9999 in 1 Dissection

This game is nearly identical to the smaller handheld Brick Game, but it doesn't have any of the same trademarks, and the gameplay is sufficiently different that I suspect it comes from a different company, rather than simply being an earlier version of the same hardware.

However, the display and keyboard are laid out almost identically, and the construction is very similar. The shape is very different, fitting more comfortably in my hands.

In many ways, this hardware is more appealing for rebraining than the other alternatives: it's cheaper, the display is larger, the case is sturdier and has lots more free space inside, and rebraining involves cleanly replacing an entire board with a few cables and zebra strips running to it, rather than drilling a hole in an existing PCB or chipping epoxy and the chip off of one.

It has a small 8-ohm speaker in the back; I think it might benefit from having an earphone jack instead.

I paid $5 for it, but a vendor at a nearby stall offers them for $4.




Keyboard

The keyboard is fabricated on a single large single-sided PCB, much like the crappy calculator, but with fewer keys, each one fabricated in a separate rubber plunger, and with a hard plastic key on top of it. The keyboard PCB layout is totally irrational; there are non-plated-through vias drilled, apparently, specifically to disconnect traces that would otherwise be shorted together; there are traces that run off the edge of the board; there are two Pause/Start buttons wired in parallel (reducing the key count from ten to nine); and the six wires of the ribbon cable are soldered through six of eight holes (although a seventh has a blob of solder on it, just for fun.)

The hard plastic keys improve the tactile feel (and, I suspect, the durability) a bit over the pure-rubber approach, but they still don't have the snappy feel of the discrete buttons in the FM radio.

The layout also isn't the 3x3 layout you'd expect for a six-wire cable running to a nine-key keyboard; that wouldn't be planar, and would require either a jumper wire or a two-sided layout, like the Brick Game's keyboard had.

The pin assignments are as follows:

Pin 1
on/off, down, left
Pin 2
up, down, pa/st
Pin 3
mute, pa/st, left, right
Pin 4
mute, fire
Pin 5
reset, fire, up, right, on/off
Pin 6
reset



Screen

The screen is connected via zebra strips, as usual; four screws compress the compute board against the zebra strips, with 22 contacts on one side and 21 on the other, if I counted correctly. It's virtually identical to the Brick Game screen, except that it has a little icon of a cup of hot coffee when the game is paused, and it has 16 pixels for next-piece display instead of 8.

Implications

The clever planar keyboard layout opened my mind to new keyboard possibilities. Deciphering this layout taught me that N wires isn't limited to decoding an (N/2) x (N/2) keyboard — no, you can go all the way to N(N-1)/2 keys if you want (15, in the case of 6 wires; you'd need 8 wires for a whole alphabet). You just lose the ability to decode multiple simultaneous keypresses, and probably to timeshare your wires with e.g. a screen, as the crappy calculator apparently does, but the reduction in wire count might be worth it. (I think decoding multiple simultaneous keypresses reliably in the AxB grid layout requires diodes on the keys, anyway; I haven't figured out if the complete-graph keyboard can decode multiple keypresses reliably that way. It can, however, do N(N-1) keys that way.)

I could stick a computer inside this chassis with a minimum of trouble, have it remain quite inconspicuous, and probably have a very sturdy machine.