THP Hacker Bio: Neil Jansen

If we were running a contest to give away a trip to space for building the most innovative open hardware project a few years ago, the winner would inevitably be a 3D printer. Times have changed, 3D printing is reaching the limits of what can be done with simple plastic extrusion, and there are new hardware challenges to be conquered. One of the challenges facing hardware designers is the ability to create and assemble electronic circuits quickly. For that, there are a few pick and place machines being developed, the lowest cost being the FirePick Delta. It sells itself as a $300 pick and place machine borrowing heavily from the RepRap project, enabling tinkerers and engineers to assemble PCBs quickly.

[Neil Jansen] is the project lead for the FirePick Delta, and along with team members ranging from software developers in the bay area, to electronics technicians and high school students, they’ve created what will become the lowest cost and most capable pick and place machine available. Already the machine has tape feeders, tray feeders, a vision system, and modules to dispense solder paste. It’s an astonishing accomplishment, and were it not for some damage in shipping, we would have a video of [Neil] demoing the FirePick at Maker Faire NY.

In lieu of that, we do have a bio on [Neil] and what challenges he’s faced in building the FirePick. You can read that below, or check out their second demo video for The Hackaday Prize:

https://www.youtube.com/watch?v=46pLeLBYVTs

Robotics, and Extreme Circuit Boarding.  The world does not yet recognize Extreme Circuit Boarding as a sport.  But it basically consists of consuming large amounts of Red Bull, listening to loud Dubstep and Electro music, and designing crazy circuits, and then building them as quickly as possible when the boards and parts arrive. It’s extreme because you have to get the designs right on the first rev, with no rework.  Kind of in the spirit of a hackathon or a Tattoo Inkmaster reality show.  Since no one knows about this sport yet, I’m the unofficial world champion for 2014.

Aerospace, GPS / GNSS receivers, radios, autopilots, transponders, and collision avoidance systems, blah blah blah…  My job is the hardware/software testing and verification of these boxes.  These days I mostly write software, but at my last job, on a typical day, I’d be doing CAD in SolidWorks, schematic capture and pcb layout, writing embedded code, wiring test jigs and harnesses, requirements capture, and lots of other stuff.  I only have a high school background, everything else is self-taught.  Even though it’s a big stuffy company, I still get to do lots of different things, which keeps me happy.  Our other team members have day jobs as well. Karl Lew works as a full stack software developer in Bay Area, California. Christian Lerche is an electronics technician in Denmark and works in a lab that works on locomotive electronics. Dayton Pid is our youngest member at 16 and is in high school.  And Thomas Kilbride is in college at Purdue University.

My passion is basically the pursuit of reality through objectivism, critical thinking, and logic, and the propagation of those ideas leading by example, living by principles, appreciating the beauty of life, and taking personal accountability for my actions.

That honor almost went to my 2009 model MacBook Unibody recently.  I have been using it as my main development platform for our Hackaday project, mostly while dual-booted into Windows 7. It’s extremely slow, sometimes requiring a few minutes to switch between different apps.  It was maxed out at 4GB of RAM, and of course, every time I needed to dual-boot back to OSX to do photo editing or whatever, that was time lost.  The battery has been dead for almost two years, so it’s pretty much anchored to the desk.  I was almost about to take it out back and destroy it office-space style, while playing the obligatory “Still” from the Geto Boys from iTunes until it wouldn’t play any more.  But instead, I hobbled together a new PC out of pieces I had lying around, and a new motherboard, CPU, and 16 GB of RAM.  My productivity towards the HaD project since has probably tripled since I’ve gotten that taken care of.

I’m OS agnostic for the most part.  On a typical day, I’m using Windows, OS X, BSD, Linux (both desktop and embedded), and occasionally esoteric RTOS platforms.  I’m forced to run Windows for CAD, but it’s my least favorite to program on, and I agree with it the least philosophically. I use my MacBook for video editing, audio editing, photoshopping, and other graphic or A/V type stuff. Linux is awesome for embedded and server applications, but I can’t run Cubify Design on it, and I feel that the A/V apps on OS X are more matured and polished than anything available on Linux.

Linux would be my absolute favorite, but big corporations that make important pieces of software still won’t write programs for it, for some reason. In my dream universe, Microsoft would do like Apple did, and ditch their underlying operating system and .NET crap, and go with something POSIX (say, BSD) for their backend, thereby making it easier to write multi-platform software applications. They could even keep their crappy desktop look the same. But we all know they’d never, ever do that.

