2 Jun
2007
2 Jun
'07
8:29 a.m.
Jules Richardson wrote:
Chuck Guzis wrote:
I'd have to say (based on experience) the repetition helps very much in
reducing the work effort.
Once you have figured out the repeating patterns (i.e. the standard gate
construction, flip-flop construction, etc.), you can in large measure stop
drawing at the discrete level and draw at the gate level. You still want to
check all the internal connections within that gate/repeating pattern for
something oddball, so yes, you are right that you do have to go through all
the connections, but you don't have to draw it all out. You do a visual
confirmation that the gate/whatever conforms to the repeating pattern and then
spend your time tracing and drawing the gate/whatever input and output circuits.
.. One can also think of it in terms of information content: the info content
of 1000 elements arrayed with a repeating pattern is less than that of a
purely random array of 1000 elements, and once you can do the compression
processing (recognising the patterns) in your head, communicating it over the
low speed link (drawing it on paper) takes less time. :)
I actually like RE'ing discrete-component digital logic precisely to find out
what the repeating patterns are (i.e. how the designers chose to implement
gates and flop-flops (in the discrete days there were other flip-flop
variations besides JK,SR and D)).
Even with analog circuits a lot of design characteristics standard to the
system cn become recognisable and make things easier.
On 1 Jun 2007 at 14:51, Jules Richardson wrote:
No - I'm saying that surely a 1000 transistor board with circuits that repeat
is no more or less difficult than a 1000 transistor board where nothing
repeats; the level of effort required is largely the same as it's tracing the
interconnects which takes the bulk of the time, and that's got to be done
fully for both boards to guarantee an accurate schematic. (All of which
probably means that the number of PCB layers is irrelevant too - the critical
thing is the number of solder points) Surely it's as much effort whether circuits
repeat or not? Don't you still
have to check all possible interconnects to make sure that you've got
everything traced right, even if a quick visual inspection might make it
appear that there's a lot of repetition?
So, you're saying that it's easier deciphering a PCB with a couple of
544-conductor house-numbered BGAs on it than working through a PCB
with 1000 discretes?