[stern-69]

Great info Due to size restraignts Im probably going to have make mine with a fixed table even though a fixed gantry system would be easier to make.

[scott0999]

The design is a tabletop fixed-gantry mill mostly made of BB ply and MDF, with alum for the Z and gantry carriages. Travels will be around 21"x14"x6". FEA of the gantry, gantry car, and Z car shows a stiffness-at-the-tool of at least 12k lb/in; the complete machine will be somewhat less. The cost less electrics should be in the ballpark of $350. So, I don't think low cost has to mean poor performance.

what all electronics for $350?

[dmalicky]

Oh, by "cost less electrics" I meant the cost without electrics (motors, drivers, power supply, router). The electronics plan is Haoyu TB6000, Kysan 57BYGH318, 36V or 48V supply + capacitor, and either an Uno+Grbl or BoB+LinuxCNC to drive it. Not including router, that's about $220. The Kysans are 252 oz-in and 2.0 mH (45V optimal; I'll probably run around 40V for the TB6600s). I need to dyno test the Kysans+TB6600 to see the torque at speed. If they can put out 25% of their rated static torque, a 1/4" lead acme can push with 62 lb.

[scott0999]

Oh, by "cost less electrics" I meant the cost without electrics (motors, drivers, power supply, router). The electronics plan is Haoyu TB6000, Kysan 57BYGH318, 36V or 48V supply + capacitor, and either an Uno+Grbl or BoB+LinuxCNC to drive it. Not including router, that's about $220. The Kysans are 252 oz-in and 2.0 mH (45V optimal; I'll probably run around 40V for the TB6600s). I need to dyno test the Kysans+TB6600 to see the torque at speed. If they can put out 25% of their rated static torque, a 1/4" lead acme can push with 62 lb.

my bad I read it wrong "less electronics" lol

[scott0999]

update - scored a piece of rectangular tubing 5x3 .25 wall 30" long @ scrap price

I'm thinking of (I know, may change.. lol) cutting holes in the gantry upright plates the shape of the rectangle, putting the rectangle through the plate, and bolting it on with angle steel vertically on all 4 corners (each end)

[RonaldoNZ]

Hopefully not to "derail" the thread but this discussion has me thinking about this SBR rail. Some thoughts Id like to bounce off David in particular. I read your excellent posts re stiffness on these sbr rails/blocks ( http://www.cnczone.com/forums/linear...l_options.html) and the bottom line that in compression they are about 4 times stiffer than in tension.

Here is my curiosity, In the common scenario of having four blocks attached to a plate with a moment acting on all of them the nett stiffness resisting deflection from that moment is going to be the sum of all of them yes. Assuming that the lateral stiffness lies somewhere between the compression and the tension values (a fair assumption I think) doesn't this render the orientation of the rails (IE vertically opposed vs horizontally aligned) irrelevant? Hopefully the two attached pictures will help clarify. Vertically opposed rail orientation vs horizontally aligned seems to be a moot point, I can't see any benefit in terms of overall stiffness. What am I overlooking?


Vertically opposed vs Horizontally aligned
Attachment 223176 Attachment 223178

[dmalicky]

Good questions! And good to hear about the SBR post. The pics are very helpful, too--thanks.

Yes, for the torque due to a lateral (Y) force-at-the-tool, in the front view pictures (A1-A4 in your pics), there is likely no net difference. (Minor correction for the vertically opposed pic--the diagonal blocks are in compression, so A2=A4=4, while the other diagonals are in tension, so A1=A3=1).

For the top view pictures (C1-C2), there are some interesting differences due to 'k' being in the denominator of F/k = deflection. I'll assume a force of 20 on each side:
Vertically opposed: C1 and C2 both deflect 20/2.5 = 8. Net angle change = 16/span.
Horizontally aligned: C1 deflects 20/1 = 20. C2 deflects 20/4 = 5. Net angle change = 25/span.
So Vertically opposed is ~36% stiffer.

