[joeavaerage]

Hi pippin,
that is correct, I have applied a load in the X axis direction, if you like 'bending the headstock sideways'. As peteeng has pointed out I'm being lazy and applying a 100N force at the corner of
the nose rather than in the middle, which mildly distorts the stress pattern, but is just so simple when running Fusion FEA.

The constraints are the two opposing edges to mimic a row of five cap screws down opposing flanges. There again not a perfect representation of reality but a reasonably
close facsimile thereof.

The advantages of using simple and repeatable settings for application of load and applying constraints is that you can run multiple FEA studies and be assured you are seeing
the results 'of that thicker flange' or 'thinner wall' or whatever it is you were experimenting with. I imagine in time I'll settle down to one model which is more realistically loaded and
constrained. At the moment 'broad brush strokes' are in order that I determine those most critical parameters and dimensions.

Craig

[peteeng]

Hi Craig - Do not apply load to a point. Create a feature so it has a "real" area to load. A point has many analysis issues and breaks a few FE rules. Make two or three components in one model so comparing is easier and keep restraints and load features the same so its apples to apples. Or I should say Kiwi fruit to Kiwi fruit. Peter

[peteeng]

Hi Craig - Heres a fusion model as a suggestion with the FE results. I'm not sure if the FE results stay in the model or on the cloud. I'll do some tests to figure that out. The aim is to get the load to the cars so the webs down the middle acheive this. I'm not sure if you want a sub plate so you can remove the backplate without removing the "saddle". But This can be made as an edge bolted assembly with dowels and it will be so much stiffer then the 100x100x9 section. It can be welded lightly and surfaces corrected. No SR needed if the surfaces are good. The 100x100x9 moved 14um the big one moved 0.4um under 1000N both are steel.... Peter

tell us how the fusion file opened and if the FE results are there, anyone with fusion...

You could weld this up and leave the front plate separate as a bolt on. That gives you access to the drive nut bolts in the "box". This puts the bolted connection close to the spindle so its loads are least... Just thought bubbles at the moment some details are needed.

[peteeng]

Morning - Since you have Fusion you also have basic generative design features. So I ran the GD on the big design. I preserved the bearing lands and the front plate of the headstock. In the minimise mass run and the maximise stiffness run the answer is the same in this case. Use the max geometry available and make it hollow until you achieve the required stiffness or mass. Peter

[joeavaerage]

Hi peteeng,
thank you very hard for this input.

I can see that even applying a load on a line is more realistic and therefore going to be closer to real world results. I will modify my models accordingly. It is not clear to
me how you apply a load over an area...but if I fiddle around for a while with Fusion perhaps I will stumble on it.

I have never used GD, and yet it looks like I need to experiment with it to gain what it has to offer.

Craig

[joeavaerage]

Hi peteeng,
looks like the zip file opens in Fusion on my desktop, but if I go to display the results of your simulation then it complains that the results have not been calculated.
That would suggest the results are on the cloud. If we want to share results I believe it can be done by you declaring that the results can be shared. I have on one or two
occasions shared Fusion files so that they could be viewed on the Fusion Forum.....but only have a vague idea how you actually do it.

Its hardly worth breaking a sweat over, I can calculate the results here just a easily. Sharing results if they took several hours to solve....well that would be a different matter.

Craig

[joeavaerage]

Hi peteeng,

The aim is to get the load to the cars so the webs down the middle acheive this. I'm not sure if you want a sub plate so you can remove the backplate without removing the "saddle".

My inclination is to have the headstock bolt to the saddle. The saddle is nothing special, just a piece of 20mm steel plate surfaced in a grinder. It's current dimensions are 275mm x 275mm and that handily covers the
cars but with little room to bolt a headstock to 'outside' of the car positions. Should it prove beneficial I can make another saddle, say 350mm x 350mm, just to get that extra wide bolting capacity.
The saddle is not sacrosanct, but shifting the cars or the ballnut block are.

Craig

[peteeng]

Hi Craig - When selecting a feature to apply a load to hover over that feature. A face, line or point can be selected. Seems the FE results get dumped. But the FE parameters are there just rerun the solver. Keep at it. Peter

[joeavaerage]

Hi peteeng,
thanks for that.

