December – Final Palette, Ordering, and Certification

Alright everyone, it’s the update you’ve been waiting for. The word prototype has become a thing of the past in our office, we’re really excited to introduce to you the production ready Palette. We’ve broken it down into the most important parts, and walked through each of them in detail.

Keep reading below to understand what’s going on inside your machine.

Hardware Updates 

I swear our engineers had nightmares refining the design of the The Scroll Wheel to hit our desired specifications. It needs to measure incredibly precise lengths of filament, in a repeatable manner, as well as detect start/stop motion of filament.
It also needed to be easily assembled, and cost-effective for thousands of units. You might remember the in-depth dive we did into the scroll wheel in our September Update, laying out the challenges we were having. Here’s the finalized scroll wheel, the one you’ll shortly be receiving with your Palette:
Top view of scroll wheel (left), outgoing side of scroll wheel (right).
So what’s new? There are four main parts of The Scroll Wheel – The rotary encoder, the casing, the Teflon, and the roller wheels.

First, we changed up the design of the casing. The scroll wheel can now be manufactured out of laser cut acrylic, instead of being machined. We have a production laser cutter in our assembly space, which means we don’t have to worry about lead times from a supplier for anything except the acrylic, something we always have lots of on-hand.

The Scroll Wheel has a very unique way of holding onto the Teflon tubes. The Teflon tubes are notched, and inserted into the scroll wheel. The notches fit into a set of tabs built into the housing that hold it in place while allowing free rotation.

The class of rotary encoder hasn’t changed, although we have put several similar models through rigorous testing to find the most reliable and cost-effective solution. The shafts will rotate about 8000 turns for every Kg of filament you use so not only does it need to be precise, but it needs to be robust.

The most difficult to design section of the scroll wheel is actually the smallest - the roller wheels. After extensive testing we were able to lock down an optimal size and profile of the roller as well as the custom hubs they fit onto. We learned that the material the roller was composed of was just as important as the geometry and that not all rubbery materials performed the same. It turned out that the roller geometry and material combination was not readily available on the market, so working with a specialized manufacturer in the UK, we had them made custom.

Months ago it all began with re-purposed roller wheels from a computer mouse. It then moved to grommets, then to machined rubber, and then cast polyurethane with plenty of steps in-between. We have learned a ton about this system during the process, resulting in a system we are excited to release.
Grooved cross section of roller wheel (left), top view of roller wheel (right).
So that covers the first closed loop feedback system…. But what about the second?

In our November update we covered a new system we implemented called the Buffer Measurement System, we refer to it as “ponging” in the office.

If you remember, the first iteration of this system had some 3D printed parts, and a limit switch. It wasn’t pretty, but it proved the concept. The next step was to take the principles in that system, and mold them into something that is integrated, reliable, and designed for manufacturing.

It took a couple days of work, but we came up with an incredibly simple system that stays inside the casing of The Palette itself. We owe everything in this system to the beauty of magnets.

The Buffer Measurement System needs to periodically home the buffer tube throughout the print. This means we need a way to trigger a sensor, and relay that back to The Palette’s controller.

We embedded a Hall effect sensor into the outgoing drive module, and attached a magnet to the larger buffer tube.  
Buffer in expanded position (left), buffer in “pong” position (right).
When The Palette is coming up to a “pong,” it will stop producing filament. The buffer will then decrease, until the Hall effect sensor reads that the magnet is present. Once this happens, the Hall effect sensor is triggered, and the information is sent back to The Palette’s controller. The Palette compares how much filament has been produced, to how much filament should have produced, and adjusts its upcoming piece lengths accordingly.

By doing so, we’re able to create a feedback loop with our own drive systems, and compensate for any inaccuracies that accumulate over time. Using a Hall effect sensor instead of a traditional switch means there is no contact involved, removing the previous stresses that were breaking splices, the issue we mentioned in our November Update.

That covers our two closed loop systems, but we’ve made a few more changes we’d like to walk you through.

Airflow and cooling have always been very important to system performance. We’re always looking to optimize the cooling systems inside The Palette, as it needs to function for hours on end.

This is one of the reasons we changed from an acrylic skeleton, to the sheet metal skeleton (covered in our September update, here). We’re always learning, and decided to complete one more iteration of the cooling system before we froze the design. Most of the changes are on the casing itself, and to the size/locations of the external vents. You can see the side-by-side comparisons below:
Before and after views of the left side (left) and right side (right) of the casing. The new casings have screen printing and enlarged cutouts.

We also made a few small changes to the internal skeleton to make everything easier to assemble. Our production engineer has been building out the assembly instructions, and helping to work in some subtle final design changes that will go a long way when it comes to the efficiency of assembly.

We just finished digging through our updates, and realize that we have not mentioned the cutting system once. This system is unique to The Palette, it’s really interesting from a design and engineering perspective. We’ve also made some changes for the production version, read below for the full explanation:

When we were working with our early prototype, The Black Box, we were using a guillotine cutter system. This system was a razor blade, screwed into an arm that would move up and down via a stepper motor driven leadscrew. It was essentially “chopping” the filament – once it penetrated the surface, it would go through in one abrupt motion.

As the cutter was closing, the system supporting it would bend elastically and then release all at once as the filament was cut. Not only was it loud but it actually shook the entire system and caused a lot of stress on other systems. It was one of the (obvious) wear points we identified early on, and was something we knew we needed to improve on a year ago.

When we moved on to our pre-production unit we designed a rotating cutter system. This rotating cutter system “slices” filament, rather than chopping it. By rotating the blade, we were able to control the cutting of the filament. This low impact approach significantly reduces the stress on the rest of The Palette’s systems and increased the lifespan of the cutting blade.

With the final iteration of the cutter system we decided to add a second bearing. Before, the acrylic would rub on the aluminum merger, causing a squeaking sound. This was somewhat annoying, and was identified as a wear point. It’s great to see, and not hear, the new cutting system function.

One of the last systems we have not yet discussed in detail is how the process of splicing works. You’ll see it all when you get your Palette, but until then, the splicing section will continue to remain a mystery…..

We completed a lot of optimization around the splicing system over the past month, mainly around improving assembly tolerances, simplifying the machining of the parts, and reducing raw material volumes. We’ll cover this system once the first few batches of Palettes leave our facility.

Production

With the new cutter system, new closed loop feedback system, new casing, new skeleton, and finalized Scroll Wheel, The Palette is just inches away from being ready to move into production.

Now, if you’ve backed a lot of Kickstarter campaigns, this might sound concerning. When you think about production, normally words like “injection molding,” and “tooling,” come to mind. There is an immense amount of time, and risk associated with these processes.

We’re huge fans of keeping it lean – The Palette was designed to be manufactured without the need for injection molded parts and complex tooling. This lets us keep the overall manufacturing setup time, and cost, relatively low.

The Palette was designed to use as many off the shelf components as possible. Those that are custom made, such as the cutter, drives, and hot tool, are either machined, or built from stacked layers of laser cut acrylic. Our structure, and casing, are bent sheet metal, and are being sourced from a top tier manufacturer only an hour drive from our facility. We already have electronics in stock to create 50 units, and are in the process of ordering enough for our next 300 units.

The lead times, and setup time for our manufacturing facility have been minimized as much as possible through strong, forward thinking design and planning. By doing so, we’ve been able to remove a substantial amount of risk, and time, from the entire process.

So what’s going to happen over the next two months before Palettes start shipping in February? Expect us to talk about these items in our next updates. 

Ordering 

We’re ordering 300 Palettes worth of off the shelf components over the next couple weeks. We know these components work well, we’ve been using them for months now. We’ve completed sample orders (around 25-50 units), so we know we’re getting solid stock from our suppliers. We’ve pulled the trigger on substantial orders, they’re currently en-route. 

We thought you might like seeing some of the orders that have already arrived and are ready for QC testing:

Bulk order of cutter bearings (left), bulk order of controller I/O shields (right).

Our custom parts are either fabricated in-house with our laser cutter, or are coming from firms in Ontario (our home province in Canada). We’ve strategically chosen these suppliers due to their proximity to our assembly space, as well as their reputation for quality. Because of their close proximity, we’re able to do a test run for 50 units worth of parts from these suppliers. That way, if there are any issues with this run, they don’t break the bank.

Everything coming from overseas suppliers for the first 50 units should be arriving at our assembly location for the middle of January. Our orders from Ontario will be arriving in the beginning of February. From there, we’re going to be working with our assembly team to get the first batch of units out our door, and on their way to our early bird backers by the end of February.

Assembly 

We mentioned in our last update that we hired an individual with manufacturing experience to take on manufacturing as his vertical within Mosaic. He’s been working on what we call the “Assembly Bible,” since July – an in depth walk through and training manual for our assembly team. Since he’s come on board, he’s been assembling, and disassembling the latest version of The Palette, and has broken it down into a repeatable process.

This is the process that we’re going to train our assembly team on when they come on in the beginning of February.

Quality Control

Quality comes before everything. We have put a tremendous amount of work into developing The Palette, we will do everything in our power to ship the highest quality product possible. We’ve been working to design processes for quality control, a series of tests to put each Palette through before it leaves our door. These steps are our way of catching any mistakes, or quality lapses in the inputs from our suppliers.

We’re completing the assembly of The Palette down the hall from our main office. The team who designed The Palette over the past 18 months will be working to ensure each product that leaves our door has been tested, and exceeds our quality metrics.

Certification 

For the past 6 months we have been working alongside a local product certification company to help us ensure The Palette meets safety and other relevant compliance requirements. You will notice that most consumer products you use will have small markings on them (or on the box they came in) indicating what types of certification they have passed. Some common ones are CE, CCC, UL, and TuV.

Certifying The Palette means we can ensure a smoother experience with international border control and more importantly safe, compliant operation of The Palette in your home, office, or wherever else you will use it. The certification process is elaborate, and complicated but we are excited to finally be approaching one of the first formal tests of The Palette. It will involve blasting The Palette with electromagnetic radiation and monitoring several aspects of its performance. We will be sure to post pictures of the testing and discuss it in our next update.

When it comes to product certification, it is critical that the product you test and certify is the final design as any small change will require a mandatory recertification. For this reason we have not been able to do the testing until now. Although we have already implemented changes to meet certain compliance regulations, there is always a risk that certain aspects of the system will fail the testing. In fact, we have been told that it is common that products will fail testing during the first attempt. The real risk lies in how long it takes to implement a corrective action. For this reason, we are scheduled to put the Palette through a pre-scan test early next week to try and discover any high-risk problems sooner than later. We will be sure to keep you informed and let you know how it goes!

-Mosaic


3D Printing