|
Yesterday, 22
March 2010, John Dore told me that USB has been
using bidirectional signalling down one twisted pair for some years. This
means that the section below, which I added to http://www.ivorcatt.co.uk/x0357.htm
yesterday, means that I was waiting for something to happen which had
already happened.
Ivor Catt
But
today, 23 March 2010, on the phone with John again, I get the impression
that USB has protocol for whether the next period of time will be occupied
with signals in the one direction or in other direction. If this is true,
then it makes my point forcefully. It will mean that lack of competence in
electromagnetic theory worldwide has led to twice as many wires being used
as is necessary (to achieve the same bit rate. But see “The Dream”, below.
I want less wires for other reasons than bit
rate).
If it is really
true, as I have thought, that the concept of sending signals down one (pair
of) wire(s) in both directions at the same time does not exist, then I had
better give more information on the design of the necessary circuits. I
would not be happy for someone else in the future to patent the idea. I shall be grateful if
those reading this take a copy and annotate the date, as protection against
exploitation by a patent later in time. I think that it would not be fair if someone read this
idea of mine and then made a lot of money. I am not against profiting from
patents. Sinclair
gave me £500,000 in money and stock options for my patent on WSI. Followed by The Kernel Machine.
Here is
a broad brush analysis of the design.
Since
signals are being sent in both directions, the 50 ohm line has to be
terminated at both ends. This means that more power is required to send the
signals.
At our
end we have a drive circuit which delivers the voltage step into the signal
line. The source is a current source which will result in a five vole amplitude
across two 50 ohm resistors in parallel, the line terminations at both
ends.
We also
have a receiving circuit. We deliver the signal we are sending (5v or 0v)
to this circuit, and put it on one side of a long tailed pair. The voltage
across our 50 ohm terminating resistor is put on the other side of the long
tailed pair. The result is that our signal is on both sides, and so cancels
out, while the incoming signal will control the switching of the circuit.
Conceptually,
it is better to think in terms of a very long connecting cable. Then having
mastered the principles involved, gradually in your mind shorten the cable
down to very short. The situation remains the same.
The
dream, or objective, would be to interconnect two modules (for instance computer
and printer) with one wire only. Unfortunately this seems impossible,
because a signal travels down between two wires, not inside a wire. The
whole irritating business of having the correct interconnecting cable and
connectors would disappear. Two modules could be interconnnected using two
wires and four crocodile (alligator) clips. The characteristic impedance,
or spacing, between the wires in not critical, if one studies the formulae
since a logarithmic term is involved.
Wireless
is not the total answer, because of hacking and also lower bit rate is a
carrier frequency has to be used.
|
