Mindset Update - August 2019
Update on our mould-making process, a deep dive into how mass production moulds are made, electronics updates, and more!
Jacob Flood, August 30, 2019
Throughout August, our factory has been working on cutting the moulds. During this time, we’ve been hard at work in Montreal on the electronics, the app, and planning for our launch.
In this update we’ll focus on the electronics progress, and the current state of the app!
The moulds are currently being cut for our first mass production batch – this process started earlier this month, and will take 40 business days. We’ll start receiving parts late next month, which will form the DV batch. More details about our timelines for this process are described below.
In parallel, we’ve received and tested our electronics batch, and it’s looking great. We shared some pictures of the complete full electronics stack, as well as our electronics specifications.
Finally, we’ve made progress in the app, implementing a goal-setting feature for forming a habit around deep work. I also share some insight I got about my own productivity last week, amidst lots of meetings.
As we mentioned in our last update, our factory has started cutting the moulds for use in mass production – usually referred to as our “tooling”. This is the major milestone that we’ve been working towards since our Kickstarter. It’s also the reason for the majority of our delays: committing to making the moulds is a >$100,000 decision that should not be taken until we’re 100% confident in the design. We’re very excited that this is underway, and eager to get our first mass production batch of units.
A byproduct of entering the mass production phase is that our schedules become more predictable. As a result, we wanted to take the time to explain this part of the process, and give more detailed estimates of our timelines. Feel free to skip this section if you aren’t interested in the technical details.
This process takes place at our factory in Shenzhen 40 days. This involves a linear series of processes that together allow for incredibly tight control of the shape, tolerance, and finish of the final mould. The steps are as follows:
Material Order for Moulds (Steel)– the moulds are cut from a single large block of steel. The first step is to order this piece, which takes several days to arrive.
Milling– the blocks are the milled, a process that uses a very large blade to shape the exterior of the mould, as well as the rough shape of the part.
CNC Programming– once the steel block is setup, a Computer Numerical Control machine (CNC) is programmed according to the specification from the mould engineers. This is a long, tedious, and very skilled process. The program gives instructions to the CNC for how to cut the part.
CNC Roughing– The first cuts are made to the moulds using the CNC. This allows for low-tolerance cutting to create the rough shape of the part, which can be later refined using more precise machinery. It’s also one of the most satisfying processes to watch, as this video shows.
Heat Treatment– the steel is then heat treated, which hardens and increases the wear resistance to the mould. This usually reaches ~1200 degrees Celcius.
Grinding– The mould is then grinded to remove imperfections, and prepare it for the EDM. This is an incredibly precise and skilled multi-step process, and largely the reason that China are leaders in manufacturing. You can read more about this process in this article.
Material Order for EDM (Copper)– during the grinding, copper blocks are ordered, which serve as an electrode during the EDM process. More on this below.
EDM electrode manufacturing and tests– the copper blocks undergo a cutting process similar to that described above, to produce the electrodes used in the EDM.
CNC finishing– the moulds then go through a second round of CNC cutting, this time with a tighter tolerance setting, to get as close to the final shape as possible with this technology.
Wire Cutting– The moulds are then cut using a wire EDM process. This isn’t necessary for all parts, but is immensely useful in some situations. This video does a great job of illustrating how it works.
Electric Discharge Machining (EDM)– The last step of the mould making is to use the copper electrondes described above to produce the final, tolerance shape in the mould using a die EDM process. This video describes this well.
Tool Frame Making and Assembly tests– Finally, the frame for the moulds (the parts that hold the mould in the injection moulding machine – are constructed, assembled, and tested. If all is well, the moulds are ready for use!
The result of this will be 17 moulds, tolerance and finished, that can produce all of the injection moulded (plastic) and cast (metal) parts in the headphone. During this period, we’re also putting in orders for batches of the other parts that don’t require tooling: electronics, fabrics, sensors, batteries, cables, speakers, and the steel band.
When the moulds are produced, the first batch of parts produced are referred to as T0. These parts are often imperfect – small errors in the calculation of plastic flow rates, temperature spread, and tolerancing cause minor issues that prevent these parts from being usable. These parts are tested, and the moulds are re-analyzed and re-cut in order to fix these issues. This shouldn’t take longer than 1 week.
Finally, the T1 parts are produced, and assembled into the units we call DV1 – this takes 1 week as well. These are fully functional, properly finished units. Summarized, the timeline for this process is as follows:
2019-08-08- Cutting the moulds (there are 17 moulds to be made, requiring up to 40 working days)
2019-08-09 - Material Preparation (all electronics, fabrics, sensors, batteries, cables, etc.)
2019-09-24 - Moulds are ready and initial parts are made (T0)
2019-09-31- Moulds have been adjusted after testing assembly/quality of T0 parts
2019-10-01- National Holiday in China (factories are closed)
2019-10-08- Manufacturing of mechanical parts (plastic, silicone and metal) to assemble DV units (T1)
2019-10-15- DV1 units are assembled and ready for testing
While the plastic moulds are being produced, we’ve managed to finish the tooling process for the electrode assembly. The image below shows the result:
While the moulds for the plastic parts are immensely more complicated than the one above, the process is very similar. An injection moulding machine – like the one seen above – holds these two parts along a very tightly tolerance guiderail. Many times per minute, the moulds are tightly forced together, and liquid plastic is injected into the empty cavity. After a few seconds the plastic solidifies, the moulds are separated, and the part falls into a container for use in the assembly line. Each part of this process, including the moulds themselves, must be tolerance to within a thousandth of an inch (~10um) for the resulting parts to function.
