Designing a better bed

[D.C.]

The standard approach to a moving gantry CNC bed seems to be a thick piece of aluminium preferably 20mm or more.
Aluminium is expensive, the cost for my bed in aluminium plate would be £338.

With that budget in mind I've been trying to design something better for less money in materials, assuming that I work for free.
Lots of head scratching and reading up on torsion boxes, stressed skin panels, constrained layer damping etc etc This is where I am at now.

http://www.mycncuk.com/attachment.php?attachmentid=15688&stc=1

The sides of the bed are 3mmx80mm steel plate.
The bottom is 3mm steel plate.
The reinforcing is 20mm x 20mm hollow steel square.

After much reading about the wondrous technology of constrained layer damping, it turns out that one of the very best 'viscoelastic damping materials' is butyl and it needs to 0.5mm to 1mm for the sort of things we do.

Which is handy, because the super spiffy viscoelastic wonder material has another name - pond liner. 0.75mm pure butyl pond liner is about £7 per square metre.
So all of the reinforcing will be wrapped in pond liner, sorry, (viscoelastic damping membrane.)

After that the bed will be filled with carbonfibre reinforced epoxy quartz and vibrated, the top couple of mm will be pure epoxy so that it self levels.

The quartz I can get from http://resinbondedaggregates.com/ they wash, grade and kiln dry quartz specifically for resin work. About 20p per kilo.
The epoxy is IN2 infusion resin with a slow setting hardener, £72 for 5kg from easycomposites.
The carbonfibre is £47 for 2kg from easycomposites.

So the original budget of £338 for 20mm aluminium now breaks down to

£90 in steel
£10 in pond liner
£140 in epoxy
£20 in milled carbon
£25 in aggregate

£285 total

I've been very generous with the epoxy as that is the scary expensive part, £140 assumes an 18% by volume mix but hopefully it will average out lower than that. I could still go to almost 25% epoxy and break even with the aluminium.

Any thoughts or comments on this approach for a DIY bed?

[routercnc]

Hi DC

Constrained layer damping sheets need an outer layer of thin steel bonded to them to make them work. I don't think you'll get much from just wrapping something in the butyl material.

They work well on large thin sheet materials such as air conditioning ducting panels etc., but you'll not get much if anything on small beams. If you do it will be at very high frequencies where the deflections are tiny, so of no practical benefit for what you need. Also by the time you flood this with the epoxy quartz that is where the benefits will come from.

I wonder if you'd be better off with a couple of 80x80 steel RHS beams across the width, filled internally with sand, then fill the whole area with sand and topped off with a 10mm ecocast aluminium plate tapped with a matrix of M8 holes . . .?

[davegrennan]

What size is your bed to be DC?

[D.C.]

It's 1320x720

[Boyan Silyavski]

Please read post #165 (http://www.mycncuk.com/threads/6619-Quite-an-Unusual-one?p=68612#post68612) (http://www.mycncuk.com/threads/6619-Quite-an-Unusual-one/page17)about my thoughts and price comparison on most available materials

There is no doubt for me UPN140 is the cheapest , strongest, stiffest... It is almost mounted on my machine and i confirm that its brutally strong and was a great idea.

basically for your bed it will cost 100 euro or less , here in Spain / 100kg ~16kg/m /

[michaelw]

(1)

Designs for any machine beds come in two classes :

(a) Those which are set up in place once only and can remain in place for service life of machine .

(b) Those which have to be more regularly re-sited .

Considering only (a) pro tem the cheapest and most rigid bed is usually a deep cast concrete block with all the precision machine components bolted down to it .

(2)

Composite construction for machine beds is usually non preferred .

Your proposed design would be better with much smaller section reinforcing rods with grip into the epoxy mix . Common ragged reinforcing rods would be ok .

Personally if I wanted to use epoxy concrete construction I would just cast a solid slab of sufficient depth but with no reinforcing at all .

(3)

There are other types of machine bed :

(a) Based on (eg) quarry slabs of rigid slate .

(Think billiard tables and DMM's)

Granite would be super excellent but probably be too expensive for your purpose .

(4)

Consideration must always be given to how machine bed sits on the workshop floor .

Concrete block cast in place on adequate sub surface is usually ok anyway but any slab type bed mounted (relatively) loose on floor needs mountings which do not distort bed and which usually need to allow for some levelling .

