[joeavaerage]

Hi,
some time ago when I was building my machine, and had pretty much run out of money, I made this headstock. Its made out of 100mm x 100mm x 9mm RHS,
and is OK for the little 800W spindle pictured but is too 'wobbly' for a bigger more powerful spindle. I'm now in a position to make a new one.

I am a traditionalist in the sense I favor steel and/or iron for machine parts, but many hobbyists favor aluminum, epoxy granite, high performance concrete among others.
I have traded 'blows', figuratively speaking, with other hobbyists about different materials and techniques. My thought was to throw the design open so that anyone who has a
favored material or technique can contribute.

The part itself is about 275mm x 275mm x 300mm, and so within reason, any decent technique should be affordable for me to actually make. I have drawn a rough sketch.While the major
external dimensions are fixed, many of the small details are not. For instance I've shown it as a 25mm thick shell, thinking that I might cast it in iron, but the drawing would be needs
be modified to include draft for casting. Feel free to modify the drawing should you need to do so for your favored construction.

I will use Fusions FEA tool to determine the rigidity of various designs. My machine is about 25N/um at the moment, and I'd hoped for better when I built it, about 50N/um. There are
things that I have yet to do on the frame that will make 50N/um achievable so this part (headstock) will need to be at least 200N/um, and preferably 300N/um so that it does not
compromise the overall stiffness of the machine.

I have a 750W servo direct coupled to a 32mm diameter 5mm pitch ballscrew so weight is not critical. I could tolerate this part being 100kg, but would prefer to keep it to 50kg
or under.

I'm guessing that if I cast this part it will cost around $1000NZD ($650USD). So I have some budget, but its not unlimited either. If someone wants to use boron fibre like was used
in the F15 Eagle.....its not going to happen. I can't afford 'pie in the sky' solutions.

Craig

[joeavaerage]

Hi,
I'm only modestly familiar with Fusion FEA but.....this simulation is for 100N applied force at the corner indicated with the part being made as a 25mm
shell of grey cast iron of 30ksi strength. The constraints are two fixed edges along the bolt holes.

The max displacement is just under 1um, for a stiffness of approx 100N/um. If the target is 200N/um - 300N/um there is some work to do.

Craig

[joeavaerage]

Hi,
I have run the same study but with SG iron as material, Youngs Mod=151GPa, for a displacement with the same load and constraints of 0.76um or 156N/um
and again but with mild steel as material, Youngs Mod 207GPa, for a displacement 0.47um or 210N/um.

It seems that you have to use high grade materials to get to 200N/um. I'm beginning to wonder if my target is too high?. This piece as drawn in steel or iron weighs 63kg....
which I would have thought was plenty, even over the top. The numbers don't lie.

It will be interesting if anyone can come up with a set of numbers employing epoxy granite than can come close.

Craig

[joeavaerage]

Hi,
I have run a model based on a 200mm x 200mm x 9mm SHS section, one of the local steel fabricators is holding offcuts in stock, with a 32mm med tensile plasma cut
base and a 20mm plasma cut med tensile nose welded together. The combination with the same load, ie 100N and the same constraints shows a max displacement of 0.63um for a stiffness of
159N/um, which is quite good for basically a welded steel structure. Especially as the SHS is available locally and the plasma cut plate is likewise readily available. Such a structure would require stress
relief after welding. A trip to Auckland and back for an item of about 40kg (approx $200NZD) and a short but hot stay in an oven while there, say $320NZD for a total of $520NZD, plus the cost of the materials,
probably around $350NZD, plus the hire of a welder, or getting someone to weld it for me, say $100NZD. Not much in it either way. Guessitmate $970NZD.

I've found that the maximum displacement is quite sensitive to any flexure in the base.....sort of makes sense, any movement at the base is much exaggerated at the nose, thus using a thick and
stiff base is indicated. But note how a good result is obtained with a thin (9mm) wall thickness. My initial models were all 25mm wall thickness and despite all that extra mass does not result in a stellar stiffness.

I beginning to wonder whether I should revisit the cast options, but be very judicious about the wall thickness and base thickness. I should also enquire at the local foundry what the costs are for casting in
steel rather than iron. I've always wanted something cast in battle steel!! Whether you can cast steel in 10mm or 12mm sections over the length I require I don't really know. I do know steel does not flow
nearly as well as iron.

