Hello, Terence here. It’s been a while, so here we have a long overdue update.
First things first. ZT Automations, of which Mike Ziomkowski and I share 50-50% ownership, had taken over the OpenBeam name and day to day operations. The transition actually started last year. This includes this website, our soon to be launched webstore, any of our merchandise on Amazon, etc. I am aware of the outstanding pre-assembled machines owed to our kickstarter backers on the Kossel project. As part of the transition, ZT Automations will be handling the outstanding kickstarter liability of undelivered, preassembled Kossel Pro printers as well. I have dipped into my personal finances and am transferring to ZT Automations money to cover a full refund of each and every outstanding pre-assembled machine ordered on my Kickstarter campaign and we will be in a position to issue refunds starting March 15st, 2017. I'll go into why I made this decision in this update.
Now, the long story.
At the end of February in 2016, I left a very cushy, relatively difficult to get fired from “iron rice bowl” job at Thermo Fisher, and joined a hot Seattle startup called Glowforge. I joined the team as one of the few people building their amazing, awesome hardware.
I ended up leaving Thermo Fisher as I didn't really see myself moving up in that company, and needed a change of pace. I was missing the fast paced product development world of a startup. So, when the opportunity at Glowforge opened up, after some soul searching, I decided that it was probably best for my mental health to move on. It was the best thing for my sanity to switch things up a bit and to go work at a company where one’s individual contribution everyday makes a difference. It also is the next logical step in my career, working on a high volume consumer electronics product.
The thing one has to understand about product design and engineering, is that in a well managed, well equipped and funded team, the camaraderie formed between the members is quite strong. Good, modern product design requires intense collaboration from a multidisciplinary, diverse team (and ultimately, it was crazy how non-diverse Thermo Fisher’s design team was - the 2016 Republican contenders for president was a more diverse group - I was the only non-white engineer, and we had zero females on the team) with each member contributing to the process. Along the way, you have to be dependent on your fellow team members - their ability to deliver gates your progress, and vice versa. And I work with some truly amazing coworkers - no one I’d want to disappoint, for sure. Overnight I went from pretending to do useful work and sitting in unproductive meetings to fill out 40 hours a week, to working probably 60-70 hour weeks, and having the time of my life. For my first few months at Glowforge, it felt like I only really saw my son on the weekends; I was eating all 3 meals at work a few days out of the week, and there were plenty of engineering catch up work to do. (I had originally planned to try to finish the prebuilt machines in the break between jobs, but Thermo Fisher decided to terminate my family’s health insurance coverage on the day I walked out the door. Given that my new Glowforge insurance wouldn’t kick in until the first full calendar month at the company, I set my start date at Glowforge to Feb 29th to minimize the amount of time my family had to go without health insurance).
Now, busy seasons comes and goes, and when I joined Glowforge I knew that I was going to be busy for a while as a long hard slog to get products shipping.. However, a few more major events happened:
1) My wife’s seasonal busy season started, leaving me as the single parent to care for my then <1 year old son.
2) We had the opportunity to buy (upgrade) our house, which we jumped on. We then rented the house back to the original tenants while they prepared for downsizing. Eventually, when they were ready to move, it would coincide with my wife’s busy season.
3) Glowforge, due to its rapid expansion, outgrew the space that it was in. (The neighboring beer brewery also had a drain rupture and flooded our old basement with sewage from their brewing plant, which during the summer made for a very unpleasant work environment). So, on top of moving houses, I had to move offices - and our hardware and engineering lab, as well.
At the time of our move, Zachary was just learning how to walk, which bumped a whole slew of home improvement projects, such as installing tamper resistant outlets, baby gates and general cleanup to the top of the priority pile.
Originally, the plan was to do a nice cleanup, move and sort, into our new home. However, circumstances from 1-4 pretty much meant that it was “sweep everything into a box and sort it out at the new place”. Even more unfortunately, we rented Frogboxes for the move and those had to be dumped out and returned. Doing this severely degraded our ability to conduct any sort of engineering work on OpenBeam / Kossel project for MONTHS.
I might have been able to handle switching to Glowforge, and maybe one of the above items. But all of those hitting at once, while still burning the candle from both ends at a startup? Let’s just say I’ve been very unresponsive to lots of people for a while.
So, where does that leave us?
OpenBeam, as you may recall, raised about $120k from the Kossel kickstarter campaign. While this seems like a princely sum to a lot of people, $120k does not go far in hardware development AT ALL.
At the time we did the Kossel kickstarter, a very well funded group out of Singapore, Bucaneer / Pirate 3D, launched a similar kickstarter. Their machine would have been built on a much simplier and much better understood CoreXY mechanism. Their team received a half million dollars grant from the Singaporean government to front load their R&D and they outraised us by an order of magnitude. Despite this, they failed to complete all deliveries.delivery. Most of their backers lost money - there was no shipment, no product, no refunds.
