PDA

View Full Version : Building from Scratch or follow some plans?



Monk
10-08-2010, 11:50 PM
Evening all, I've had an avid interest in CNC for about 3 or 4 years, bought a CNC machine off a chap that never arrived and got my fingers burnt. Hey ho... I've got the itch to build myself one now and would like to make it out of Aluminium Extrusion rather than mdf.

It would have to be a work in progress machine as I can't really afford to go out and buy everything at once and then sit piecing it together. Do you think it would be better as my first build to follow some plans, or adapt some plans? (assuming I can find some suitable to adapt). I'm quite practically minded and not too daft when it comes to tools. I've only found one (well one free) set of plans that use extrusion and thats all in imperial, which is a bit difficult to get here :thumbdown: The only thing I'm really unsure of is the mechanics and making sure I get it right. I don't have the ability to make nice diagrams to show you my ideas, i'm old school with a pencil, ruler and a piece of paper :heehee: I can alter them quickly then too!

I'm after a cutting area of around 1000x800mm and 3-axis. If I designed my own I'd probably start on the X and Z axis. I get the feeling I should start from the ground up and build the base first then up and up, but I'd like to tackle the hard parts first.

So anyone reckon they could guide a newbie with some extremely daft questions? :eek: :heehee:

So far I would opt for Hiwen Rail and Bearing blocks on an aluminium extrusion frame... opinions?

Many, many thanks!

GeorgeD
11-08-2010, 12:19 AM
Can't help with the mechanics of the build,Monk as I'm in the same boat as you at the moment,but here's a site for all and every piece of extrusion you might want for the build.
You have to completethe formto download two seperate PDF's one for Pricing and tother for profilesof what extrusion they do.

The drawback is...Minimium order is 100

http://www.valuframe.co.uk/

Ross77
11-08-2010, 12:47 AM
I'll help if I can.

Can you scan and post your scribblings to give us a rough idea of where you are at?

Sounds like your on the right track with ali extrusion and linear rails tho,

Good luck

ecat
11-08-2010, 01:58 AM
We find ourselves in much the same boat: similar position, similar design goals and similar decisions- as far as they go :) I can offer little in the way of construction help as I've never done this before, but I wish you luck.

I'm still at the point of deciding if the end justifies the cost (1), to that end I need both a cost and an understanding of design limitations and to that end I need a design based on my budget. All a bit chicken and egg (2), lol.

To get the ball rolling I've downloaded a couple of CAM packages to look at and a trial copy of Alibre Design, collecting models of various parts that may make it into the final design (3) - testing parts is cheaper this way ;-). I hope this will get my head into an engineering mind-set and playing with some design software feels to be the right place to start, after all there is little point building a computer controlled machine if you have zero understanding of the software required to make it all work :)

(1) Cost... Ouch. The old mantra, buy cheap buy twice, haunts me every time I look at ball-screws. Hiwin rail and blocks just make me cry and I know a lower speed spindle will be a must and.... I'm probably preaching to the converted here.

(2) Silly phrase, evolution of course dictates the egg must have come first.

(3) What the ....? Register, register, register, everywhere I go it appears the same. I'm trying to give you guys some money, why do you make me go through loops and hoops to get a model of some profile or even just a price list? Grr. "We want to track all our prospective customers". Fine, do that, but isn't email address enough? Name, address, telephone number, colour of socks, so old fashioned, so last decade </rant>

routercnc
11-08-2010, 08:43 AM
Hi Monk,

You might find the following post useful. I listed some terms which might help, plus others have added helpful comments:
http://www.mycncuk.com/forums/showthread.php?2048-Wood-cnc-router-for-hobbyist

Maybe we need a glossary of terms section in this forum which anyone can add to...?

Anyway, you can always go to the local library and get a sketch scanned in and email it to yourself, to then post here. Picture paints a thousand words as they say. By the way, if you need to cut aluminium profile, several of us have found out you can use a chop saw with metal cutting blade. Do some trial cuts on wood to get the blade as square as possible.

Hiwin profile rail and ali profile make a great combination. Almost as good, but cheaper, is the supported rail and open bearing. 16mm is popular, as sold by Zapp and others.

Post away . . .

Monk
11-08-2010, 02:08 PM
Thanks for the advice guys. I have a scanner at home so I scan my sketches/drawings/doodles in and post them up. Seemed everyone was doing stuff in CAD and making it look really smart and professional, then you'd look at mine and go erm... :heehee:

I will try and get a drawing up of my proposed Z and part X axis up in the next few days, I'm away for the weekend and got loads to do before hand so time is a bit tight.

