5 May
1998
5 May
'98
5:33 a.m.
I have just returned after a long weekend to find my mail had got set to
Postpone again. This has probably all been said but I thought I'd put
my bit in anyway.
Tony:
< A question occurred to me today : Can you have an embedded analogue
< computer, and if so, how many op-amps are needed to have one?
Allison
Yes, and that's very common. None, a low pass
filter(RC) performs a
function and can be considered analogue.
A filter made of passive components or otherwise I wouldn't consider a
computer, although I agree it is definitely analogue (Americans may omit
the ue where appropriate).
I would say that an analogue computer:
(a) combines two or more signals
(b) does so in a more complex way than by simple addition or
subtraction. (But this could be A+dB/dt, for example)
An analogue computer need not be electrical at all - quite complex
analogue functions can be implemented in cams, for example. A good
example of a simple embedded analogue computer is the ignition
distributor on a petrol engine. This:
Takes two inputs - camshaft angle and manifold vacuum;
Differentiates camshaft angle to get engine speed (centrifugal
weights on springs);
Applies some non linear function to engine speed (cams attached to
the centrifugal weights);
Adds together camshaft angle, function(engine speed) and constant *
vacuum level;
Compares the result with a reference angle to generate pulses of a
given width for the ignition.
I claim that is a simple (but actually quite sophisticated) analogue
computer.
< I was looking at the service manual for my Micropolis 1203 hard disk, and
< I read the circuit description of the servo electronics. It's a fairly
< complicated array of op-amps, which combine integral and differential
< forms of the position error, positioner current, etc. I would claim that
< is an embedded analogue computer.
Valid claim, also a good example of a fairly complex
function.
Agreed 100% (As Tony would say). An excellent example of an embedded
analogue computer.
Another example is the convergence circuit in a colo(u)r television.
This takes the two timebases as inputs, multiplies them and their
squares/ first derivatives etc. by user settable constants, and feeds
this back onto the deflection systems of the tube.
< On the other hand, I think it would be stretching the definition to call
< a simple op-amp wired as a voltage follower an analogue computer.
Correct. However often the buffer is between some
function or follows one
so it's part of the analog system.
Agreed it could be part of an analogue computer. But I think Tony's
point was that it does not by itself make one. Otherwise practically
any analogue circuit becomes a computer (One of Vonada's axioms, I
think: All circuits are amplifiers)
Other analog systems common to computers:
cassette IO (low pass filter on output) and complex filter/differentiator
edge/peak detectors for input. Some of the acients used PLLs for clock
recovery (KANSAS City is one).
Analog to digital conversion (quantification).
Digtial to analog conversion (filtering)
Disk/tape systems have several layers of analog function for data and
control.
Definitely analogue systems. But some are merely filters, not
computers. (I think a PLL almost qualifies as a computer, though...)
But in general, I agree with you both - analogue computers are often
small, simple and embedded, and they're a heck of a lot more common than
most people think.
Philip.