I’ll admit. When I saw the Othermill for the first time I thought it was just another mill with cheap Chinese hardware inside sold as a premium. I’m ashamed to say that I even trash talked it a little bit. It gave me another chance to relearn that I should always do my research before being a jerk, check my assumptions thoroughly, and even then it’s not recommended. Other Machine Company was kind enough to let me swing by the office in Berkeley California. [Danielle], the CEO, led me through the design of the mill as well as the challenges in running the operation.
The Othermill is a serious machine, and with the recent release of the Othermill Pro, it’s only getting better. The components are not bargain basement. This is something that could be more obvious, but it’s almost entirely made from US sourced parts, including the custom stepper motors. There aren’t any ball bearings that will start to make strange noises in a year. It can now cut 6mil traces in a PCB all day long. To put it into perspective. The Othermill Pro costs a third of the price of an equivalent machine from LPKF and has the same capabilities.
The Othermill started as a DARPA grant researched at Otherlab. They wanted a cheap, long-lasting, and easy to understand CNC for every classroom, something with the same capabilities as a laser cutter but none of the toxic gasses or fire hazard. It resulted in a rather odd-looking machine. The machine worked exactly like a vinyl cutter with a spindle rather than a blade. Sheet stock was fed into the rollers and it moved the material back and forth until it was finished. I have some doubts about the design but, [Danielle], assured me it ran pretty well. Since she has a PhD and is the CEO of a CNC machine company, I was inclined to believe her.
However, this machine is not the machine we see today. The government, capricious as always, saw a newer, shinier button on the floor and waddled over, dropping the Othermill research project from its sticky fingers as it ran. With nearly all their funding gone, Other Machine Co should have given up, instead they restructured, took on some jobs just to keep the lights on, and worked towards a Kickstarter.
The next iteration of their machine, and the one that was shown in the Kickstarter video, began its transformation into the Othermill we see today. Interestingly, the machine was fastenerless at the start. This was a cool design choice, and had some advantages, but not enough over the use of fasteners. The machining was more expensive and the machine was harder to service.
Throughout the development to final Kickstarter release the machine got a lot of upgrades. It grew handles. It got an enclosed build volume. The wires were nicely managed. On top of that they added a really nice software stack. The level of polish is impressive.
In the end, this worked. Other Machine Co didn’t go under. It worked its way to enough financial independence to split off from Otherlab and get its own facility in Berkeley.
What’s The Difference? Low-End vs. High-End?
To the hobbyist stumbling onto the Othermill it’s hard to get a grip on why it costs what it does. Ebay is flush with those 3020 CNC mills from China for one-third of the price. Why should someone drop the extra cash on a machine with very similar on-paper specifications? Those specifications are nice, but all specs are written when the stars are all aligned, the head engineer has just spent three days tweaking the device, marketing is a little drunk, and the CEO is irritable.
Paper specs will give you an indication of what to expect, but evaluating the real world performance will give you the truth. Even top-of-the-line CNC machines with exactly the same specifications can be behave differently in cutting, noise, vibration, surface finish, speed, and more. If this wasn’t the case they wouldn’t spend endless hours pestering companies with salesman. They’d just post the lowest price and people would buy.
Vibration is a subtle thing in CNC machines. By its nature, the machine makes a lot of noise, flings chips around, and generally makes a mess. If you put a good CNC beside an equivalent bad one and run them side by side with the same settings, you’d often be hard put to spot a truly obvious difference right off the bat. However, when it’s time to fit two parts together or inspect the surface finish of the part the truth becomes obvious. Vibration matters.
It’s really easy to see the effect of vibration in a 3D printer, a sister machine, as the picture to the right shows. The dynamic forces on the frame all add up to a repeatable ringing in the plastic. You can actually see the nodes and antinodes in the wall of the print. By changing the speed and acceleration settings the user can reduce these forces on a printer until the print comes out smooth.
Now, add in a spinning endmill seeing an intermittent load of varying magnitudes depending on a whole bucket of variables, from material type to the phase of the moon, and the problem with vibration in a CNC mill becomes easier to see.
Mentioning the 3020 mill again, here’s a great slow motion video of the spindle occasionally deflecting and vibrating back to zero. It would be very difficult to spot this behavior with an unaided eye. Keep in mind, this device is promising practically the same positional accuracy as the Othermill, however, it is often deflecting by more than .01 inches from the loads on the spindle.