Mountings in any case have to sit on floor and sub surface of sufficient rigidity .

[Robin Hewitt]

Aluminium is expensive, the cost for my bed in aluminium plate would be £338

But it is instant gratification, square edges, easy to work takes a thread and looks good. Plus working with unfamiliar materials can go wrong very quickly. Better to have a big lump of aluminium to sell or scrap rather than something you pay to be removed :smug:

[michaelw]

Always worth trying a bit of lateral thinking on jobs like this .

Why do you need a bed at all ??

If , just for example , work envisaged was routing out shapes from large sheets of plywood a moving gantry machine could be designed where the main slide rails were bolted direct to the workshop floor and the sheet of plywood just mounted on a few packing pieces also bolted direct to the workshop floor .

Floor would need to be rigid but any normal concrete floor of reasonable thickness would be ok . Slide rails could be levelled on packing if needs be .

This set up would not do for precision engineering but for simple routing out it would do all that was nescessary .

Improve the set up by levelling the slide rails more accurately and installing a grid of accurately levelled embedded work holding positions and very good work could then be done .

Purely for interest :

Some engineering works that I visted years ago had jigging floors specially installed . These were large floor areas which were trued and levelled and provided with large numbers of fixing down positions . Used for trial setting up of large machinery and for all manner of ad hoc machining of large components . Sometimes jigging floor was provided with a pit for turning flywheels and a moveable compound slide .

[davegrennan]

DC the reason I asked about the size of your bed was your comment about 20mm aluminium being some kind of standard approach, surely the thickness of the bed would be proportional to the size? e.g. a 20mm plate would be overkill in 400 x 400 but not good enough for a 4m bed. I'm considering options for a 800 x 600 bed and one thought is to make a grid, with around 75mm spacing and top it off with a full 800 x 600 drilled and tapped plate. This suits me since I have loads of 15mm plate, maybe a similar approach could work for you?

[D.C.]

Tried to edit and failed, ignore this one, read the next.

[D.C.]

Thank you everyone for your thoughts.

I'm currently clearing out a garage to convert into a workshop and literally sweeping the cobwebs off the ceiling. In a week or two the paint will be dry and I'll need to put an order in for some steel to start building work surfaces, tool storage etc.

MichaelW - I had thought about levelling a concrete platform, but any cost saving would be outweighed by me explaining to prospective future buyers that yes, this was once a garage but now it is an excellent place to park your lathe. Sorry about the giant lump of concrete, how are you with a hammer drill? The machine will need to be be theoretically moveable, even if that means hiring a forklift. Realistically it will probably never be moved and reused. If I really needed to dismantle it, the expensive bits electronics/rails/etc would be for reuse/ebay and a better shiner machine frame built.

Routercnc - I've been reading every paper I can get my hands on to do with CLD, (so many are pay-walled, grrr) as far as I can tell, a thin layer of steel is not required. What is required is a 'viscoelastic damping layer' sandwiched between two comparatively rigid layers and strongly bonded to them. As long as 'pond liner' can not move in relation to the rigid layers, it has no choice but to convert the vibration into heat, which is where the damping effect comes from.

Boyan - You are right about the channel section strength but in Britain steel costs a lot than in Spain. There is also the problem that when you hit a piece of steel with a hammer it rings like a bell. That sound is massive shockwaves travelling through the material and forcing air out of the way as it flexes. The 'perfect' cnc machine would be completely silent, there would be zero vibrations that you could introduce into it that would result in a measurable distortion.

DaveGrennan - I think 20mm seems to be the standard approach around here because of the z-axis. The vast majority of people have a similar requirement for z-axis movement and 20mm alu plate is recommended for multiple parts of the machine. It thus makes cost effective sense to order one sheet of 20mm plate and use it for everything, including the bed.

Because I need some steel anyway before I start my cnc build, and because I know that I will end up using epoxy levelling for my rail mountings I think some testing is in order.
I will buy a bit of 80mm square steel, some pond liner, adhesive, reinforcement etc, and do some comparative tests. Deflection is easy, vibration can be done on a smartphone with a seismometer app, weight, easy, cost easy.

http://www.mycncuk.com/attachment.php?attachmentid=15702&stc=1

This is a beam design I have being considering. In terms of machining and CLD, it was first done in the 1960's but still crops up in the 2000's as the optimal design.
Bare metal beam VS sand filled beam VS reinforced CLD/EG beam. I'll post the results and go from there.
Any other tests I can do with peasant level equipment and any advice in setting up a fair test?