For the extrusion I have found a place KJN Ltd off a search on here that does good prices, also Marchant Dice does extrusion but is pricey. However for small bits it may be cost effective and they cut to requirements! I've looked at Zapp as well for rails and bearing blocks. I kind of know where to get parts, its making them a) fit together and b) work :smile:

irving2008
11-08-2010, 02:57 PM
Thanks for the advice guys. I have a scanner at home so I scan my sketches/drawings/doodles in and post them up. Seemed everyone was doing stuff in CAD and making it look really smart and professional, then you'd look at mine and go erm... :heehee:

This is a well tried and trusted technique known affectionately as COC or "Crap-o-CAD" but sometimes its the easiest route..... failing that, Google Sketchup is a great tool for playing with ideas... and its free!

Wobblybootie
11-08-2010, 03:51 PM
KJN sell their own extrusion and also Bosh Rexroth stuff as well, if you look closely each item will show which is which. I have found that for some of the Bosch stuff can be got cheaper from Hepco ( http://www.hepcomotion.com/en/mcs-machine-construction-system-pg-14-get-31 ) You have to juggle between the two sites doing a bit of maths. KJN also do a whole lot more sizes than you can see on the site ... just email and ask.

Monk
11-08-2010, 07:15 PM
Ok no sniggering at the back! :heehee:

I had some time at work to mess around. Ok it only shows three components but its a start! This is the x-axis I was going to make out of 4080 extrusion. Fixed together in a L shape with the rails on the top and bottom. Will this be strong enough over approx 1200mm or should the extrusion be bigger. Also when the bearing blocks are fitted, do I build off the blocks to account for the ballscrew between the rails, or should I mount it behind and build the z-axis to sit acutally over?

If I fixed the aluminium plate or extrusion for the z-axis to the bearing blocks it would sit quite close to the x-axis.

irving2008
11-08-2010, 08:58 PM
Its not a question of strength... that'll hold quite a weight... its a question of rigidity... how much will it deflect, vertically and horizontally, in the middle under load?

In the catalogues you will find design tables which shows how a particular type of extrusion will deflect and that will give you a clue. If you look at page 49 of the Hepco catalog they give the formulae (use the simply supported arrangement as a worst case) and page 50 the parameters for their 80x40. If you work the math it suggests a deflection of 0.00084mm in the 80mm direction (vertically) and .003mm in the 40mm direction (horizontally) for each Newton of load for a 1200mm length. (someone check these, they seem a tad low)

Your 'L' structure is slightly more complex. The vertical deflection is inversely proportional to 0.5 * w * h^3, where h is 120 and w = somewhere between 40 and 80 (I'll hazard a guess the effective value is around 55). So if a simple 40w 80h deflects X vertically, your complex beam will deflect less, roughly 25% of that. Similarly in the horizontal the ratio is roughly 30%... these are very much rough and ready, others (Ross, routercnc, et al) can do the more detailed analysis... and tell me I am wrong!

Typically the horizontal load for a wood router is due to cutting forces and 5 - 10N would be a good allowance, giving a deflection of roughly 0.003 * 30% * 10 or 0.01mm horizontally. Vertically you have the weight of the z-axis and the router/spindle - that could easily weigh 5 - 6kg or 50 - 60N, so a deflection of .0008mm * 25% * 60 or 0.012mm in the centre of a 1200mm run.

Incidentally the deflections for a piece of 80w x 120h would be .003mm horizontally and .008mm vertically, better but at the expense of added weight.

Ross77
11-08-2010, 09:59 PM
these are very much rough and ready, others (Ross, routercnc, et al) can do the more detailed analysis... and tell me I am wrong!

never Irving, I'm not that brave, Ive looked in to composite beam design an it gets a bit messy, the main factor that I have a problem with (calculation wise) is that once the hiwin rails are bolted on the ali structure is made a lot stronger,

Im still struggling with the spindle design math, I think I would just model the structure and put it in FEA as I'm getting lazy...:heehee:

Ross77
11-08-2010, 10:13 PM
Irving have you got a link to that catalogue? I cant find it,

I Looked at the individual section specs and the dimms are in mm's but the Ixx and Iyy is cm^4 and vol cm3, could be the problem with the decimal places.......

routercnc
11-08-2010, 10:15 PM
Hi Monk,

Firstly just looking at the section sizes, given that it will be 1200mm long, would seem a reasonable start. It will depend on how far down the cutting is taking place away from this section, since this will lead to a twisting torque, along with the deflections Irving mentioned above. Any ideas on the height down to the workpiece?