I have to keep in mind that 90% of my machines use is making circuit boards, and really the machine is vastly bigger, more powerful, faster and totally overbuilt for its
main application. I have over the last week accumulated the best part of $3000 in commissions to build, test and program various boards for a range of customers.
Here I am spending time and money on a headstock that I want,as opposed to what i need, a fourth axis that I wanted, but really don't need, and a fifth axis I want, but don't really need.....
The fourth and fifth axis have cost me something like $5250NZD thus far....and yet that investment was only possible because I can make a living building circuit boards!

You may recall in other posts my servo driven spindle has been damaged. I allowed it to sit in a manner where coolant got into the encoder and wrecked the encoder. Very annoyed
with myself for allowing it to happen. The encoder is not to my knowledge available as a spare part so I trying to buy another Allen Bradley servo to replace it. I've found one in the UK for 200GBP.
By the time I get it, with GST and freight, its going to cost me $600NZD...and that hurts! While I don't use the servo spindle a lot it is indispensable when I do, so I feel obligated to repair it.
It will certainly teach me not to be so cavalier about how I store and operate it.

Craig

[peteeng]

Hi Craig - Working for oneself is always a juggling act. My want and need list is big. Anyways here's another go at the headstock. Its a cast E50 grout pyramid with steel end plates. The 100x100x9 reference moves 31um, the big steel box moves 1um and the grout moves 2um. So comes down to whats easier, steel or grout... Peter

[joeavaerage]

Hi peteeng,
starting to get better and more consistent results now.

I'd like to experiment a little more before I commit to any one strategy, after all the point of the thread is to explore a number of options.

To date I'd say that a 200 x 200 x 9 SHS section with welded base and nose plates, a solid block of Al, a fabricated steel beam, a cast grout headstock, a cast iron or steel headstock are all viable.
I'd like to run some models with epoxy granite, although would guess they'd be comparable to E50 grout and maybe glass or carbon fiber headstocks.

One thing is for certain, it will not be hard to beat the existing 100 x 100x 9 welded headstock that I'm currently using!!

Craig

[joeavaerage]

Hi,
OK, slightly more sophisticated drawing. The restraints are two slightly raised lands at either side, to model a string of cap screws down each vertical flange.
A hollow pyramid shape with a 12mm wall thickness and a 20mm nose thickness, and a 32mm thick flange, in regular 30ksi gray iron. the flexure is a hint
over 0.3um for a stiffness of 325N/um. This easily exceeds my target, and casting this in iron is very doable. Hardly any need to consider the same design in SG iron or steel,
if gray iron exceeds my target.

This is not ten-fold better than my existing 100 x 100 x 9 headstock, its better by 100 fold!

Craig

[peteeng]

Hi Craig - So why not make in steel? 2 days and you'd have it? Peter

[joeavaerage]

Hi peteeng,

So why not make in steel? 2 days and you'd have it?

Quite true, and maybe that's what will happen in the end. I suspect that I'll never be happy unless the steel is stress relived, and that is about $8NZD/kg, and the part is 35.4kg,
plus transport to and from Auckland, not unreasonable expense but not inconsiderable either.

The only real thing to recommend gray iron over steel is the potential damping. If the casting process is too expensive then I would be dis-inclined to pay all the extra
to get a damped material. It may prove that casting is not prohibitively expensive either, in which case I would pay extra for it...but not over the top.

I would like to go through a few different designs to get a feel for what each offers. I've only been a few days at this yet...after a fortnight or so I will know where I'm heading.
Given the aforementioned $600 expense to resurrect my servo spindle, this project is delayed anyway.

Craig

[joeavaerage]

Hi,
essentially the same design, ie pyramid shape of the same external dimensions with a 20mm nose and 20mm flange, 12mm walls in mild steel but retained at the four corners
where the cars are on raised pads results in a deflection of only 0.18um or a stiffness of 550N/um. Very impressive. It sure is hard to beat steel for stiffness.

Craig

[peteeng]

Hi Craig- If you fabricated it in steel and faced the front and back and it stayed flat (which is my guess it will) then your home and hosed. It's a simple part not like a space frame with lots of different internal pulls and pushes. The internal stress is in equilibrium so does not affect performance. The only reason to SR is so it does not move when you remove lots of material, which in this case is minimal I believe. Set the connection on epoxy so it has full contact plus the epoxy acts as a damper...