Thanks,

D.C.

[Boyan Silyavski]

Boyan - You are right about the channel section strength but in Britain steel costs a lot than in Spain. There is also the problem that when you hit a piece of steel with a hammer it rings like a bell. That sound is massive shockwaves travelling through the material and forcing air out of the way as it flexes. The 'perfect' cnc machine would be completely silent, there would be zero vibrations that you could introduce into it that would result in a measurable distortion.


You have got it wrong here. If it was a sheet or just box section profile, yes, its like you say. But not in that case. As it happens that i make various musical instruments and trying to make some from steel, all my designs of machines are made so that sound is canceled/ no vibration.

I am telling you here not abstract theory. I have already mounted that UPN 140 profiles and definitely there is no vibration. As the C form is doing just that-cancelling vibrations. There is no ringing whatsoever.

I am sure in UK also the UPN 140 is the cheapest way to go compared to all other variants. Just check a supplier, not local welder. Cause for example if i go to a welder he would sell it to me for 2-3eur kg instead of 1euro kg . Just cause he cut it. hense the Rage saw you see on the pictures of my build. It payed on the first machine i made. Just from ordering directly the 6m material. The funny thing is that the supplier transport of a whole truck of steel costs only 10 euro.

You could fill that C at the back with cement and here you have a winner. Dont over complicate things. Its right here in front of me, tested, super rigid, cheap and does not ring. What more?



Michael,
a simple aluminum plate is a nice solution, however if the hold down of it is not designed properly and not braced where necessary it will ring like a drum skin. Know from experience as i had a machine that did just that because was not well thought.

[cropwell]

For the sacrificial bed part of my machine, I an considering Apollo Slabtech (from Sheridan Worktops*). It is a cast and machined stone filled acrylic. I have the offcuts from a kitchen refit. It machines easily and is waterproof. You could easily machine a coolant gutter. On my small machine I have an MDF sacrificial bed with a matrix of holes which have furniture T-nuts in recesses underneath, so I can hold work using M6 screws (I use nylon ones, they are kinder to cutters if you crash into them). I plan to replicate this system in the Apollo Slab.

*If I can get a 700 x 900 mm piece at a reasonable price.


Cheers,

Rob-T

[cropwell]

There is also the problem that when you hit a piece of steel with a hammer it rings like a bell. That sound is massive shockwaves travelling through the material and forcing air out of the way as it flexes. The 'perfect' cnc machine would be completely silent, there would be zero vibrations that you could introduce into it that would result in a measurable distortion.

D.C.

What would be the magnitude of these vibrations ?

[D.C.]

What would be the magnitude of these vibrations ?

Tiny, but enough to cause chatter, wear tools and affect a finishing pass I think?

[D.C.]

You have got it wrong here.
As it happens that i make various musical instruments and trying to make some from steel, all my designs of machines are made so that sound is canceled/ no vibration.


Thanks Boyan, do you have any links I can read for how to design to minimize vibration?
I don't have access to academic papers at the moment and 95% of the interesting stuff on google scholar is behind a paywall. :(

[Robin Hewitt]

To counter sympathetic vibration think stub and unequal helix tooling.
Obviously nobody can talk you out of this so I say go for it, build this monster and be sure to tell us all about it :thumsup:

[Sven]

There are a few concepts and definitions in this thread that in my mind make the discussion a bit clouded, as it often does when it comes up.

Hope I can point all of it out, here it goes.

First and foremost, a cnc machine needs to be stiff (rigid). Everything else comes second.

It can be good if things are strong but it is very hard to build a machine that is stiff but not strong enough.

Then people often say it needs "dampening" and/or "weight" and then go on using either in the wrong way, in real life and in conversations.

If you add bitumen within the construction you will add "dampening" to the structure. It will make any vibrations die out quicker.
But there is a big risk if you add bitumen or similar within the construction.
Chances are you will reduce stiffness.
I am unsure if adding bitumen to a cnc construction is totally useless but I am sure it is extremely hard to do it so that it does dampen vibration without compromising stiffness.

