[mkhoward]

Hello everyone, I've just about finished designing a Tab & Slot style frame that will bolt together for my mill. I haven't ordered any of the parts yet and wanted some feedback before I commit to the build.
Specs for the parts I've included in it:
Avid CNC electronics bay and 4hp MTC spindle
Nema 34 motors except for on the 4th axis
counterweight for the z axis (not modeled yet)
HGR rails and HGW bearings (probably gonna source from aliexpress?)
MACH4 for the firmware
should weigh around 500 lbs assembled, maybe 1000 lb if I fill voids with epoxy resin and gravel, like the actionbox guys.
before shipping and taxes the build should be less than $5000.

My questions:
Do you think this could cut steel reliably and accurately using a 4hp Hiteco spindle and the frame just bolted rather than welded?
Is sourcing the linear rails and ball lead screws from someone more reputable like misumi really worth the extra money?
Are Nema 34 overkill or just right for a machine this size?
Can I spend $5000 elsewhere and wind up with a 4 axis mill capable of cutting steel and can be moved into a basement down a hill with no hoists? (most difficult requirement honestly)

Thank you in advance for any feeback!

[joeavaerage]

Hi,
yes that design certainly looks the part. I can tell from personal experience that the rigidity required to cut steel is extreme.
My guess is your design will cut steel but slowly, maybe slower than you hoped. My machine is very similar, that is it cuts steel but with less authority than I hoped.

Few points:
If you use 5mm ballscrews then the chances are you will not need a counterweight. The mechanical advantage that you get with 5mm pitch is so high that you might require less
the 0.5Nm from the stepper to hold it up......so the counterweight is not really required.

Nema 34 size steppers tend to have good torque but high inductance and will severely limit speed.
23/24 size stepper look for 1mH-2mH, 1mH preferred and reject anything over 2mH
34 size stepper look for 2mH-4mH, 2mH preferred and reject anything over 4mH
Use at least an 80VDC supply and 80V capable drivers.

I would have guessed that low inductance (1.5mH) 24 size steppers of say 400oz.in to 500oz.in (about 3Nm-4Nm) would be good.

Wasting your time filling with epoxy, as vibration damping goes it does very little. If the machine vibrates its because you did not make it stiff enough. Spend the money that you would have spent
on epoxy making the sections thicker and thereby stiffer.

The trick to cutting steel is high torque but low rpm. For a 12mm carbide tool you'll want 4Nm (min) and spin it at 2500rpm. High speed (18000 rpm-24000rpm)low torque spindles are a joke in steel.
I have two spindles, an 800W 24000pm spindle that I use for small parts in aluminum, brass and plastics and for making circuit boards. Had many thousands of hours from it....but its hopeless in steel
except with really small tools. My second spindle is home made with a second hand Allen Bradley servo driving a cylindrical ER25 tool holder in matched P4 angular contact bearings. It is 1.8kW, 6.1Nm (cont),
18Nm (overload) at 3500rpm. Use it for steel and stainless. Unfortunately I had it lying in the bottom of the enclosure and coolant got in a buggered up the encoder, so its out of action until I can afford to fix it.
Very stupid of me.

Ideally you'd use C5's screws, but they are ten times the price of C7's. Be careful there are quite a few Chinese suppliers claiming to sell C5 and C3 but scarcely more than C7's....total BS. There is proper
C5's and C3's and then there's rubbish.

The are stacks of Chinese suppliers selling Hiwin knock-offs. They look like Hiwin, same part numbers, same colours, same everything but they are crap. Buy genuine if you want good, but don't buy cheap and
expect genuine....it doesn't work that way.

For your budget you could buy a drill-mill. This is an example:

https://rongfu.com/mill-drill-machin...-machine-rf45/

Does a nice job in aluminum, and with care a fair job in steel. Its not CNC, but you could convert it. Your machine will have to be at least as rigid as this to do steel, and that is a tall order.

