[FoRK] Cheap Electronics Dissection Project

[Kragen Javier Sitaker]

On Sun, 17 Sep 2006 19:12:57 +0200, Dave Long wrote:
> This is an interesting approach for a dot clock: use the SPI interface  
> to serially push pixels out, yielding 8 pixels per load and allowing  
> one to "unroll" the display update.
> 
> http://www.serasidis.gr/circuits/AVR_VGA/avr_vga.htm

That's a pretty nifty idea --- I had been wondering if I could make
that work, in particular for audio output.  (I can't find my notes on
that at the moment, though.)

I was just reading about how Steve Wozniak used a similar approach for
his TV terminal and for the Apple ][ color video; from
http://www.woz.org/letters/general/91.html:

    My only affordable output device was my Sears TV because it was
    free. Going to characters on the screen from a game on the screen
    was a natural step. I already had the TV timing circuits minimized
    and just had to use the horizontal and vertical counters to drive
    some memory with the characters stored. I chose what I figured was
    the absolute least cost memories, some dynamic shift
    registers. They were old (PMOS) and came in tiny 8-pin packages,
    and were half the size of 14-pin chips. My design goal was to have
    the fewest chips in the end. More correctly it was to have the
    smallest board space used. So these tiny serial shift registers
    were the winners for me. Some low level engineers or designers
    might be scared off by dynamic parts, but I just thought it all
    out and made sure that they never stopped shifting.

"Dynamic" here means that the memory circuit was astable --- if you
let it just sit there for a couple of propagation delays, it would
forget everything.  In Jeri Ellsworth's lecture last year about
reverse-engineering old hardware (presumably in this case she meant
the 6502), she had a few words about how this sort of thing is kind of
a pain to reverse-engineer --- you can't single-step it!

    But a VERY significant role was how my hardware BREAKOUT
    experience at Atari led me to a unique and creative goal for my
    Apple ][. I'd designed this machine to have color, which was
    totally based on an idea that I'd had one night working on
    BREAKOUT over at Atari. The idea was to use digital chips, or more
    accurately a single chip, to rotate a code around at a multiple of
    the color subcarrier rate of NTSC TV signals. From 16 starting
    codes, you got totally different digital patterns out of this
    shift register. If it was synchronized with the TV via the color
    burst technique of NTSC then it would be 'similar to' true color
    info. I was pleasantly amazed on the Apple ][ when it worked.

That doesn't sound too hard, does it?  Although that only gives you 16
colors.

Some of this information was also in the Woz "Founders At Work" page:
http://foundersatwork.com/stevewozniak.html

> Better yet, the following sites show that despite a small LCD display,  
> a machine resembling a glorified Speak'n'spell with a built-in BASIC  
> can be compelling for (at least the geekier elements of) the videogame  
> generations:
> 
> http://rasterweb.net/raster/computers/vtech.html
> http://elevate.wordpress.com/2006/03/12/first-computer-program/
> 
> The company that made these is still around, and currently offers a  
> "vtech Genius Notebook"*, with a ~64x32 bitmap LCD, for $30.  Looks  
> like they have devices with screens about half the resolution at $15.

Awesome!

> BASIC no longer seems to be included, perhaps due to lack of demand (or  
> lack of desire to field the support calls?), but considering that they  
> also offer manufacturing services:
> 
> http://www.vtechcms.com/
> 
> they might be perfectly happy to build cheap programmable boxes if  
> someone else were willing to shoulder the market risk.

That sounds pretty cool.  Maybe they can teach me a thing or two.