Ohmic Resistance and Logic gates

[Delta2]

Main Question or Discussion Point

Is ohmic resistance a "necessary evil" in order for transistors to be able to function as logic gates?

I mean, I have seen some circuits for the NAND gate in RTL and TTL and they both seem to involve ohmic resistances. Can we make a super conducting NAND gate that will have total ohmic resistance zero?

 

[Baluncore]

Can we make a super conducting NAND gate that will have total ohmic resistance zero?

Yes. But it would be a superconductor, not a semiconductor. It would also be slow to respond. The resistors used to make logic gates provide the current sources and sinks needed by the transistors to function as current switches.
https://en.wikipedia.org/wiki/Superconducting_computing

 

[KD5ZXG]

Transistors and gates are sometimes helpful, but not the only way to perform logic.
You might spend some time to look up "Adiabatic Quantum Flux Parametron", which
is a Superconducting Josephson Junction majority logic device. Distantly related to
Eichii Goto's much earlier Parametron Computer.

Those earlier devices made of LC resonant tanks excited by varying the parameter L.
Accomplished by saturating the inductor in such a way that it doesn't interact beyond
changing L. Wound to saturate at a 90 degree angle like a magamp, not as a forward
transformer. According to your profile, you already know electromagnetics. But you
might not have heard them used for oscillating computation. Parametrons are a good
read.

Well, the old LC version oscillated in one of two binary phase shift key'd phases related
to 1/2 the exciting clock. The quantum one, I'm not sure if it oscillates or is DC flux?
I think it inherited the name Parametron from Goto's team working on it. But aside from
both being gateless transistorless magnetic majority logics, they work quite differently.

At least now you know what to search for...

 

[KD5ZXG]

You already know NAND and NOR are universal gates that can make any other logic.
Now add to that universal list: MINORITY3. Minority being the inverse of Majority.

The losing vote of A, B, Zero is NAND. The losing vote of A, B, One is NOR
Therefore twice proved universal. That's the sort of logic Parametrons do.

 

[Delta2]

Ok so the answer to my question is that
"Yes we can make superconducting logic gates but not from transistors (which are semiconductor devices) but from devices that have superconducting Josephson junctions"
MINORITY3 is the name for a universal gate that uses superconducting josephson junctions?

 

[KD5ZXG]

Transistors, depending the type, have transresistance or transconductance or some messy mix of both.
Just how they amplify. At room temperature, I don't see any way around it. Even Josephson Junctions
have a small Voltage drop before they superconduct. Voltage * Current = Power. Nothing is free.
I guarantee that running the fridge isn't free either...

And a lot of the energy lost in logic gates has nothing to do with that. But Joules stored in capacitance,
dumped, and recharged. If lines are longer than 1/8 wave, they need terminations on both ends. You
might wrangle a gate that superconducts, transistors or Josephsons, but what of those other losses?

Majority is the proper name of an often forgotten logic. Minority is it's inverse. Minority is universal.
Majority might not be universal, but still extremely useful. The Majority of A,B,C computes Carry.
The Majority of A,B,C,/Carry,/Carry then gives us addition's Sum with a delay of just two.

Most Majority logic devices offer differential inputs and outputs. You will usually have both options.
Twist a differential pair between devices to invert, thus inverters are never needed. This is not unique
feature of oscillating Parametrons. For example: plain old transistor ECL gates that don't oscillate
were differential and could use the same wiring trick. AND, OR, NAND, NOR, all from the same gate.

Full addition using ordinary familiar gates takes somewhere around eleven devices, with a delay
of five is it? To be fair, a DPDT relay is another forgotten gate that can do full addition just one
device, also with a delay of two. So, Majority gates may be great, but aren't all that amazing...

Don't get me to lying about how Josephson junctions work to perform logic, only that they do. There
are plenty of gatelike ways to use a Josephson that they are not limited only to use as Parametrons.
I've read plenty, but can't say I understand enough to be explaining them yet. Ask me about old LC
tank logic. Search out Hitachi's Hipac, a computer built of shirt buttons. OK, so those buttons were
magamps. The variable L parameter would pump energy into in an oscillating LC tank, like a child
on a swing. Information stored in which direction the child was initially pushed.