The spindle on the 3020 is a cast and lightly machined bearing housing press-fit into the front of a brushless motor housing. Any vibration incurred inside the motor goes right to the part. Also, the rolled sheet metal housing does not have enough metal to provide an adequate hold against the forces. This video explains how some serious design flaws in these spindles can lead to a large run-out. It also mentions that it’s possible to ruin these spindles in about 4 hours with cuts that should be expected for a milling machine. There is a reason most, but not all, of the sellers market these are CNC engravers instead of mills.
The Othermill has a custom machined spindle housing with a top-of-the line collet holding the bit. The spindle has a properly matched set of precision ball bearings which are properly pre-loaded.
The motor is decoupled from the action of the endmill by using a belt drive instead of a direct one. These belts have to be carefully selected and will eventually wear out from the heat and mechanical stress of everyday operation. This small trade-off (the belts are cheap and easy to replace) is worth it. The Othermill also recently got a custom spindle motor manufactured for them, offering higher and quieter performance than the R/C motor they were using before. I’ve been told that their new motor meets the same dynamic balance standard as a hard disk drive.
This all adds up to a spindle that is quiet, precise, long-lasting, and can take a surprising amount of load for such a small machine. The real-world run-out for this spindle is significantly lower than that of the Chinese spindle.
Alright, but the Chinese mill is aluminum and the Othermill is just some crappy plastic. Surely aluminum is a superior material for a mill. This is not true.
The frame Othermill is constructed from machined HDPE. Depending on the aluminum alloy it is compared against, HDPE can be stronger than aluminum while providing many distinct advantages.
HDPE dampens more than aluminum. All these tiny vibrations add up in the machine. HDPE is able to convert some of those vibrations to heat. Aluminum, instead, transfers most of those vibrations rigidly, acting like a spring. This means that the forces and oscillations of the cutting and movement of the machine have to go somewhere. Typically this is into the joints of the machine; that is the end mill and the bearings. This can cause anything from poor surface finish to premature failure of the machine if not handled properly. For the Othermill this results in a quieter and more accurate operation.
They also use the greater flexibility of the plastic cleverly by using a compliant joint to align the smooth rods. One rod is held fixed in the frame and the other is allowed to move left-to-right, but not up and down on the other end. This is actually a very stiff construction when compared with the loads the machine will see and prevents any of the out-of-alignment stiction issues or premature wear that can happen with a linear guide bearing set-up
HDPE is much lighter than aluminum. For the spindle carriage and other moving components in the mill as well as for the mills portability, this is a good thing. The heavier the carriages the stronger the dynamic forces will be on the frame. The gantry design of the chinese mill is moving quite an enormous weight for such a small machine. This will cause a lot of vibration, noise, and more deflection when the machine is traversing quickly.
Of course, as seen in the spindle assembly above, where the more rigid properties of aluminum is needed, Other Machine Co has chosen it as the material. HDPE would make a terrible spindle housing. It expands too much when hot. It deflects too much hold a bearing accurately.
HDPE is cheaper than an equivalent aluminum alloy. This brings the cost of the machine down. Since the plastic is cheaper they can just use more of it. In volume alone the othermill uses significantly more material in its frame than the 3020 mill. This results in a very strong, light, and enclosed frame for the machine.
Not only is HDPE cheaper, it cuts like butter on a properly set-up machine. This allows them to do very complex cuts that would simply be too expensive to do in metals. As we’ll see later. They can even afford to cut slots into the walls of the machine for very neat wire routing.
With a solid frame and spindle the only significantly contributing mechanical element left to compare are the linear movement mechanisms of the respective CNC machines. The 3020 mill is somewhat famous for occasionally offering ballscrews and linear guides at its price range. However, a ballscrew is a famously difficult object to manufacture and as many have experienced, not worth much if it won’t turn.
The 3020 mills vary between sellers. Some offer ACME leadscrews, others offer, typically, C7 precision rolled or ground ballscrews. This added complexity of the ballscrew on such a small mill is not necessarily a selling point. One of the advantages for the ballscrews, especially in larger machines, is the lower friction under load. However these larger machines have drives so large and are moving weights so substantial that things like the inertia of the screw and the vibration and noise from them become negligible. Typically ballscrews are also more shielded from the particulates generated by milling than they are in the 3020 mills.
The Othermill uses teflon coated precision rolled lead screws from a USA company, Koco motion. Paired with a properly matched anti backlash lead nut, these can match the performance of a C7 ballscrew without any of the complexity drawbacks of the ballscrew. The teflon coating eliminates most of the driving force difference between the two.
