Hello from Oakland, California!
It’s been just over a month since we last wrote. This week, we’ve gotten a few gentle, friendly reminders that we’ve been a little quiet and should be sending out updates more frequently. Please keep doing that.
When you’re hard at work on a project and the main thing you have to report is that things are going more slowly than you’d hoped and you haven’t hit a big milestone, it’s very, very easy to talk yourself into waiting just a bit longer before sending out an update. That is, of course, exactly the wrong thing to do.
In this update: where we are with manufacturing, engineering progress and challenges, schedule estimates, some photos from Maker Faire, and SF Bay area office hours.
We have a half-dozen quotes in hand from potential manufacturing partners. A few more are trickling in. We’ve turned down three quotes from manufacturers that are very clearly not the right partners. Right now, we are still considering bids from a Taiwanese company that manufactures in Taiwan, a Taiwanese company that manufactures in Shenzhen, and a Chinese company that manufactures in Shenzhen. One of the quotes we’re waiting on is from Flextronics, who’d be manufacturing in North America.
The good news is that most of the quotes have been in line with the pricing we’d expected. (They’ve actually been a little better than we’d expected, which was a very pleasant surprise.)
Since we’re still negotiating, we don’t really want to say too much about the quotes or the process just yet, but we’re hoping to talk about it in a fair bit of detail in a future update.
Over the past month, we’ve been hard at work on engineering for the Model 01. We’ve brought in a couple of folks on contract to help out with both the mechanical and electrical engineering.
As we’ve mentioned in previous updates, we learned a lot about the Model 01 as we watched nearly a thousand of you interact with it for the first time during our Kickstarter road trip. On our trip to China in August, we learned a ton about what we need to do to make the Model 01 easy to manufacture. Over the past month, we’ve started to update the electrical and mechanical designs based on all that learning.
You can find a mostly-up-to-date set of notes on the changes we’re making here: https://github.com/keyboardio/Model01/blob/master/design_changes.md
We’re making progress, but it hasn’t exactly been smooth sailing.
On the electrical front, we’re combining the keyswitch boards and LED boards, as well as swapping out the LEDs, IO expanders and some of the USB support circuitry. We’re also adding header pins (or a place you could solder header pins) to the left-hand PCB to give you access to all of the pins you’d ordinarily find on an Arduino Leonardo. We’re adding a hardware interlock inside the Model 01 you’ll be able to use to prevent firmware updates over USB. It looks like we’re going to be able to switch the Model 01’s USB jack to Type C, the new reversible design that Apple and Google have started pushing.
We’re bumping up the fuse on the keyboard so that you could make the Model 01’s LEDs draw more current from your (over-spec) USB port. (We continue to limit power draw in software). We’re also moving the LEDs from the centers of the keyswitches to the lower edge of the keyswitches. Right now, the sliders in the keyswitches block most of the light the LEDs are putting out. We’re currently testing APA102C LEDs to replace the WS2812B LEDs we used on the Kickstarter prototypes. Right now, the biggest worry about the LEDs is that by mounting them underneath the keyswitches, they’d be hard to repair or replace. That’s mainly an issue during assembly–if an LED fails after the keys are soldered onto the keyboard, it’ll be relatively difficult and costly to swap out. A better design would be to mount the LEDs upside down on the bottom side of the PCB, but upside-down mount RGB LEDs are few and far-between. As yet, we haven’t found even a single reasonable option.
One other source of frustration has been IO Expanders / Key scanning engines. These are the chips that tell the Model 01 which keys you’re pressing and when you release them. They’re supposed to offload all the hard work of watching for keypresses from the main microcontroller. That means that the code on the microcontroller can be much simpler and will be much easier for you to modify without making the keyboard miss keypresses. The Semtech SX1509 chip we used for the Kickstarter prototypes claimed to be able to do all the key scanning we need. As we dug into it, we discovered that it detected key press events just great and…didn’t actually report key release events. We were kind of bummed out that the manufacturer didn’t understand why we might want to know when keys got released.
