OK team, we’re back again with another post!
We’ve made substantial systems upgrades in the past few weeks. These changes have started turning Palette into a much more heavy-duty (dare we say tank-esque?) machine.
As always, this is a pretty lengthy post. If you’re looking for the TL;DR version, skip to the last few paragraphs for a summary, timeline and next steps.
Our number one goal as a company has always been to deliver you an amazing product. This goal has driven our decisions over the past year, and has driven our product’s progress since the launch of our campaign.
As you may know from some of our earlier Kickstarter updates (April, May, June) we’ve been having some trouble with our drive systems. Palette’s drive systems control the movement of filament through the system, and are essential to Palette’s reliability and functionality.
The majority of this post will be focused on improvements we made to these drives, the outcome of the tests we’ve been running, and what this means for you and your Palette.
Until now, our drives were created from layers of laser-cut acrylic. This acrylic was stacked in sheets, bolted/pinned together, and screwed into the stepper motor(s).
Here you can see a picture of our acrylic drive systems:
For a number of reasons (covered in this post), we’ve upgraded to machined aluminum drives, as pictured below:
Aluminum drives without the top piece. Here, you can see the Teflon connections, gears, and idler bearing mount points.
Drives with top on: this is how the drives will be installed into Palette. Note: the pink 3D printed pieces will be replaced by laser-cut acrylic.
This is a substantial upgrade to Palette’s systems: we went out and made the most troublesome subsystem one of the strongest.
So… What problems did the acrylic drives have?
Using acrylic for Palette’s drives (1) caused issues with consistency and (2) resulted in a design flaw.
A very important aspect of Palette’s drives is repeatedly and reliably completing the same actions, thousands of times in a row. This consistency comes into play in two different aspects – inter-drive (multiple drives compared with each other) and intra-drive (a drive compared with itself).
When being tested, we’ve had some acrylic drives run thousands of splices without issue. Other Palettes we’ve been torture testing had issues occur around the 1,000-1,500 splice mark. A Palette would work great all afternoon, but late into the night drives would begin to slip (which would ruin splices and ultimately caused failed prints).
When we’d shut that Palette off, we’d remove the drives to investigate. One of our team members would be able to tighten the screws on that Palette about half a turn. These loose screws were a big issue: they suggested that the acrylic was warping over time. This was unfortunately not something that our old QC processes would have captured.
This degradation is evident in some Palettes, and not in others. This wear is unacceptable, and is one of the biggest driving forces behind the change of our drives. This issue was driven mainly by the material choice (acrylic). This pushed us in the direction of a more rigid and durable material like aluminum instead.
In 3D printing there are two types of common drives: (1) fixed-force (spring) and (2) fixed-distance. (These are also referred to as “constant-force” and “constant-distance”).
Most extruders in the market are fixed-force extruders – the extruders have a spring that pushes the idler bearing into the filament, pushing the filament into the drive gear, which drives the filament into the extruder. This design is referred to as “fixed-force” because the force is essentially constant, via the spring, and the distance is variable – i.e., the arm attached to the idler bearing can move back and forth.
Here you can see the drive on a Printrbot Simple Metal. The spring exerts a force on the arm, driving the roller bearing into the filament. This force pushes the filament into the gear, which drives the filament into the extruder.
Fixed-distance extruders are found on a lot of industrial machines, as well as a few consumer machines. The idea with fixed-distance is that the idler bearing and the drive gear are fixed at a certain distance away from each other. This distance allows the idler bearing to force the filament into the drive gear, which pushes the filament forward.
Here you can see a picture of our fixed-distance drives. The spacing between the idler bearing and the drive gear is fixed; there is no spring.
Originally (back in 2014), we used fixed-force gears – specifically for our ‘Black Box’ prototype featured in our Hardware Heaven blog post here: Link.
We quickly realized that the fixed-force approach we were using in the Black Box did not give us accurate enough control over the distance the filament was being driven. We made the decision to go to a fixed-distance approach to give us better control over the ultimate length of driven filament(s).
Working with fixed-distance drives introduces a lot of potential issues – the spacing of the drive gear to idler bearing must be correct every time if you do not want to have issues with slipping (gear/idler too far away from each other), or skipping (gear/idler too close to each other).