For a longitudinal (X) force-at-the-tool, the analysis is more complex and the difference becomes greater. I'll use the side-view picture in this post: http://www.cnczone.com/forums/diy_cn...ml#post1408004
Since the lower bearings are the fulcrum, Fb will always be greater than Fa, usually by about 2x. If Fc is to the right as in the pic, the lower B bearings are in tension, they deflect a lot, and the cutter deflection is amplified to be rather huge. This is a leveraged version of the above C1-C2 calcs. Of course, reversing the cutter load direction puts the lower bearings in compression -- but once I saw a weak mode, I gave up on that approach. (The xls assumes equal stiffnesses for A and B, but it could be modified to calc for Horizontally aligned to see the numbers.)

[dmalicky]

It was an easy update--attached is the xls that calculates longitudinal stiffness-at-the-cutter accounting for different stiffnesses of the upper and lower bearings. I also updated the inputs to just accept the single bearing stiffness (instead of K of the bearing-pair as in prior versions... I kept forgetting to double it.) This xls is useful for SBR Horizontally aligned, and also anytime double-bearings are used on the lower rail (which often makes sense since it takes ~double the load).

For SBR Horizontally aligned, assuming AB=BC=7, the stiffness-at-the-cutter is
23k lb/in for longitudinal cutting force to the right (puts the lower bearings in tension)
49k lb/in for longitudinal cutting force to the left (puts the lower bearings in compression)
Here's the screenshot for the lower SBRs in tension:
Attachment 223280

For SBR Vertically opposed (guessing a lateral bearing stiffness of 130k lb/in), the stiffness-at-the-cutter is 52k lb/in, in both directions.
52k lb/in is on track for an overall machine stiffness of about 5k lb/in -- excellent for cutting wood. A total of 4 bearings on the lower rail might improve that to ~87k lb/in, on track for 9k lb/in overall (those 4 bearings might not give 2x the stiffness, since the rail moves with them.)

For 1/3 the $, similar stiffnesses can be obtained with a CNCRP-style carriage using either double 608s, heavier duty 6200s, or similar. I've shelved SBR-TBR for now and am working on the max-stiffness low-$ machine mentioned earlier. It's up to 15k lb/in now for the Z, Y, and gantry. I just got some 628 and 638 bearings to test stiffness.

[RonaldoNZ]

David thank you for the spreadsheet, brilliant! I tweaked it a little for my case (using four blocks per rail) screenshot attached. I'm using the LUU double length variant that has two bearings in one double length housing, punched in my other numbers and the results don't look half bad. Which is just as well really since I'm half way through building it.



I went with the sbr system for X and Y as a stepping stone to upgrading to LM at a later date, finances permitting, the Z is Hiwin20
With that in mind I've been aiming for as stiff a gantry design as possible with the materials I have to hand that I can work, steel. Large section aluminium extrusion is the stuff of a mad mans dreams here in NZ else I might have gone that route.

Attachment 223422
Attachment 223416
Attachment 223418

Everything orange on the gantry and Z is steel, I got a good deal on some two pack epiglass, but only in "emergency orange". Old stock nobody wanted to paint their boat in it seems, can't imagine why. Its 5 and 4mm RHS for the most part with sundry bits of 8 and 5 mm angle and flat. Getting the gantry faces as parallel as possible I'm doing with some adjustable jigs knocked up for the purpose. A lot of carefull tacking and not building too much heat in any one area at a time. Gently gently.

[drieslaas]

Hi Dave,
Been lurking for a while, and did a very similar spreadsheet to yours a while ago. Not as neat though.
Will double check my work from yours, but we are thinking about it exactly the same way.

I currently have my gantry with two mild steel square tubes laying in the same horizontal plane ( instead of one above the other, as seems to be most popular.)
It works well, but my big gripe is that it limits access to the spindle. I am not sure that the vertical arrangement is better.

I will be doing some work to verify the optimum before deciding on a way.

[Bl@ckrat]

dries . would like to see a pic of yours ....

having the square tubes above each other spaces the rails further apart ...

