The Palette - Progressing to Production

The Palette - Progressing to Production

We’ve been told that one of our most interesting blog posts is Hardware Heaven, where we talked through our different prototype iterations. In the same tone, we wanted to take a second to do a bit of a deeper dive into our recent tech development.

A quick throwback to our old Black Box Prototype featured in Hardware Heaven

 

So let's dive right in…..

We’re in the process of going from our pre-production unit, to our production ready Palette. This evolution has a lot to do with reliability, ease of assembly, and putting the final touches on everything to go from a unit in our lab to thousands of products that can be shipped to countries all over the world.

We’ve made quite a lot of modifications over the past system's design. We’ve implemented our new casing, new machined parts, new internal structure, new I/Os and have gone through a full scroll wheel iteration. We also added in a few nice to haves, as we think they’re worth the investment.

The goals of this process are very important to understand. First, we want to eliminate as many inputs from our BOM [bill of materials] as we can. Reducing the number of inputs will make our inventory management, ordering, and assembly operation easier. Having 46 different types of screws means we need to pick, and lay out all of these different screws every time we want to assemble a Palette. Using just one type would make the entire process a lot simpler, more efficient, and cause fewer QC issues, as you couldn’t mess up what type of screw goes into what section.

For us, its important to try and use as many off the shelf components as possible. Everything we add to our BOM that is custom increases costs substantially, and makes ordering more difficult; therefore, the replacement of any custom parts with off the shelf components is extremely valuable all around.

Oh, one more thing. All of this needed to be accomplished without suffering any loss of quality, as quality is the most important aspect of our product. Shipping a sub-par product is not something that helps build a sustainable company. Quality comes before all.

So now you have a bit of information as to what our goals were, and what our main constraint was (high quality), we’d like to take you through some of the actual design changes we went through.

We made the decision to increase the size of two of the 7 stepper motors used inside the Palette. To accommodate these changes, the casing needed to become a bit taller. One of the motors we chose to upgrade controls the cutting system. This extra torque will be useful in future proofing the Palette to work with new materials that might be harder to cut. Now all seven steppers inside The Palette are the same type – helping us simplify our BOM.

Pre-production (left) vs. new Palette (right)

Something to recognize is that not all design is meant to be practical, products should be a blend of beautiful aesthetics to compliment their functional purpose.

With this in mind, we thought The Palette could do with a bit of a facelift, because who doesn’t like sleek looking lines and vents. So we decided to update the external aesthetics, which you can see above.

Venting:

The modifications we're about to introduce were made mainly for long term reliability, both in the sense of operating long prints, and running The Palette on and off for years.

You’ll notice the updated lines on the side of the new Palette (right) that separate the top and bottom portions of the casing, as well as the new venting. Not only do these improvements clean up the visual design, they also close off the fan systems inside The Palette to remove the chance of broken blades from external interference (such as fingers or filament, two of the things we know from experience can get in the way).

We made additional changes to the vent layouts to optimize the airflow on the inside of The Palette. After monitoring the electronics area of The Palette we learned that it can get considerably warm. Because of this we decided to add additional venting. This is the new vent on the bottom right, pictured below.

We made the move to a sheet metal internal structure (covered in detail later in this post). The new sheet metal internals make the entire structure inside the Palette act as a heat sink. This update lets heat diffuse much more efficiently, and (best of all) works passively without any additional modifications.

By making changes to the passive and active heating systems we’re able to diffuse heat more efficiently, which will lead to less strain on the system, specifically the electronics, as time passes.

Side by side of the pre-production and new version of The Palette

We’re in the final steps of putting together this new version, we’re waiting on our production laser cutter to arrive (should happen early next week). Once this is in, we’ll cut all of our new drive modules, and start to test the new system in its entirety. We may need to make one or two small revisions to better accommodate airflow, but the casing you see above is very close to what we will deliver to our customers.

