Product Update - February 2018
Welcome to our blog! In the upcoming weeks you’ll find some great content made by our team. Take a peek around our website, and let us know what you think!
Jacob Flood, February 19, 2018
The last batch of prototypes didn’t reveal any fatal flaws – this next batch will focus on more minor tweaks to audio, fit, and signal quality. We’ll be using the next batch to set the standard requirements for our production units when they come off the chain.
Since the factory is on holiday for Chinese New Year, we’ll be working mostly on getting our tests set up, so that we can move quickly from the next batch to tooling. So far we’re keeping the same timeline – this next prototype will tell us whether we need to run another iteration, or we can move straight to tooling, certification, and production. In the meantime, check out the questions at the bottom of the update, and let us know what you’d like us to focus on!
Happy Chinese New Year!
During the month of February, most of China slows down to celebrate Chinese New Year. In particular, all the manufacturers and suppliers go on holidays - they started winding down their operations about two weeks ago and came to a full halt this week. We saw this coming, and are making sure to keep busy while production halts.
An art display in front of our Shenzhen office, celebrating Chinese New Year
This update will focus mostly on the tests we are currently running (and plan on running). These tests will set the requirements for our next batch of prototypes, which – barring any issues – will be the last batch before we cut the tooling.
Ready, set, let’s go!
In the previous update, we have reviewed the latest engineering prototype displayed at CES. Since them, we have gone back to China to implement the various changes we talked about. We are currently doing one more iteration of engineering prototype before creating the molds for the various parts. The primary goal of this batch is to decrease the potential for errors as we move forward with tooling production, as well as to set the standard specifications for the production runs. Once we start the tooling process, it becomes increasingly difficult to correct manufacturing issues, so it’s critical that we diagnose any problems now, before making the leap.
Shenzhen is a ghost town during Chinese New Year!
The next prototype will be built to evaluate how well the device are meeting our requirements. Specifically, we will be looking at:
- Active Noise Cancelling (ANC) tests and tuning
- Audio quality tuning
- Head-fit tests
- Force distribution tests
- EEG signal quality tests
- BOM readjustment
In addition, there will be a few notable changes, such as the way that the fabric is laid onto the ear-foam, to avoid infringing on certain existing patent. More on that later – let’s start with the Audio.
ANC Tuning and Validation
We intended our previous prototype to be able to validate the ANC solution, however, the poor tolerancing of some key parts created leaks in the ear-cup cavity, which prevented us from properly testing the ANC. The next prototype will tackle this problem directly by increasing the tolerance requirements, and choosing a supplier that can deliver better quality parts. In addition, sealing material will be applied between the ear-foam and the ear-cup to avoid air leakage, which affects both the quality of the sound as well as the ANC.
A key part of the ANC design is the assortment of microphone holes: on the outside of the cup, the inside of the cup, and on the control panel near the USB-C port. In order to optimize ANC functionality, each of these microphones must be precisely positioned. In particular, two of the microphones must be placed along the axis that connects your ear and your mouth, to ensure that the microphone that you use for voice calls works properly while using ANC. This step was a large part of the design of the previous prototype, which was not mentioned in last month’s update.
Our current manufacturer – Grandsun – has tremendous experience in developing, integrating, and tuning ANC headphones. As a result, they have several hardware solutions for us to choose from, that will function properly for our product. For those interested in the details, the big debate is between pure analog ANC, versus a digital solution. While digital solutions are more flexible, and sometimes sound better, they consume significantly more battery - roughly 50% less battery efficiency compared to an analog design. Digital solutions are also much more expensive - about 3$ difference per unit, at scale. There is no immediately obvious answer.
We are currently exploring which solution makes more sense to adopt. To make our final call, we will have to experience the difference between the quality of the ANC with both analog and digital solutions – we will be running this test over the break this month.
In the next prototype, we will be properly tuning the sound response for the first time. With the changes made to the ANC, including proper sealing, and changes to the way the fabric is laid, we can expect the sound to be easier to tune, and have more punch.
When it comes to the actual sound signature, we will be able to provide more details after the tests are completed. Right now, we’re aiming for a “natural-sounding” response – some frequencies, particularly on the low end of the spectrum, will be overemphasized.
Last year, during our headphone tests, we tried out several high-end headphones – both high fidelity and natural sounding. High-fidelity defines a device with a flat frequency response – a mathematically “perfect” sound signature. Natural sounding headphones, to contrast, overemphasize certain frequencies in order to provide a warmer sound, more aligned with most music tastes. Good HiFi headphones tend to sound flat, while good natural sounding headphones sound vibrant.
We found that most people – us included – much prefer natural sounding headphones. If the key to having great sound was a perfect audio response (high-fidelity), then most headphones by now would sound identical. But that’s not the case - the magic in a great headphone comes from creating a sound signature that is exciting, vibrant, and fulfilling. Partnered up with Onkyo, this is what we plan to do as well.
