On Jun 9, 6:16�am, John Davis wrote:
On 6/7/2010 5:49 PM, K6LHA wrote:

I think we are now on the same page.

We aren't even at the same bookshelf.

I'm not endorsing Microchip. They are very busy making new PICs of
all kinds and lots of other IC types. Also making lots of good
technical information on how to use them.

To use those, especially for the task of generating the control word
for an Analog Devices single-chip frequency synthesizer (serial or
parallel word, a choice) for any stable frequency from near-DC to 60
MHz, takes a DIFFERENT route than copying analog circuitry out of a
1960s QST project.

For example, Instruction Sets. I've got five different micro
instruction sets running around in my head from past microprocessor
adventures and Microchip has this RISC or Reduced Instruction Set
which is mostly different than any of the other five. Add to that at
least two dialects of Fortran and six more dialects of Basic, yet the
procedural creation of this software is the SAME as what I was doing
back in 1972. One needs to CONCENTRATE on the Instruction Set being
used and, hopefully, the source code editor will trap any typos or
wrong syntax. It becomes harder and harder as the semiconductors get
smaller and smaller. It is like trying to translate something while
riding a bus along a pot-holed street with lots of nosy riders. It
can be done but it takes CONCENTRATION.

But .. To play around at the component level in today's VLSI world.. You
need a clean room, lasers and things to control them.

NO. A "foundry" is NOT needed. Not even if you are only doing
digital stuff.

"VLSI" is just more of the same of "LSI" and that was just more than
medium-scale integrated. Ever try a "STATE MACHINE" project? One
could do that using an EPROM. The essential part of an Apple ][+
floppy disk controller used one and Wozniak did his all by hisself.
Fascinating.

How about a Logarithmic Detector such as Analog Devices makes (several
models available but they don't hand out "samples" readily now).
Analog thing, just a chain of successive-detection blocks inside,
designed so that the summation of their outputs has a logarithmic
function of the input. Can make a very high dynamic range spectrum
analyzer.

If you think you NEED foundry facilities, then read Hans Camenzind's
small book on "Designing Analog Chips," a wonderful look into the
basic guts of doing just that, complete with all the specialized info
on junctions and hanging them together. He writes very well. He made
IC masks from the start, cutting Rubylith by hand. Designed the most-
sold IC of the analog kind, the 555.

73, Len K6LHA