Craig

[peteeng]

Hi Howard - The tab and slots are good if you intend to weld. But then the frame will distort and you will have to finish machine the critical parts & lands. If you intend to bolt only, then do not use the tab and slot method as it does not help you. Mill the faying edges square and flat and use long stiff screws. This will make a very good machine. Plus use engineering dowels on critical edges and its a good machine. Epoxy filling is a band aid and if you design the machine right in the first place filling the parts is not needed. Go up a plate thickness and its cheaper then the epoxy but miles stiffer. Don't place the rails on the outside of the column. This will make it near impossible to get them parallel. Place them in the front so the lands can be machined in a single setup. Plus 4 rails on a single axis is not needed it does not help you and adds a lot of cost. Square cars should be moderate or heavy preload for a mill and square cars are very stiff so hardly any load will get the the back bearings. I have modelled this type of thing extensively and the rear bearings never help... Round rail bearings are another matter. The tab and slots are very good for brazing if your into that. I have done that with a few parts and its been very successful. Peter

[pippin88]

Overly complicated

Doubling the rails on the Z axis does very little. Makes it very hard to align all four rails

[mkhoward]

Craig, Peter, and Pippin, thank you for the excellent feedback! I will have to process it all and give a more thorough reply after work.

Craig, I’ll also have to look at those drill mills, my thought on them was that the 1hp motors will be in the same slow steel cutting category and I’m unsure how easy it would be to connect the spindle to avid cnc’s controller board. I was also looking at grizzly, only brand I’ve typically seen.

[jaguar36]

My questions:
Do you think this could cut steel reliably and accurately using a 4hp Hiteco spindle and the frame just bolted rather than welded?

Assuming you use enough bolts that are large enough, its fine.

Is sourcing the linear rails and ball lead screws from someone more reputable like misumi really worth the extra money?

The problem with getting components from a random vendor on aliexpress is you don't really know what you are gonna get. It may be perfect, even better than expected, or it might be absolute crap. Likely it will be fine, but its a risk and you don't have much recourse if it goes wrong.

Are Nema 34 overkill or just right for a machine this size?

There is a bit of overlap between the biggest NEMA 23 and the smallest Nema 34 motors. I think you could probably make do with either, but I would go with Nema 34 in case you decide you need to go bigger in the future.

Can I spend $5000 elsewhere and wind up with a 4 axis mill capable of cutting steel and can be moved into a basement down a hill with no hoists? (most difficult requirement honestly)

If its all bolt together, you can always take it apart to move it. (although aligning it again is a pain). However at 500lbs it should be easy enough to move down a flight of stairs. I've moved much larger machinery into my basement.

As for cutting steel, I built one for around $500 using a wooden frame that can cut steel.... slowly. It just depends on how fast you want/need to go. What thickness plates are you using for your frame? I assume they are steel? I think you're gonna have some issues with the way the Z axis mounts to the column. Those plates are going to twist and using 4 rails will add alot of expense and complication in alignment without much benefit. I'd instead double up the thickness of the plates and only use 2 rails. Also I think you need some more reinforcement at the base of the column. The plate inbetween the X and Y bearings looks pretty flimsy and is going to be pretty critical. I'd consider doubling its thickenss as well. Wouldn't add a ton of extra weight/cost but would give you alot more rigidity. The whole thing doesn't look particularly stable either, your CG is very high without a very large base.

Lastly, you haven't given many details on your 4th axis, but they can be tricky to build and stay rigid enough without any signfiicant backlash issues. Also keep in mind that there are very limited cheap/free CAM options for 4 or 5 axis work.

[mkhoward]

Okay I’ve had some time to dwell on this and think the appropriate answer is to change my budget or change my expectations, in addition to the design tweaks that were mentioned.

If I adjust the goal to be an aluminum milling machine that can do general tolerances of +/-.005” is that reasonable? I would be happy outsourcing my steel if it came down to that. However, some tolerances I would like to reach +/-.001”.

The design changes in summary that I need to focus on:
-Potentially break the frame down into two welded halves (column & base) so that I can still break those down for moving.
-Up the plate size from 1/2” to 1” on at least two sides of the column and probably all of the base, and taper the column so the center of gravity will be lower.
-use only two Z axis rails and put them on the front surface of the column.
-mill the front surface of the column and the base and make sure they’re trued in respect to eachother. Hate this one since I don’t have a mill still.
-beef up the bed, possibly with some welded steel tubing
-drop the counterweight
-probably scale down to Craig’s Nema 23 requirements

And jaguar36, I was going to use a prebuilt 4th axis with a belt which I read helps with backlash a lot. I’ve also found some videos on rebuilding those axis’s and improving them if I need to. For the CAM I only plan on doing wrapping tool paths for now, which Fusion 360 includes with their basic package.