If you add sand to a gantry you will also add dampening but as described above would be missing a great opportunity.

Pouring loose sand in a construction will indeed reduce vibration, in that it helps vibrations to die out.

However, if you add sand with epoxy mixed in, it will become part of the structure, adding stiffness and at the same time adding dynamic stability to the construction, making the amplitude (size) of the vibrations smaller, giving a better finish to the product.

[D.C.]

Obviously nobody can talk you out of this so I say go for it, build this monster and be sure to tell us all about it :thumsup:

Actually your comment about screwing up and still having a sheet of aluminium to sell for scrap did alter my thinking somewhat, don't give up on the subtle prods informing me that I'm an idiot, I might eventually get the message.

I've finished, clearing, cleaning, painting, realising that brushes don't work well on crapcrete blocks, spraying, spraying again, cleaning, etc ordering steel for a worktable and testing.

I'll post back when I've gotten round to ordering the other bits and testing some beams as best I can and then mull things over.

[Robin Hewitt]

don't give up on the subtle prods informing me that I'm an idiot

I'm good at that, maybe there is hope for me yet :beer:

[Sven]

The sides of the bed are 3mmx80mm steel plate.
The bottom is 3mm steel plate.
The reinforcing is 20mm x 20mm hollow steel square.

After much reading about the wondrous technology of constrained layer damping, it turns out that one of the very best 'viscoelastic damping materials' is butyl and it needs to 0.5mm to 1mm for the sort of things we do.

Which is handy, because the super spiffy visc

£90 in steel
£10 in pond liner
£140 in epoxy
£20 in milled carbon
£25 in aggregate

£285 total

I've been very generous with the epoxy as that is the scary expensive part, £140 assumes an 18% by volume mix but hopefully it will average out lower than that. I could still go to almost 25% epoxy and break even with the aluminium.

Any thoughts or comments on this approach for a DIY bed?

Forgot to reply to the general idea...

Here are some thoughts

- Epoxy adheres well to steel but may not stay bonded due to differences in thermal expansion. That would make the steel tub a bit overkill
- If you use that much epoxy it is likely to form a layer on top of the casting.
- Using milled carbon as well as steel tubing is a bit like suspenders and belt to keep your pants up. And still, 20.-- seems a small amount.
- using tubing instead of bars lowers the mass/volume which I think is a good idea.
- I would use round tubing though as I think it would leave for a stronger EG mass.
- You can probably get away with using pvc piping instead of steel if the plate is thick enough

[D.C.]

Thanks Sven, after Robin's comment about having a big chunk of ALU that can be sold if things go horribly wrong I have to admit that I'm rethinking things a bit...

I'm not too concerned about thermal expansion because I'm building a CNC machine in a garage that is attached to a house and the temperature doesn't won't really fluctuate more than 15C over the course of a year.
I was banking on the epoxy getting vibrated to the top and selected the epoxy especially to do that as it will self level.
The milled carbon amount is based on the best practice mix on the easycomposites website, 10% by weight compared to the epoxy. Obviously the quartz/granite/etc in the mix doesn't count, this results in a surprising small amount of milled carbon fibres that add greatly to stiffness and dimensional stability.
I'm not sure exactly where the optimum lies between tube and square, I selected square so that I could weld it together easily, circular tube would be much more difficult to fabricate.

I've been arguing the toss with a structural engineer and insurance company about the fact that my house has giant cracks in the extension so for the next couple of weeks/month I need to back burner the CNC which is very annoying, I am relying on it to do a lot house upgrades. :(

[Sven]

I' ve been reading up on EG as my own ideas are taking shape.
As far as I' read, vibrating EG does not work unless you use a mix with too much epoxy for a small size casting. Too much meaning it will yield an end result that has some shrinkage and maybe some warp.

The best seems to be a mix that is "very dry" compared to what most people on forums seem to be using: in the 4% area, or a bit wetter where it needs to adhere to another material.
Also, best to use various grain sizes, the next size 1/5th of the previous.

Put a measured amount in a jar, then add the smaller size and shake until the volume increases and you know the ratio.

A dry mix like that will need stamping.