Craig

[ardenum2]

upload the whole c frame, in 3D, if you want any real suggestions, with carriages and rails.

modern casting techniques don't need a draft, I'm guessing your casting guys cast old school, maybe take a look at a chinese company that specializes in machine tool castings, I think I've seen prices around $5 per kg for a meehanite casting. I'm sure you'll be sceptical about those claims though. There were a few that cast in steel or even stainless, on alibaba.

[joeavaerage]

Hi,
investment casting requires no draft but greensand casting does. Investment casting is way WAY too expensive. Its a great technique for hard molds for steel and stainless
but not cost effective for iron.

I think I've seen prices around $5 per kg for a meehanite casting. I'm sure you'll be sceptical about those claims though.

Actually no, $5/kg would be about what you would expect if you were doing production runs. One-off parts are more expensive.
The price for my axis beds for instance includes the price of the pattern. The pattern was pretty simple, I think it was about $400 of the total $3000.
Molding is expensive when done by hand such as the one off jobs that I'm likely to do, whereas production molding is mechanized and cheaper over a
production run.

upload the whole c frame, in 3D, if you want any real suggestions, with carriages and rails.

I don't have a single 3D model. Is there something wrong with the dimensions I have given?, do they not make sense?

Craig

[ardenum2]

I don't have a single 3D model. Is there something wrong with the dimensions I have given?, do they not make sense?

then just the column, I don't like working in a vacuum, otherwise just make the headstock as wide as the column, if you want to maximize the stiffness in it. Material doesn't matter, you might wanna do a test headstock from a grout before going to permanent iron, there really isn't much to do conceptually, it's just a block, it's final appearance will depend on what you're attaching to it.

Are you updating to a bt cartridge+servo? where is this going? might wanna include models for these components too.

if you're casting at the local company can you make the mold yourself? if you make a foam model yourself than you'd only pay for the iron, the question here is the ceramic shell, which might require an oven.

[joeavaerage]

Hi,

otherwise just make the headstock as wide as the column, if you want to maximize the stiffness in it.

The column width is 235mm and the rail spacing (centers) is 200mm. The saddle of the Z axis is 275mm x 275mm. I came to the same conclusion, that is that the headstock
would be stiffest by taking advantage of the widest dimension of the column, no matter the material used.

Material doesn't matter, you might wanna do a test headstock from a grout before going to permanent iron, there really isn't much to do conceptually, it's just a block, it's final appearance will depend on what you're attaching to it.

Good idea. I have been using my machine quite happily for 18 plus months with the headstock I currently have, so this is not an urgent project.

The few models I've done to date all point to the wall thickness being less important to stiffness but the any flexure in the base has a major effect overall. Basically if I use gray iron I end up at about 100N/um,
SG iron with intermediate (151GPa) modulus about 160N/um and if I use steel i get 210N/um. The modulus of the material is (approximately) linearly proportional to the stiffness of the model overall. No big
surprises there I guess.

Are you updating to a bt cartridge+servo? where is this going? might wanna include models for these components too.

Yes, that is exactly where I'm going eventually and is part of my reason for having the headstock retain square or rectangular surfaces that would be more convenient for subsequent design
work......but I don't have a design for those parts yet. I do have the large servo motor I intend on using, but nothing more. For instance the existing headstock is 350mm from base to nose,
but my drawing shows this design as 300mm base to nose, to allow for presumably a bigger diameter cartridge on the end.

if you're casting at the local company can you make the mold yourself? if you make a foam model yourself than you'd only pay for the iron, the question here is the ceramic shell, which might require an oven.

These are valid questions also. When I had my axis beds made I used the pattern maker whom is a regular at the foundry. I could have made the pattern myself but only at the expense of depriving the guy
of a small but legitimate job for which he is trained and skilled. I decided to pay the pattern maker. This was a cast iron job, so a soft mold in greensand. There is considerable skill in molding
and that is the province of the foundryman, aside from anything else I don't have the molding boxes, I think they call them 'copes'. I took the approach that if I were prepared to pay for the foundryman's
and pattern makers time and skill I would get best possible service. I think it was the right approach, although not the cheapest approach.

