On 8 Nov 2000, Eric Smith wrote:
John Lawson wrote:
> ALSO: Aircraft compasses are generally of the flux-gate or flux-ring
[snip]
You've got me very curious. I'm familiar with flux-gate compasses which
do not use light, and with laser ring gyros which do (but, of course,
[snip]
You'd think after five months in India I'd get over the jet lag. ;]
Sorry for fluxing this up. I of course *meant* laser-ring gyrocompass.
But this has drifted off the edge of the topic horizon, and folks might
have little patience left for it. Private e-mails on the subject are
requested hereafter, if anyone wishes.
* * * * * * * * * * * * * * * * * * * * * * * * *
ObClassiccmp:
On the Hybrid Computer topic - since they are now quite rare, not many
of Us are familiar with just what the word 'Analog' refers to when
speaking of computing, thus the confusion over the role of the active
devices, especially tubes (valves). Actually it's pretty simple: Digital
computers operate strictly on numbers, and numbers alone. The allowed
states in a digital machine are 1/0, on/off, yes/no, asserted/negated,
etc. We then use patterns of these on/off states to stand for numbers,
characters, quantities, what have you. I'm sure I'm preaching to the
choir here.
On The Other Hand: Analog computing devices use varying and measurable
physical quantities to represent the data... i.e. if you need to know how
much a 10-gallon tank weighs having 3 gallons initially and adding 4
gallons, you set up devices (usually 'Operational Amplifiers') whose
parameters are related (are analogs of) these quantities... One Opamp can
have it's initial quantity at 3 volts for the starting gallons, another
can have a circuit configured to multiply the 'gallons' voltage by the
factor required to turn 'gallons of water' into 'pounds', and a third can
add the sum of these (plus, say, a fixed voltage of 40.5 volts to
represent the dry weight of the tank itself) and direct the output to a
meter, which reads from 0 -> 100 volts, but which we calibrate in
"pounds".
Now, having set up these amplifiers, we can use a knob with voltage on
it to 'fill' the tank, and watch the resulting 'weight' change on the
meter, in real time. This may sound cumbersome today, but consider in
that in the mid fifties, digital computers were very, very expensive,
fairly rare, and every single problem required much analysis and coding,
and the expense of the CPU Time alone was huge. BUT: any electrical
engineer worth his/her diploma could easily set up the equations to model
all but the most difficult of simulations, and with most large-scale
analog machines, could get meaningful and accurate results in a few hours
or days, as opposed to the well-known "code your problem, punch your
cards, submit your deck, wait for your print-out, debug your code,
re-punch your cards, submit your deck....repeat, repeat, repeat..."
Not to mention that analog circuit design was fully mature at that
point, and analog computers are relatively simple to design and construct,
having no 'specialized' devices like memory elements, digital mass storage
devices, electromechanical I/O, etc.
BTW: at least one list member with whom I corresponded was engaged in
building his own modern analog computer, using today's advanced
components. I wonder how that worked out? I used mine, and a friend uses
his several large analog machines, in conjunction with our older
electronic music synthesizers, to generate new and strange sounds based
on obscure mathematical regimes...
Cheers
John
PS: Anyone *really* interested in getting into analog computing might do
well to consider scarfing up a Heathkit EC-1 when possible.. they are
simple, small and portable, and have easily replaceable tubes, and having
(IIRC) 9 opamps, one can get quite elaborate with them.