniques you can devise within your budget, and steel is pretty much the top of the list in terms
of material properties, cost, and manufacturing versatility. There are any number of engineering companies offering services like machining, heat treatment, profile cutting for steel and or iron.
Such services do not come cheaply but achieve excellent results and must be considered cost effective.

I would highly recommend steel construction for your project, it versatile and as tough as hell.

Craig

[peteeng]

Hi Thunza - No point in epoxy levelling then trying to machine it, Instead of buying expensive epoxy buy a good reference straight edge. This will allow you to tell if a piece of metal is straight. Steel RHS can be "crowned" or "cupped" Aluminium extrusions can be flatter. Again look at your application, I agree square rails can jam on surfaces, been thru that a couple of times. So consider round rails but get the medium priced ones not the cheapest ones. Cheaper then square but very good for making timber parts. I'll try to find a link. We still lap things flat. If you can get a good flat piece of metal say 600mm long you can lap a surface very flat with patience and wet & dry. If you look up 3 surface lapping, you can lap something to extremely flat easily just takes some time and effort. Peter

Better quality round bearings

https://www.aliexpress.us/item/32568...andl_shipto=US

[peteeng]

Hi Thunza - Steel is your best bet in terms of stiffness. I suggest you look at heavy channels vs hollows. These can be bolted up hard vs hollows which are hard to bolt to. They are the same stiffness as hollows but are less stiff torsionally. But good frame design handles the torsion. Peter

[peteeng]

Hi Thunza - I looked at 1000mm + 4 cars on BST site. TBR20(round) $198USD Hiwin20 square $270USD BLH20 square (bst house brand) $176 and SBR20 (round) $117USD seems the house brand BLH is the go for value and performance. I've used the BLH for a couple of years now with no complaints. Latest machine I've used the SBR just to give them a go....Peter

[jckstrthmghty]

I have the BLH20 for my x & y and they are excellent. I should've bought them from the beginning. There should be no concern for any wood cutting application.

[peteeng]

Hi Thunza - Consider making a UHPC base such as attached. You can have steel or aluminium foundation plates made, these would have levelling screws built in. Then you level the foundation plates and seal the edges to the concrete top. Pour epoxy in slowly through the holes until filled and then let cure. Chase the threads and then you can assemble the machine. Concrete machine bases are very stiff and very damp. There are many examples in the forum. Making a mould from MDF or plywood is straightforward and does not require it to be to the micron. As the foundation plates can be made from flat and straight material and on a mill they are accurate and only need aligning... You could make the gantry the same way....

https://www.cnczone.com/forums/diy-c...ml#post2538998 XYZ-spec machine easily cuts aluminium... concrete (UHPC) is low cost and a very good cho9ice for this application. Peter

[Thunza]

Thanks a lot everyone, this is far more help than I expected. I have some thinking and more research to do for now.

Perhaps I'll make a built log thread in future once things start in physical terms.

[hanermo]

It´s not hard to level steel surfaces to machine tool rigid rail accuracies like Hiwin HTD, whatever.
It´s quite easy and feasible to sand them straight and level, for the last 0.2 mm - 0.5 mm or so.
If You have bigger errors, use a flap wheel of good quality and a 125 mm hand grinder.

Achieving 0.02 mm accuracies = machine tool accuracies, per 200 mm. Fairly easy.
1-2 days practise. Newbies make hills, valleys, and shiny parts, and go deep too fast.

0.01 mm for ultra precision rail is also doable.

You just need a wide sander, much wider than the rail, and a reference, aka the precision ground straight edge.
You then mark with a permanent marker every 10 cm how much each dip/hill is, more or less, at first, and lean on the big wide heavy belt sander.
Make 2 parallel marks where you want to remove more material.
Use 40 grit belts.
The marker is about 0.02 mm thick, and quite consistent.
Just sand trying to keep the surface flat, not smooth, and pretty soon you will get a very flat, very shiny, surface.
Angle the wide belt sander to 45 degrees, each side, goes much faster.
Use dust mask !!

Pretty and shiny is NOT FLAT !

It´s hard work, noisy, unpleasant, dusty.
It´s also very fast and productive.

The end result is extremely good and made very quickly.
A 12.000$ scraping job equivalent is done in half a day.
I use a Festool BS 105 wide belt sander - very wide and powerful and unfortunately very heavy and noisy.

The whole point is that the 105 mm wide belt wont round the surface for the 35 mm linear guide rail.
You really want to use a wide belt sander.
A smaller thinner sander will go much faster, but the underlying surface will be round.
About 20% of both the edges get rounded with a belt sander.
So 105 mm wide belt gets me about 105 - 20 - 20 % == 60 mm of VERY FLAT surface.