And off course, make sure to do trials :)

[Robin Hewitt]

If all you want is to flow a flat surface, why not use Wood's Metal? That goes very runny and sets hard. You could even reflow it if you moved the machine.

[D.C.]

If all you want is to flow a flat surface, why not use Wood's Metal?

Not a bad idea but Wood's Metal / Cerrobend etc looks to be about £40 per kg, so it would cost a lot more than epoxy levelling I think?
Have you ever tried it if you have some lying around?


I' ve been reading up on EG as my own ideas are taking shape.
As far as I' read, vibrating EG does not work unless you use a mix with too much epoxy for a small size casting. Too much meaning it will yield an end result that has some shrinkage and maybe some warp.
The best seems to be a mix that is "very dry" compared to what most people on forums seem to be using: in the 4% area, or a bit wetter where it needs to adhere to another material.
Also, best to use various grain sizes, the next size 1/5th of the previous.
Put a measured amount in a jar, then add the smaller size and shake until the volume increases and you know the ratio.
A dry mix like that will need stamping.
And off course, make sure to do trials :)

For larger aggregate This place does washed, graded and dried quartz specifically for resin bonding:
http://resinbondedaggregates.com/category/resin-bound-aggregate

For the smaller sizes I think it would have to be blasting grits:
http://www.stacey-processing.com/glass-grit-and-shot-blasting-media.html

Then Aluminium Oxide powder and milled CF.

[Boyan Silyavski]

Linear or volumetric thermal expansion coefficient is irrelevant in constucting a CNC.

First of all cement, steel and concrete have very similar thermal expansion. Even if the epoxy has higher thermal expansion its irrelevant at 1m and temperature changes. Plus the epoxy cement or epoxy granite will more similar expansion to the concrete

http://www.mycncuk.com/attachment.php?attachmentid=15884&stc=1


EG is same like all, do it properly and it will work well.


There are better options of course and not so expensive- purpose made epoxy concrete, especially made for the purpose. Problem is finding them close to you or in your country at all.

But those who search will find :-).

Key words:

epoxy concrete cast machine bed, ultra-high performance concrete


https://www.youtube.com/watch?v=ftopsaGUZzs



These guys http://durcrete.de/ can do machine bed design or cast inhouse for you or as far as i remember 1 ton of the stuff was like ~400euro

Gues what i will be bringing home if i go to Germany :-)

[Robin Hewitt]

10 kg Cerrobend £162

http://www.ebay.co.uk/itm/Cerrobend-Woods-Metal-Bismuth-alloy-10KG-/261983313460?hash=item3cff6c1e34 (http://www.ebay.co.uk/itm/Cerrobend-Woods-Metal-Bismuth-alloy-10KG-/261983313460?hash=item3cff6c1e34)

[Sven]

I may have had my expansion reasons mixed up...

A guy I know built an epoxy-concrete machine with steel bars at the surface to connect the rails to.
They were flat when casting started but the end result was not.

If you cast EC in a steel tub, it may end up not being part of the construction but just a liner and not a reinforcement.

[Boyan Silyavski]

There are many ways to not do it right. Apart from choosing the best epoxy for deep casting, is good idea to read the documents of the said epoxy and follow them instructions to the letter. best is to call the epoxy manufacturer and speak with them for the most suitable epoxy for the purpose.

Cause at the end of the day they know their stuff and even can mix a custom epoxy for your purpose. I have talked with one company technician and he knew exactly about that situation and explained me the differences between the 3 suitable epoxies they had.

[D.C.]

The http://durcrete.de is quite interesting.

What I was aiming for was finding a sweet spot between price/performance and ease of construction given that I only have not very accurate tools to work with.

http://www.mycncuk.com/attachment.php?attachmentid=15902&stc=1

The Durcrete site has some vibration damping data available and a demo:

http://www.youtube.com/watch?v=pCU7H6Yyrmw

[Boyan Silyavski]

Definitely they are the most promising ones. I remember spending 2 weeks of my free time / when i had one/ browsing and searching. And thinking of it its not at all so expensive as it could be with the epoxy. I had somewhere their offer and it was relatively cheap. from their site:


Do It Yourself

Durfill is a simple and quick solution. So we have a lot of clients all over Europe, which use this high strength grout for filling their machine components. Without additional dealers, we export the material in full palettes including the required delivery directly to your company.