This is also my first post, do I need to be doing something different so people see that I replied to them?

Thank you everyone!

[peteeng]

Hi Howard - If you weld you will need to have the parts stress relieved. So you need to find a local heat treater who can do that and figure out costs and logistics. If you don't stress relieve the parts will change shape when you machine them and you won't get straight parts. Do not believe a welder who says they can weld and keep it straight. No disrespect to them its just physics and the demands of machine building. Straight for a machine is different to straight for a structure. I have worked as a welder and as a welding trainer and have tried to make welded parts with no post machining and have failed. I'd stay away from welding... unless you can SR the parts.

You need to look at mills that are accurate to your 0.001" and see how they work and what parts they use. To build a machine that can achieve 25um means everything has to be built to 1/10 that so its a 0.0001" 3um build. Thats getting way out of a home build zone. But to achieve +/-0.001" repeatability is doable. You need to figure out what you actually need there. All of the drive parts come in various tolerance grades. Cheap parts are not that accurate. High tolerance parts are very expensive, like high tolerance ballscrews are 10x the cost of hobby grade stuff. You will need to learn about part tolerances and suppliers and cost... Then you need to machine parts to a commensurate tolerance. The project balloons from there.

Your replies are fine. N23 or N24 motors are fine with a 5mm pitch screw. Aim at cutting steel, where your heading is good. Keep at it. Peter

[joeavaerage]

Hi,

If I adjust the goal to be an aluminum milling machine that can do general tolerances of +/-.005” is that reasonable?

That is a very much more achievable goal, and its not the tolerance, I think with care you could do ten times better that that, so 0.0005 (5 tenths), but about doing aluminum rather than steel.
Aluminum requires a rigid machine also, but way way less than than doing steel.

-Potentially break the frame down into two welded halves (column & base) so that I can still break those down for moving.

Highly practical idea. The only problem with welding steel is residual stress which will cause small but troubling movement in your machine in years to come, especially if you machine a surface.
To overcome this you should thermal stress relieve the structure once its all welded up. I priced thermal stress relief here in New Zealand about a fortnight ago and it worked out at $6.70NZD/kg ($4.00USD/kg),
so its not cheap....but once its done steel is a perfectly stable and uber stiff material to make machines from. If your machine parts weighed 250kg then would $1000USD stress relief scupper the project?

-Up the plate size from 1/2” to 1” on at least two sides of the column and probably all of the base, and taper the column so the center of gravity will be lower.

Good idea, in fact why stop at one inch? The sides of my column are 32mm. Wouldn't bother with a taper, anything that adds stiffness is worthwhile, anything that does not is pure fluff.

-mill the front surface of the column and the base and make sure they’re trued in respect to each other. Hate this one since I don’t have a mill still.

This is almost mandatory. When I got my cast iron axis beds machined it cost me $6,500NZD, I almost cried.......but the company that did them made an EXCELLENT job, detailed care at every step
on the very best of machinery (near new Okuma) by a guy who knows what he's doing and takes pride in what he does. In short while the cost seemed high, I get the benefit of it every day and every time
I use my mill. My advice is don't be put off.....this step will determine the squareness and accuracy of your machine. Do it well, whatever the cost, and you'll get the benefit for the twenty year life of your
machine.

-beef up the bed, possibly with some welded steel tubing

Possibly. Use Fusion 360 to model that part of the design. As Jaguar points out it is critical, and that's where good FEA modelling really comes into it own.

For the CAM I only plan on doing wrapping tool paths for now, which Fusion 360 includes with their basic package.

I don't think that is correct. I believe that Fusion 360 Basic does 3+1, ie indexing fourth axis only. Fusion 360 Basic ($780NZD/year) plus Fusion Machining Extensions ($2200NZD/year)
does simultaneous four axis, simultaneous five axis, collision avoidance and toolpath editing....ie the full works....not cheap, and that what Jaguar was referring to.

Craig

[joeavaerage]

Hi,
there is a wrinkle to stress relief. For centuries thermal stress relief has been the norm, but raising a steel or iron part to 650⁰C letting it soak and then slowly cooling it over a period of 24 hours or so
requires a s***t ton of heat.