OpenBeam was able to deliver a vast majority of its rewards because of the existing business in selling aluminum bars. Without the steady influx of income from OpenBeam sales, we would have ran out of money long ago, and there wouldn’t have been a Kossel Pro 3D Printer either. But we’ve not been diligent in keeping up our marketing, our web presence, or our product line refresh, and sales have suffered as a result. We’ve put a tremendous amount of engineering resources into Kossel - resources that really would have taken OpenBeam to a new level if we had spent it wisely on its core products, and sales have suffered as a result. Today, as ZT Automations take over the OpenBeam brand and operations, my first order of business is to revive and strengthen our core product offerings.
Kossel Pro and its related development will still continue, but at a much slower pace as before. (I will share progress at the end of this post and more in depth in the upcoming months). It is our hope that Mike Z will be able to take over most of the business day to day operation and I will handle R&D, product design work and blogging where I am much happier and where my strengths lie.
I’ve finally dug myself out of most of the hole from moving. My computers are now finally back up and running. I just rebuilt the optical inspection fixture for K-heads (having to order new parts to replace what was lost in the move). I can actually find my calipers and thread gauges and inspection tools. As part of the move / upgrade, I’ve moved back all the prebuilt machines and all of Kossel’s R&D back into my garage. It’s been hell trying to find time to go to our shop in Lynnwood, while caring for my son. More than once I started transporting goods, only to have to unload the car to make sure that a stray 3D printer subassembly won’t bash him in the face when he’s in the car seat in the back and I take a corner a little too aggressively.
We’ve been building out our infrastructure for documentation and testing as well. The home office is finally somewhat organized, and we now have dedicated photography area to facilitate technical documentation. (Before, this had to be set up and torn down at various parts of the house that was deemed “not in use”).
Where we are going with the Kossel Pro Platform:
As far as Kossel kits are concerned: We delivered a kit that can be assembled into a printer. People have assembled and printed successfully with it. Of course, it's not perfect, and the documentation could have been better. But as far as commitments for Kickstarter rewards goes, I considered them fulfilled.
However, with a preassembled machine, the user has every right to expect that the machine would work, flawlessly, 100% of the time. And, to be honest, this is where I feel that we fell short. Our machines print, but this is not the user experience I would consider acceptable as a turn key consumer electronics product. (Working at Glowforge had skewed my perception as to what constitutes a *good* user experience. I wouldn't feel right releasing what I have right now to someone who may not have had any experience with a 3D Printer at all)).
Some of the biggest frustration with the Kossel Pro lies in the following areas:
- We engineered our own controller PCA for motion control. While electrically the board performs well, we chose an off-the-beaten-path processor and as a result for some people, the USB communications can get a little bit wonky. On top of that, the code base for Marlin is a freaking train wreck, the moment you stray off the beaten path.
- Knowing what we know about delta robots in general, it is very difficult to achieve perfect parallelism between the end effector and the build plate unless the underlying mechanical subsystems are built to tight tolerances. Our mechanical auto levelling probe, which sits offset to where the end effector is, exacerbates this problem.
- The world had generally moved on, and we engineered a very tightly integrated and pretty closed off system. To do what is right, we need to make it so that 3rd party parts are more easily integrated with the Kossel Pro.
Here are a few things that our existing Kossel Pro users can do to mitigate some of the issues that we’ve identified above.
1) We've been testing our preassembled machines with a Raspberry Pi configured with Octoprint. We noticed that the USB communications issue seems to be hardware dependent, this way we can at least mitigate the hardware risks by controlling both ends of the USB communications.
2,3 ) I’ll write another update on what we are doing to fix some of the parallism issues in our machines, and the design changes that we are hoping to execute in the molds to bring the existing install base of machines closer to open source hardware standards. These changes include designing out some of the tolerance stackup in the mounts, releasing again our machined end effectors to allow for 3rd party parts to be integrated with Kossel easier.
4) Mike had been doing a bang up job keeping stuff in stock on Amazon; and hopefully in a few weeks we can put more inventory for Kossel spare parts back onto Amazon.
5) Matthew Wilson, one of the original team members, have been experimenting with FSRs and running Reprap Firmware on the new Duet controller board, with great results. You can see pictures of his machines. Matthew administers a fleet of 4-5 Kossel Pros (spread across Google's sites here in Seattle) as well as at home.
We’ve had a rough 2016, for sure. I’m not out of the woods either, as we prepare to ramp up at Glowforge for our product launch. Mike and I, however, are in this fight for the long road, and we’ll keep at this.