Mindset Update - July 2019
Product renders, production challenges, and timeline updates!
Jacob Flood, July 17, 2019
Montreal’s hot and humid summer has contributed to a productive month of June at Mindset. We’ve got a lot to cover with this update, so let’s dive right in!
The last month has been dedicated to the DMF/DFA process – below we describe in more detail what that entails. In particular, we made some major changes to the upperband and cabling solution that will ensure a reliable, scalable production. The DV batch will be ready in late August as a result, instead of July as intended. Once we validate these changes, the next step is mass production all the way.
In addition, we’ve received our first mass batches of electrodes, which look amazing. We’ll be ordering a batch of 1,000 next week, which will enable environmental testing and the production of our first mass manufactured batches (DV and beyond).
We’ve started making high-resolution renders of the final industrial design for marketing purposes, and they look stunning – scroll down to check it out! We’ll be sure to share more as we continue the process.
Finally, the app is running the new UI skin, and it looks gorgeous. We’ll be prioritizing robustness over the next months, gearing up towards the launch. We answer a few questions below that we got from backers last update.
INTRO – DMF, DFA, and the DV batch
Between now and the last updates, there has been quite a lot of developments on the hardware end. Over the last few months, the big milestone we’ve been pushing toward is to produce our first off-tool units for beta testing.
This off-tool batch – which we’ve been referring to as DV – would be produced using the same materials, manufacturing processes, and supply chain as all future batches, and represent a huge step forward towards shipping. At the same time, it represents a huge risk, as the moulds cost upwards of $100,000 and therefore make any do-overs absolutely infeasible. Put simply, we get one shot to make this work. We’ve spent a lot of time and energy as a result working on optimizing our design, to ensure that once we move to mass manufacturing nothing gets in the way of shipping your pre-orders.
As we described in our April update, after our previous EV4 batch, we found some major reliability issues, and a few more minor issues that needed to be resolved before we decide to pump out units. For the past months, we’ve been iteratively dealing with these problems. We had predicted that the process would yield units by late July. Due to this design work, and in part to the negotiations we’ve been undergoing, August is a more realistic timeline to finish the DV batch.
As eager as we are to get this going, it simply isn’t a smart decision to force into mass production, knowing the design still has a major reliability issue. In this update, we’ll discuss in more details the challenges we faced going through the Design for Manufacturing (DFM) and Design for Assembly (DFA) optimizations, how we resolved them.
In particular, we’re going to focus on the upperband, the cable assembly, and the electrodes.
Aside from the EEG stack, the upper band is the single most complex part of the headphone assembly. This is due to its multi-faceted role: it has to maintain structural integrity of the headphone, provide attachment and sliding friction for the armbands, hold the lower band and EEG sensors, and provide the right clamping force for the ear cups, all while acting as the aesthetic centerpiece to the industrial design. Despite ours, our factory’s, and Onkyo’s team’s extensive headphone design experience, this part has proven very challenging to get right.
Through an extensive design review of our EV4 design with the factory’s engineering team, we found an issue in the upperband that affected the function. With the current mechanical implementation of the upperband, it was found that the clamping force was not very reliable – given that the clamping is finely tuned for comfort and audio performance, this design had to be solved.
Because of the specific look of the industrial design, it has been quite a challenge for the mechanical engineering team at the factory to implement a reliable working solution. The upperband’s latest design uses plastic with a silicone over-mold at both ends. Inserted into this piece is a springsteel band, which provides the appropriate clamping force and additional “springiness” that plastic can’t replicate. What we found is that in many prototypes, the combined force of the metal and plastic ended up being too stiff, and the resulting clamping force too strong.
The root cause of this problem is somewhat difficult to describe. In order to attach the two parts, the metal was fixed at both ends to the plastic upperband. As we open and closed the headphones, the metal and plastic - which were curved at different radii - were forced to move together. The metal part, to compensate for this, was designed to slide with respect to the plastic piece, maintaining the expected curvature as the headphones opened. In design, this worked; in practice, the pieces did not behave as expected.
In addition to this issue, our bending test – the process of opening and closing the headphones many thousands of times to evaluate its longevity – discovered another issue. The junction between the upperband and the bracket compartment (where the armband slides in and out) was discovered to be weak in fatigue. This means that in the wild, a certain number of the headphones would break at this interface a year or two into use. As a premium product, this is not acceptable.
