Russ Nelson : chordite

Working on the Mark XIII keyboard. This one will have the keymodules connected together with a matrix of stiff-ish copper wires, covered over in polycapralone. The theory is that people will be able to change the shape of their keyboard by heating it up using a hot air blow dryer. That will soften the polycapralone, which gets quite soft, and won't hold up its own weight, hence the copper wires. The keyboard owner can push the keymodules around until they're in the perfect position for them. The copper wires will hold the shape until the polycapralone hardens, making the keyboard both flexible and stiff, soft and hard, as needed.

Here are a set of keymodules, soldered to a small PC board, with a brass strip used as a barrier against the polycapralone. Without the barrier, the polycapralone would stick to the keys and/or jam up the key movement.

Stuart writes to ask if you couldn't put a strap around the wrist to sense the movement of the tendons through the wrist, by way of making a keyless chordite.

I think that would be dangerous. The tendons run inside little tunnels past the carpal bones (hence the name carpal tunnels -- oh these doctors can be SO creative sometimes). Squish the tunnels and you can inflame them. If they get inflamed they can swell up which keeps them inflamed, etc etc etc. Bad news.

No, better to go looking in the muscles further up the arm for myoelectric signals.

But note that it can be more fatiguing to be moving your fingers in mid-air than to be pressing them against something. The Chordite is designed to put the keyswitches at the finger's neutral point, so that relaxation puts your fingers on the keyswitches. It then takes only a little bit of pressure and movement to press the key, but because the key has hysteresis, you can feel it when you have pressed the key.

Status report on the free chordites over in my '770' category. Readers of the main blog may be confused, but pyblosxom allows me to create multiple rss feeds. The 770 category gets syndicated on Planet Maemo. But if you only ever look in the Chordite category, you won't see it.

Several people have asked me what coding system is used for the Chordite. There is no standard chording system for chording keyboards, because they're all different. Some use two hands for chording. For example, Douglas Englebart's Augment used a chord keyboard and mouse combination. Hold a combination of buttons on the chord keyboard (five switches, one for each finger) and press one, two, or three mouse buttons. Or for example, the Twiddler uses a large set of buttons, three for each finger. Or I saw a fellow at a Linux World Expo who had a wrist-mounted keyboard that had four buttons under each fingertip and three under the thumb.

The Chordite has at least two possible codings. The one proposed by the inventor, John McKown, involves pressing buttons until you've gotten the correct combination of buttons, then releasing at least one button to generate that key. Another one is to use only combinations of two buttons, let's say A and B. The two of them generate four keys by changing the order in which you press and release A and B. I haven't (yet) experimented with both to see which works better.

Johan Helsingius suggests that I compare the Chordite and the Twiddler. There are two possible comparisons: between concepts and between implementations. So, the concept first.

The Twiddler uses several three keys for each finger, and two keys for your thumb. Various combinations of these keys generate keystrokes. The Chordite uses the thumb to hold the keyboard. Since the Chordite keys are pressed by the knuckles rather than fingertip, each finger may press one or two keys.

In both systems, each finger can generate two bits of information. Twiddler can press Left, Middle, Right, or no key. Chordite can press Distal, Medial, Distal and Medial, or no keys. However, the Twiddler includes two buttons for the thumb, so that makes the chordspace four times as large. The Chordite makes up for that by providing for sticky shift, control, and alt prefixes.

The implementations differ because the Twiddler 2 is a USB keyboard, and my Chordite is bluetooth. Since the Twiddler 3 will be bluetooth, there's no obvious benefit to the Chordite.

MJ Ray suggested that I explain how the Chordite differs or is similar to the CyKey. Well enough! Under "similar" I would put: chording, unfamiliar, pocketable, one handed, wireless, battery operated. Under "different", I would say that the CyKey must be used against a surface whereas the Chordite can be used in mid-air. The CyKey works well for either hand, whereas the Chordite is handed. The CyKey uses infrared, which restricts the positioning of the device, and which restricts the devices that may be used, whereas the Chordite uses Bluetooth, which is widely supported and doesn't require any special positioning. The CyKey's infrared is cheaper than bluetooth if you already have infrared, but is more expensive if you lack it. With the CyKey, you type with your fingertips, whereas with the Chordite, you type with your knuckles. With the CyKey your thumb moves from one key to another, whereas with the Chordite, your thumb holds the keyboard and the other fingers simply move up and down; no hunting for keys. Finally, the CyKey is an off-the-shelf product whereas a Chordite (currently) must be custom ordered to fit your hand.

