Mindset Update - October 2018
October has come and gone! We have a dense update this month, with lots of questions we’d like your feedback on.
Jacob Flood, November 16, 2018
This update covers the progress we made over the last two months. In addition to the hardware, we’ll be sharing some updates on the software we’ve been working on, including a few really interesting applications that we haven’t yet shared publicly.
We’ve fixed the cabling, Bluetooth, and PCB layout, ANC, and button problems we mentioned in our last update. We’ve also commissioned our first tooling, a huge milestone! We’re considering making the armbands out of plastic – please read below and voice your opinion in the comments.
Our timeline is to finish tooling by early January, finish DV by February/march, and start ship soon after.
Below we describe the software interface, and announce our neuro-adaptive music system: a dramatic innovation in focus music. We also have several questions we’d like feedback on, so please comment with your responses!
The last few months have been difficult for the hardware team – being so close to the mass production precipice, it’s been a constant negotiation of how much time we want to spend optimizing parts before committing to the design and cutting the tooling. We’re excited to say that as of today, we’ve begun preparing our first tooling parts!
Below we’ll discuss some of the optimizations we’ve been working on, and then dive into what the tooling means for the current timeline.
In the last update, we mentioned that the large number of cables what causing a lot of unforeseen issues – mainly, it added too much friction to the arm band sliding, and had trouble bypassing the ear cup pivot.
Since then, we redesigned the cable harness to pass a flat cable assembly on the outside of the earcups rather than on the inside. This allows us to use special cables – shielded, and material-optimized for EEG signal quality – that will ensure there are no signal quality trade-offs in the product.
In order to implement those changes, we worked with our industrial designer to adjust the earcup and the headband clip to allow a cable to pass through, bypassing the arm band. We have tested three different solutions, and decided to move forward with the following:
The cable exits the top of the ear-cup parallel to the arm to avoid any twisting. It then bends once, and enters the top band from the bottom of the clip.
Other alternatives we tried had the wires exit from the side of the ear cup (instead of the top) and enter the upper band after clip, bypassing the slider entirely. A fatigue stress analysis on these designs showed us that after many cycles of wearing the headphones and adjusting the slider – our allowable limit currently is 20,000 – these variants risked breaking one of the wires. This will be further validated through a bending test.
Each cable we use provides a specification that details the way they are allowed to bend. In our case, the primary concern is the bending radius: the smaller the curve over which the cable bends, the higher the stress on the cable, and therefore the higher the risk of a wire breaking. In order to ensure the product has a long lifespan, we need to respect the cabling specifications. We’ve done preliminary validations with single-unit test wires, and the system behaves as we expect. Once we receive the off-tool batch of cables, we will be running a bending test to confirm how long the cable assembly lasts before it stops functioning.
The picture below shows a test assembly we made to see how the cable would bend in practice. We used this to evaluate the design before ordering a real sample.
Once the bending test is complete, the remaining step will be to test the assembly, verifying that the soldering process doesn’t overburden the assembly of the rest of the headphone. When our final cabling sample arrives, we’ll be able to validate this point and freeze the design of the cabling.
Hidden in the cabling update is some very good news: the cabling sample ordered above marks the creation of our first official tooling!
This is a momentous step in every hardware products life. The tooling – the name given to the moulds used to mass product parts – represents the move from engineering the product, to the production process. As the tooling costs upwards of $100,000, the commitment can’t be understated: once we start the tooling, design becomes fixed, and any adjustments must be made under the constraints provided by that design.
The first tool we’re cutting is for the cabling solution described above. As we validate this first mould, we’ll be ordering the tooling for more and more of the parts over the next weeks. We’re incredibly excited to be moving into the production phase of the product, where we’ll be able to start batch-production, beta testing, and – of course – shipping units to backers.
We want to take the time to reiterate that this process would never have been possible without the support of our backers. We’re incredibly grateful for your continued support, and we can’t wait to get this product into your hands so you can share in the experience with us.
As we head to mass production, one of the last tweaks we need to undergo is shaving weight off the headphones. As it stands, the headphones are heavier than we would like.
The major contributor to weight, unfortunately, are the armbands. The industrial design has the armbands made of aluminum, which is significantly heavier than plastic – as a result, these pieces contribute nearly 100g to the final weight of the headphone. In order to cut down on weight and improve comfort, we are considering changing this part to plastic.
The picture above shows the armbands and earcups. Currently, the armbands are aluminum, while the earcups are plastic. Changing the armbands to plastic would help us match the color and finish of both parts better and drastically reduce the weight, while keeping the metallic finish that the industrial design demands.
We’d like your opinion on this: do you prefer we keep the aluminum and have a heavier headphone, or move to plastic to improve comfort? Please answer in the comments!
The last weeks have been focusing on the assembly of the electronic stack: making sure that each part can be fit into the headphone snuggly, without interference from ground loops or mechanical failure.