Hakko FX-888 soldering iron.  I do everything from SMT to soldering 4-AWG wire, all with the same tip, out of laziness, and because I can. I couldn’t even imagine having to use a crappy soldering iron.  Yet, I can’t say it makes me a snob, I’d have to have a $500 Metcal to say that.

MOS Technology 6581/8580 SID6581. I have a small collection hoarded in my closet somewhere.
I tend to gravitate towards the C-like languages, because I can pick one that suits any level of application I need.  I routinely code in C/C++ for embedded stuff, but only really use the basic object-oriented things like classes and inheritance.
I’m starting to really like Javascript, it’s so ubiquitous and useful for high-level stuff, even if it’s a decidedly evil language.  And I actually really like the concepts and paradigm of Java, even if it never seemed to work well in practice.  I don’t really like Python even though I use it occasionally (the tabs and general treatment of whitespace bugs me).  I also have an aversion to Perl, Lisp, Ruby, Haskell, Clojure, etc.  Not because they’re bad. But because I don’t have the time to understand them, and they make the open source software ecosystem a fragmented mess.

  •  Create an open-source micro-factory, capable of creating amazing things locally and sustainably.. Ideally this is what Karl and I hope FirePick will evolve into.  We see this as the logical progression of 3D fuse-filament printing, 3D metal printing, laser cutting, and SMT component assembly.  If we can make it profitable for companies to at least make prototypes and small runs in America (or pretty much any other country for that matter), rather than everything coming from China, then we can say that we’ve done something truly empowering.
  • 2000px-HAL9000.svg“Hard” Artificial Intelligence (as opposed to the status quo “soft” AI).   When I was a kid, I dreamed of pioneering in this field.  Google claims to be working on this sort of stuff.  These days, I’d be happy to be a cog in the wheel of some grand decades-long project to accomplish true AI.
  •  I have to at some point finish a few pointless projects that I put off to work on this project… If I get the pick and place working well, then these will be a slam dunk.  I want to finish my Nixie Tube wrist watch, that uses some Burroughs B-4998 tubes, aka, the smallest Nixie tubes in the world.  Second, I want to make a really tiny robot (40mm x 40mm) with two wheels and a camera, that can sort a pile of M&Ms by color.  Hopefully with a powerful ARM processor and lots of sensors, running SLAM technique, kalman filtering, etc.  Third is a telepresence robot with two robotic arms and gimbal stereo vision on a segway syle inverted pendulum, controlled with a LEAP motion controller and an Oculus Rift.  Then I could just mail my artificial self around the world and see cool places, and get to interact with stuff, while staying on schedule.

I would say that almost all DIY pick and place machines were created by someone who had to manually place lots of SMT components onto circuit boards, all by hand, probably with tweezers and a microscope if they were lucky.  Most people find the process of doing it by hand very tedious.  Many of them, myself included, have decided to engineer their way out of the problem by building some sort of homebrew machine to make the process faster and less error-prone.  I’ve seen countless DIY pick and place machines, both manual and automatic, on the front page of Hackaday over the years.  They all shared a common idea and problem, and most of them shared a similar fate.  The DIY machines were never completed, because the people who built them were smart and busy with whatever project they needed the placement machine for. The problem was that the their DIY machine was just a means to an end, it was never an end goal for them. That’s where our machine differs.  FirePick Delta is our only project, that we plan on working on for many years to come.  We plan on refining it, and adding new SMT component feeders and other modular tools.  And eventually, if everything works out, we’ll probably make the world’s first open-source SMT assembly line, with reflow ovens, conveyor belts, and stencil printers, all released under open source HW and SW licenses.
I’ve actually had the idea to use a delta mechanism for pick and place applications since some time around 2003, but didn’t have the manufacturing resources to build it.  Around January 2014, a friend from a DIY forum asked me if I wanted to borrow and put together a kit for a RepRap Huxley 3D printer.  I built it in a weekend, and quickly built five Prusa Mendel i2’s on it. One of those Prusa i2’s printed parts for my FirePick Delta prototype, which will hopefully print other FirePick Deltas one day.  It’s incredibly cool when you think that the little eMaker Huxley was actually printed from other machines, that were printed from machines, that trace their way back to the first Darwin 3D printer, invented by Adrian Bowyer… which was itself printed from a Stratasys in Bath University, UK.  So the hundreds and/or thousands of machines that I’ll be building over the next few years can all trace their heritage back to the original Darwin (the eMaker came with a certificate and everything, so I’ll likely have certificates for mine as well).  The ideas of robots making robots intrigues me.  They sort of have a pedigree, which is really neat.
Anyway, I had just assembled the first prototype of the FirePick Delta (back then codenamed Project Bismuth), about a day or two before the Hackaday Prize was announced.  I was probably one of the first people to create a project after the contest was announced, because i remember frantically typing it out that night.  The timing couldn’t have been better.  We had always intended to keep it open source, but entering it for the Hackaday Prize sure has forced us to do the dirty documentation work that no one ever does.  Kudos for having a requirement to do a system-level diagram.  That’s one thing that I’ve rarely seen from an open source hardware project, but is sorely needed.
I could use some advice on where to find a good web app developer with node.js, Bootstrap, AngularJS, and Express skills, that would work for free and has lots of time to devote :)  We had originally intended to have our html5 gui for the pick and place machine completed by now, but we didn’t take into account was how hard web app development really is, and how few people out there really have a mastery of it.  We’re stuck with a Java Swing UI instead, for the PnP functionality :(

