Re: LZ: Memory Expander

Subject: Re: LZ: Memory Expander
From: Mel Tsai <tsaimelv@pilot.msu.edu>
Date: Sun, 22 Sep 1996 22:38:38 GMT
In-Reply-To: <199609221613.QAA12511@lnd.com>

On Sun, 22 Sep 1996 16:11:50 +0000, you wrote:
>So far, the only type of serial expanders use one line for data, and=20
>the other for the clock.  Would it be too hard or slow to use both=20
>lines for data and have different pulse widths for 1 and 0??
>
>-Ed
>

This would be possible, but it would complicate the circuit in one of
two ways:

1.  The hard way would be to use a Phase Locked Loop circuit, and this
would generate the necessary timing after sending a few cycles of the
data.  However, unless we keep some sort of internal timing, this
would quickly fall apart.  Note that this is how most of the modern
"Cable TV descramblers" work.  After some of the video timing
information is sent, the rest is rebuilt by the PLL circuit and your
screen gets unscrambled.

2.  The easier way would be to externally generate the clock signals
necessary to determine whether the pulse width was a "1" or "0".  This
could probably done with a 555 one-shot timer.

It's a very good idea (and I wish it could be done), but unfortunately
it will be extremely difficult to implement.  Keeping a synchronous
connection is difficult to do with a direct bus connection, much less
2 data lines!!!  Also, the clock signal is necessary to work with the
serial flash expander, and it couldn't get any faster even if we had
10 data lines.

-Mel

References:

LZ: Memory Expander

From: Ed Plese, Jr. <eplese@lnd.com>