But yes, early days keep evaluating and cogitating. If it's that stiff damping is a none event compared to the rest of the machines vibs like the column for instance....Peter

[ardenum2]

over $5 grand for, what are these? cam rollers or worm gear reducers? if cam then eh its okay, wear on cams is low but worm gear not so much, what are these rated for? 6000h?

you make weird purchase choices, you could get two direct drive torque motors for that amount and not worry about any wear, ever.

[peteeng]

Hi Craig - Now you have a model do a modal analysis. When you start an FE study pick "modal" rather than static and follow the same workflow. Now Fusion will figure out the vibration modes and their frequencies. If the freqs are all really high and not at freqs you'd expect from the motors or the spindle than its all good.... Peter

sorry can't do a modal at my level of payment so will take it into simsolid. Peter

[joeavaerage]

Hi ardenum,
they are worm servo reducers, one has a lash of <2arc min and the other <1arc min, both are new old stock and in perfect order.

The slightly larger of the two has a peak torque of 400Nm, and a continuous rated torque of 120Nm. The smaller one has a peak torque of 180Nm and a rated torque of 80Nm.
By the time they were in New Zealand I paid about $600NZD ($370USD) each for them.

I want you to try to find a direct torque motor of even 20Nm rated, and matching drive and tell me how much its worth. For their torque output they are expensive, I know because I looked and I looked
and I calculated. There is just no way I could get the torque I needed at reasonable cost with a direct torque motor. Aside from anything else...do you know how much those things weigh?? They are
bloody heavy, even the freight to New Zealand would be more than a servo reducer.

If you can find a direct torque motor to suit either a fourth or fifth axis then good for you. I could not, so I looked for the best solution I could afford and these worm reducers are it.

I have two 750W Delta servos and they are $1000NZD each. They start off at $438USD each but then $156USD freight or $594USD=$1000NZD each.

These Items I actually have. The fourth axis is complete and is in use. The fifth axis is largely complete. I need the timing belt which is on its way to NZ and a couple
of T slot tools and they too are on their way. I've bought and paid for the steel for the trunnion table, I have to pick it up next week. I need to get my servo spindle running
(another $600) in order to machine the steel for the trunnion. I still need a small piece of 4140 shaft for the offside trunnion spigot and a piece of 12mm plate to mount the whole lot on.
Maybe another $100 worth of steel, and a couple of bearings for the offside trunnion spigot and I'm done with the 'spend money bit'.

Perhaps if you think my purchases are weird you'd be kind enough to show some of the things you've bought within the last year. I'm certainly interested to see the linear servos you linked to in another post,
of course they were 400N rated so way WAY too small for my machine, but I'd love to see what you've built with them.

Craig

[joeavaerage]

Hi peteeng,
the original headstock did not get any stress relief, and I have no complaints. Firstly the 100 x 100 x 9 SHS was as it comes from hot rolling and would not expect the huge stress than welding
can induce. The base and the nose were skimmed flat in the mill and that was it. Even if it does move...so what? Its not like its moving around with residual stress while I'm doing a job, except
cutting forces of course.

If I were to fabricate a pyramid style headstock there would be a lot more welding. By your argument it may not be significant, which is an encouraging thought, but I would rather that it were
stress relieved.

There is still plenty to model, so I do not have to decide at this instant whether to stress relieve or not. The company that did my axis beds have a vibration relief machine, and that is so much cheaper,
and its about a 3-minute drive to get there. The foundry is 5 minutes away (less in low traffic) in the other direction. The shop with the big surface grinder is less than two minutes away.

I have not tried a modal analysis, but suspect it's the sort of thing you actually have to pay for. There is a distinct possibility with flat pyramidal sides that there could be annoying vibrations.
On the other hand with a stiffness of 500N/um it highly probable that any vibration of the headstock would be vanishingly small relative to the 'twist' of the column.

I modelled the AL headstock. I retained the same external pyramidal dimensions but with an increased wall thickness of 25mm and a increased flange thickness with four munting pads
per the earlier steel model. The result was good, abou 0.3um deflection for abut 300N/um. Hollowing out the inside of a 60kg lump of aluminum sounds rather daunting, and I might just leave
the internal volume alone. The only thing that counts against AL is the price of the orignal 60kg lump, about $1200NZD otherwise it too could be a very viable solution.

Craig