I was really intrigued by the metal 3D printer that was one of the first projects to be tagged with TheHackadayPrize.  I think it would have required a crazy amount of metallurgy skills to pull off successfully, and I’m not sure they had said skills, but the concept was still really cool.

We actually have an incredibly complicated project, compared to say, OpenMV (which we love by the way).  We’ve had problems with our delta arm linkages, backlash / accuracy problems, Hierarchical BOM generation, and feeder problems.  However, we’ve got some really innovative ways to mitigate them that will hopefully be documented on our site by the Sept 29th milestone.  We’d be a month further along if we didn’t have hundreds/thousands of BOM items to document, so, uh, thanks for making us do all of that :)

Thanks to the FirePick Delta core team, and all the people who skulled our project and believe in our idea.  You guys rock.

13 thoughts on “THP Hacker Bio: Neil Jansen

  1. Awesome product. I love it. At what point does it make sense to run a batch of parts? In other words, there seems to be a lot of time involved in setting up all of the pick and place parameters, calibrations, ordering tape and reel (large number of ICs per tape) and so on. Is it 30 boards? vs. having it done by a professional outfit. Of course, they also have a set up charge.
    keep up the good work. this is excellent.

    1. We’re going to make it easy enough that even loading in parts for a single board will be well worth it. Soldering small TQFPs, DFNs, and QFNs by hand isn’t a skill everyone has. I can do it well, but don’t really enjoy it. This machine will place those parts with ease, along with BGAs and other stuff. Loading 20-100 boards would certainly be better to do it in this machine than by hand. We’re running into exactly that problem as we try to assemble prototype boards for our beta testers and developers. Really wish the machine was done enough to run the boards, but sadly, it’s not, which is going to create some real headaches.

  2. I like the idea so far. I realize you’re trying to make it inexpensive, but it seems like the 3D printed plastic T-nuts with metal nut inserts is going to compromise the rigidity of the machine vs. full metal T-nuts. It would also seem the machine time & fuss of 3D printing all those T-nut parts and then installing the washer & nut, the cost of a 100pk of metal T-nuts starts to be a wash. Misumi wants $28/100pk, which is expensive, but it seems the total ledger for the alternative is ignored. There may well be alternative sources for full metal T-nuts as well.

    1. The idea is that you can use either…We’ll have several top-level assemblies for cheap DIY and commercial / industrial versions. We actually prefer the printed ones a lot, they don’t slide around as much and they’re much cheaper… But we’re certainly not making anyone use them.

      1. Understood, thanks for answering.

        I’d like to start building one of these in collaboration with some people at the local maker space. Is the frame & mechanism pretty well nailed down for this version? Your updates read like it’s mostly the software that needs work.

        1. We’re still working on feeders, and a new version of the carbon fiber linkages, end effector, and delta arms. Software is very close but not in a completely usable state yet. We’re still chasing down bugs in our FUSE layer that’s making in difficult for us to get images from the camera.. and we’ve got a lot of modules to implement inside of OpenPnP for feeders, vision, motion, etc. Lots of other stuff too, I’d better get back to work :)

  3. So Brian Benchoff screwed my bio up and hasn’t fixed it, after emailing him. The first paragraph about circuit boards should have been under the “What is Your Hobby” section title. The second part about aviation should have been under “What is your Profession” section title. And the next paragraph should have been under the “What is your Passion” section title.

Leave a Reply to Luís OuteiroCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.