The sanded surface is pretty close to the milled surface from commercial machine tool manufacturers.
Not as good, perhaps half, but then You are not using precision high preload rails and blocks, so it wont matter.
The surface is better than the hiwin 0.02 mm guidelines for installing linear rails of std precision.

I´ve done this perhaps 20 times on 60 rails.
Std tests like ground machinist straight edges (light test), mounting blocks per hiwin with dti, etc, all show consistently the same results.

Proof:
My VMC frame is mostly accurate to about 0.02 mm throughout, measured with 2 x 4-sided precision machine builders precision levels of 0.02 mm / 1 m.
Most of it trends to 0.01 mm, and ends trend to larger errors towards 0.02 mm (acceptable according to hiwin).
Vertical rails of 1000 mm. 12 x 35 mm HTD-W blocks.

About 1000 hours extra work to disassemble, transport, strip, 2-prt epoxy paint, reassemble, realign.
4 king bolts lost in transit (70x70x600 mm solid tool steel billets, 45 kg each).

Any of the 35 mm wide Hiwin HTD-W carriages move easily under light hand pressure.
The assy of 4 carriages each side, vertical, z axis, front, that is double-constrained, moves with moderate pressure == 6-8 kgf plus plate mass (30 kg).
The torsion plate assy is about 10-40x more rigid than a single track with 35 mm carriages, and about 4x more accurate due to geometric averaging.

Since light hand pressure will move the carriages, each with 4000 kgf carrying capacity for 2 M meters, and they will bind significantly against rails that are not "flat", the assy is sufficiently "flat" or straight.
And 4 carriages rigidly mounted into a 300 x 450 mm x 10 mm steel plate will not bend or squirrel when running up and down the vertical track.
Hiwin states this as an alternative assy method, so do other linear manufacturers.

[peteeng]

Hi - I agree with Hanermo - but I use long laps vs power tools. Start with 80G or 40G (use roll abrasive and lap say 500-600mm long wet is best) use a surface contrast (available at spray painter supplies) or a mist of contrast spray paint or cross hatch of felt pen. Lap along foundation to identify high points. Then a little work on that point then contrast then 45degs until scratch is consistent, then -45degs until consistent. Then down to 120G more contrast etc. Do not work on a "spot" but over a large area. Doesn't take long to come to flat. And shiny is not flat. If you make 2 laps or 3 and lap the laps flat first so you know they are flat you can get very very flat surfaces... I've done this with steel, aluminium and timber surfaces. Once really down you can use a thin screed of epoxy putty across surface to fill in the small hollows if you don't want to bring entire surface to flat. Peter

I got quotes for machine laser levellers that started at $25000USD so I could map this but that's way too many $$$ for me. One system was for $43k...

[joeavaerage]

Hi peteeng,

I got quotes for machine laser levellers that started at $25000USD so I could map this but that's way too many $$$ for me. One system was for $43k...

Yes laser interferometry is not cheap. If you had a round that you did 15 CNC machines for certification or similar that might be different.

There is another method which while not a good as laser interferometry, is a lot older and yet still pretty damned good, is an autocollimator.
This sort of thing:

https://www.ebay.com/itm/17535372797...Bk9SR6z-3Y3AYQ

They can measure angles down to several arc seconds. This is what they use to certify granite surface plates.

The one pictured is pretty cheap, most of the Hilger & Watts, good old English stuff, in good working order is commonly around $1000USD on EBay. I used one in the Mechanical Engineering lab
at University once. Once you cotton on to how they work you'd be amazed how simple and accurate they can be. I've always hankered to buy one ever since.......would be a great idea if two or three
or four blokes got together and bought one. Its not like you use them all the time.

Craig

[peteeng]

Hi Craig - I started looking for those some time ago. I also need a 90deg mirror to square the gantry. All rare these days. Peter

[Thunza]

It´s not hard to level steel surfaces to machine tool rigid rail accuracies like Hiwin HTD, whatever.
It´s quite easy and feasible to sand them straight and level, for the last 0.2 mm - 0.5 mm or so.
If You have bigger errors, use a flap wheel of good quality and a 125 mm hand grinder.

Achieving 0.02 mm accuracies = machine tool accuracies, per 200 mm. Fairly easy.
1-2 days practise. Newbies make hills, valleys, and shiny parts, and go deep too fast.

0.01 mm for ultra precision rail is also doable.

You just need a wide sander, much wider than the rail, and a reference, aka the precision ground straight edge.
You then mark with a permanent marker every 10 cm how much each dip/hill is, more or less, at first, and lean on the big wide heavy belt sander.
Make 2 parallel marks where you want to remove more material.
Use 40 grit belts.
The marker is about 0.02 mm thick, and quite consistent.
Just sand trying to keep the surface flat, not smooth, and pretty soon you will get a very flat, very shiny, surface.
Angle the wide belt sander to 45 degrees, each side, goes much faster.
Use dust mask !!

