This week the Phy team got our hands on a new PCB milling machine called the Othermill from the folks over at Other Machine Co. (OMC) — a San Francisco startup determined to bring desktop PCB manufacturing back into vogue. And this was no loaner…let’s be clear: we’d purchased the Othermill cash (ok CC) at full MSRP, after getting a live demo onsite at the OMC HQ in San Francisco.
So this early review (a more detailed look will follow in subsequent posts) is not biased by their generous offer to outfit Phy’s labs. In fact, at $2199 + tax, you can bet we’d be more than a bit hot-headed if this little device didn’t perform as advertised. But just when I thought I was over milling for good? Well, the Othermill just reignited a latent frustration since the last time any of us had milled PCBs on the bench: the inability to build real electronics, reliably, in a day. And the Othermill may just go a long way toward solving that conundrum for a whole new generation of the hardware-minded.
And just before we get into the technical mish-mash — let us also stress that for what it’s worth, we have owned or operated on the order of at least 4 PCB milling machines over our respective careers. And no slouch, home-cooked models either, but rather 1x LPKF, 2x Roland (MDX / Modela models), and access to another higher end LPKF mill thru a company I was working for many, many moons ago. All told, including the Othermill, that makes 5 machines spanning something like 15+ years.
Othermill appeared on our radar back in early 2013 after launching a highly successful Kickstarter campaign to build “An easy to use, affordable, computer controlled mill”. And with a target of just $50,000 USD the team at Other Machine blew it out, raising instead in excess of $300K. The item on offer? A $999* desktop mill to take “all your DIY projects further with custom circuits and precision machining.” (*special Kickstarter pricing at the time…it seems with experience the OMC crew would realize this just wasn’t sustainable at this price point, at least not if they intended to meet their target quality requirements.)
The premise was and remains a good one. After all, what aspiring entrepreneur or maker wouldn’t want the ability to fast track electronics at a pace more familiar to the likes of software engineers? Likewise, we hardware professionals (hackers at heart) have long sought the ability to design, build, & test a PCB in the same day, much less before lunch!
So, coming off the campaign, the OMC team labored thru more than 10 iterations and 8 months, dialing in their manufacturing process, before finally filling those Kickstarter orders…and fast forward another 12 months to Dec 2014 when it was finally time to release the Othermill V2 to anyone that might want one. Being the gadget masters that we are (ok, me personally) we had to have one.
Firstly, the Othermill isn’t something you order and arrives 2 days later, via Amazon Prime. At least not yet. Instead, the mill is on backorder (trusting that’s a good sign) and though they’re working hard to get the machining right and build quality gear; the average wait time for a mill is running a painfully loooong 6-8 weeks…Which for a $2200 gadget, feels like nothing short of eternity as you discover all sorts of nails just waiting for your hammer to arrive.
And at first glance, the Othermill V2 doesn’t look like any benchtop mill we’ve seen before. It’s compact, measuring 10.25 × 10 × 12.75 in. It’s also lightweight, weighing in at 16.8 lb, and yet feels sturdy, despite it’s modest stature. The handles imply of course that the mill is portable, which – with a proper generator – just might prove incentive to finally attend that family reunion at the beach (at 150W, best not to drive it from the car battery).
The mill area is a modest 5.5 × 4.5 × 1.25 in, however for PCB blanks, the size is probably more than sufficient. Only caveat being that most blanks don’t come cut to the bed size, so you’ll be buying blanks from the OMC gang (their prices are pretty competitive) or you’ll be picking up smaller sizes and making do with them (which should prompt you to think up a clever jig you can use to ensure flipping them is done while maintaining proper alignment). Either way, you have fairly limited Z-axis travel to work with, but it should suffice for small enclosures, gears, panels, small molds, etc…In addition to PCBs. (Nevermind your jelly molds and action figures.)
The assembly of the mill was truly painless. The Othermill comes preassembled with 4 plastic protective shields (windows) and a spring collet that needs to be installed. Other than that, it worked out of the box. And included are two endmills, one 1/64 and one 1/32 flat end mill, both ideal for PCBs.
