What's new
What's new

Starting a small machine tool manufacturing business

Generic Default

Aluminum
Joined
Jan 23, 2016
Location
Wilmington / Long Beach
The premise is that since 2010 there has been a massive increase in the number of "makers"; technical people who enjoy building stuff as a hobby. Many start out with a 3d printer or hobby level lathe or mill, and quickly outgrow it. There is a substantial market for inexpensive CNC machines that did not exist 10 years ago, and recent advances in electronics have dropped the costs of building these machines to a fraction of what they were before.

SwissMak - The Mill Turn Center for your machine shop by Overlord Robotics Corporation —Kickstarter
2174481519373640219.png


So I built a couple prototypes of a machine design that I found to be very useful. This is a mill-turn layout, which lets it make typical lathe turned parts, 3 axis mill parts, 5 axis parts, 3+2 and anything in between. It's optimized for bar work through the spindle, and anything in the size range of a 2 liter soda bottle or smaller, although it can still swing much larger diameter parts. This style of machine also allows sub-spindle/backside work, regular tailstock work, and even removable trunnion table workholding between centers. It fulfills the needs of people who are making complex metal parts but cannot afford or justify a full sized mill turn center.

Most of the internet skepticism has been directed toward the solid aluminum structure, on the assumption that it will ring during cuts. In a nutshell, the structure is very stiff but doesn't dampen vibration within the aluminum itself; rather it just passes vibrations through to the vibration damping pads that the structure sits on top of. This method works just fine since the spindles are only 2 horsepower. It's obviously not the machine for roughing big inconel blocks.


So far the Kickstarter has over $134,000 pledged, and there is approximately a 1:4 ratio of people who actually pledge for a machine versus people who say they're waiting until after the Kickstarter batch is shipped to place orders. The plan is to use the Kickstarter funds to set up shop and produce the first ~30 Kickstarter machines, then produce them continuously a la carte. After the Kickstarter batch is done we'll introduce a heavier built cast-iron version with larger motors and servos on all axes for industrial customers.

So in the coming weeks and months I'll be posting a few more threads here on PM, like a New Machine Day thread as soon as the big VMC hits the shop floor. Let me know your thoughts on this, and any advice you have on what to do/what to avoid based on your own experience running a machine shop. If you have any questions, ask away, but read the Kickstarter first!
 
You don't want to hear this, but:

It's a very nice design, but if any repeatable accuracy (or any reasonable metal-removal capacity) is expected it needs to be cast iron. Aluminum is a low-modulus material and deflects under one-third the load for iron (or steel).
 
Not just the deflection (which is extremely important for accuracy), but vibration damping as well. This machine are giant interconnected spring-mass-damper systems after all. Cast iron has excellent damping properties which is one of the reasons why it is so widely used in machine tools (and engines!). This machine is going to be like a tuning fork at the smallest amounts of excitation.

6061-T6 damping capacity: 5
2024-T4 damping capacity: 2.5
Mild steel damping capacity: .9-1.4
Grey cast iron damping capacity: 30-90
Source - https://faculty.engr.utexas.edu/sites/default/files/jmatersci_v28n9y1993p2395.pdf

BUT, like you said, it's targeted for a hobbyist. So I'm sure you've weighed the optioned and determined that your customer would rather have the lower price and a lighter weight machine. If it's priced competitively I'm sure you'll get some interest, but keep in mind this forum is largely focused on production/industrial machining and not hobby.

Either way, pretty cool design and good luck.
 
So $4700 USD gets me a mill-turn centre with Al-Al ways?

"It takes little effort to hold overall part tolerances to within a few thousandths. The machine alignment and positioning resolution are well under one thousandth of an inch. Rating the overall precision of a machine tool with a single number is kind of like rating the picture quality of a camera as megapixels."

:skep:
 
It's pretty. And ambitious. Pretty ambitious...

I'm curious how you control and interoperate the multiple axis. I went to the ODrive page, and it doesn't indicate how many simultaneous axis it can handle. I also didn't see whether the linear drives are "all thread" equivalents, polymer nut/stainless screws, or ball screws (I think that the last isn't happening at this point).

So could you explain the controls side better? How much is ODrive, how much is your adaption?
 
This is a mill-turn layout, which lets it make typical lathe turned parts, 3 axis mill parts, 5 axis parts, 3+2 and anything in between.