There is another technique called Vibratory Stress Relief. It's a machine that shakes the bejesus out of the part with a variable frequency. Its way WAY cheaper. There is quite a lot of papers about its
effectiveness in the academic record. It seems to work well with large structures but is less effective with thin sections, so the jury is out with regard to welded steel structures with thin sections, say 1/2 inch or
thinner. For such structures thermal stress relief is still the way to go.

My cast iron axis beds were vibratory stress relieved, the company that did the machining have such a machine and use it themselves, and it cost about $300NZD to have three cast iron beds of 115kg each
done. Had I got then thermal stress relieved it would have cost $2000NZD.

If you are looking to weld a structure the vibratory stress relief might save you a swag of money. It did for me....mind you the thinnest part of my axis beds is 50mm, so the technique is known to work well with
parts of that type.

Craig

[peteeng]

Hi - Vibratory stress relief is good for any thickness material. Its the relative thickness of the part and any free edges that are a problem. Say you have a square box structure with a thin tab on it. The tab will vibrate violently and bend, so it needs to be supported somehow through the process. The company I do lots of machine analysis for does lots of VSR and we have found that sometimes things like ram ears and tabs can crack at welds when VSR is used. They now tack on braces to tabs and cantilever bits then cut them off after. If the company is good at VSR they will know these things and accommodate for the issues. Peter

[joeavaerage]

Hi,
have you given any thought to casting the major parts?. Casting is expensive, and benefits from thermal stress relief, adding to the cost over and above the machining required,
but the results are superb.

You may have seen from the pic of my machine I had three identical axis beds cast in grey iron, weighing 115kg each. Cost about $3300NZD ($2000USD) to get them cast. They were vibratory
stress relieved for another $300NZD. They are superbly dimensionally stable, stiff as an old boot and damp vibrations rather well. I can certainly say that I got what I paid for and extremely
happy with the results. Just as a reality check I paid $3200NZD for the two 750W Delta servos and one braked 750W Delta servo. Since then I've bought another two 750W delta servos
for $2000NZD for my fourth and fifth axis. So the axis beds (just the casting and stress relief) cost less than the servos I use.

Just for clarity I bolted the axis beds to a rather large (1400mm x1100mm) 'L' shaped steel frame, the sides are of 32mm med tensile steel. Being able to bolt them to a frame allows me to
use shims to 'square' the Z and Y axis beds to each other, that is to say I did not have to get the frame machined, and thereby saved having to sell off my first born to finance it.

I understand the cost....but do a little research, you might be surprised what you can do for a reasonable sum of money. The other thing to remember that if you do a good job of this this machine will be
in the family for decades. What seems expensive at the moment becomes very modest over the years to come......but that ONLY applies if the results are good enough, or better yet, superb, that you'd keep it for decades.

Craig

[mkhoward]

Craig, if there is a similar service here in the US that charges $2300 for casted and vibratory stress relieved grey iron, I will do it in a heartbeat. Do the surfaces where the HGR rails go not need to be machined? Do the parts not also need to be set for years to let the internal stresses relieve or is that an urban myth?

Thanks,

Mason

[joeavaerage]

Hi,
casting iron is an age old process and is practiced in the US as much as anywhere. You need to find a foundry who would do it. Do some research, I have no doubt you will find foundries
who will help. Once upon a time every town and city had a foundry, these days not so much. What you are looking for is a Master Craftsman foundryman, and when you find one treat them like
gold.

Do the surfaces where the HGR rails go not need to be machined?

Yes, they do. But then whether you make it out of steel or epoxy granite or UPC you will still be up for machining, so most techniques require finish machining.

Do the parts not also need to be set for years to let the internal stresses relieve or is that an urban myth?

Myth. No doubt that years of weathering would be a benefit, but cast iron suitably stress relieved is as stable as can be. Prior to the advent of granite surface plates what was the choice of material
for surface plates....cast iron?. Still today if you buy a cast iron surface plate how much more would you expect to pay verses a granite plate of the same size? Cast iron has always been more costly.