In response to these issues, the design of the upperband was adjusted. The main changes were removing the silicone for reliability reasons, strengthening the junction between the band and the bracket, and decreasing the excessive clamping force that resulted from the metal and plastic being pinned at both ends. These changes have been implemented and we are currently waiting for the newest prototypes to arrive in order to validate that all major issues are resolved.
The upperband is currently the rate-limiting factor for making a mass-produced unit, which means that if this next version works, we will be able to get the mass production DV batch started immediately. Barring any more major issues, we will have 20 units ready by end of August.
We will provide more explanations and photos about the manufacturing processes we are using in the next update, but to give you an idea, a mold for a simple part looks something like the image below. Our goal is to have 19 such molds for our parts nearly ready by next update!
Following last batch, another source of concern that our manufacturer raised is the cable assembly.
The original cabling solution included a bundle of 22 individual wires, which were hand soldered to their respective PCBs. This requires more labour than is typical in a regular headphone, but at the time was deemed manageable.
However, our manufacture recently set up an assembly line for another headphone project which also has many hand-soldered cables, and ran into a dramatic number of quality issues. Essentially, the more cables a product has, the higher the risk that one of these wires is improperly soldered. This would increase the scrap rate – the number of units that are rejected during quality assurance testing – as well as the risk that during use, headphones would spontaneously stop working. Both of these realities present large risks.
In order to avoid this issue, our manufacturer adapted the wire stack to retrofit an FPC (Flexible Printed Cable) into the current design. While the design cost to use this solution is higher, this alternative would avoid soldering entirely: the circuit could be plugged directly in to each PCB during assembly. We decided in June that spending the time to implement this strategy was worth the cost.
Luckily, changing to the FPC has minimal effect on the rest of the product’s design. We are currently testing this new cable assembly to ensure there are no major issues. With this cable assembly, we expect to decrease the scrap rate significantly, and prevent many issues in the field – aside from the obvious benefit of a better product, this also means we should be able to scale up our manufacturing faster. If you’re reading this, wondering what an FPC connector looks like, it looks like this:
To the user, this change will not be apparent. However, in practice, it improves the product quality dramatically.
Beyond the scrap rate, this change has further implications. In a previous update, we discussed that the cable harness was inflexible, which caused force imbalances issues on the ear-cup. This was due to the types of cable used inside, as well as the stiffness of the material covering the cables. In order to solve this issue, we had used an additional process on the cable to transform its shape and counteract the force imbalance.
The downside to this added process was that it created a distinctive feature on the cable – a kink in the wire – and diminished the quality of the surface finish. In the image below, you can see an example of this defect, wherein the cable would kink at the bending point. This defect was due to the way the cable was shaped: clamping the cable and immersing it in hot water. By doing so, the material became soft, and would subsequently maintain the deformed shape when cooling down. Unfortunately, this risked causing visual defects, as shown in the picture below.
The new FPC will instead be wrapped in silicone, which is significantly softer than the previous material. The end result is that the added process is no longer necessary. In addition, while the surface finish of the previous cable was not smooth due to the type of material required to manufacture it, with the FPC cable’s silicone wrap this is no longer an issue. We should also be able to remove the indented line running across the cable for the next batch of prototypes, a subtle but satisfying win.
The lesson learned in the process runs counter to Occam’s razor: in this case, the simplest solution was not the best. We’re happy to have learned this lesson, and happier still to have found a reliable, beautiful solution for the wiring harness that will enable us to effectively mass produce Mindset. In our next update we’ll be able to report back on the results of the wiring tests, and demonstrate the assembly of the now-finished design.
In parallel to the wiring and upperband changes, we received our first batch of the electrodes that fit in the upperband. We’ve been running quality test to validate that their material properties, plating thickness and resulting signal quality meet our requirements.
With those tests, we will be able to better understand the best and worst-case scenarios of electrode lifespan. Our goal has always been to design electrodes that will work for the lifetime of the product and beyond. Despite that, we’re aware that in certain harsher environments, the signal quality of the electrodes may reduce quicker than expected. It is for this reason that we have made the electrodes replaceable – no matter what, you’ll always be able to access what you need to make Mindset work great.