Pretty and shiny is NOT FLAT !

It´s hard work, noisy, unpleasant, dusty.
It´s also very fast and productive.

The end result is extremely good and made very quickly.
A 12.000$ scraping job equivalent is done in half a day.
I use a Festool BS 105 wide belt sander - very wide and powerful and unfortunately very heavy and noisy.

The whole point is that the 105 mm wide belt wont round the surface for the 35 mm linear guide rail.
You really want to use a wide belt sander.
A smaller thinner sander will go much faster, but the underlying surface will be round.
About 20% of both the edges get rounded with a belt sander.
So 105 mm wide belt gets me about 105 - 20 - 20 % == 60 mm of VERY FLAT surface.

The sanded surface is pretty close to the milled surface from commercial machine tool manufacturers.
Not as good, perhaps half, but then You are not using precision high preload rails and blocks, so it wont matter.
The surface is better than the hiwin 0.02 mm guidelines for installing linear rails of std precision.

I´ve done this perhaps 20 times on 60 rails.
Std tests like ground machinist straight edges (light test), mounting blocks per hiwin with dti, etc, all show consistently the same results.

Proof:
My VMC frame is mostly accurate to about 0.02 mm throughout, measured with 2 x 4-sided precision machine builders precision levels of 0.02 mm / 1 m.
Most of it trends to 0.01 mm, and ends trend to larger errors towards 0.02 mm (acceptable according to hiwin).
Vertical rails of 1000 mm. 12 x 35 mm HTD-W blocks.

About 1000 hours extra work to disassemble, transport, strip, 2-prt epoxy paint, reassemble, realign.
4 king bolts lost in transit (70x70x600 mm solid tool steel billets, 45 kg each).

Any of the 35 mm wide Hiwin HTD-W carriages move easily under light hand pressure.
The assy of 4 carriages each side, vertical, z axis, front, that is double-constrained, moves with moderate pressure == 6-8 kgf plus plate mass (30 kg).
The torsion plate assy is about 10-40x more rigid than a single track with 35 mm carriages, and about 4x more accurate due to geometric averaging.

Since light hand pressure will move the carriages, each with 4000 kgf carrying capacity for 2 M meters, and they will bind significantly against rails that are not "flat", the assy is sufficiently "flat" or straight.
And 4 carriages rigidly mounted into a 300 x 450 mm x 10 mm steel plate will not bend or squirrel when running up and down the vertical track.
Hiwin states this as an alternative assy method, so do other linear manufacturers.

Thank you, i'll consider getting a second hand one, i've never got round to buying a belt sander. Do you do any checks to make sure the rail isn't twisted from end to end?

When I flattened a bobbin sander plate I used an angle grinder with marking blue and just tapping light touches. I'm happy to go slow and do it by hand to as it's a one off job.

Do you think I need the beams to be flat as well as parallel and square. I guess using leveling legs means being parallel on the bottom isn't necessary. I haven't got as far as deciding on a frame design so not too sure if i need a 90 degree say inner side too. Perhaps not as I will have some adjust ability due to bolting and epoxy. Right not i'm thinking ball screws for all axis, rack and pinion might need additional sides flat unless you were to just build upon the existing flat top.

[joeavaerage]

Hi,
you certainly can do it by hand, or at least with a belt sander, but I don't really have that sort of patience.

About four blocks from here is a toolmaker whom has a 1m travel surface grinder. To put 2 or 4 lengths of 50 x 50 x6 or similar section
which could all be laid alongside each other and ground flat would cost a box of beer and a plate of date scones....the boys seem to like my scones.
Even if I paid them full price it would take less than an hour, say $100NZD. Flat and straight on one side as-near-as-dammit for $100, no brainer.

Craig

[joeavaerage]

Hi,

Pretty and shiny is NOT FLAT !

Well maybe not, but pretty and shiny is still PRETTY AND SHINY.

Craig

[Thunza]

Hi,
you certainly can do it by hand, or at least with a belt sander, but I don't really have that sort of patience.

About four blocks from here is a toolmaker whom has a 1m travel surface grinder. To put 2 or 4 lengths of 50 x 50 x6 or similar section
which could all be laid alongside each other and ground flat would cost a box of beer and a plate of date scones....the boys seem to like my scones.
Even if I paid them full price it would take less than an hour, say $100NZD. Flat and straight on one side as-near-as-dammit for $100, no brainer.

Craig

I'd do the same as you if I had the option. I don't think we have the same small odd job type machine shops here in the UK, at least I can't find anything locally.