[peteeng]

Hi Craig - I have looked at that in the past. Although they are replete (visions of grapes and togas) within themselves once you get to the machine it becomes another layer to deal with. I think its better to design each machine component for its own end vs trying to generalise the design. Every time you generalise something this is a compromise. You can buy unitised drives like attached from various companies but they do cost quite a lot compared to separate bits. Comes down to what the Maker wants to do. Since all of my machines are designed for serial production in theory I don't see the value in such a general solution better off optimising the design as far as possible. If the product then sells, the volume sorts out the rest. I choose the best solution within the budget and have access to big machines if I need surfaces trued so I'm not in a corner like some. Peter

PS its sort of what I do with the timber beds. See attached scoot image. I now have a source for harmonic drives and slew bearings so Epoch5 needs to be detailed out.... Peter

[joeavaerage]

Hi peteeng,

I think its better to design each machine component for its own end vs trying to generalise the design. Every time you generalise something this is a compromise.

I'm not sure I agree. I designed my axis beds with 'no compromises', but to reduce the cost that is when I decided to make them all identical. One pattern....but three castings.
The only difference in the machining between the three axis beds was that the X axis had T slots milled into it. This again saved money....and therefore I was able to afford
a 'no compromise' design, whereas if each axis were individualised it would have cost considerably more OR I accept would have been forced to compromise to meet a realistic budget.
Not to say that making three identical axis beds with a 'no compromise' design was cheap, but it would have cost a damn sight more had I done it any other way.

I'm sure you can imagine also that three identical axis beds bolted to a large 'L' shaped steel frame resulting in a column type machine is a great, easy, 'no compromise' way to
employ the three beds. They are sort of an ideal match. If however I decide at some later time to change the arrangement then I can unbolt the axis beds and mount them in some new
frame. Most importantly I am able to retain the three axis beds, because they cost a fortune. $10,000NZD for the casting, stress relief and machining. Then add to that the linear rails,
the C5 double nut ballscrews and three 750W servos, one braked, so I spent something like $15000 (total) on these three axis beds. I doubt I will ever make another axis bed again,
they have another twenty years life in them which will probably see me out.
Craig

[peteeng]

Hi Craig - Everyone's best design and best set of compromises is different and leads to their own "best outcome". But the grail is out there... Peter

[peteeng]

Hi all - I have been working on a 5 axis router called Epoch5. I'm tossing up between a steel trunnion and an aluminium one. The AL one would be cast. I have a favourable quote at $30AUD/kg from the caster (included heat treat to T6 which I prefer). So now I have to detail out the two concepts and cost them out for a good decision to be made. I do want to get involved in aluminium casting as I see it as the way forward... Today I did a quick FE on a 35mm slab of AL vs the 50mm deep 12mm thick steel version (both 200mm wide). F360 materials went really weird today so I sent a strong note to autodesk that the way materials are handled in F360 is the worst way I have encountered in any CAD or FE system ever. So the AL 35mm thick weighs 11kg and the steel weighs 15kg so AL wins there so we shall have to wait for the costing to decide... They deflect nearly the same in a plunging load case. Peter

[joeavaerage]

Hi peteeng,
the base of my trunnion is 32mm thick steel. It was plasma cut then placed in a machine to buzz around the edges....but no heat treatment or any messing around. The 32mm thickness allowed me to bolt the
20mm uprights on and use plenty of torque on the bolts to ensure that the upright remained square. The greater the thickness of the bottom plate the less deflection might be expected from the uprights.

Cost $122NZD +GST and weighs 20kg. Cost me a box of beer to have the edges milled square and true, my own machine is not big enough or rigid enough.

Quite frankly by the time you bend a piece of 12mm plate with a return on each edge it will cost more than just a single piece of 32mm (or 40mm or 50mm...)....just to save a few kg????

Second issue is that you have shown the stepper/servo of the fifth axis concentric and parallel to the fifth axis bearing. That will make you fifth axis very tall and consequently the walls of your machine will need to
be very tall also, counting against stiffness. You need to minimize the height of the fifth axis stepper/servo, gear reducer whatever it is and the bearing stack. This in turn suggests you need a right angle drive
so that the stepper/servo be parallel with the underside of the trunnion table.

Last thing I would comment that I would design the fifth axis such that the platter is below the centerline of trunnion even if only by a few mm. If you have a control that you know is capable of TCP then maybe you don't care......but if you
don't then the center point of the machine, ie the intersection of the A axis (trunnion) and C axis (fifth axis platter) be exposed that you might touch off to it. You will absolutely need it to set up the stock to be machined.