Once our prototype is tuned at the factory, our audio partner Onkyo will be reviewing the sound quality to ensure it meets their criteria. This implies that the device will go for a two week in Shanghai where engineers will test the device, and propose minor changes. After some alterations, the device will then be sent in Japan to be tested by a team considered to have “golden-ears” (Onkyo’s words, not ours!) This will be the final step to ensure that the sound is exactly as it should be.
From there, this device’s sound profile will become our reference. Each of the headphones we build will be compared to that reference and must be identical.
Head Fit Test
One of the important test we must conduct with the next prototype is how well it fits on various heads. This has been a core property of each of the design iterations so far.
In order to design the headphones, we used MRI data from head scans to recreate head models and obtain various key measurements. Using those, we created a mathematical model that would mimic the way headphones open, close or adjust. Now, the goal will be to put the theory to the test and see how well the headphones fit on different heads – specifically, verifying the electrode contact and comfort.
First, we will test the fit on foam heads representing maximum and minimum head shapes to see how the headphones can adapt to these extremes. Then, if the test passes, we will move on to testing on as many people as possible – different shapes, sizes, and hair types.
Force Distribution Test
At the same time as testing for the fit on different heads, we will also measure the contact force for each electrode, and the pressure distribution on the head. This way, we can ensure that the weight is evenly split and that the headphones are comfortable at all points. This test will allow us to fine tune the contact profile to keep as high signal quality as possible without compromising comfort.
The different rigs we’re testing to equalize force between the electrodes
We will likewise be tuning the compressing force of the headphones on the side of the head. Too high, and the headphone causes headaches – too low, and the sound quality, ANC, and overall comfort diminish. Luckily, Grandsun has a thorough understanding of the effect of this parameter on each of these variables, so tuning the compressive force will not take more than one iteration.
Signal Quality Test
Once each of these tests are complete, we will be able to test the signal quality of the EEG stack.
Naturally, the data acquisition system involves many layers: attenuating external noise (like 60/50Hz generated by power-lines), diminishing internal noise from the electronics (generated within the system), and protecting the overall signal from when it’s first measured on the head until it’s digitally converted. Since each of these noise sources affect the overall signal quality differently, we’ve had to iterate on our electronics stack several times to get it right.
In addition to electronics, the impedance of various components of the system – including what we refer to as the contact impedance – is critically important for the signal quality. This is a proxy metric that indicates the quality of the electrode contact – it’s essentially a measure of how difficult it is for the tip of the electrodes to “capture” the electrical activity happening in your brain. If the contact is not good enough, it’s a bit like listening to a radio through a bad antenna – it’s very difficult to discern any important information from the signal. Together, these measurements allow us to compare how significantly our hardware has improved from the previous iterations.
Our signal quality rig, with an electrode setup. It’s fancier than it looks!
The final test, naturally, is to validate the accuracy of the hardware by using it with our algorithms to measure and predict concentration. The better the hardware, the better we’ll score on that test. We’ll be performing each of these tests on our final hardware with the next prototype, to give our factory the stamp of approval to move forward with mass manufacturing the design. Very exciting!
Each iteration of the hardware allows us the continuously update the Bill of Material (BOM) for our product. This is a crucial data point to keep in mind – the final BOM cost can legitimately make or break companies.
Each iteration allows us to better understand the price of each component and assembly. From there, we iteratively make decisions on where we should spend time reducing cost, in order to ensure that the product will be viable in the market when sold in brick-and-mortar stores.
Complex designs have a tendency to inflate the BOM as they are developing – as issues come up, you have to spend more on parts in order to achieve the same desired goal. Removing parts, simplifying designs, and switching to better materials all help achieve this goal, while simultaneously making the final product more reliable, both before and after assembly. This is a big priority as we near the final stages of the design process.
While we don’t have any specific numbers to share, we thought it’d be interesting for you to know about all facets of the design process, including the financial parts. The more you know, as they say!
Things are Moving
Things are continuing to make a lot of progress – we’re really excited to get the next batch of prototypes and share results from our tests with you all.
In parallel, I’d love some feedback on which elements of the design you want us to focus on. Are you more interested in the mechanical parts, or the software? Do you like hearing about the product, or also the business? Would you like more detailed explanations about the design, or would you prefer to hear from the team about what we’re focusing on? Let us know in the comments so we can tailor our next updates to your interests!
The Mindset Team saying hi, from all over the world
So far, we’re still keeping the same timeline. When our factory gets back from their holiday, we’ll be making the next batch of prototypes – from there, we’ll be able to see whether we want to make one last iteration, or move straight to tooling. We’ll keep you posted on how that affects our schedule moving forward once we get some test results back.
In parallel, we’ll be reaching out with surveys soon, focused instead on you: why did you buy the product, what do you intend to use it for, and how can we better fulfil the experience you’re hoping to get. This time the questions will be qualitative – you can ramble about the different features you’d like, and how Mindset can help improve the way that you work. Please take the time to fill them out – the more you write, the better the final experience will be!
As always, much love!
- The Mindset Team