I/O Updates:

We updated both the power and Scroll Wheel connectors to make the plug-in points more robust, and ready for mass production. Our pre-production version had the scroll wheel plugged into the board via a few exposed wires (seen in the picture below). Although this method did make for quick and easy detachment, and even though we never had any reliability issues, it was susceptible to damage due to strain on the wires and connection point.

Scroll Wheel connection point on the pre-production Palette. Note the exposed wires.

To make this connection more reliable we decided to add an external port for The Scroll Wheel. This new connector will let you detach your scroll wheel for storage and travel much easier, and is a more heavy duty design. Check out the new approach below:

New mounting system for the scroll wheel system (top-right).

When you’re working to get a product functionally ready, you focus on the areas that matter, and sometimes deal with things that are a pain, because you (and the rest of the team) are the only ones using the prototype. The next section is about one of these pain points that we’ve wanted to change for the past couple months.

This design update is also a bit less obvious, at least to anyone lacking the eyes of a hawk. The screen is now mounted to the sheet metal frame on the inside of The Palette, you can see the new mounting location below:

New Screen Mounting Location

The screen was previously mounted directly to the casing from the inside, meaning the screen would need to be disconnected to take off the top section of the case.
We made the decision to move the screen mount because it was something that would often frustrate our team. The frame-mounting approach will allow for easier maintenance by the user, easier troubleshooting, and quicker assembly. The internal sheet metal structure is a more robust mounting point than the casing, and keeps bolts from coming out of the front of The Palette.

Pre-production version of the screen mount

Now we’d like to tour you through the internal updates we’ve made….

In our Kickstarter campaign we mentioned that one of our final steps was to optimize the materials inside The Palette. The pre-production Palette has its internal structure made of PMMA (acrylic) that was laser cut, and jigged together. This worked well, but “well” has never been good enough for our team. We received the first iteration of our sheet metal internals about a month ago. They were a great test, and it was really helpful to have a version in our hands, but that design wasn’t ready to be built into The Palette.

First internal sheet metal design – this design wasn’t ready for installation.

Working with our favourite metal shop, we made modifications to the design, and added a few new features.

WE GOT THE NEW VERSION LAST WEEK. It works beautifully. (If you can’t tell, we’re really excited about this, as its something we’ve been trying to get right for awhile).

Disassembled sheet metal internal structure. Note the new bent metal on the corners, as well as the three bridging points between the two sides, instead of one.

So why do sheet metal internals matter?

This change improves reliability, decreases the number of items on our BOM, and makes everything easier to assemble.

Think of the sheet metal internals like The Palette’s skeleton. The sheet metal internals are the bones of The Palette, they are what everything is mounted to, what lets The Palette hold its form. By replacing PMMA with sheet metal we’re able to make the entire system stronger/more rigid, easier to access/maintain, reduce the part count, and make it quicker to assemble.

Semi-Assembled internal structure shown off by our professional hand-model.

The pre-production Palette relied on the external casing to give it extra rigidity – the new version doesn’t use the casing as a crutch.

And like we mentioned before, this entire structure now works as a heat sink, which will help reduce the requirements of our active ventilation system.

Oh yeah – one more thing. As much as we wish they were perfect, we know from years of experience with printers that sometimes the electronics need maintenance.

Because of the relatively compact design of The Palette, we needed to put a lot of thought into where the electronics went. But we still wanted to give the user easy access if a connection were to come loose in shipping, or after heavy use.

So, in order to make maintaining, checking, and double-checking as easy as possible we made the electronics tray removable. Check out this GIF of how it works:

We know we’re in the stage of nice to haves, but a product needs to have all of the small points considered, and built in a way to give the user the best experience. Even when things (inevitably) go wrong.

The Scroll Wheel

The Scroll Wheel is what lets The Palette stay in the loop with the user’s printer. The Scroll Wheel is what lets The Palette measure the actual usage of filament by the printer, and adjust each length accordingly.

It works with our calibration, and correction algorithms to ensure that the calibration is always on, layer after layer. The Scroll Wheel is, as far as we know, something new to the industry of desktop 3D printers. It was designed to overcome the initial challenges we had with calibrating different prints, different printers, and different printer settings.

