[vcf-midatlantic] vacuum tube computers, was Re: Museum Report 2017-11-26 November 26, 2017

Dave Wade dave.g4ugm at gmail.com
Mon Nov 27 09:33:58 EST 2017



> -----Original Message-----
> From: vcf-midatlantic [mailto:vcf-midatlantic-
> bounces at lists.vintagecomputerfederation.org] On Behalf Of David
> Gesswein via vcf-midatlantic
> Sent: 27 November 2017 12:55
> To: Jeffrey Brace via vcf-midatlantic <vcf-
> midatlantic at lists.vintagecomputerfederation.org>
> Cc: David Gesswein <djg at pdp8online.com>
> Subject: Re: [vcf-midatlantic] vacuum tube computers, was Re: Museum
> Report 2017-11-26 November 26, 2017
> 
> On Sun, Nov 26, 2017 at 10:34:27PM -0500, Jeffrey Brace via
vcf-midatlantic
> wrote:
> > >
> > > I'll jump in since I do have some interest in the education side
> > > though this isn't one of my better areas of expertise. Others will
> > > probably correct.
> > >
> > >
> > Thanks for the for the explanation David! Let's start with the most
> > common question that everyone asks: "How was this computer used?",
> > "What could you do with this computer?", "What was it capable of
doing?".
> >
> I don't know enough history on either of the specific computers we have to
> comment on how they were used. I can only comment on the general class.
> 
> Hopefully you can pull out what you need from this.
> 
> The what was is capable of is solving systems of differential equations.
> Since since teaching calculus probably won't go over well just having the
> equations for some problem and saying the equations represents the
> example (bouncing ball, car spring & shock absorber
> etc) may be all that can be done. The complexity of equations that can be
> solved  is limited by how many hardware block (integrators etc) the analog
> computer has.
> 
> http://chalkdustmagazine.com/features/analogue-computing-fun-
> differential-equations/
> https://www.nsa.gov/news-features/declassified-documents/tech-
> journals/assets/files/why-analog-computation.pdf
> 
> The power is that many useful real processes can be transformed into that
> type of equations.  Physical motion, chemical reactions, aircraft flight,
> 
> The analog computer can solve the equations standalone by setting the
initial
> conditions and then showing how they progress with time. For example the
> bouncing ball say you scaled 1V = 1 foot you start with 5V so ball dropped
> from 5 feet and can watch the height of the ball with time.
> 
> You could also feed in an analog signal representing stimulus such as for
the
> suspension example a signal representing the bumps in the road or for
flight
> gusts etc perturbing the flight.
> 
> Sometimes you just want the final state of the system so would let it
> progress until you get a stable answer and then read off the voltages and
> then convert them back to the original units. Other times you want to
watch
> what happens over time. You can scale time also so the time "running" does
> not have to be equal to the physical problem. For viewing the time history
an
> oscilloscope could be used and I think I remember a CRT in the Philbrick
so it
> likely was used like an oscilloscope.
> It was said that our Philbrick had a plotter. That directly generates a
hard copy
> graph of the value vs time. Some were chart recorders which can plot
> multiple values at once. Don't know what type we have.
> 
> The analog computer my mother worked with modeled the flexural modes
> of vibration and critical frequencies of complex mechanical systems like
ship
> hulls and shipboard machinery assemblies.

Can I say that analog computers continued in use for a long time because
they solve time domain problems in real time.
(or sometimes in scaled real time, the speed of the solution can be
increased when modelling long term processes, say glacial flow, or slowed
down, e.g. for some problems involving nuclear collisions)
This contrasts with digital solutions where the speed of calculation depends
very much on the complexity of the equations. 
So for example the very popular SPICE program which models electronic
circuits, essentially solves the same equations that an analog computer
solves to calculate voltages and currents in a circuit over time.
As it's a digital solver, it solves the equations at discrete points in
time. As it evaluates the equations it estimates the error and if it finds
large errors it will reduce the time step thus maintaining accuracy but
increasing the solution time...

An analog computer does not do this, making them popular for applications
such as flight simulation, where, if the pilot performs a sudden actions,
the last thing we need is to have to wait because the equations have become
complex...

Dave

 





More information about the vcf-midatlantic mailing list