Errr No it's not. I see indexing for B not Feeding. Pretty cool though, maybe I'll get one for me Daughter. :D It actually looks like a decent layout, so long as it isn't meant for anything heavy.

FYI, bragging that it'll broach ®Delrin isn't exactly awe inspiring.
 
I like it, I think you could sell a bunch, but.... the type of people that would want this are not likely to be G-code geniuses, so including a really easy to use cad/cam conversational type program is important in my opinion.
Modelers and lab guys wanting a quick complex prototype part are not going to want to spend hours programming.


Sent from my iPhone using Tapatalk Pro
 
I think it's neat.

Lots of people are doing cool stuff with 3D printing, which offers atrocious materials, accuracy, and surfaces until you step up to extremely expensive printing setups. I think Tormach has proven that there is definitely a healthy market for inexpensive machines, and LOTS of dudes who started out with a Tormach are now buying real gear and making great incomes.

3+2 machines are probably the most optimal for the prototyping market in a lot of respects, because it's easier to get people to wrap their heads around done-in-one techniques than it is to go through the long, somewhat painful learning curves of multiple part setups.

My only critique is that I would have skipped the aluminum, been a bit more ambitious with the price points, and gone for a $10k 3+2 machine with a cast iron base. Tormach has proven that there is a market for what's essentially a modern Bridgeport in the $20k price point (and brought it down to the 440 at the $5k price point).
 
I saw that on Kickstarter, and it looks like a really good start. A couple of things would need to be improved for me to buy into it though. First, as mentioned before, the aluminum needs to go. Steel, or better, iron all the way. Second, I need better than .001", since on a multi-axis .001" repeatability per axis stacks up fast to around .005" part accuracy. Third, it needs a controller capable of dynamic work offsets and tool length offsets. I don't want to have to regenerate and repost my code every time I replace a worn tool.

Good job though, very ambitious!

Oh, and have you looked at Clearpath servos?
 
Weldments of mild steel will do just fine.

Outsource locally to any number of shops, that can burn & bend 1" plate.

Have them weld & anneal, machine surfaces for mounting linear ways.
Deliver to you.

You bolt on, and wire, and align.

Much like other small machinery builders have done.
 
Not easy to find a source of iron casting, except,,, china.

There are iron foundries in the US. I printed a pattern and had State Brass in Salt Lake City make me two grey iron tailstocks for an old gear hobbing machine. Nice parts, good machinable iron, and inexpensive enough to be viable for a small machine tool, probably comparable to solid aluminum plate/bar stock considering you would core it out.

On edit: The constructional standard the OP needs to meet is that of the better manufacturers of simple machine tools of the WW2 era, such as Walker-Turner or Delta. Making it from aluminum is about like printing it from ABS—great as a visualization aid...
 
I'd echo the concern over the aluminum ways. I'd be more concerned about getting them gouged up in short order due to chips getting stuck in the slides. Perhaps hard anodizing would help alleviate that, or better, an excellent baffle system to protect the ways completely from contamination.

The lathe turret looks a bit spindly to me, and is likely to be a source of vibration. It may even be overkill where a QC multi-fix could be used for a hobby guy. He can also defer some initial cost by buying fewer toolholders to begin with. Or, he might want to rig up a gang tool which doesn't require a tool turret.

I could not tell from the video if the tailstock can be coupled to move in unison with the headstock so that longer components can be supported during lathe operations. It should be, to be useful.

Congrats on a really nice looking machine. I can't believe you could make any money at that price, which may take a lot of the fun out of the whole project.
 
Big list of responses to everyone below:


Oldwrench,
indicator-repeatability-limit-gif.gif

linearrepeatabilitygif.gif


Just a couple GIFs of an indicator test for linear motion. The limit switches repeat to within a couple tenths and there's really no stick-slip. Backlash was indicated at about 0.0003" when adjusted and 0.0007" without adjustment. Given that all machining is done within a few inches of travel on each axis, the speed benefits of ballscrews are negligible. The linear thrust assemblies use opposing roller bearings.


AzDrake,
The 6061-t6 machine frame can't build up resonance once it's on the damping pads. My first v1 prototype did resonate, which is why I took measures to get rid of it on the V2 prototype you see in the first post.

Metfab,
Cast iron is no doubt a better material for heavy users. Rather than go to China for castings we'll just mill the parts from Versa-bar or Dura-bar here in the US.