They final design includes a button endpiece, which clips into the headphones. You can see a small sample of the electrodes below:
So far, we are very happy with the test results. Next week, we will be ordering a batch of 1,000 electrodes to run more extensive environmental tests, and start producing our first batches. As the EEG remains core to the value proposition of our product, it’s been incredibly satisfying to see this part of the product scaling as we imagined – we can’t wait to get them in your hands, so you can experience it for yourselves.
In addition to the major issues described above, we found several more minor issues that we resolved with subtle design tweaks.
One example is that during assembly, we found that one of the walls of the speaker plate was too easily breakable. We subsequently increased the wall thickness in that area to reinforce the part, and ensure we do not get too many defects.
During the last month, we also finished negotiations with our supplier. This was a long process – we went into more details about it in our last update. We’re happy to say that we finally came to a mutually beneficial agreement between us and the supplier.
While we won’t be showing fully assembled units until the DV batch is ready, we are currently preparing the materials to assemble them. Those includes the plastic parts, electronics, speakers, electrodes, custom made fabrics, and the rest of the assembly tools required to make the headphone work. As part of this design freeze, we’ve started producing high-quality renders of the final industrial design to use in our marketing materials.
We’re so excited to see the fruits of all of this labor – we think that the final design looks even better than when our kickstarter went live. Barring any catastrophic issues, the headphones you see above will be identical to the ones you receive in the mail.
Moving forward, as the manufacturer starts making more units to nail down any potential issues, we will start doing more user testing. Our goal, now, is to fine tune the experience between using the headphones and the app, and ensure that everything is as seamless as possible.
In parallel the hardware improvements, the software team’s been working hard to implement the UX we showed last month.
We still have a few features left to implement, but as a whole the app is looking and feeling incredible. The team’s been competing lately to see who can perform the most deep work – so far David & the China team have been winning.
The next months prior to launch will be spent building up robustness in the app: making sure that account creation, login, and the music integration is as seamless as possible. For more details about the app, be sure to check out last month’s update.
In particular, last month we got a few questions that were not clearly addressed. Firstly, related to platforms: the app at launch will support MacOS, Windows, and most Linux builds out of the box. We’re prioritizing the desktop apps to ensure that the UX is seamless – this means that the mobile apps will instead follow up post launch. App development takes time, and we don’t want to rush the experience - our goal is to cover all platforms with an equally powerful UX during next year.
In addition, we want to make clear that the app will be free at launch, including all the functionality we’ve spoken about, and all ongoing updates. We’re thinking about potentially adding a premium subscription for additional music content later on, but this is not a current priority – only once we’re 100% satisfied with the full offering of the free app will we start working on paid content.
Finally, we will absolutely continue to push our additional updates as time goes on. We’ve built into the app the ability to update on-the-go, so you never have to think about checking for the next version. We have a long list of really cool experiences we want to build – alongside all the tools that our community develops through our API – that will keep us all engaged and focused for long to come!
Before we finish off, I wanted to get some quick feedback on our updates from you all.
Over the past months, a few of our updates have been posted late – often the 2ndweek of the month instead of the first. This is typically either so the product team can focus on time-sensitive work (for example this week David pushed the update so he send the order for the new upperband samples quickly as possible), or because we know that if we wait a week we’ll hit a milestone and have much more interesting details to share. Our #1 priority is getting the hardware out as fast as possible, but we also feel it’s very important to share as much as possible with you all – sometimes, navigating that dichotomy has led us to post an update later than expected.
A few backers expressed frustration at the uneven update timing – some suggested that short, consistent “TL;DR” updates bi-weekly would be preferable to the more narrative, information-dense summaries we typically post. As a result, I wanted to poll the audience: what do you think of our updates, and would you prefer we nix the details in favor of a very short status report instead?
Share your thoughts in the comments, and we’ll discuss!
As always, feel free to follow up with any questions in the comments, and we’ll be happy to answers them.
Likewise, I’m going to be doing another round of interviews to get a better sense of our backers’ needs and wants over the next weeks. If you’re interested in being part of this research, leave a note in your comment, and I’ll follow up with you!
Lots of love,
- The Mindset Team