The three axis beds weighed 115kg each and cost $3000NZD (excluding tax) for the three parts, including the pattern, the molding, pouring and fettling, working out to $8.70NZD/kg or
$5.50USD/kg. The parts are superb and the guys at the foundry treated me very well, I'm happy and likely to do more business with them as a result.

A foam model and ceramic mold are for investment casting and is quite expensive for large parts.

I have a customer whom 3D prints plastic parts and then has them cast by the investment process. There is a company in a small town a little north of the city that specialises in investment casting.
They have a stellar reputation both within New Zealand and internationally for their stainless castings. New Zealand has a huge dairy industry, and this company specialises in components in the
milk processing industry. The tolerances this company can achieve are simply fantastic. My customer incorporates allowances into his 3D parts and gets better than 0.1mm as a cast part. The
ceramic mold is heated to red heat prior to being poured and is part of the technique that is used to keep the casting tolerances as tight, but that is a far cry from casting in iron in a greensand
mold.

I have just emailed the company to get an estimate of what an investment casting might cost.

Craig

[peteeng]

Hello Craig - Geometry is key here to acheive max rigidity. Make the section as big as possible. Steel at 200GPa will give you the most rigid answer. You appear to be loading a "corner" make a feature that gets loaded vs loading an edge or a corner. Make the model as real as possible. Also don't restrain the model over a large face, this is over stiff. Make a real or near real feature like 4 washers where your bolts will go and support those. Flanges are quite compliant and will need webs at the bolts. Or machine pockets. You may note that commercial machines have pockets at the bolt holes to maximise the size of the part and to make the local area very stiff in shear. Keep at it early days. I suggest you don't use SHS or RHS sections as this will tie you down to particular sizes. Design what is needed then figure out how to do it vs starting with something that will immediately confine your result. Looking at the saddle, its bigger then the spindle plate. Construct a pyramid shaped part. Parts with parallel sides have a tendency to vibrate more then pyramids do. Pyramids don't fall over. I'm sure you can make this yourself even if its a prototype then go for your beloved casting. Peter

edit- and which direction is the wobble? One thing to check is that the faying surfaces match. There's a lot of discussion about scraping and lapping at the moment. if the surfaces don't match then the wobble could be just a mechanical thing. So the surfaces need to be blued and checked for fit. Then they need to be matched or use epoxy to set the connection.

[joeavaerage]

Hi,

Make the section as big as possible.

Pretty much have done that. Using the widest dimension of the saddle is the most rigid.

You appear to be loading a "corner" make a feature that gets loaded

Yes, it was the most convenient location to load the part in Fusion. I need to refine it, as the stress is slightly skewed as is.

Also don't restrain the model over a large face,

No, I did not restrain over a face but rather the two edges. As you say this is still probably over-stiff, however a line of five cap screws down the flange will be a close(ish)
approximation to the model restraint I have applied. There again it was easy to apply an edge restraint in Fusion, and at this stage I'm talking 'broad brush strokes' while
I get a feel for the most important parameters and dimensions. Thus far I've found that wall thickness is less important than flange thickness and broadly speaking the
stiffness of the part is proportional to the stiffness of the material.

I suggest you don't use SHS or RHS sections as this will tie you down to particular sizes.

Yes, that is correct, however it turns out I can have a piece of 200mm X 200mm x 9mm here this afternoon and for very modest money, so you lose flexibility by virtue of taking
whatever section you can get but gain in availability at reasonable cost. I did not do the model, nor make the phone call with a view to actually doing it, but I can add it to the list of
'possibles'. The post welding stress relief adds to the cost, but worth thinking about.

Construct a pyramid shaped part. Parts with parallel sides have a tendency to vibrate more then pyramids do.

The reason that I've gone with parallel sides is simplicity at the earliest stages with respect to 3D modelling. I also at some time in the future I will add a servo driven ATC spindle and
square and/or flat surfaces would be convenient for that future plan, but not sacrosanct.

and which direction is the wobble? One thing to check is that the faying surfaces match. There's a lot of discussion about scraping and lapping at the moment. if the surfaces don't match then the wobble could be just a mechanical thing. So the surfaces need to be blued and checked for fit. Then they need to be matched or use epoxy to set the connection.