Due to the cabling changes above, we made some last-minute tweaks to the mechanical design of the earcups that resulted in updating the layout of the main PCBs. This iteration included specifying the connectors, wiring requirements, and assembly constraints for the boards to make sure nothing comes undone or decays during use.
The last test we need to run is an evaluation of the effect of the magnetic field of the speaker on our electronics. While a sanity check we performed showed that the effects were negligible – back before we ran the Kickstarter – we want to make sure that the layout and assembly we’re committing to doesn’t interfere in any way with the signal we’ll be collecting, regardless of the environment it’s used in. Once this is confirmed, we will freeze the PCB design.
Once those PCB shapes are fixed, we will manufacture a new test batch of boards and mechanical parts to do full-assembly tests, and verify that everything fits well together. If all is well, we’ll give the green light to start cutting the rest of the tooling.
While testing the EV2 units, we ran into some technical issues with the Bluetooth range: we couldn’t listen to the music properly farther than one meter from the source device. We’ve since discovered that this was due to a firmware issue on our manufacturer’s side, and a loose antenna component in the prototype. The issue has since been fixed and tested.
EAR CUP ELECTRODES
We’ve talked a lot in past updates about the electrodes at the top of the head. In parallel to these, we’ve been iterating on designs for the ear cup electrodes.
Previously, we’d talked about the conductive fabric we had been using for these electrodes. This concept came from several papers we read, which used conductive fabric for high-comfort EEG measures with children. In practice, conductive fabric provided a wide surface of contact and a low contact impedance – a great start for an electrode.
We put the ear cup electrode on the backburner while we worked on more pressing issues: the top electrode design, the cabling solution, and the rest of the industrial design changes. In order to focus on the biggest issues, we deprioritized the two points that remained to test with the fabric electrodes: the assembly, and the material properties.
The assembly is critical, since this dictates the mechanical properties of the electrode, and therefore can dramatically affect signal quality. Firstly, we would need decide where to place the electrode to maintain contact with the skin. Once this was decided, we’d need to choose how we wire the electrode to the amplifier PCB, making sure that the connection is robust, shielded, and easy to assemble.
The material properties were equally important, since dissimilar materials create battery potentials which can increase noise in the signal. Firstly, we would need to compare the material used in the fabric electrode to the top electrodes, making sure that the material profiles are similar enough to minimize voltage noise differences. Once this is decided, we would need to run a corrosion test to ensure the longevity of the electrode, since corrosion can hurt the signal quality over the long run.
In the end, we decided that the risk of delay associated with these processes outweighs the benefits of using fabric for the ear cup electrodes. Instead, we are manufacturing a new generations of button-shaped cushion electrodes that will rely on the tests we already conducted for the top electrodes, and a design we already validated.
We feel that it’s important for us to acknowledge when we are wrong, and avoid making decisions biased by the sunk cost fallacy. While the appeal of the conductive fabric was alluring, our priority is creating a quality product, and we feel that we can do that faster, and with less risk following this approach.
We’ll be sharing pictures of the new button-shaped cushion electrodes over the next weeks, as we quickly prototype, test, and finalize this piece of the design.
We mentioned last update that we found the function buttons on the headphone difficult to locate, since they lacked an obvious reference feature. This had been an issue we discovered while testing the prototype with users.
As a result, we experimented with a few design variations that improve the ease of use of this feature.
Raised volume buttons, and lowered center button.
Raised + volume button, and distinct circular feature on the center button.
Let us know in the comments which design you prefer!
Having listened to the audio from the latest prototype batch for dozens of hours, David is now deeply satisfied with the quality of the audio. We will need another audio tuning iteration post-tooling to account for small fit changes, but no major edits.
We have since discovered, however, that the passive audio through the AUX port does not perform as well when the device is not powered by the battery. This is due to the audio signal processing performed on-device, a second layer of audio optimization in addition to the mechanical tuning.
We are working with our factory to ensure that the audio performance does not suffer when the battery is OFF. This will likely require the addition of a few passive components on the audio PCB, a relatively painless solution - we don’t anticipate this causing any issue.
In parallel, we mentioned in the previous update that possible air leaks could affect the ANC performance. We have since resolved the issue by adding a step in the assembly process to seal the ear cup, preventing such leaks. We don’t believe any design changes will be necessary.
Following our last update, we got a lot of questions about our current timeline. We want to clarify that for you.
We have 2 major milestones left before shipping out units:
As we mentioned above, we’re starting this step now. We anticipate the tooling will take two months, from last week through until early January.
2. DV batch
This is the first pass production batch. We wanted to have DV done by end of year, but because of the cabling problems and Chinese New Year, we think that February/March is more realistic. Environmental certification, final audio tuning, and the last design tweaks take place from this batch. The delay between finishing tooling and getting the DV batch is mostly due to lead times on some of the parts, which is out of our control.