Now for our sort of designs 1/64th was still far too wide an endmill for the PCBs we were building. So we ended up ordering a handful of 1/100 endmills good enough for 10mil tracks and 10 mil spacing, but alas not quite good enough for most BGAs or QFPs.
Up and running
Anyone who has owned a mill or operated one, learns the hard way that going from design to milled PCB is riddled with fiddly steps…Each of which unpack into a timeline that is almost never what you’d expected. Sometimes painfully so. And so much of this has to do with finnicky details like zeroing the mill, determining the bed height, locating brackets / fixtures, configuring material parameters, setting up endmill details in the software, and converting whatever design files you have to something that can actually be milled out, to name a few. Not so with the Othermill!
Above all else, our hats off to the folks at OMC for solving what is a complex software problem in an elegant and intuitive way, and in a single user interface with a single primary panel driving the whole occasion. In fact, my first design was on the mill and running in under 30 minutes. That was just remarkable considering the history we’d all had with all manner of other machines whose software looks more like something written by the creators of Bryce3D than something made for engineers building hardware. <cough…Roland…cough>
The second example I ran thru was a real PCB — an Arduino clone they call the Otherduino. All-up, the machine took roughly 4 hours to mill both sides of a board measuring 2.7 x 2.0 inches, and did a pretty solid job of it, though the thru-hole design was clearly not intended to push the limits of benchtop milling. Still, the design sailed thru the machine and the accuracy was spot on. Limited burring, very little, if any delamination, and nice routing paths.
So 10 mils you say?
Next up, we’d decided it was time to push the limits of the machine, which OMC says are 10 mil features with 10 mil spacing. The machine’s actual resolution is much smaller – ~0.001” or 1 mil…but these limits have to do with endmills breaking more than locational accuracy. Reality is, that even hardened carbide endmills simply lose mechanical strength when diameters drop to below 10 mils and even 10 mils is super small and prone to breaking as we re-discovered with the Othermill. (From personal experience, I’d broken 10 mil bits at the rate of sometimes 3-4 per day on my LPKF and my Modela…despite my best efforts to go slow and give the machine time to plough thru the material.)
Unfortunately, this limit is something that may never go away and as such, may well prove a deal breaker for anyone using some of the fine pitch parts increasingly common today.
This next design followed the Arduino theme, though using the ATMEGA328P-AUR in the MS-026-ABA JEDEC package on a little board we’d designed custom in Eagle..
The first thing we noticed was that following the JEDEC package spec still wouldn’t work out of the box because our pad to pad spacing was going to end up less than 10 mils (recall this is the area between pads, not the pad pitch). So with a few mods to the footprint, we managed to narrow the pad width a bit and increase the spacing to something large enough for the endmills we had on hand.
We also routed this board using 10 mil tracks in some areas and 16 mils in others, depending largely on what was nearby. We used a few vias with 20 mil diameter holes and decided to throw a polygon on the bottom side of the board for good measure. Lastly, we dropped the thru-hole caps, instead replacing them with 0603s and used a tight CSTCE16M0V53-R0, 3-pin resonator with integrated caps to top it all off. Altogether a reasonably standard design by today’s measure but something that would certainly keep ‘em honest.
And how did it fair?
The Othermill made reasonably short work (1 hour or so) of this design which measured 30x30mm, with only a few things we’d classify as “unexpected,” given our experience with benchtop mills. Still, as some folks are new to this space, it’s important to know what you’re getting when milling PCBs in the lab and we’ll detail a bit more on this later on.
Never the less, the Othermill did a great job at 10 mils. Minimal burring, only a few “overmills” and zero delamination. We used three endmill sizes including the 1/32, 1/64 and 1/100 in endmills — all flat end.
You can see we had some alignment issues which unfortunately are part and parcel of not having a more sophisticated jig setup when milling things to these tolerances. Though the supplied bracket did a world of good in making the result look at least respectable.
The Othermill is an amazing little device. It is well designed, well constructed, and just an altogether improvement over previous mills we’ve used. Topping the list again is the software. Heaps of machines can get to the resolutions of the Othermill, but what blew us away was the ease with which we could get going.