Not all Majority logic devices are gates. When I say Parametrons are not gates, I mean they typically
have four I/O ports, and no direction. When active, they just amplify whatever small signal might be
present to full scale and hold that state for as long as powered. Changing inputs after the "vote" has
no effect. The device has to be powered off to forget. Note: two or six ports are also common.

To make logic flow between parametrons, a wave of three clocks push (actually pull) it forward.
Devices of a phase group shut down to prepare for a new vote, and then new votes will flow in as
power comes back on. Shutting down an "output" to prepare it also stops it from messing up the
previous group's vote. Input vs output is decided by only by clock timing. These clocks are not to
be confused with the much higher frequency clock that pumps and phase locks each LC tank,
but instead are periods when this pumping activity ceases.

Oscillation offers the possibility of zero voltage and/or zero current switching at the zero crossing.
Thus Parametrons could work around the energy lost when transistors must dump capacitance.
Ordinary (non-majority) logic that oscillates and always moves in one direction like a gate could
potentially do the same, its just that no examples come to mind.

 

[KD5ZXG]

Just re-read my above post. Even I can't make easy sense of it.
Perhaps too many unfamiliar topics have been strung together
without adequate context.

Would be more understandable with commented drawings. I've
already done so at great length in another thread. Far too much
to copy it all here.

https://hardforum.com/threads/gateless-majority-logic.1954426/

 

[Baluncore]

Is ohmic resistance a "necessary evil" in order for transistors to be able to function as logic gates?

It depends on the type of transistor. BJTs, Yes. FETs, No.
You are new to switching logic circuitry. Now is not the time to be distracted by esoteric technologies such as superconducting logic gates.

It is not necessary to use superconducting gates to have low power. You may notice that CMOS logic uses very few resistors and uses very little power. That is because the resistors are replaced by complementary transistors that source or sink current only when required. When the inputs are static, the gate current consumption can be turned completely off. Current is still needed to charge and discharge the input capacitance when there is a change of inputs. Using lower speeds and lower voltages reduces power consumption.

Take a look at CMOS logic to see why it uses such low power; https://en.wikipedia.org/wiki/CMOS

 

[KD5ZXG]

Gates and switches are neither the only or fastest way to do logic.
Well, maybe they are best for now, but as things get smaller...

Completely normal computers have been made without switching logic.
Once you get your head around a few of the alternatives, transistors
become optional glue of convenience, not so much a necessity.

Its a blind spot like the first time you ever saw a Guanella balun
and couldn't get past thinking the short circuit had to be a mistake.
Only later you are grateful to know such a useful thing, even if its
not for every situation.

 

[Baluncore]

Completely normal computers have been made without them.

How can a "completely normal computer" not have logic gates made from transistors?
How do you define normal?

Yet we still have a huge blind spot for the alternatives.

You are speaking for yourself there. Some of us know the alternatives and reject them immediately for quite valid reasons.

Engineers are risk averse, and provide reliable solutions by using available technology, today.
Conservative engineers are still employed tomorrow.

 

[KD5ZXG]

I define normal as NEC once sold them for general purpose use like any other.
A brochure exists in English if you like. Visit my other thread for attachments.

They used logic, just not in the form of gates. And definately had no transistor.
Accomplished logic with Parametrons, therefore gates weren't necessary.
Switching diodes weren't needed either...

A manageable small number of vacuum tubes served as clock oscillators.
Correction: There might also have been a few more used in interfaces to
other systems, such as the core memory. Some could have been gates.

I define smaller and faster as quite valid reasons. Upwards of 20GHz recently.
Of course, no tubes or shirt buttons this time. Quantum has new purpose for
some of this almost forgotten tech...

What are you talking about? Conservative engineers get laid off all the time.
They tend to shave the top end of the pay scale and keep only the starvation
wage know-nothings. I'm one of the latter.