After a bunch of digging around, we found a similar part from ISSI, the 31IO7326. This chip claimed to do everything we needed, including reporting those all-important key release events. As an added bonus, it accepted 5V power, so we were able to remove some support circuitry we’d needed for the 3.3V SX1509. Fast forward to last week, when Scott Perry, who’s been helping us with an Arduino library for the 31IO7326, discovered that the chip actually swallows key up events under some circumstances. We’re still working through the right solution, but the current leading contender is replacing the 31IO7326 with a pair of ATtiny microcontrollers. They’re essentially the simpler, lower power version of the ATmega32U4 microcontroller we’re using as the brains of the Model 01. The upsides of a design like this are that we’ll be able to implement exactly the NKRO-compatible key scanning engine we want and that the ATtiny chips are actually cheaper than the 31IO7326. The downsides are that we have a bit of extra engineering to do and that we’ll have the complexity of 3 chips with reprogrammable firmware inside the Model 01, rather than just one.
On the mechanical front, we’ve been working on tweaks to the butterfly shape, key layout, and keycap shapes. By massaging the butterfly’s wings, we should be able to make the Model 01 a bit narrower, a bit lighter and a bit more aesthetically pleasing. Right now, the outline of the keycaps and the shapes of the butterfly’s wings…have very little to do with each other. Where one is curvy, the other is angular and vice-versa. As soon as we have the new shape settled, we’ll let you have a peek.
We’ve discussed most of the key layout changes before, but as a recap:
- We’re tightening up the thumb arc to make the inner- and outer-most keys easier to reach for folks whose thumbs can’t stretch quite as far.
- We’re moving the palm key out and down a bit to make it easier to hit.
- We’re extending the Esc and Butterfly keys down to make them easier to hit with your thumbs, if you want to. (This change also eliminates a potentially fragile ‘tongue’ of wood on the enclosure.)
- We’re going to try out a slight reduction in front-to-back key spacing to make the number keys easier to hit for folks with small hands.
Most of the changes to the keycap shapes are to make them easier to type on. By reducing the sizes of the key tops, we’re able to dramatically reduce the typo rate and to improve comfort. (Microsoft made a similar change in the latest Surface Pro 4 Type Cover.) We’re also resculpting the thumb keys to make them more comfortable to press and to chord. Jesse’s been typing on our first round of updated keycaps for the past month and is much happier with them than with the keycaps you saw during Kickstarter. There are still improvements to be made, but, with the exception of the thumb keys, if we had to lock the keycap design down today, we’d be happy.
We still need to settle on a design for the Model 01’s feet and redesign the bars that connect the two halves of the keyboard. Right now, they’re still too hard to put together, fall apart too easily and are too loud when they get dropped.
The engineering task list is longer than we’d like it to be at this point, but we’re moving at a good clip and it’s really important to us that we deliver you a good keyboard.
As of this moment, we believe we are still on track for an April 2016 delivery, but we’ve definitely used up a lot of our engineering margin. Until we have a signed statement of work, we won’t have a firm realistic delivery date.
As we’ve been talking to manufacturers, the way we’ve explained our April delivery date is that it’s very, very important that we deliver a high-quality product and merely very important that we hit our April ship date. We want to ship you a high quality keyboard in April 2016, but if we’re going to have to choose one or the other, we’ll pick “high quality” over “April 2016.”
Some manufacturers have told us that the schedule’s tight, while others (with a fair bit of bravado) have asked us how early we’d like to deliver.
Two weeks ago, Kaia flew out to New York to show off the Model 01 at Maker Faire. Atmel, who make the microcontroller inside the Model 01, very generously hosted us in their booth.
Bay Area Office Hours
We’ve been fielding a fair bit of email from folks who live in the Bay Area and, for one reason or another, missed our roadshow event at Eventbrite. So we’re going to hold office hours.
Come hang out, play with the Model 01, and chat with us about keyboards!
Oct 13, 2015 - 5-7 pm, http://www.coffeebarsf.com/bryant-st in the Mission
Oct 20, 2015 - 5-7 pm, http://www.lindgrenscafe.com in Berkeley