We’ve been investigating the best ways to construct The Scroll Wheel reliably, as any variance in how it’s put together will have a huge impact on its accuracy.

Now we’re going to walk through the different Scroll Wheel iterations we’ve been through recently, and some of the lessons we’ve learned from each version.

Pre-production scroll wheel

If you’ve seen us run a demo, this is The Scroll Wheel we were using. It is built with a 3D printed casing, a rotary encoder, and a rubber wheel from a computer mouse (among a few other things). This version of the scroll wheel had issues when first feeding filament through, as the filament would sometimes get jammed on the outgoing Teflon tube. It was also printed, which is not the most scalable manufacturing method. It doesn’t take an engineer to figure this out, but we knew we could make it more accurate if we didn’t use rubber from a computer mouse….

Cannibalizing computer mice isn’t the most scalable manufacturing method. It would’ve also added a few parts to our BOM….. so we took our knowledge of The Scroll Wheel and moved on to our next iteration, pictured below.

First go at a production ready version

This version of The Scroll Wheel is made out of laser cut acrylic. It represents a big leap over our pre-production version, but still had a few issues we want to solve before the design is deemed ready for production. This version of The Scroll Wheel has two machined rubber rollers that contact the filament. The groove that was machined into them was too wide, which allowed the filament to move up and down as it got pulled through. This movement meant that we didn’t achieve as high an accuracy level as we were striving for –it didn’t hit our level of quality for the final product.

Our machine shop had a number of failed attempts at machining the soft rubber wheels. These failures were a huge red-flag, so before we even got this version in our lab we knew we needed to find another method for creating the roller wheels.

There are a few really good takeaways from this version – first, we validated a new method for mounting Teflon tubing to the structure. Second, we learned that we needed to make the filament groove deeper to allow for better constraint. Third, assembling this many pieces of laser cut acrylic takes a lot of time, and is not repeatable enough. Fourth – we learned that thicker Teflon was the best way to go, as it holds its shape much more, and experiences less wear. Finally, the roller wheels are too large, we could further increase the precision of The Scroll Wheel by reducing their size.

So we took these lessons and applied them to our latest version of The Scroll Wheel, pictured below:

Sizing up the fit on our latest iteration with a 3D printed version

So what’s new with this version?

First, it was designed to be CNC machined, instead of crafted from laser cut acrylic. We’re not sold that CNC machining is the most efficient way to manufacture the casing, but it’s something we feel the need to investigate.

Our team has started testing out a new method for producing the roller wheels – casting. We’ve seen some really promising results with this method, as it produces high quality parts, and is scalable, repeatable, and cost effective.

This new iteration is on its way to being constructed, we’re really excited to see the results.

Ensuring the Palette and your printer stay in perfect calibration over long prints is critical. In the perfect world this would be an easy problem to solve but in the real world we know that printers mess up and don’t always print consistently. In order to design and optimize our calibration algorithms, we have had to do a lot of extensive testing on different printer models, filament types, and even on the accuracy and repeatability of The Palette’s filament production. Below you can see some test print blocks where you can tell they vary in size, filament type, and complexity.

We wanted to share a picture of our always-growing family of test prints

This data has proven to be extremely valuable as we have used it to design better solutions to some ongoing calibration challenges.

There’s a few more things we wish we could share with everyone here – mainly our new machined internals. We’ve done some really cool things with the internal layout, new machined parts, and wire management – but we don’t want to spoil the surprise.

So what’s next for us?

The detailed list would take forever to write on paper, our to-do lists are longer than ever. But here are some highlights:

  • Testing, testing, testing
  • Finalize the scroll wheel design and manufacturing method
  • Order off the shelf components for a run of 50 Palettes
  • Software features, continue developing our stand-alone application
  • Certification and testing
  • Firmware updates
  • Production optimization

We’re in the midst of outfitting our production space in Toronto – check out the before version.

New Space awaiting setup

If you ever have any questions, or want anymore information drop us a comment below, or shoot an email to info@mosaicmanufacturing.com.

-Mosaic