Re: TI-H: ti-turbo


[Prev][Next][Index][Thread]

Re: TI-H: ti-turbo



On Tue, 22 Oct 1996 18:35:35 EDT, you wrote:
>
>On Tue, 22 Oct 1996 07:34:05 EDT wesleymcgrew@juno.com (Robert W McGrew)
>writes:
>
>>I can't see why not, if they all have the z80 processor, then TI put=20
>>something in to slow 'em all down.
>
>This leads to the obvious question: Why did TI slow them down so much?
>
>This is the way I see it (please, bear with me... =3D)
>
>The Z80 is a chip that can run at almost any speed, assuming it doesn't
>get too hot or some other theoretical limit is reached.  (Same as with
>the 80x86/Pentium Intel Chips).  Now, from my experiences working with
>them whilst open, none of the chips ever got more than warm while doing
>intense calculations (A.k.a. Game Playing).  Therefore, since the 85 =
runs
>at 6 Mhz (and to my knowledge, minimal thermal trouble...) why not have
>it automatically sped up to say, 18 Mhz?  Surely TI would have noticed
>that normal graphing operations (Some o which take 10+ mins. including
>ZFit)  would be sped up immensively...
>
>Ok, I'm done.  If I'm missing something (or if anyone actually
>understands what I'm trying to say...) let me know.
>
>	- J. Bishop
>




That's the thing, though, theoretical limits ARE being reached here.
As the transistors in a chip get smaller (i.e. the sub-micron
processes that intel is now boasting), three things happen.  You can
fit more of them in the same space, they heat up less, and (most
importantly) they switch and transmitt signals faster.  Now, with the
Z80, I doubt they're using "cutting edge" manufacturing techniques, so
each transistor can only switch so fast.  Therefore, there's a
definite ceiling to the maximum speed you can clock it to.  If it's
too fast, certain paths and transistors won't be able to keep up, and
you'll have a big mess.  Heat was never really a problem on the older
chips, because the darn things were so big that you couldn't fit
enough of them on there to have generate a lot of heat.  However,
newer processors can have 15+ million transistors on a square
centemeter, and now heat is a BIG problem!


I really don't see why TI decided to go with an RC clock circuit.  A
crystal would have allowed for timer-dependant applications (like on
the HP-48's).  The only advantages to an RC clock is that it's
cheaper, and it consumes slightly less power when the batteries are
getting low.  Frankly, I just think TI's too cheap!


Really, though, who said that the TI-85's clock is dependant on the
battery life?  Are you guys sure?  There could EASILY be some sort of
TTL crystal clock circuit built into one of the prefab chips, one
that's dependant on an RC circuit for calibration only.  This would
make the clock relatively constant at any battery level.  If it's a
non-fundamental mode crystal, this could explain why changing C9 has
such a dramatic effect on the clock speed.


-Mel


<pre>
--
The TI Memory Expansion Homepage
http://pilot.msu.edu/user/tsaimelv/expander.htm
</pre>


References: