|Re: Nothing much|
Message #1 Posted by Les Bell (Sydney) on 27 Aug 2002, 11:17 a.m.
Steve Borowsky wrote:
"This is good to know, but i'd have to know more about the specific models
to tell whether they would have been a good or a bad influence on the
acceptance of RPN. The NS models for example were so crappy,
comaparatively, that they may have had a negative influence".
I remember the old Sinclair Scientific, from around 1974 or so. I used to
build the earlier Sinclair calcs from kits, and then typically sell them
that evening to the first person who saw me use one in the bar where I
worked during student vacations. So when the Scientific came out, for
GBP14.95 in kit form, of course, I had to have one!
Trouble was, after I built it, it wouldn't work. I had no schematics, of
course, but probing around with a scope, all I could see was DC everywhere
- there was no clock running. I remember walking back to the university
Hall of Residence where I lived, and despondently fishing the calc out of
my parka pocket, only to discover it now worked - for a few minutes.
A little experimenting revealed that the calc would work for a few minutes
after being cooled down in the icebox of the fridge next to my room - so
for a few days, I used it like that to do Bode/Nyquist calculations: grab
the calc from the freezer, work the numbers like crazy until it stopped,
and then back into the icebox!
Eventually, a phone call to Sinclair revealed that TI had supplied a faulty
batch of chips, and they would rework the calc if I sent it back, which I
However, the Sinclair's implementation of RPN (three level stack, from
memory) was fatally flawed by the absence of an X<->Y key, so that
intermediate results had to be written down. Nonetheless, it beat the
*crap* out of using a slide rule, so I persevered, until the bar job during
the following Easter break provided the funds to buy an HP-45.
--- Les Bell, CISSP