New Crystals To speed up my mac

[Mark Glauser]

I have heard about over-driving your PC by putting in a slightly
faster crystal. Is extra heat build up the major concern? I
realize that I can't replace my 16 mhz crystal with a 100 mhz.
But what else should I consider when doing this. I am planning
to speed up my MAC LC (16 mhz) to 20 mhz. Is this the limit, I
suspect that the processor is the limiting factor, am I correct?

My News Reader is really flakey, so I am using my advisors account.
please send any non-net replies to sto...@bart2.larc.nasa.gov

Thanks,
Sheldon

PS is there any interest in discussing vacuum tube designs out there
(seriously, I'm not joking)?

[John F. Woods]

gla...@sun.soe.clarkson.edu (Mark Glauser) writes:
>I have heard about over-driving your PC by putting in a slightly
>faster crystal. Is extra heat build up the major concern? I
>realize that I can't replace my 16 mhz crystal with a 100 mhz.
>But what else should I consider when doing this. I am planning
>to speed up my MAC LC (16 mhz) to 20 mhz. Is this the limit, I
>suspect that the processor is the limiting factor, am I correct?

One of many. The support logic on the board also tends to be a limiting
factor; propagation delay through a gate array isn't going to change by
one picosecond just because you swapped crystals, which means that signals
which arrived just in time before would now be late. It is rare for ICs
to be exactly as slow as the spec sheet says, but competant designers* must
assume that to be the case**, so you usually have some leeway -- but how much
leeway depends on how close to the edge the design was, what the shape of the
speed distribution for the manufacturers' processes, and even what manufacturer
of ICs the computer vendor was using that week.

Heat usually isn't a major consideration (though there was one Mac model
released early this year with a 68LC040 (no FPU) which could be upgraded
to a full 68040 by just plugging it in; it didn't require a heat sink, unlike
the equivalent full-power Mac model, because the '040 at 20MHz doesn't need
a heat sink while at 25MHz (I think) it does).

* Hence those readers with PC clones may ignore this part :-).
** Unless they're willing to convince Manufacturing to 100% screen all incoming
parts for meeting the requirement; better check that with the actual part
manufacturer to make sure they don't change their process in a way that cuts
down on the number of fast parts. For low-volume products, this may make
sense; for high-volume products, it would be better to try to get the
manufacturer to speed-test parts and send you the fast ones.

> PS is there any interest in discussing vacuum tube designs out there
> (seriously, I'm not joking)?

Sometime I have to get around to wiring up that magic-eye tube I've got as
a loadaverage meter.

[Aaron Lung]

In article <glauser....@sun.soe.clarkson.edu> gla...@sun.soe.clarkson.edu (Mark Glauser) writes:
>I have heard about over-driving your PC by putting in a slightly
>faster crystal. Is extra heat build up the major concern? I

Probably not...

>realize that I can't replace my 16 mhz crystal with a 100 mhz.
>But what else should I consider when doing this. I am planning
>to speed up my MAC LC (16 mhz) to 20 mhz. Is this the limit, I
>suspect that the processor is the limiting factor, am I correct?

197z
Depending on the components used in your computer, the overall design
may be the limiting factor. By increasing the clock rate of the
processor, signal timing usually changes by the same amount. Whether
or not the rest of the logic on the board is designed to run at a
faster rate will determine whether your attempt to speed things up is
successful.

The only way to really figure this out is to get a hold of a set of
schematics/databooks and do a timing analysis on the whole thing. :-(
Or try a faster crystal and see how far you can go empirically.

aaron