Craig

[peteeng]

Hi Craig - point by point
1) This is a prototype router not a mill. For plastics, timber maybe aluminium. So stiffness requirement is not near a mill
2) I want to avoid a flat plate and bolted flanges. So bending allows a couple of things that "flats" don't
3) I looked at right angle drives and they are very expensive (low lash units) plus did not save much height. Agreed trying to minimise machine height is important... One design V12 has the trunnion low with a fwd gutter for the motor to swing thru. We shall see how V14 goes now I have settled out the trunnion options. Harmonic drives are very flat but very $$$ and I'm unclear how some of them anchor. Waiting for more details about those from the supplier.
4) I have not looked at the stack height of the system yet, been struggling with getting the trunnion concept right. I think a laminated steel bent design will be fine. My Z axis costs around $75aud to be laser cut and bent and it has proved to be very stiff. Its 3mm steel laminated to 6mm or 9mm. In the trunnions case I think 3 layers is good. But the top would be a very big dust/swarf collector so I'm thinking an upward flange is not a good thing. But the direction is clear and I'll detail an AL version to see cost deltas.
5) Getting the geometric relationships correct is now the big effort... I shall have to start reading up on F360 posts to see what its capabilities are.. Peter

I looked at bent thick Steel and aluminium plate, cast concrete, fibreglass/carbonfibre, steamed laminated ply and eventually cast aluminium came out where I wanted it but then laminated steel came back up... Once my mind had settled down to a reasonable level of activity and direction

edit - I tracked down the Z axis thickness calc. My Series3 laminated I section from 3mm steel is equivalent to 32mm of solid steel in bending or 46mm of aluminium so its light and stiff... My series 2 was made from 6mm steel and it was equivalent to 40mm thick of steel its flange depth was 55mm but it cost quite a bit more due to paint and or made from stainless. This one will have zinc plated parts so steel thickness is not that critical..

[joeavaerage]

Hi,

I shall have to start reading up on F360 posts to see what its capabilities are.

Its not the capabilities of Fusion that need concern you but your motion control. Fusion can produce either plain vanilla five axis Gcode without TCP, OR can produce Gcode which
is then interpreted by a TCP capable motion controller. It is the motion controller that calculates the required kinematics for TCP. This separates the 'men from the boys'. Most hobby CNC
solutions do not have TCP kinematics. The notable exception is LinuxCNC, but even LinuxCNC aficionados do admit that it is not that easy, but certainly possible.

You can see from the attached pics that in order to reduce the height or equivalently the max swing radius of the trunnion I used a right angle gear reducer but elected to drive it by
a belt that allowed me to place the servo alongside the gear reducer and minimize the swing radius as best I could. Of course there are even better solutions but the costs just spiral out of control.

I would suggest to give yourself and any potential customers of yours the ability to use this machine WITHOUT the benefit of TCP. Should you or they have a TCP controller then that is great,
but the machine should still be useable without TCP. For that purpose having the centre point of the machine a few mm above the platter is advantageous.

Craig

[joeavaerage]

Hi peteeng,
I did briefly consider casting the trunnion, but in cast iron rather than aluminum. What put me off was not so much the casting itself but rather the machining afterwards. It would in effect require a five axis machine to do
in one setup OR at least a four axis horizontal. As my trunnion table is 370mm long that restricted the possible machining facilities to the very best in the city and the most expensive.

I decided to fabricate the trunnion with one reasonably thick base and uprights bolted to it. To date I'm very happy with it. Any compliance I've noted in the is not in the uprights as I thought it might.

Craig

[peteeng]

Hi Craig - Yes machining it is tricky but my machinist has a large 5 axis mill but using that takes $$$. My thoughts at the moment is that the pressed metal steel is the way to go for the project at hand. But I keep looking at alum casting Peter

[joeavaerage]

Hi peteeng,
when I first thought about casting I was going to cast the trunnion stub axles and table in one piece. That sure does put the 'cat amongst the pigeons' when you go to machine it. If you were doing
it yourself and did not account your time, that would be fine.......but all the large 5 axis machines around Christchurch cost $200/hr and more.

In order to keep the budget under three figures you pretty much have to assemble the trunnion out of smaller parts that you can make yourself or your budget will be entirely consumed by some tricky-dicky machining,
at least that if what I found.

Craig

[peteeng]

Hi all - Today I sorted out how I will drive Epoch5's gantry. The 16mm ballscrews have a 15mm stack height delta to the bearings, always a juggle. But I'm happy with the solution. I shall cast the spacer using steel fibres so its damp... I've ordered the trunnion bearings so the build has begun!! I settled on using Unitised class 2 bearings from a Nissan Z. Using std double row (decided on 40mm) or matched tapered bearings was going to cost at least $80aud then I had to make the flange. These cost $46aud and I get the flange as a bonus!! They are matched tapered (43mm inside dia) so will adjust nicely.. Should run a million km.... I may use the same bearing for the rotary platter. When I get these two modelled up I can detail the trunnion then decide on the rotary details. Peter

I also sent a suggestion list to F360 support. See what happens.

[joeavaerage]

Hi peteeng,
the unitised wheel bearings are a practical solution, just cut a hole and bolt them in.

Not quite so easy as the platter bearing. You really need a pair and the further apart the better. A single bearing will rock a little even when done up tight.

You will no doubt have seen mechanics jack up a car, grab the wheel and 'feel' the play. You always expect some, say 0.5mm at the periphery of the wheel but less than 2mm or so.
Even when the axle nut is done up tightly, very tightly even, and the bearing is hard up against the 'crush sleeve' it assumes its 'preload state', but on a car axle that is just short of 'no movement'.
For CNC purposes you might adjust them closer than that, even tight. A wheel bearing is heavily loaded and spins fast whereas a CNC bearing is very much more lightly loaded and glacially paced by comparison.
A wheel bearing requires a little bit of clearance otherwise it would overheat whereas a CNC bearing will not.

