Re: gEDA-user: Outsourcing PCB layout

[msokolov@xxxxxxxxxxxxxxx (Michael Sokolov)]

> Why not use RoHS chips?  You can use those with leaded solder, as long
> as they're not BGA or CSP.

Yeah, that's what I do when SnPb components aren't available, but if
they are available, I use them as a matter of principle.

My understanding is that if one uses RoHS parts, then tin whiskers can
still grow even if one uses good SnPb solder.  I am bothered by tin
whiskers because I believe that computing machinery should outlast the
average human lifespan and should be passed from father to son to
grandson.  The word "obsolete" is banned from my vocabulary.  That's
actually the root of my beef with RoHS: lead in electronic devices can't
cause any harm to the environment if those devices will never, ever,
ever be thrown out, so the whole RoHS concept is fundamentally based on
the premise that electronic devices will be tossed at most a few years
after they are made because they'll be "obsolete".  That ideology is
anathema to the core of my being, I still use 1970s computing technology
and plan on continuing to use it well past year 2038, meaning that some
time soon I'll have to change time_t from a signed to an unsigned long.
(Extending it to 64 bits would break too many things I'm afraid.)

> Double-check the SRAM timing; 70ns seems too slow for a 25 MHz cpu.

Well, the M68K bus protocol gives the target device two clock periods
(80 ns at 25 MHz) to respond with the data when set to 0 wait states,
but then of course there are setup and hold times, propagation delays,
transmission line effects and other "real hardware" stuff that makes my
software&logic-minded head hurt...

I actually plan on populating a 16 MHz MC68302 part on the first board
because that's what I already have on hand and because it'll be plenty
fast for Flavor B operation, and even if I were to run it at 25 MHz and
the SRAM is too slow, I could program it for 1 extra wait state, but you
are right, eventually I will want to run it at 25 MHz with 0 wait states
to get the full performance for the on-the-fly protocol converter mode
of operation, and at that point 70 ns *might* turn out to be just a tad
too slow when all the annoying "real HW" factors are figured in.  I
already have 70 ns SRAM parts on hand, but I should check to see if I
can get faster ones in the same footprint.  5V SRAM parts are a problem
unfortunately...

> Put two 0603's between circuit and chassis ground.  For my furnace
> controller, I populated these with a 10M resistor and a 0.1uF
> capacitor.

Thanks for the advice.  So this arrangement is better than having the
two grounds connected with a 0 ohm resistor or disconnected altogether?
What's the underlying rationale?

And I'm still looking for someone to do the layout for a price I can
afford...  (So far I have no clue as to what price range to expect, not
even an order-of-magnitude estimate.)

MS


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user