[RonaldoNZ]

After more consideration of all the preceding information,running some FEA of my own and looking at the stiffness calculator found here ...

cnc machine stiffness calculator

(Credit to routercnc of mycncuk.com)

I've revised my gantry design to a single large piece of RHS 250x150x6mm for the beam bulkheads installed. The beam is 925mm in the span. Also decided to not waste my time with the y axis sbr, bit the bullet and sold them.. it's just too problematic, regardless of the configuration there is always going to be a weak axis. Im useing them for the x only for the time being simply because I've already mounted that set but will upgrade them to a 30mm hiwin or similar further down the track.

Attachment 226102 Attachment 226104 Attachment 226106

Lesson learned..again..never buy anything till the plan is solid.

[drieslaas]

dries . would like to see a pic of yours ....

having the square tubes above each other spaces the rails further apart ...

I bet you say that to all the girls....;-)

No Blackrat, your machine is so cool that I shudder at the thought of showing anyone pics of my lame attempt.
I have considered this, and I think that the moments around the combined neuttral axis of the two beams is lower when both the beams lay in the same horizontal plane IE next to each other. I will run a quick fem when I get time to check if there is a clear winner, I suspect it will be the one I used previously.
The big disadvantage is access to the y-carriage to fit other stuff like dust extraction, and to do tool changes. ( Although this will become somewhat easier with my new chinese watercooled high speed spindle :-D)
Of course you can space the rails further apart vertically as you mention without losing x-travel, so it may not be as clear cut.
I tend to overthink these things in any case, engineers disease.....
I will gladly share pics of the way I am thinking of building the upgrade.

[drieslaas]

Hi Ronaldo,

In the case of a single beam the horizontal cutter tip deflections are resisted by the beam working in pure torsion, while the two-beam gantry ( if I can call it that) puts both the beams into bending to resist the bending moment resulting from the cutter tip force.

By the way, when you mentioned doing fem, would you share your loadcase, and constraints and general geometry. I would like to do the same, having trouble guessing what the sideway force at the tooltip will be.( I use Mechanica for FEA work, will do a quick shell and beam model, probably.)

[pippin88]

I've arrived at a similar conclusion about gantry for my future build. Big steel tube.

30mm linear bearings are massive and very much overkill. 15mm are usually more than enough if you do the numbers.

[RonaldoNZ]

Here you go drieslaas, I used 500N in the X of the machine only, fixed contraints at beam ends.

Attachment 226226 Attachment 226228 Attachment 226230

Find attached three detailed reports for each

Edit: here is a side loaded case for comparision

Attachment 226232 vs

And a single large beam with bulkheads

[drieslaas]

Thanks Ronaldo.

How do you propose to fabricate that tube with the bulkheads in?
If you can't stitch the bulkhead all the way round, you lose much of the advantage.

Suppose you could fold two lasercut plates with slot details around some bulkheads with tab details, and then weld on the tabs from the outside.....

[RonaldoNZ]

Plug and slot welds apart from the bearing face.

[drieslaas]

Hey Ronaldo,

I had a quick look at your fem results. The twin beam and the single beam has very similar deformations in the region of 20 micron.
The diagonal web reduces this by approximately half to 9 micron.
Have you had a look at the cutter deflection of a typical cutter at that side load (500N)
Deflection Force Length E Dia I

8.38E-04 500 0.04 2.00E+11 0.006 6.36173E-11 0.84

At 500N, a 6mm cutter sticking out 40mm will deflect more than 840 microns (calc is for a solid round bar, cutter will be less stiff) that is 200 times your deflection because of torsion.
For what it is worth, the normal stress at the root is circa 940 MPa, which means that cutter will fatigue off in short order in any case.
This is why I have been asking about load cases in the first instance, as I think a more realistic load will be in the region of 200N???

Please check and see if a had finger trouble, it is late here and I should be getting into bed.

[drieslaas]

A quick check:
Side-by-side beams are 76 square tube, 2mm wall, 300 mm apart, 300mm to point of force application at bottom of cutter
whole z-axis/cutter,y carriage is modeled as very stiff beams, so only deformation because of two tubes is really seen.
500N side-force
Deflection is 0.3mm