The issue we encountered with our acrylic drives [which were fixed-distance] is that the distance was not precise to passable tolerances.
So… When we started evaluating a new system to replace our acrylic drives, we wanted to design around this fixed-distance hurdle (the precise tolerance required for the distance between idler/gear). We wanted to get the best of both worlds – the low tolerance requirements of fixed force, and the consistent drive performance of fixed-distance. The solution was to build adjustability into the system using an approach we call the “adjustable fixed distance” drive.
A quick note about the Printrbot drive system from above:
You’ll notice the bolt on the left side of the drive, running down the middle of the spring. This bolt is used to adjust the tension in the spring, either increasing of decreasing the force the spring puts on the idler bearing’s arm.
This drive is an example of adjustable fixed-force. It gives you a lot of the advantages of fixed-force, but allows the user to increase the pressure from the spring if needed.
When designing the new version of our drives, we found a way to harness the best parts of fixed-distance (consistent filament distance), while also allowing for adjustability.
We shared a version of this picture of the new drive design at the beginning of this post:
Take note of the two dark areas circled in green - these are bolts.
The bolts mentioned above force the aluminum to bend around a point, giving us control over the distance between the drive gear and the idler bearing.
These adjustments have removed slipping and skipping drives, and allowed us to adjust the distance each drive pushes the filament. This then allows us to better match drives to each other, which results in better consistency.
Although pre-calibrated, these bolts can allow you to adjust your drives should they ever slip or skip after significant use of your Palette.
Machined Aluminum Drives
The key advantages of this new system come in a few main areas:
3) Assembly time
Our previous drives were made of acrylic, a polymer that is flexible and has the potential to warp when subjected to enough stress. By making the change to aluminum, the drives will become much more rigid, will better dissipate heat, and will be much more resistant to breaking down over time.
If you’ve been keeping up with our Kickstarter updates, you may be familiar with the issues we were having with drive gear sizing: Link. There are two important numbers when determining the distance a drive pushes a given piece of filament. First, drive gear bite in (effective distance). Second, gear to idler distance.
By designing out adjustability in our old acrylic drives, we were aiming to keep the distance between idlers and gears constant across Palettes. In practice, when we started assembling our first 50+ units, we started to discover that our previous system – while quite precise – was not precise enough to achieve a standard of reliability that we’d be proud to ship.
Because our new drive systems are adjustable, we’re able to tune every drive system’s gear-to-idler distance, effectively controlling the drive gear bite in. By controlling this distance, we’re able to increase the range in sizes of drive gears that can be used together effectively in each Palette.
All-in, assembling, testing, and categorizing acrylic drives used to cost around $30/Palette. This figure was high given the failure rates for in-circuit QC drive tests. Each ingoing drive pair required matching two motors (2 pairs of 2 ingoing drives for each Palette). For every drive pair that passed QC, an average of two to three motors had to be swapped to find matches that met the required specs. On top of that, around 33% of our drives flat out failed, and could not even be fixed by swapping the gears/motors that were paired together.
Check out these pictures for an understanding of what our fail/pass numbers looked like with the old drives:
On average, the “Passes” required 2-3 motor swaps per pair. These swaps were completed to find pairings that would have acceptable variances between the two motors and used to take an average of 5-10 minutes to swap/test.
This cost did not even include the number of hours required to replace slipping drives in Palettes once they left our production facility, and entered QC.
Over time, these numbers add up to become quite substantial, both in terms of opportunity cost and capital requirements.
Prototyping and Producing Aluminum Drives
We began investigating this approach about a week after we called the all-hands meeting we mentioned in our last post: Link. We reached out to our machine shop in Ottawa, as well as our contract manufacturer in Shenzhen to begin the process for procuring aluminum drives.
Our shop in Ottawa really came through for us – they shipped us the first iteration of the outgoing drive within two days of us placing the order. We want to give a shout out to Chris Nooyen, one of our team members for making this happen. Chris has been instrumental in making Palette move as quickly as it has; we’re incredibly grateful to have him on our side.
We are currently in the validation phase and are 2-3 days away from placing the order for the first 100 ingoing drives (50 outgoing drives), which should be here in the next two weeks. This is enough to ship 50 Palettes – all of which are currently waiting for the drives to arrive. When we are confident these drives will result in passed QC tests (and we have every reason to believe they should), we’ll quickly order more.
