Assemble the Fleet!
Look at the time, it’s March already, and with another month comes another milestone. This time around, it’s the test rig, arguably one of the coolest. So without further adieu, let’s check it out!
The Assembly Process
We designed Tiko with manufacturability in mind, and not just the components, but the assembly process too. It’s easy to say that something is designed for assembly, but a whole ‘nother thing to experience it. Over the past few weeks we had the opportunity to experience the assembly process first hand as we assembled the test rig.
Cool, How’d it Go?
We found that many of the assembly steps were indeed quite straightforward (even when rushing), just as we had designed them to be. Tiko is mostly axi-symmetric on the inside, but every component still has a correct orientation and we implemented lots of little tabs to keep things aligned. Choices like that worked out great, most of the components can be assembled properly with everyday tools and almost no training.
Nonetheless the process was quite enlightening, and we discovered numerous ways to speed up / combine various steps. We also found steps that were sensitive to assembly error, but more on the those later. All in all, doing this ourselves has laid the foundation for a robust and consistent assembly process.
So, How’s the Test Rig?
See for yourself! :) It's just a sneak peak, but is it cool or what!
It's mind-blowing in person. For now, we'll tease you with that preview, as the big reveal will happen once we start printing with it. :)
Now, cool as it is to see 100 Tikos show off some basic motions, they aren’t yet fully operational. We found some problems in a couple of components / assembly steps, and had to hold off on completing assembly until they’ve been addressed. We’ll be running tests that work around these, but in the meantime let’s talk about our results thus far.
What We Found
Back in July we posted a photo of Tiko’s flexible-PCB based accelerometer. The design worked perfectly, and so we left it alone. Come November, we confirmed they were safely RTMable and placed an order. Unfortunately, during those few months the project had changed hands and the new production team (still the same supplier) made a simple omission. They didn’t apply the hard backing underneath the accelerometer (see below), as it wasn’t explicitly stated in the PO. Simple mistake, nobody’s fault, but the consequences were significant.
The problem is that we tape Tiko’s accelerometer to the Bowden tube, which bends the now completely-flexible accelerometer head. The tight diameter of this curve causes solder joints to regularly break free from the tiny little traces, which ruins the connections and renders the units unusable.
This happened so frequently in fact that we stopped assembly of this component, worked out a plan with our supplier, and are now awaiting a fixed batch. This is why the test rig printers do not yet have Bowden tubes installed – because the accelerometer cannot be added in later.
Unfortunately, this gets in the way of testing auto-levelling, the liquefier (which we’ll start testing later this week on a smaller scale) and a few other functions. However, we have many other tests to run and can work around this issue for now. It's a shame, but hey, at least we caught it before shipping. It’s amazing how such tiny errors can have this big of an impact, and we’ve since incorporated many additional QC (quality control) steps into our assembly process as a result of this discovery.
Back in November we told you about the linear rails/bearings, and that we were perfecting the tolerances on them. The following update, we told you how we had finalized them. Indeed, we did, and it was great… or so we thought.
Those sub-assemblies were assembled by hand, so with a certain degree of care. We knew we’d use an adjustable torque electric screwdriver in production, so we figured that’d give us the level of consistency we needed. Oh boy, were we wrong.
During test-assembly, we found that these assembly screwdrivers were not precise enough, leaving some units too loose (hurting print accuracy) and others too tight (causing binding/jamming). Once we powered up the test rig we found that a fraction of printers skipped steps, while others had loose-feeling kinematic assemblies. We marked the former with yellow cards.
Now, it might sound like we’re in deep trouble, and we would be… if we didn’t have a plan B.
Back in January, we actually started testing a new type of bearing. It started as purely experimental, but during testing numerous benefits became apparent. It proved to be quieter, more accurate, easier to assemble, and most importantly – seemingly un-jammable. So we took a gamble and continued its development. Shortly after Chinese New Year we decided to initiate tooling, even though we weren’t sure it would make it into production due to upcoming deadlines.
However, it’s now clear that it will. Tooling won’t be finished for another couple of weeks, but it could have been a lot longer if we didn’t get that head start! This is a textbook example of how your phenomenal support has allowed us to build a better Tiko. Every time we have a new idea, we can prototype it, just in case. That’s why in situations where others make compromises, we make improvements. :)
More of an inconvenience than a problem, we found that if the infeed tube (the one into which you insert filament) was not correctly inserted into the extruder block, it left a gap that makes it difficult to insert filament. Not impossible, but difficult.
Like the carriages, this is something that really only occurs during mass assembly, so we didn’t see it coming. Luckily the modification required to fix this is relatively minor, and we’re already adjusting the tool accordingly.
OMG Tiko Is Doomed!
Nah, it’s not so bad. Sure, it's inconvenient, but these are just growing pains. We expected this when going from prototype to pre-production, but we didn’t necessarily expect trouble when going from pre-production to mass production.
It’s terrifying to think that we could have said “we’re ready” from a handful of pre-production samples. We can only imagine how many start-ups have made that very mistake, but thanks to the encouragement – and often valuable advice – that you’ve given us throughout this journey, we’ve managed to avoid this trap.
Now we just need to implement these fixes (while searching hard for any other problems) and we’ll be underway. Ahh you can practically smell the PLA in the air. :)
What’s the Lesson Here?
The test rig has already proven to be a valuable tool in our pursuit of perfection, so our only regret is not having built it sooner.
Now, if you’re making a product, here’s a tip we’ve learned first hand. Even if you can’t realistically make 100 pre-production units, just try putting together your prototypes with barely any care. Just throw them together as fast as you can. Seriously, it’s odd, but that’s how it’s going to be done on an assembly line. Your product has to be resilient, even in the face of assembly errors… so don’t baby your prototypes too much. You need to get them ready for the real world. :)
First we have to receive and replace the offending components, then move on to the next phase of testing. In the meantime, we’ll prepare for production.
One of the key objectives of the 100 printer test-assembly was to study the per-printer assembly time and from there, determine the rate at which we could assemble and ship your Tikos in production. During the campaign, we estimated 3000 units per month, but that was then and this is now.
Now that we have the data, we have a clearer picture of how quickly we can assemble Tiko, and as it turns out, we can do a lot better than 3000 units per month. We’re not comfortable specifying an exact monthly capacity just yet, but what we can tell you is that the gap between early backers and late backers is going to be quite short. So if you’re in one of the later tiers, relax and breathe a sigh of relief. :)
Meanwhile, there are certain assembly steps that we know are “safe” (ie soldering of certain components) so over the coming weeks we’ll get a head start on those while we wait for test results to come in, that way part of the assembly process will already be finished and we’ll be able to produce the first few batches even quicker. Woohoo!
All in all, we know we’re close. Our focus at this stage is to iron out any remaining issues so that when we do ship out your printer, we can be proud to put the TIKO name on it!
Besides that, it’s the same old grind for us. Test, improve, repeat until Tiko is up to our standard. And boy, what a standard that is.
In the meantime, we’ll be in the comments section as always, keeping you in the loop and answering your questions, so join us! Also, if you’re still having any issues with the backer survey, just PM us and we’ll get you up and running.