Othermill takes a BRD file (PCB file) from Eagle directly as one of the input methods. That said, all we had to do was muddle our way thru Eagle (Altium guy, sorry) and when we were finished; we had only to load the board file into the Othermill software and we were ready to go.
Similarly, the calibration step is about as automated as it could ever be. With power to the collet the Othermill will slowly lower your endmill on tool changes and detect the surface of the spoilboard automatically (the spoilboard is that metal plate on which you intend to mill). Much cleaner than doing this with arrow keys on the Modela and much faster as well.
Using this same method, it will also identify the boundaries of the supplied bracket; so again, making this far easier than anything we’d used in the past.
We did break one 10mil endmill in the process of milling the PCB but this is to be expected at this resolution and is hard to avoid with endmills this small.
Though to be fair, we have to say that thus far, it’s been largely Othermill gear, with Othermill materials, and all working in a pretty insular environment. Still, this early success is key and if it continues as we step it up this coming week, we’ll have plenty more to be excited about (either way, we’ll have more to write about ;). Like Microsoft Executives have said for years…supporting every conceivable piece of hardware and software developed by a range of folks with varying skill levels is a hard task. So the fact that we’re working in the controlled environment is not the best example of how the Othermill will perform when let loose on any variety of probelms. Still, if it were to fail on these fundamentals (cough…makerbot…cough), it would be a disaster.
Tell me where it hurts
A few frustrations do exist though. Firstly, the software is currently only running on Mac. <sigh> As an engineer who grew up on CAD tools, I almost exclusively use Windows. It is just the standard for top-end PCB tools and thus using Windows is not optional. Thankfully I also have a Mac so this was a no brainer to install and get running. However if you didn’t, the added cost of getting some Mac HW to drive the already $2200 machine is something to consider.
Regarding the cost also — the Othermill didn’t come cheap. However consider how many kids have makerbot footstools under a bedroom desk. I know that’s where mine sits and know many folks who also share in my “most expensive ottoman I’ve ever owned” sentiment. I only wish the Othermill had come out sooner as it may just have spared me the 3D printing frustration as I would have been far too busy building boards to half-print distorted blobs of PLA.
Some final considerations when milling
So Othermill aside, there are some realities easily lost in the generation gap between those of us who were part of the first wave of benchtop milling in the late 90s and early 2000’s, and this latest generation and these things are definitely worth mentioning…Consider it the wisdom from the elders of PCB:
- Keep as many connections on the same side as your components in the PCB tools. This will minimize layer changes and save you a ton of manual via work (see below)
- Vias are just two rings on opposite ends of a board, with no conductive plating along the Z axis to join them (mills don’t plate). What’s more, the lack of hole plating makes it all the harder to plug vias on milled PCBs. Consider a strand of wire or a component pin to create surface tension that draws solder into the hole if you get stuck.
- Along those same lines, consider that thru hole parts with tracks connecting to pads on both the top and bottom sides (same pin) are not connected unless you solder that pin on both the top and the bottom. This can be particularly complicated for connectors that may obscure the solderable area on one side or the other.
- Alignment can be an issue when flipping the board. Plan to use a jig or at least a bracket to ensure alignment. Othermill ships with one.
- There’s also no soldermask on milled PCBs and that will become painfully obvious if you’ve only ever used fabricators for boards. Feel free to “spread it around” as you will find that solder can quickly get away from you and start spreading down the length of tracks or polygons. Just go with it. A tip is to dope the area along the track with a bit of solder just to prevent too much run-off.
- Same lines, be exceedingly careful of bridging as you don’t have the same solder resistant surface supporting you. Flux is your friend, be sure and use it.
- Endmills break and the smaller they are, the deeper you cut, the more prone they’ll be. Have multiples of all smaller endmills on hand and cut only as deep as you need to. Remember: just because you can draw it, doesn’t mean you can build it.
- Use a mix of tool sizes including larger tools for clearing large areas. Spare your smaller tools in the process!
- Avoid acute angles in PCB routing and with polygons. This will prevent the mills from routing into areas you expect copper to be removed (the angle will approach a width which may be smaller than your smallest mill).
- Avoid FR4. The substrate material contains fiberglass and the dust is terrible for you. Use FR1. OMC provides only FR1 blanks for purchase.