This keyboard is finally salable. It has worthy electronics which will give you a nice long battery life. It's sturdy. It's replicable on a reasonable basis. Right now I'm still building it on a custom basis, but I have some ideas for how to fit people with different size hands.

Unfortunately, I haven't been able to get past the custom firmware hurdle. I'm using a driver on the Linux side which pulls in seven different keys and implements the chording algorithm. It works well enough, but it's restrictive to need a driver. Can't sell into the Windows or Mac market without modified firmware. Blue Packet has offered to modify the firmware for a stiff price. Unfortunately, that requires a larger committment than I can put forth given my current lack of understanding about how to fit multiple people.

Plus, not only is the fit a problem, but everybody (everybody, everybody) thinks it's hard to learn how to use. It isn't, because the most common keystrokes are also the easiest ones to make. Given the cheatsheet, you can type your name within five minutes. It's really not that hard, but it's so unfamiliar to people that it looks hard.

Here's the front of the keyboard, folded for pocketing. Notice the classy 1/4" plywood and ground-off wood screws. This is for strength. Relative to the stresses on the keyboard, the 1/4" plywood is quite strong, and the hinges ensure that the wood meets up with a hard stop at the limit of its extension.

With the keyboard unfolded for use, you can see the whole wood and brass steampunk thing going on here. The previous keyboard fell apart in several ways. This one won't, not even if you throw it into a soft suitcase and take it on an international trip.

You can see how the upper piece of wood hangs off the knuckle of your first finger, and how your thumb rests on the top of the AAA battery box. The piece of wood at the bottom rests against the base of your palm, and provides one end of the lever that allows your finger knuckles to reliably press against the keys.

Here's how you grip the keyboard. The thumb only holds the keyboard. You press the keys with the knuckles of your four fingers.

Here's how it looks from the edge, or top, of your grip.

Thanks to advice from Mitch Sun of Blue Packet, I fixed the problem with the Mark XII electronics! I didn't realize that a pull-down on a diagnostic serial input was required. Soldered an additional 10K resistor onto the board, and it's working fine. "hcitool scan" prints this:
00:12:A1:61:0E:21 BluePacket Bluetooth Wireless Keyboard

Now to design a PC board appropriate for the product!

The Mark XII keyboard is as yet unbuilt. I'm happy enough with the Mark XI physical design. With a little bit more work, it's manufacturable, except for the electronics. For that, I need something which consumes less power, is smaller, can implement chording, and implements the HID profile. That perfectly describes the Broadcom BCM2042 bluetooth keyboard controller.

Unfortunately, that chip is a BGA chip, which is hard to work with. Fortunately, several companies implement Broadcom's 92042 bluetooth keyboard module. Blue Packet is one of them. They sell the 2042 on their BP20422 bluetooth keyboard module (to which I can't link directly, but it's off the drop box.) They haven't (yet) committed to modifying 8051 firmware with the necessary chording algorithm, but I'm confident that I can talk them into it. Talk, yes, and money. Unfortunately, the module on the PC board isn't working. I suspect that I toasted it while soldering it to the PC board.

Speaking of the PC board, my friend DJ Delorie very graciously answered my questions about how to use PCB, the Open Source PCB package. Then, he even volunteered to make the board for me. Ahhh, it's to nice to have college buddies. They make your life so much easier. This first cut at a board is just a footprint adapter. I wanted something from which to hang discretes, as well as have good access to the pins. Plus, Blue Packet hadn't yet sent me the schematics.

I have ordered another ten modules from Blue Packet. That should be enough to get a good prototype working, even without the chording firmware. I'll only be able to type 7 keys without chording, but I'll have all the electronics in order at that point.

Been sitting on this version of the keyboard, but I want to get it out there. This one is much smaller and stealthier. It also puts the keys on a flexible PC board. That's much more manufacturable than the previous versions which used the key's mounting straps. The keys flex a little along with the PC board, but that's not a big problem. Because the keyboard is cupped, that gives it much more strength than if it were flat. And the pressure from the fingers is always directed against the direction of the cupping, so physics is my friend here.

This version is just the flexible PC board, mounted in a chunk of shapelock. It doesn't pocket as well as the Mark X keyboard, whose thumb rest folded. But, it's sturdier, lighter and fits well in my hand. The back view shows how the big ugly electronics (hopefully the Mark XII electronics will be smaller), the palm support, and the way the thumb rests on the battery box. You can see the PC board just a bit here. In the back of the hand view, not much shows beyond the fingers.