In our last update, we announced the start of shipping. It's been a while, and things have been a bit quiet, so let's dive straight in.
“No battle plan survives first contact with the enemy” – General Helmuth von Moltke the Elder
After all of the tests we ran on the TikoWall this spring, we felt certain we had a reliable printer. We had thoroughly tested each system and component for a variety of use conditions, user inputs, and outright longevity. We were confident that we had a winner, and so we initiated production. We told the world that Tiko was ready. But little did we know what could still go wrong.
We shipped two batches of 100 printers, ran into issues, and in total went two months without a formal update. It’s been a trip, so let’s get you up to speed.
How It Started
The logical place to start is by asking “Why did we ship those first 100 Tikos?” To answer that, let’s first look at the situation we were in. We had just completed the testing of the TikoWall, which had flushed out a plethora of bugs and was working quite well. The components were finalized, manufactured, and ready for assembly. The founders were in China and set up a Tiko-run operation including staff and facility, and built 100 printers. As part of QC, every one of them ran a successful print before packaging and shipping. What could go wrong, right?
Well, we’re no dummies. We’ve learned first hand just how many things could go wrong. We would have rather shipped those first 100 units to HQ, tested them, and then continued production.
So, Why Didn’t We?
Well, around the same time, a vocal minority began to suggest that Tiko was vaporware. Words like fraud and lies were starting to appear. Some even went so far as to claim the entire TikoWall was a computer rendering. Sure, these people were fools and represented a tiny minority – most backers believed in us, cheered us on, and told us to take our time and get it right – but another thing we’ve learned this past year is that one rotten apple can spoil the barrel, jeopardizing the whole project for everyone.
So, having 100 printers in hand, all of which could successfully print, we had a tough choice to make: Ship the first 100 printers to HQ and test them before shipping to backers (a process that could easily take two months) and risk the allegations getting out of hand -OR- shoot these rumors down once and for all, and ship the first-run printers at the risk of an unknown flaw publicly popping up. We made our decision. Here’s what happened next:
The First One Hundred
Being a first-run batch, we expected a minor bug or two, but nothing like what actually happened. Here’s what we found:
As you can see, a quarter of the printers had a problem – not a good way to start! Luckily, many were easy to address and were solved quickly after the first batch shipped. Here they are in more detail:
Damaged in Transit (14%) – Many printers arrived with broken filament trays and/or completely separated print chambers. Some also had gear racks coming off the rails. All of these had one common cause: poor quality adhesive. The adhesive we used was similar, but not identical to the one we used in drop testing. As it turns out, it was not equivalent.
Solution: We imported and used the adhesive used during drop testing at HQ.
Liquefier Jamming – More on that later.
Power Jack Solder (2%) – Due to a number of design considerations, the only feasible way to power the electronics is through a wire soldered to a power jack located at the bottom-rear of the chassis. Unfortunately, this relies heavily on the skill of the person soldering the wire to the jack. It turned out the connection for the first 100 was inadequate, and the solder broke on a number of printers.
Solution: We lengthened the power wire, and reinforced the connection to make it more robust.
Bowden Tube Flare Popping (2%) – In Update #18 we mentioned this was an issue that was solved, and it generally was, but the flare was still by far the weakest link, holding back the rest of the system.
Solution: We redesigned the twist-lock fitting and extruder block to instead use the highly-reliable retaining ring system commonly found in push-fittings and already in use in Tiko’s liquefier heatsink.
Punctured Bowden Tube (1%) – The material used to make bowden tubes (PTFE) is relatively soft, whereas PLA and other common printing materials can be quite hard. Sometimes, when cutting/breaking filament, the remaining tip can be quite sharp. If this sharp tip is inserted into the extruder, it can dig into (and even puncture) the Bowden tube.
Solution: We stiffened the spring at the twist lock to straighten the path of the filament in the tube. We also added a warning in the WIT to remind users to check their filament is not sharp, as punctures are still possible if sharp-tipped filament is inserted.
The Second Hundred
After receiving feedback from the first 100, we moved quickly and implemented solutions to the issues that came up. In less than two weeks, we already had a second batch that we were more confident in, and felt could be a precursor to continuing production. Here’s how it went.
A failure rate of 14% is still quite high, but it was impressive just how much of a difference a couple of weeks could make. This time around there were just a couple of bugs remaining.
Liquefier Jamming (8%) – More on this later.
Damaged in Transit (6%) – After finding out just how bad the first adhesive was, we tried the already tested solution – an expanding vinyl-based glue. Unfortunately, due to the higher humidity of our factory compared to the office, the glue expanded more aggressively than expected and imparted stresses on the print chamber. During shipping, these latent stresses combined with the forces of shipping and resulted in cracks along the top of the print chamber.
Solution: We have since experimented with dozens of adhesives (local and international) and selected ones that strike a perfect balance of strength, ease of application, shelf life, and aesthetic appeal.
And that’s it. Those are the only hardware issues we had left in the second batch. That said, both batches did experience a variety of minor issues, which for completeness sake we’ll list out below.
We also encountered a number of minor issues. Some of these were simple software/firmware bugs, others were inconveniences in hardware, and some were one-off cases. To keep this update from getting too long, here’s what they were in a condensed format:
Slicer – The print settings were nowhere near optimized at the time of launch. It was a cause of inflated embarrassment, but luckily, it's easy to solve. In-house testing has already demonstrated dramatic improvements, and further development will continue well into the future.