If you used just one bearing the platter would rock, not by a great deal certainly, but some, and much more than you'd be willing to tolerate I'll be bound. If you had a pair of these bearings
say 150mm apart the 'rocking' would be very much reduced.

Craig

[peteeng]

Hi Craig - I have 2 on order for the trunnion and maybe a pair of double race deep groove for the rotary. Not sure yet, will have a play with the Nissan bearing and see how it feels. Auto wheel bearings have clearance suitable for running warm as they expand and as you say CNC runs slow and won't get hot. Will know more in a weeks time.... I may have to put a candle under the trunnion. Do they call them trundles in NZ? Peter

[joeavaerage]

Hi peteeng,
for twenty years I worked as a mechanic. Over that time I have fitted hundreds of wheel bearings. My recollection is that they all had a tiny bit of clearance even when the axle nut is torqued to
250Nm. Wheel bearings need it, but I don't think you'd be happy with it as a platter bearing on its own. A pair of them several inches apart...yes, that would be fine but adds to the construction and alignment challenge
and overall height.

As you've seen in the pics the gear reducers I used have two opposing taper bearings, approx 40mm ID and 150mm OD. The worm wheel stubs are a press fit into the ID of the bearing. The worm wheel is through bored
to 28mm (3 size platter reducer) and 32mm (4 size trunnion reducer). To date I have yet to detect any compliance in the bearings. My guess is therefore that properly adjusted opposed taper bearings of 40mm ID and about 120mm
apart offer adequate rigidity for machine such as ours. I suspect the 28mm diameter shaft of the platter to be the most compliant part, not that I can say I seen or measured it.

Craig

[peteeng]

Afternoon All - I'm impatiently waiting for the bearings so I can start detailing Epoch5. But I also want to make it half the weight (or close to) that of Frankie. So I worked through the usual material densities and stiffness's and find that F17 formply is a formidable material. If I look at equal weight basis then I need to use 4mm aluminium sheet or 1.5mm steel sheet. If I calculate the in-plane stiffness of these they are basically equal. A 100mm wide strip of F17 is 0.000029mm/N 4mm AL is 0.000033mm/N and steel is the same as AL. So to halve the weight I need to use 2mm thick AL which I can work at in my workshop or get 1.5mm steel laser cut. Now if I look at F17-17mm in bending it has a rigidity of 818,800mm4GPa. The 4mm AL is 37,300mm4GPa and the 1.5mm steel is 5625mm4GPa... So the F17 is the clear winner, in bending its 818/37= 22x stiffer then AL... Plus its easy to edge glue and screw into a stiff connection. The 2mm or 4mm AL I'd have to make brackets not difficult but its something to do... Now back to the half weight target. Unfortunately they don't make 8mm formply.. But 12mm pine ply is just over half the sheet weight. So seems 12mm BB ply achieves the weight target but not the stiffness target. So I looked at a laminate of 0.8mm AL each side of 12mm pine ply which gives the same stiffness as the F17 formply. But F17 is $34sqm and the laminate is $199sqm plus labour to make... Every time you try to loss weight it costs more!! There is a falcata ply that is very light but does not take edge screws well. Its pine skins with balsa core... So seems I shall use 12mm BB ply for primary areas with doublers at thru screws and 9mm for secondary structures... I also have to paint the ply but then I'll use epoxy or vinyl ester resin so its waterproofed and better glued together. Onward Peter

[cadpete]

A 'trick' with balsa and screws is to soak the balsa with thin CA glue around the screw area which hardens it with little weight penalty - used in the RC aeromodelling field all of the time. May be worth further investigation.

[peteeng]

Thanks fellow Pete - I don't haver CA glue but I can do the same with thin epoxy and wait overnight. But falcata is not really stiff enough. I went to the shed to do a screw test in 9mm ply and just looking at it gives me the insecurities.... so I think the main parts that are big and the screw is going thru it will be 9mm but the secondary bits that I edge screw into will be 17mm and I shall sculpt the webs etc in an engineering truss manner and try for a 50% reduction.... I rebuilt the E5 base into V14 using 9mm and dropped the weight from 21kg to 11kg so that was encouraging....will have to do some simulation to compare the two as well....Peter

[peteeng]

something like this--- still 11kg Peter

[peteeng]

Hi All - Been too hot to work in the shed and we had a little threat of a flood with the Qld cyclone so I've been busy prepping for that. But I have done some more on Epoch. Styled it a a bit and light weighted it. Happy with its progress. The saddle will be a fibreglass/carbon fibre hybrid part that look's like our aluminium saddle. I'm tossing up whether to go the std steel Z axis or making it composite. First get the trunnion sorted then the rest can follow. Hopefully the bearings arrive monday Tuesday then I can finish the "bottom end" Peter

[peteeng]

Morning All- Getting closer, by V20 should be polished....Peter