My plan for my mill was that I would make axis beds using the best, stiffest, best damped material I could find, and that is still cast iron. With the casting, stress relief and machining they are expensive,
but then they form the heart of the machine. They are bolted to a steel frame. At the current time the steel frame has not been welded and does not therefore require stress relief. When I designed it
my intention was to have two thick (32mm) 'L' shaped sides and then 8mm or 10mm plates welded top and bottom making a rectangular tube. When it came to it my budget had been nearly exhausted
and so I have the two 'L' plates bolted together with 20mm separator plates. I needed to get the machine operational as I use it for work.

Some day I will get around to stripping the mill down and cut the 8mm/10mm plates and weld them in. Then the structure would require stress relief. The nearest oven big enough is Auckland, 1500km away and on a different island,
so freight is a significant component of that cost, aside from the 250kg of steel, or $1675NZD of thermal stress relief.

Then my machine will be at design stiffness, whereas it is about 40% of design stiffness. It still works, and especially for aluminum, brass and plastics more than adequate. It even works quite well in steel, but less well than I had designed it for.
This is just the reality of home brew CNC, you have limited resources so you have to do the best with the budget you have. In this case I can revisit the structure to improve stiffness when time and budget
permit, but the axis beds do NOT have to be revisited, and they will be with me for years to come.

I don't have $1675NZD lying around, but an even bigger issue is that I can't really take my machine out of service for even as long as a fortnight, I use it daily for work. I'm not quite sure that I'm ever going to get the time
to do it.

Craig

[jaguar36]

Hi,
casting iron is an age old process and is practiced in the US as much as anywhere. You need to find a foundry who would do it. Do some research, I have no doubt you will find foundries
who will help. Once upon a time every town and city had a foundry, these days not so much. What you are looking for is a Master Craftsman foundryman, and when you find one treat them like
gold.

Sadly, most of the foundries in the US have dissapeared. The building I used to work actually was originally a foundry making single castings that were over 50 tons. Its generally too expensive too do though these days and has mostly given way to welded steel construction or outsourcing over seas. There are a few 'backyard' foundaries popping up again though such as Windy Hill, but this might be too big for them, and is likely going to be very expensive.

[joeavaerage]

Hi,
the US is the third largest producer of cast iron castings (after China and India) with 11million tons in 2019, and averaging over 10 million tons over 2018 thru 2020.

The foundry I used is not dissimilar to that that Jaguar linked to. It's run and owned by a Master Craftsman foundryman. They did the pattern and molding but sent the mold to Timaru (150km distant)
to pour, as the Timaru foundry has a 1 ton capacity whereas my local shop has 65kg capacity. My next casting is likely to be done by a company about 30km north of the city. This company specialize in
diary equipment casting in stainless, using investments, lost foam and greensand....so the whole show. 250kg capacity.

You really need to do some research as these companies are somewhat specialized and are not going to be widely known. Ask at any engineering workshop, there are many whom perforce get bits made,
pump impellors and that sort of thing. They are bound to know whom it is who still does such things, the guy's name right down to his shoe size! As I say these remaining Master Craftsmen are highly valued and
respected in engineering circles.

Craig

[peteeng]

Hi All - Internal stress created by the manufacturing process, thermal changes or metallurgical changes or welding in a machine part, even how its held in a mill to rough it can introduce strain, this does not matter as its in equilibrium at its finish point. When it does matter is after this point when you machine it and change its shape, now it's not in internal equilibrium and will change shape again to achieve equilibrium. So when you machine lands for the rails or finish machine a casting it will change shape. This is why we stress relieve the part prior to final machining so when its machined it stays in the intended shape. In regards to leaving metals for years to age that's possible but most metals can be heat treated to accelerate that process into hours vs years. Cast iron and aluminium have internal metallurgical processes that generally involve compounds that move from the grains into the grain boundaries. This changes the volume of the material slightly and creates internal strain or the material tries to minimise its stress as it sort of re bricks itself. This is sometimes thought about in aluminium alloys as the grains get "bonded" together with hard compounds that fill up the gaps and increase its strength. Plus the aging temperature stress relieves the structure.

Granite was made around 30 million years ago so its had time to settle down and become stable. Thats one reason its excellent to use for machines. Thermal stress relief works on the mechanism that at some temperature the metal will become soft and any strain higher than that hot strength will stretch (steel 650C red hot, that's why steel buildings fall down in fires) and when cooled that area will then have less strain in it. I say strain because stress is a mathematical object and actually does not exist physically. Strain does exist physically, and we use stress as its easier to do math's with. But it does not help to visualise what's happening....

