[sdiy] The future of synth DIY

[Scott Bernardi]

How useful this process is would also depend on the type of transistors
that could be manufactured.  If they're talking about doing logic with
it, it's probably some form of MOS. Can bipolar transistors be made? I
don't think you can get a decent exponential response out of a MOS
transistor - it's more of a square law.
Will they really be able to shrink the size of the components? Somehow I
don't think you'd be able to match the resolution an X-ray lithography
machine with an ink jet printer. Still, It may be acceptable for analog.
The philosophy for designing IC circuits is very different than discrete
design. In discrete design, your most expensive devices are you
semiconductors, then capacitors, then resistors (high quality capacitors
can be more expensive than transistors). In IC design, "cost" is
determined by the amount of real estate a device takes, which makes
capacitors the most expensive, then resistors, and transistors the
cheapest. 
Tolerance is also an issue. Typically you can only get +/-20% on
resistors and capacitors. Where the IC process really shines however is
in matching between components - better than 0.1% if laid out right.

I'd settle for a readily available, cheap source of good npn and pnp
matched transistor arrays! Why is it that these always seem to go
obsolete? AD821, uA726, THAT co.... even the LM394 is hard to get
nowadays.

Martin Czech wrote:
> 
> :::    It would be a dream come true....
> 
> or a nightmare. Analog design for logic process is a pain in the a**.
> Even if you have a process more suitable for analog purposes
> you still need to fight down all kinds of problems you'll never
> encounter in discrete circuits. There is nothing like resistors
> from the shelf, capacitors, and neglegible parasitics.
> 
> Only advantage you have is matching, and that is a science for it;s
> own.
> 
> m.c.

-- 
Scott Bernardi
sbernardi at home.net