Leveling – The auto-leveling system isn’t as bulletproof as we hoped. It worked well on the TikoWall, but not in production. We’re tackling it from a number of angles and think we can get it into the 90%+ reliability range, but for now, manual leveling is still the best choice. Luckily, the WIT makes it quite easy.
Carriage-Rail Separation – The intentionally removable motor carriage subassembly sometimes became dislodged during shipping. Supports have been added to the filament tray to prevent this.
Nozzle Came Undone – A nozzle was inadequately tightened and came loose during a print. The assembly process has since been improved.
Difficulty Loading Filament – Sometimes, it can take a bit of force to load filament. A complete fix is not feasible at this time, but conditioning the filament tip helps. It may take some force, but every Tiko can load filament, as this is part of the QC test.
Fit-and-finish – There were small gaps between the chassis and its adjacent parts due to an imperfect cut. Fixed by designing a new saw for the chassis.
The Mystery Problem
With all of these bugs solved over a month ago, it would seem we were ready to get back into the swing of things. Unfortunately, it wasn’t that simple. Despite the extensive rounds of TikoWall testing, we heard stories of backers experiencing liquefier jams. So we shipped some production liquefiers to HQ for testing, and indeed, many of them were jamming. Often, it was erratic and unpredictable, and we couldn't pinpoint the cause. Printers would often alternate between jamming and not jamming. So, we spent another month working on it.
We could write an entire update about this topic alone, but here are some of the symptoms we thought could be behind the jamming:
Heater calibration inadequate – Theoretically results in nozzle operating too hot/cold. Fixed by developing a revamped calibration method.
Heater partially slipping off nozzle – Theoretically results in incorrect heating and therefore excess back-pressure. Fixed with a snap-ring.
Tight retaining rings – Theoretically the retaining rings became too tight (due to supplier change) and were constricting the tube, making it difficult for filament to pass through. Fixed by redesigning retaining rings.
Bowden tube damaged during assembly – Theoretically deformed the tube in some printers, constricting it and making it difficult for filament to pass through. Fixed by improving assembly jig and process.
Extruder motor current too low – Theoretically motor power was insufficient to consistently extrude. It was lowered intentionally to protect the tube flare, but new design can handle more force. Fixed by raising power in firmware.
Bowden tube kinked at heatsink – Theoretically a kink forming at the heatsink inlet was increasing resistance to filament travel through Bowden tube. Fixed by adding spring just like the extruder-side of the tube.
Aggressive filament drive gear – Theoretically the filament drive gear was biting too aggressively into filament, deforming the filament and making it difficult to push through the Bowden tube. Fixed by switching finer-toothed drive gear.
With all of these solutions prototyped, we ran two distinct sets of tests last week and over the weekend. One set consisted of printers in China with the new-and-improved liquefiers, running current-production firmware/software. The other set at HQ consisted of printers with current-production liquefiers, but focused on testing software/firmware based solutions.
What Did the Tests Reveal?
After hundreds of print hours of testing in China, we found that all of the little hardware improvements added up to a greatly improved liquefier - one that is finally consistent.
We found that either no printers jammed, or when pushed hard they all jammed at the same place in the same print. So, while they’re not jam-proof, they are consistent and have worked well under normal print conditions.
Meanwhile in Canada, we found that certain printing/extrusion settings practically eliminated jamming on the buggy production liquefiers, and next week we'll begin testing these optimized settings on the improved liquefiers. One can't help but expect good things here.
Overall these results mean we can tackle this issue from two fronts. Both hardware and software have shown promise, and so we are cautiously optimistic that we have this final problem cornered.
It looks as though we have the liquefier issue cornered. With all of the other bugs completely resolved, Tiko’s hardware appears ready for full production. This time with an improved assembly process, more robust QC system, and general knowledge of issues to look out for. It should be a much smoother ride.
However, there is still the very real possibility that the software development will not yield results, or the performance envelope will turn out to be too narrow, or some new long term problem will come up. If that happens, we could be in for a longer wait. We’ll continue hardware testing and development throughout this period, just to hedge our bets.
Expect to hear from us in around a month with news – either of large-scale shipping or, gulp, trouble.
Sorry for not updating you sooner. Honest. We shipped the second mini-batch believing we were ready to get back into production. We had an update already written explaining the setback, ready to post. But then the liquefier happened and we waited for concrete test results. One week turned into a second, which turned into a third, and so forth. We're stubborn about solving problems. Sometimes our strengths can become our weaknesses.
It's truly amazing and humbling just how many problems we could experience moving from pre-pro to production, but this gives us a greater-than-ever appreciation for how companies can spend hundreds of millions of dollars on R&D. It's just so full of surprises. But it doesn't take money so much as patience.
On that note, we deeply appreciate the kindness and patience many of you have shown. It's easy to be a critic (and look what road that led us down) but it takes class to see our stumbles and still root for us. And we notice it. We made a commitment to you to change the status quo of accessible 3D printing, just as you did to backing us. Each and every morning we get up and work to honor that commitment. It doesn't matter if you fall. As long as you get back up and keep going.
Lastly, we want to thank the first 200 Tiko owners for their grace and feedback. You guys have made invaluable contributions to the campaign and improved Tiko for all future owners. Who knows where we would be without you. Thank you again, and although we've hinted at it before, we'll come out and say it: We can't wait to ship you all another Tiko - but once it's ready this time. :)
Cheers guys, talk soon!
P.S. We love to get involved in the comments section and share the inside scoop on progress with our most active backers. If you ever wish to dive into the details and latest developments, but don't enjoy scrolling through comments, then check out our comments directly at https://www.kickstarter.com/profile/tiko3d/comments