RC Mech,
The boxways will be hard-anodized with PTFE in the pores for the production version. The sliding contacts are adjustable polymer pads, similar to this: https://www.igus.com/wpck/17352/drylin_t

Milland,
Each ODrive board only controls 2 servos. The main control board for the SwissMak can run up to 12 step/dir outputs, and on top of that things can be controlled by a CAN bus. No allthread either, the production version will be running precision leadscrews coated with tungsten disulfide against long polymer nuts.

ripperj,
The medium term goal is to have a conversational touchscreen that makes it super easy to write 3+2 programs. Longer term goal is a drag-n-drop workflow that analyzes 3d models and writes the program for you. This will need some serious software investment, for now the machine just runs regular GCode from generic CAM posts.


gkoenig,
The "done in one" aspect of a mill turn center is the big selling point. The reason I went with 6061-t6 for both prototype machines was because it's available, movable by one person, and it doesn't rust. Plus, I had to build the 2nd prototype (Kickstarter showpiece) on a $2,000 budget. Everything you see in the video was done on that budget.


mhajicek,
Tool offsets are already in the firmware, non issue there. Clearpath servos are compatible with this machine. It's designed for mounting compatibility with Nema 23, 24, and 34 motor frames as well as ODrive servos.
0.005" is terrible in terms of repeatability and positioning accuracy. When I say the machine positions to within a few thousandths, I didn't mean that each axis stacks up like that! Main concern here is thermal growth from uncontrolled air temperature.

gkoenig again,
Yes, I've been seriously considering an epoxy granite base. Still on the table.
 
I'd echo the concern over the aluminum ways. I'd be more concerned about getting them gouged up in short order due to chips getting stuck in the slides. Perhaps hard anodizing would help alleviate that, or better, an excellent baffle system to protect the ways completely from contamination.

The lathe turret looks a bit spindly to me, and is likely to be a source of vibration. It may even be overkill where a QC multi-fix could be used for a hobby guy. He can also defer some initial cost by buying fewer toolholders to begin with. Or, he might want to rig up a gang tool which doesn't require a tool turret.

I could not tell from the video if the tailstock can be coupled to move in unison with the headstock so that longer components can be supported during lathe operations. It should be, to be useful.

Congrats on a really nice looking machine. I can't believe you could make any money at that price, which may take a lot of the fun out of the whole project.

The kickstarter text mentions "Hard anodized boxways for long term wear resistance".

I wonder if something like turcite could be used on the ways, applied to the aluminium substrate?

EDIT: nevermind! seems like you just addressed all that in a post while I was typing.
 
Taken from the kickstarter page...

[FONT=&quot]We have hundreds of hours of machining experience[/FONT]

I hope that is not referring to yourselves but to actual machine run time of the prototypes :o
 
RC Mech,
The boxways will be hard-anodized with PTFE in the pores for the production version. The sliding contacts are adjustable polymer pads, similar to this: https://www.igus.com/wpck/17352/drylin_t

You'll have to use good wipers, otherwise chips will start embedding in the polymer and score even hardcoated ways.

Milland,
Each ODrive board only controls 2 servos. The main control board for the SwissMak can run up to 12 step/dir outputs, and on top of that things can be controlled by a CAN bus. No allthread either, the production version will be running precision leadscrews coated with tungsten disulfide against long polymer nuts.

I'm not familiar enough with CAN bus operation, will it allow multiple controllers to coordinate? And if so, at what data rate? Where can one read up on this?

BTW, a little confused on the plug-in operation claim. It looks like you offer up to 5Hp in spindles alone, setting aside the drive electronics and linear/rotary motors. Are you really saying this machine can function from a 120V/15A outlet?
 
The main control board is the hub for all other boards, in addition to step/dir outputs and a CAN bus for the ODrives, it also has a ton of auxiliary pins and ports for touchscreens, LCDs, solenoids, thermistors, ect. Can't remember the data rate off the top of my head but it's pretty high.

The plug-in operation is exactly as it sounds. The SwissMak has 2 DC power supplies, one 48v 1500w and the other 24v 350w. The big one is for the spindles and servos, the small one is for the control electronics and steppers. Both power supplies work at 50/60 Hz and 110-240 volts single phase. The reason the spindle PSU is 1500 watts is because that's approaching the safe limit for 120v 15amp home circuits. Above that, there is a risk of tripping the breaker. You can't run all of the machine spindles at the same time at maximum power without triggering the power supply auto shut off.
 








 
Back
Top