Aluminum Drive Test Results
We received the first 10 of our new drives late last week (final week of June) and have been testing all weekend and throughout this week. We’ve noticed a few things that we want to catch everyone up on.
For our first 10 drives, we made a small dimensional change, which caused an alignment error between the bolt holes on the drives and those on the casing. On our 100 ingoing/50 outgoing drive order, we’ve fixed this error.
BUT. We do have some awesome news. Our first 10 drives arrived on Wednesday morning, and we inspected and installed them by Wednesday evening. Thursday morning we made some changes to our splicing algorithm, and Thursday night we setup a print.
What you are about to see is the largest, longest print we have ever run.
The print you see here was 508 splices, and over 160 meters long.
…..it was supposed to be 524 splices, and 168 meters. It failed right on the top few layers due to a broken splice. It ran for approximately 30 hours in total, with no signs of degradation (until the broken splice at the end).
To give you a bit of scale from the prints in our last post:
The Hilbert Cube is ~3x longer than the longest of the other prints (160 meters compared with 55 meters), with 200 more splices than the Tiki Head).
While we’re never happy with broken splices, we are seeing continual advances in splice consistency and reliability. Splices are only going to get better as we continue to tune the new splicing algorithm (which can also be updated with firmware updates).
So what’s next?
Drives and Shipping
We have a lot of tuning to do for the new splicing algorithm. We’re going to keep working on this to ensure splice strength and reliability to do not cause any issues.
As mentioned we have 2-3 more days of validating the 5 production prototype Palettes to prove that we have solved not only the tuning but also the degradation problems that were present with the acrylic drives. Following that we will pull the trigger on the first drive order (50 Palettes’ worth) from our Ottawa supplier; these drives should arrive by July 19. From there, we’re all hands on deck getting these drives into Palettes, testing, and shipping them out to you (aiming for 20-30 by the end of the month if at all possible). Once we’re confident that they’re passing QC, we’ll be able to accelerate the rate at which we’re shipping Palettes.
This is definitely a lofty goal; however, we have (literally) hundreds of Palettes in our facility waiting for drives to be installed. We’re ready to calibrate, install, and QC test the drives when they arrive.
We’ve been working on entirely new versions of software and firmware for everyone.
Our new application is set to be released on July 29th. Our software developer, Brandon, is going to be putting together a post for everyone to walk you through the improvements that have been made the system.
Check out a sneak peek screenshot below:
Firmware: Better UX/UI for Palette
We’ve been working on new firmware release as well, something we’ll be talking about in one of our upcoming blog posts. This new firmware has a host of features to make your lives easier when operating Palette.
New commands include: Jog Mode, and Tool Cleaning Mode, among others. Much more on this to come.
Calibration + Pinging Update
Part of the new firmware is substantial upgrades to calibration and pinging. We’ve had some feedback from our early backers stating some issues with ping detection, and calibration algorithms. As we mentioned in our last post, we’re going to dive deep into new ping detection methods, and the calculations that happen once a ping is picked up.
We promised this as our next post, but once we made the decision to swap our drives we wanted to relay this information to you as soon as possible (hence today’s blog).
What to expect at the end of July
The end of July is a big deadline we’re setting for ourselves. Please expect the following from us at this time:
1) Three Blog Posts – one covering ping detection, one covering corrective algorithms, and one covering software + firmware. These updates will be released sequentially, one each day for a period of three days (similar to the mega Kickstarter update we split up in the middle of March -March Update 1, March Update 2).
2) Shipped units – we’re aiming to push the next 20-30 units out the door at the end of this month. These units will have aluminum drives and new firmware. Our updated software will also be available then.
3) If you have already received your Palette we will be in touch to ensure you have the option to send your Palette in for upgraded drives and new factory calibration if you desire.
We’ve been working on these goals for a few months now, and things have been progressing incredibly well – we’re doing everything we can to begin shipping Palettes again at the end of the month.
We have a monumental amount of work ahead of us, but we’re so close to getting this operation to finally click.
We really appreciate the amount of support and patience we’re receiving from our backer community – we promise you Palette will be worth the wait!
You will hear from us very soon…