This marvel of engineering is what permits the types of consumer electronics we use today. In the case of the electrode above, the assembled pins are then electroplated in gold, tested, and assembled into the headphone.
We’ll have more images of our moulds to come as they’re finished next month!
FABRIC & EAR CUP
Earlier this month, we ordered our fabric for our DV batch. This will be ready by the time the DV units need to be assembled.
In our recent prototypes, we have been using the correct type of fabric - technical fabric for upperband, and microfiber for lowerband and earcup - however the samples used have never been precisely the right color. This is because the colors used in our industrial design are non-standard: we need a custom fabric color in order to match the design.
This is not a complicated process - it takes approximately one week to order the customized one. We are currently in the process of doing exactly that. We want to make sure we get the color right, because the minimum order quantity for this type of fabric is at least 300 meters. For context, it will take less than 7 meters for the 40 test units - the rest will go towards future mass production batches. It is therefore important to ensure everything is done right, as we do not want to waste any material in the process.
When customizing fabric, suppliers uses the Pantone color coding system, and adjust the fabric according to that color. For instance, if we wanted Pantone 430, we would need to make customized fabric because their standard color is much darker. We’ll be ordering this fabric this week!
In addition to ordering fabric, we made a small tweak to how the ear cup electrode is attached to the ear cup fabric. The picture below demonstrated the before and after. This fix ensures that the electrode placement more closely matches our ergonomics testing, and won’t interfere with comfort or ANC performance.
In theory, the connector shouldn’t have been that high to begin with, but there was a lot of variations in the manufactured part, and the assembly method wasn’t great prior to the change. After the change, everything looks and feels better - we’re glad we caught this early. The electrode is still replaceable with the change.
Once the DV units are received, we will begin the phase of testing and analysis described above.
The DV units are primarily used for a variety of functional validation tests:
ANC testing and tuning
audio testing and tuning
sensor contact ergonomic tuning
testing full assembly process, with proper materials and off-tool parts (since the molded parts are always slightly different from CNC prototype parts, so its needs to be tested).
understanding the defects and defects rate in order to adjust the molds
We will be ordering a batch of 40 units for the DV run. We expect a scrap rate of ~25% for this batch, which will yield 30 functional units.
Approximately half of the units will be used for the tests described above. The remaining units will be used in-house for development, as well as beta testing. We will be reaching out to a few backers once DV is ready in order to coordinate for beta testing. We’re really excited to get this party started!
In parallel to the mould cutting, we’ve been testing our latest electronics batch. This is the version that will end up in the DV units when we receive them from the factory.
The latest batch has been received and tested, and adheres to our requirements. Given the very sensitive nature of EEG electronics (since an EEG signal is typically <20uV, which is very small), we’ve had to go through several revisions of the electronics in order to get it up to spec. The primary specifications of our stack are as follows:
250Hz sampling rate
<1uVrmsof noise injection
24bit ADC resolution
No packet loss during transmission
±180mV input voltage range
We’re happy that with the latest batch we’ve been able to reach these specifications. As we work towards mass production we will be iterating on the PCB layout (the way that the circuit is manufactured) in order to optimize for noise and scrap rate, but for all intents and purposes we consider the electronics finalized.
You can see our full stack laid out, from electrode to Bluetooth, in the images below:
In parallel, our software team has been hard at work debugging the latest version of the app.
The latest version adds goal-setting to the deep work sessions. Each day you can set your target amount of deep work, and strive to achieve this goal throughout your day. This feature helps you to build a deep work habit, all the while training your focus during each session through our neuro-adaptive music.
We also implemented the settings page of the app, where users can modify and personalize the way their Mindset app interacts with their headset and their data. Still a few aesthetic tweaks to go, but we’re happy with the progress!
I wanted to share this anecdote from using the app over the past months. The screenshot below is my data from this past week. Much of my time this week was spent working with David on the electronics, which is collaborative, and therefore incompatible with headphones. This is clear from my data: during my mornings and afternoons, I often accomplished very little deep work.
This was hugely insightful – while I was aware that last week’s meetings were interfering with deep work, seeing this data gave me clear insight into how and why I was losing productivity. In particular, I often spent my best hours (which I know from previous weeks’ data is 1:30 to 2:30) in meetings, which is suboptimal. This has motivated me to shift my schedule for next week, to ensure that I maximize deep work in the mornings, and save meetings for later in the day.
We want to build as much value into our deep work sessions as possible, so that you can gain deep insight into your focus and productivity. I’d love to get your feedback on this: what information other than focus levels would you like to track in your deep work sessions?
Comment your answers below!
That’s all for now!
Feel free to share your comments or questions below. In particular, we’d love your feedback on the app question highlighted in bold above, to help guide us in making the app experience as useful as possible.
Take care, and lots of love,
- The Mindset Team