I need to dash off now, so will get back to you with some proper analysis. Really you need to use what is called the 'parallel axis theorem' which works out how to combine the two shapes (or any number of shapes), and find the neutral axis. This will be slightly below 'half way' in your L case. Then you can calculate the total section 'I' value, second moment of area, in mm4. From here it is then possible to get the deflection, and given a bit more data from you, the twisting.

The Hepco 40x80 shows an Ixx value of 61.4cm4, which if my brain is working is 61.4x1e4 mm^4. This is a start, hope to be back with some better answers tomorrow unless someone jumps in . . .

Ah, Ross, you jumped in while I was typing. Here is the Hepco catalogue link:
http://www.hepcomotion.com/en/literature-mcs-machine-construction-system-pg-16-get-31

Yes, watch those units. Should be consistant really.

irving2008
11-08-2010, 10:51 PM
Irving have you got a link to that catalogue? I cant find it,

I Looked at the individual section specs and the dimms are in mm's but the Ixx and Iyy is cm^4 and vol cm3, could be the problem with the decimal places.......
routercnc gave the link in his post above... yes they use inconsistant units but in the formulae they throw in a compensating 1e4 on the bottom to tidy up :rolleyes:

Monk
11-08-2010, 11:24 PM
Hoping for about a cutting height of 150mm ideally. I never realised there was so much to consider. Are these forces transferred through the z-axis or exerted on the z-axis as well?

Basically I need to design a couple of different x-axis then work out the deflection to minimise it? Ooo the planning is going to take a while!!

Ross77
11-08-2010, 11:38 PM
The answer is both, the z axis also needs to be rigid enough to transmit the forces, which will then be transfered to the y axis and will be max at full z travel. This will cause a moment on the y axis, (both directions) so like routercnc said your main concern is the torsional stiffness, beam deflection will mainly be from the weight of the x axis and carriage (static and dynamic) but some uplift could arise if you are plunge routing....

Maybe wait for Irving or routercnc, they are better at explaining than me.

irving2008
11-08-2010, 11:40 PM
both!

Think about where the tip of the cutter will be when cutting...it'll be about 200mm below the y-axis (depending on how much y-axis clearance over the workpiece you want. So any sideways force on the cutter will be trying to bend the cutter.. but if that was solid it would be trying to bend the z-axis... but if that was solid it would be trying to twist the y-axis - the worst scenario being the twist in the X direction perpendicular to Y. Each of those components will deflect to some extent and transmit the forces back up to the component above them in the chain... and then the y-axis transmits them to the gantry sides and thence to the x-rails...

The easiest way to get a handle on the forces is to draw stick diagrams... some forces can be ignored generally, e.g. the twisting motion of the Y-axis on the Y-axis due to a sideways force on the cutter parallel to Y, although that force is also trying to push Y sideways and that is trying to rotate the bearings on the x-rails and that can't be ignored... at least initially... good bracing and gussetting helps, but you need also to keep the moving weight down...

Monk
12-08-2010, 12:10 AM
Irving where did you find the force allowances for wood? I could do with investigating aluminium as well, as I can almost guarantee I'm going to end up wanting to make something in aluminium!

irving2008
12-08-2010, 12:26 AM
ummm... off the top of my head i dont recall... google for 'specific cutting power'... its 17 for aluminium... that allows you to work out for a given material removal rate how much power is needed and therefore also, given revs the torque, which given a cutter size, gives the force applied and thats a guesstimate for the cutting force at the cutter... its rough and ready so we've generally doubled it... some hardwoods are nearly as tough as ali!

Ross77
12-08-2010, 01:11 AM
See what I mean Irving's much better than me. I can see it but I'm buggered if I can explain it :whistling:

With regard to cutting forces look here (ignore the spindle stuff)

http://rogercortesi.com/portf/spindle/spindle.html

Scroll down to the Niagara site as well and I'm sure there is a section on wood too, only just found it so haven't had a Reilly good look but there is plenty on metals :smile:

Going back to your machine have you looked at the 80 x 80 box stuff. over the long distance you need torsional stiffness, and as Irving pointed out lightness. (after all you have to move it) the box offers the best resistance to torsion, has equal strength in both axis and is light as there is no/little material in the center.

With the composite beam you have a lot of material (weight) close to the neutral axis (no stress) which again going back to parallel axis theory you want all the material at the extremes, like a tube or next best thing a box.

hope this helps

routercnc
12-08-2010, 08:48 AM
Hi Monk,

Firstly a quick check on Irving's first calc from the calalogue giving 0.00084mm deflection per Newton. I think this value is OK, as I'll show below.