From there, each subsequent batch will include shipping units. The PV batch will be shipped to beta testers (we’ll follow up by email), and subsequent batches will be shipped to all backers. We don’t have a shipping schedule yet, since this will depend on the certification, and some limitations set by the freight exporter from China. The moment we have a schedule built for shipping, we’ll be sure to share it.
“We designed Mindset with a single goal: to create the perfect work headphones.”
This was one of the first lines of our kickstarter campaign. We then described the EEG, the noise cancelling, and the concentration features, both in our campaign and our updates since. That vision, however, still rings true: Mindset is about more than improving concentration. We want to give you the tool to get your work done, so you can bring your own creations to life.
In light of this, we want to clarify our north star: Mindset helps you maximize and optimize the time you spend in deep work, so you can spend less time on distractions and more time on what’s meaningful to you.
Most of our updates focus on the headphone: from the audio to the sensors. In reality, however, the software we ship is equally important: the app is the brain that controls the Mindset experience.
While the hardware team has been busy working on the headphone, the software team has been refining the app experience: building out the app infrastructure, improving our data science, and building out new features to help you improve your focus. For the first time, we want to talk about some of those features.
At the core of the app, is the idea of deep work: single-minded, focused, cognitive tasks that push our limits and stretch our minds. This concept was most recently popularized in Cal Newport’s book Deep Work, where he extends the following claim:
“The ability to perform deep work is becoming increasingly rare at exactly the same time it is becoming increasingly valuable in our economy. As a consequence, the few who cultivate this skill, and then make it the core of their working life, will thrive.”
During the research I did writing my own book Study Smart, I came to find that the most successful students across domains embodied the idea of deep work intuitively. We consider deep work to be the starting point of all learning and knowledge work, and we consider it our job to help maximize and optimize your deep work.
When you start a deep work session, you’ll have the option to customize your experience: choosing your audio landscape, muting notifications, setting a timer, and identifying the task you want to complete. This process helps you get in the state of mind necessary for a focused work session.
Once you’re in deep work, the app becomes secondary to the work you’re doing. Your sole focus will be your work, your thoughts, and our neuro-adaptive music.
We’ve hinted at this project before, and we’re now ready to talk about it: we’ve built the world’s first neuro-adaptive music system.
When we pitched the Kickstarter, the idea was that audio feedback – chimes and volume changes – would signal decreases in concentration, and help you refocus. While the science behind this feedback is sound, in practice this kind of feedback often feels chaotic and obtrusive.
In addition, getting into a flow state goes beyond simple audio feedback. Ear fatigue, often referred to in academia as audio habituation, often pulls you out of flow when music becomes too repetitive. An uncontrolled music landscape also prevents us from leveraging any of modern day understanding of audio entrainment, and the associated benefits for focus and stress. Finally, simply delivering feedback on top of existing music leaves a very small parameter space for optimization and individual personalization over time, which is a key part of Mindset’s appeal.
To improve upon this experience, we’ve been working with sound engineers, composers, and audio researchers to design an experience that’s optimized for deep work from the ground up. The result is a first of its kind music system designed to improve your focus, and get you into flow.
It works by creating modular audio tracks, that can be triggered, changed, and combined in real time. Non-linear looping events and beat-matching decide what combination of instruments and sounds to use, and how to shape these sounds to match together. The result is a unique soundscape that integrates modern-day understanding of the neural effects of audio, that never repeats, and that can evolve over time.
Finally, EEG metrics – attention, motivation, and fatigue for example – further modify the musical landscape. Instead of sending a chime when you get distracted, we can increase the intensity or warmth of the music. By integrating the feedback into the core structure of the music, the resulting experience is smooth and subtle.
We’ve been working really hard to bring to life what we think is the pinnacle of neurotech. We’re testing out our first neuro-adaptive music system in-house as we speak, and it sounds amazing. We’re incredibly excited to share this news with you, and can’t wait for you to try it out for yourselves.
MAXIMIZING DEEP WORK
In addition to the neuro-adaptive music, we’re working on other features that help motivate and maximize the deep work sessions. We’re testing out task-tracking, calendar integrations, brain training games, and other productivity-centric features to see what combination best motivates us to get our best work done.
We’d love your feedback on this part of the app. In particular, please comment with your suggestions:
What music tracks help you focus, so we can use them as a reference?
What metrics would you want to track on a session-to-session, day-to-day, and week-to-week basis, to motivate you to do more deep work?
What tool integrations (to do list, calendar, notes, email) would help motivate you to do more deep work?
We really appreciate your feedback, so please comment-away!
That’s all folks!
Again, please comment your answers to the 3 software questions to help us test out new app features!
As always, lots of love,
- The Mindset Team