I have not noticed if there is any particular direction that the existing headstock wobbles. My feeling is that the principle directions of compliance are in the X and Z axes, and I base that on the
surface finish properties. The Z axis saddle is ground flat in a surface grinder and the existing headstock base is milled flat. It is far from impossible that the surfaces are not true and therefore
the headstock rocks. Even if the mounting were perfect the 100 X 100 x 9 section is still not enough, and it needs be replaced whatever.

I plan on running a few models that have internal webs and try that out.

I'm in no desperate hurry to do this, I need to finish my fifth axis before I start to spend any money on this, although no harm in thinking, talking, drawing and simulating it.

Craig

[peteeng]

Hi Craig - If 200x200 is the available space then a 200x200x9 is only taking advantage of 31% of the available rigidity vs a solid 200x200 solid section. 90% of the available stiffness is 200x200x42mm thick. So I think you need to think thicker. I did a quick calc for aluminium and a solid 200x200 piece will be stiffer then the 200x200x43 and lighter....Peter

200x200x43 steel is 210kg/m and 200x200 al is 108kg/m so half the weight... Al is easy to machine...Ops did that too quick, the solid aluminum is less rigid then the 200x200x43 steel. Seems steel is the answer for max stiffness.

[joeavaerage]

Hi,
I enquired about thicker walled sections but they are unavailable. 200 x 200 x 16 is a standard steel section but is very uncommon here, and not available locally.
The second issue is that even a 9mm walled section (32mm base and 20mm nose) results in a headstock of around 160N/um and not far away from my target.
So while a thicker wall may be optimal what I can actually get is close enough to being acceptable without a vast increase in weight or cost.

A solid aluminum section may well be a viable solution. I will enquire if any is available.

Have a company that sells offcuts of AL and just checked their site:
6061 203mm thick 230mmwide and 400mm long, 50kg
6061 254 mm thick 270mm wide and 300mm long, 60kg.

Have rung and the guy is going to check just to make sure either or both pieces are actually there and email me a price.

Craig

[peteeng]

Hi Craig - what is the size of the headstock back plate? plus car hole positions? or publish a fusion file or step file here. Plus how long (outreach) will it be? Peter

[joeavaerage]

Hi,
the size of the saddle, or backplate if you prefer, is 275mm x 275mm. Outreach is 300mm.

The cars are 200mm apart on centres. I'll draw up a sketch, I've lost the file I used for drilling.

Just had a reply, both pieces of AL are still available.The slightly larger piece, ie 254x270x300 is $1000NZD and $60NZD to get it to Christchurch from Auckland plus GST for a total of
$1219NZD. So AL is not especially cheap but is otherwise readily available. It should be on the list of 'possibles'

Craig

[peteeng]

Hi Craig- Thats $18NZ/kg. I pay $17AUD/kg so not bad. As noted on possible list. Peter

[joeavaerage]

Hi peteng,
the last time I priced from this crew it was $14/kg about three years ago. It was a very close call when I cast the axis beds, aluminum was just that little bit more expensive
for an equivalent design. In the end I decided to cast in iron, not based on price but rather that cast iron is lesser volume and it has such nice damping characteristics,
as I say it was a close call.

Craig

[peteeng]

Hi Craig how thick is the back plate? Peter

[joeavaerage]

Hi peteeng,
20mm steel prior to be whipped over the surface grinder, say 19.5mm.

Craig

[joeavaerage]

Hi,
I have just run the simulator for solid Al, ie 254x270x300. Works out to about 0.8um deflection for 100N load constrained at two edges or about 120N/um.
Sort of surprised, I would have thought a solid section would do better, but its actually less rigid than a piece of steel 200 x 200 x 9 SHS and welded base and nose plates.

No matter which way you cut it steel is as stiff as hell. Not saying it cant be beat but steel always seems near the head of the pack.

What I did notice about the FEA that the bulk of the Al block deflected vey little indeed but the deflection in the immediate vicinity of the load was quite high. Thus
if the load is distributed so that local deformation is minimized THEN solid AL looks good, very good even.

Craig

[pippin88]

Something sounds sus (suspicious, not right) about this.

Which way are you applying the load?
Worst cases scenarios for this part will be cutting across the face (cutting in X axis in the usual nomenclature) or plunging in Z.
An axial load (in line with the SHS wall will not be particularly informative)