While both the Othermill and the 3020 use round linear guides the 3020 typically employs roller bearings while the Othermill uses the more expensive teflon impregnated acetal bushings. Not only are these bushing just as or more precise when properly preloaded. They, again, have considerably less noise issues, wear issues, maintenance, and vibration issues than the roller elements. This is visible in a comparison video below.
It can be easily assumed that a $ 600 dollar mill is going to have very cheap electronics. The electronics for all of these mills are designed to run on a parallel port computer. So along with your purchase you’ll have to find a suitably old computer to actually run the software, Mach 3, for this mill. (LinuxCNC is possible as well. )
Aside from the higher power draw, higher noise from a lower PWM frequency, neither Mach 3 nor Linux CNC currently support S Curve acceleration without some serious hacking. As this video shows, this can make quite a difference for a machine in motion. Not only will the load on the tool be more uniform, but the machine will vibrate less.
TinyG is also a well proven board with a lot of support. The Chinese mill will come with electronics which have gone through the heaviest of bean counting with only the most cursory of tests. It is likely that they will break under normal use. We’ve covered one user’s experience with the 3020 mill electronics. One of the first things he had to do was completely replace the power supply. It also took some digging and hacking to enable features, like spindle rpm control, which should have come enabled.
Lastly the electrical connections are just more reliable in the Othermill. The 3020 mills usually lengths of wire that need to be routed and terminated in the cheapest possible DIN connector they could find. These are hardly reliable connections for a milling machine.
With the Othermill’s goal of being a portable machine a lot of work has gone into having the correct cabling properly routed and terminated. There’s a lot of work that can go into this. For example, if you just attempted to use standard Belkin Telecom cable or ribbon cable for a CNC machine you’d quickly find the wires breaking connection intermittently (Looking at you Makerbot!) wherever they are moved. Cabling has to be designed for flex applications or they’ll break. This costs money, companies do everything from special plastic formations and complicated wire braids to putting teflon powder between the conductors to achieve a flex cable.
The final difference is one of QC. The import mill will definitely need some work by the end user to come anywhere near the specifications promised. We’ve seen users do anything from replacing all the electronics to replacing the spindle with a more expensive one. The machine is likely to be out of square, and adjusting one of these machines can be quite difficult.
As with most higher-end/domestically manufactured products, the Othermill goes through significantly more testing and calibration before leaving the factory. The mill is trammed and the bed is milled to be exactly square with the travel of the machine. The electronics are all checked and everything is tracked. Other Machine Co offers another level of support than can be had from a Chinese maker.
Othermill does deserve a final mention for its software stack, especially for PCBs. It’s one of the selling points they tout the most. Having seen it in action. It definitely works as advertises and will especially make someone’s life easier if they are making circuit boards.
However, in the case of regular machine work, LinuxCNC or Mach3 isn’t that hard to master. At the end of the day, owning a CNC of any sort still requires a bit of technical know-how and the ability to read documentation. Once it’s all set-up, it’s basically click-go for both.
What it all means:
At the end of the day all this adds up to a machine that will cut the same way every time. As an added bonus, it will also be significantly quieter than other options. I was able to sit next to one while performing a circuit board routing operation without hearing protection. A properly built machine is a durable one; so far I’ve seen four Othermills in the wild and they’ve all been in heavy use. They can mill a wide variety of materials precisely, with occasional maintenance they should last a long time before a significant failure
The import mills, on the other hand, can theoretically be brought into the same specifications as the Othermill (thereby matching the specs they promise). However it will come at a significant time and money investment.
Mechanical engineering is a whole different beast. It was interesting, while performing the research for this article, to see how even some assumptions I had before writing it were less correct than I imagined. I thought that a ball screw of any class would outshine an ACME thread, for example. I’m sure there are many of you out there with experience and education that exceed mine. Looking forward to the comments.
Disclaimer: In a search for something to write I asked Other Machine Co. if I could visit their offices. I do not at this moment own an Othermill nor has Othermill given me any sort of boon for writing this article. All the views, subjective and objective, in this article are my own as a working engineer who happens to write for Hackaday. Hackaday does sell the Othermill in its store and has one in the Supply Frame design lab, but that is not related to my writing. I do think it’s quite rude that no one has ever thought to at least offer me a comically sized bag of cash (to turn down, of course) for a positive review. It’s just polite.