One thing that has occurred in the threads is that when final machined the lands have come out wrong. A good machinist will clamp a part in a zero strain condition. If you bolt down a part and force it into a shape even 0.1mm, that shape will be mirrored in the part when released. So make sure your machinist understands what a zero stress or zero strain set up is when it comes to the final machining. A zero strain hold down is checked by releasing a clamp and checking the part does not spring or sit at that point. Then check the other clamps as well. There's lots that's can go wrong in making a machine... Peter

[joeavaerage]

Hi,

Granite was made around 30 million years ago so its had time to settle down and become stable. Thats one reason its excellent to use for machines.

There is no doubting its stability, just a shame its no stiffer than aluminum. Many of the granite machines are low stress machines, thing like CMMs and grinding machines, where stress is low and stability is prized for
accuracy and to hell with the processing cost.

If you want to make a machine stiff then cast iron is a better choice (110GPa vs 70GPa) for stiffness alone and is pretty damned good for stability. Steel is much stiffer again (205GPa) and with thermal stress relief and care is both stable and stiff.
Within the limits of what can be welded steel is still the low cost and versatile go to material for stiffness at lowest cost.....not that steel or steel processing is cheap, but bang-for-your-buck its bloody hard to beat.

Craig

[mkhoward]

Update: I've found a foundry a 3 hour drive away that I'd be happy to drive to to pick the parts up to save on shipping. They also do heat stress relieving, not sure about vibratory. I've modeled the base and the column (picture below) to be each under 350 lbs and the foundry said they would accept a pattern made by me which could save on labor. The surfaces that aren't modeled with any fillets are what I intend to be machined and trued to each other. I should get a quote back tomorrow but I expect it to be higher than the New Zealand prices, Craig. The two halves I currently have modeled with x4 5/8" bolts binding them together. Does cast iron drill and tap well?

If they give me a quote that is in a doable range I'll model the bed and spindle carriages too.

On the Fusion 360 topic, the wrapping style toolpaths work with the basic package. If you require more than 3 axis simultaneously cutting together then yes you need multi axis, but wrapping doesn't need that. I've linked the video that autodesk tech guy sent to me with this explanation.

https://www.youtube.com/watch?v=GTV9vCTPiZo

But, because I'm a pessimist about the cheapness of the cast iron I'm still trying to figure out the best way to go the other route. Even welding and truing surfaces seems expensive so I'm trying to brainstorm ways around that. Going back to the first design I uploaded it seemed unanimous that I don't need 4 rails on the z axis, and it was suggest that I remove it from the sides. I originally had them on the side because I could then do finite adjusting of the rails to true it to the bed on at least one axis, without machining. Yes it would be a pain but potentially less painful than paying thousands?

[peteeng]

Hi Howard - 1) a cast iron frame needs to be finished machined so that cost is same for steel/aluminium or CI. 2) if you make your own pattern you need to discuss shrinkage allowances with the caster and various details so the cast is good. Every technology has its own nuances 3) CI machines well 4) since you have stress relief handy, welded steel will be cheapest solution 5) having rails on the sides is doable if you can have the lands machined parallel. if the sides are not parallel you will kill the cars very quickly by the misalignment forces. Look up the tech manual on the rails/cars and they have geometric specs to meet 6) making a good machine costs $$$. You can't beat a mass produced machine that does what you want if it exists. A simple commercial mill may work out cheaper than a bespoke simple mill.

Looking back at your original spec list, I'd go welded steel as this will be your cheapest (steels cheap) lightest (welded steel parts are hollow) stiffest (steel is the stiffest material). Since you have TSR handy steel is the "best" route. Don't worry about damping make it super stiff and damping is a not on the radar. There are millions of steel machines out there and they don't have a vibration issue.

The machining costs always surprise people and there must be 1000's of thinking years spent on trying to sidestep this issue on this forum. Unfortunately, that's a given unless you have a family member or mate in the trade and can access cheap machine time. You always have to use a technology level up from the technology you want to achieve in this sort of endeavour. Peter