1. For simplicity work out the equivalent RHS:
size Ixx Iyy
Profile: 40x80 614000 170000
RHS: 40x80x3 558332 208049
RHS: 40x80x3.5 636869 184292

Choose 40x80x3 to be on the safe side

2. Vertical bending, and deflection due to twisting:

20kg z axis (router, motor, metal brkts etc.)
length 1200mm
150mm to tool
40x80x3 alum RHS

Z bending deflection is 0.183mm (Irving got 0.164mm for 20kg using catalogue profile data, so think Irving's calc is OK)
X bending deflection is 0.028mm
twisting deflection is 0.003mm

3. You have an 'L' which is part way between the following results:
40x120x3 Z bending is 0.06mm, X bending is 0.020mm, twisting is 0.0019mm
80x120x3 Z bending is 0.043mm, X bending is 0.004mm, twisting is 0.0005mm

I need to create a spreadsheet to work out the true value, but it is between these. For a wood router it's looking OK to me.

If your z axis is less then 20kg, then you can ratio the Z bending deflection down accordingly (but not the twisting, or X bending).

If you go for just 80x80 profile:
Z bending 0.112mm
X bending 0.006mm
twisting 0.001mm

So you are better with what you have. Make sure they are well joined to each other, otherwise you will get back much less than this as they will work on their own.

4. Final note:
If you stack the profiles to get 40x160:
Z bending 0.031mm
X bending 0.015mm
twisting 0.001mm
which is the best of the lot (in bringing down the highest value) for no weight gain

You also asked about ballscrew location etc. I'll get back to you . . .

EDIT: added X bending results for completeness

Monk
12-08-2010, 09:37 AM
Hi Monk,

Firstly a quick check on Irving's first calc from the calalogue giving 0.00084mm deflection per Newton. I think this value is OK, as I'll show below.

1. For simplicity work out the equivalent RHS:
size Ixx Iyy
Profile: 40x80 614000 170000
RHS: 40x80x3 558332 208049
RHS: 40x80x3.5 636869 184292

Choose 40x80x3 to be on the safe side

2. Vertical bending, and deflection due to twisting:

20kg z axis (router, motor, metal brkts etc.)
length 1200mm
150mm to tool
40x80x3 alum RHS

bending deflection is 0.183mm (Irving got 0.164mm for 20kg using catalogue profile data)
twisting deflection is 0.003mm

3. You have an 'L' which is part way between the following results:
40x120x3 bending is 0.06mm, twisting is 0.0019mm
80x120x3 bending is 0.043mm, twisting is 0.0005mm

I need to create a spreadsheet to work out the true value, but it is between these. For a wood router it's looking OK to me.

If your z axis is less then 20kg, then you can ratio the deflection down accordingly (but not the twisting).

If you go for just 80x80 profile:
bending 0.112mm
twisting 0.001mm

So you are better with what you have. Make sure they are well joined to each other, otherwise you will get back much less than this as they work on their own.

4. Final note:
If you stack the profiles to get 40x160:
bending 0.031mm
twisting 0.001mm
which is the best of the lot for no weight gain

You asked about ballscrew location etc. I'll get back to you . . .

That's great thanks, I've just been sat looking at the graph on the Hepco Catalogue doing a few sums. With either option (although the stacked 40x160 looks my best option) I would assume with the Hiwin Rails attached it becomes even stronger?

I think I have worked the ballscrew mech out. Just working out how to mount the servo/stepper motor and connect them to the ballscrew.

routercnc
12-08-2010, 11:27 PM
Hi Monk,

I've created a spreadsheet to work out simple arbitrary shapes I values - will post it sometime.

Anyway, to recap, your L shape was somewhere between:
40x120x3 Z bending is 0.066mm, X bending is 0.020mm, twisting is 0.0019mm
80x120x3 Z bending is 0.043mm, X bending is 0.004mm, twisting is 0.0005mm


Actual value for L shape is:
40/80x120x3 Z bending is 0.052mm
which is closer to the 80x120 than the 40x120

But oddly this is the same deflection as:
60x120x3 (0.052mm) which is half way between width 40 and 80. I wasn't expecting that, and neither was Irving with his guess of 55. Double checked it. Can't find anything wrong.

Enough maths, looking forward to seeing the ballscrew options. . . .

Ah, yes, adding rails will increase stiffness but don't bother working it out, get the structure right underneath and you're away.

Barry