Resistor, Capacitor, Inductor, now the 4th element: the Memristor

democrates

Check it out
http://blog.wired.com/gadgets/2008/04/scientists-prov.html

According to R. Stanley Williams, one of four researchers at HP Labs' Information and Quantum Systems Lab who made the discovery, the most interesting characteristic of a memristor device is that it remembers the amount of charge that flows through it.
Indeed, Chua's original idea was that the resistance of a memristor would depend upon how much charge has gone through the device. In other words, you can flow the charge in one direction and the resistance will increase. If you push the charge in the opposite direction it will decrease. Put simply, the resistance of the devices at any point in time is a function of history of the device –- or how much charge went through it either forwards or backwards. That simple idea, now that it has been proven, will have profound effect on computing and computer science.

Well done HP.
The standard specs of max and mins of voltage and current in each polarity as well as limits of resistance and charge would be easy enough to deal with in terms of speccing reliable circuits, but how to model a resistor whose value is the sum of previous currents? I wonder could the memristor be modelled in terms of a resistors and capacitors, that would save a whole lot of effort.

CathalMc

More well done Chua I think myself. He gave an outstanding talk at a plenary session at ISSCC in Cork 3 years ago, a real towering intellect. He described the model of the memristor in his 71 paper as a mutator (an active circuit) in parallel with any one of a nonlinear R, L or C component - so it is not much intuitive help.

democrates

CathalMc wrote: »

More well done Chua I think myself. He gave an outstanding talk at a plenary session at ISSCC in Cork 3 years ago, a real towering intellect. He described the model of the memristor in his 71 paper as a mutator (an active circuit) in parallel with any one of a nonlinear R, L or C component - so it is not much intuitive help.

+1 for Chua in fairness ok Cathal, credit where it's due.
Bummer on my theory about modelling it with RC, revealing my years out of electronics there, and/or onset idiocy :mad:
My main interest in this will be the mooted analog applications, eg climate models etc, and the challenges posed for software and data storage - one for the software threads really after a dig around citeseer, looks like another spatio-temporal DB type problem area.
Meanwhile here's an EETimes take on it with more info:

The hold-up over the last 37 years, according to professor Chua, has been a misconception that has pervaded electronic circuit theory. That misconception is that the fundamental relationship in passive circuitry is between voltage and charge. What the researchers contend is that the fundamental relationship is actually between changes-in-voltage, or flux, and charge. Such is the insight that enabled HP to invent the memristor, according to Chua and Williams.

ressem

In fact, you can use any fab to make these things right now, but somebody also has to design the circuits and there's currently no memristor model.

The way memristors handle current and voltage is startlingly similar to the way synapses between brain cells do, says Chua. Both build up voltage to a threshold before firing and letting a current pass.

This'll keep the AI guys (and possibly the tabloids) busy for a bit. It'll be interesting to see whether enormous grids of artificial synapses built in labs with this are more easily trained to do recognition work than the smaller setups currently in existence.

democrates

ressem wrote: »

This'll keep the AI guys (and possibly the tabloids) busy for a bit. It'll be interesting to see whether enormous grids of artificial synapses built in labs with this are more easily trained to do recognition work than the smaller setups currently in existence.

Yes, no doubt there'll be great excitement for those chasing the AI touchstone. The moral dilemma with AI hinges on whether it is regarded as a sentient being. If it is, then there is the dilemma of keeping it prisoner versus putting humanity at risk by letting a superior intelligence (the goal surely) loose online, one for the philosophy forum I guess.

As for there being "no memristor model", from page 1 of the link in the first post:

Up until today, the circuit element had only been described in a series of mathematical equations written by Leon Chua, who in 1971 was an engineering student studying non-linear circuits. Chua knew the circuit element should exist -- he even accurately outlined its properties and how it would work. Unfortunately, neither he nor the rest of the engineering community could come up with a physical manifestation that matched his mathematical expression.

If the mathematical model for the circuit element is down, I presume they mean that this remains to be added to existing LRC model to yield the LRCM (or whatever letter will represent the memristor).

folan

read about this yesterday, very interesting stuff, and the implications of this are huge! I really cant wait to see this in action, really get to grips with its benifits!

Chua modelling this so long ago is amazing, and fair play to the HP Labs crew finally getting it! Something this important hasn't come from the Labs in a long time (if ever! this could be the most important development from there).

And after so many were giving out that the Labs had been made smaller, its great to see a win like this coming from the garage!

democrates

IEEE are selling the 1971 paper here if anyone finds an unencumbered copy please post a link.

A Wikipedia page has also begun and has some basic math (under 'physics'(!)):
"The memristor is an element in which the magnetic flux Φm is a function of the accumulated electric charge q in the device. The rate of change of flux with charge



is known as memristance. Here q is electrical charge and Φm is magnetic flux."
Way to remember "Defy em over d queue".

There's summer sorted for a lot of people

backboiler

As one of the comments on EE Times asks, what is the unit of memristance?
If Wikipedia is to be believed M = dphi/dq, so Weber/Coulomb, which is Volt-second/Amp-second, which is Volt/Amp, which is Ohm.
Now these HP lads and gang are obviously a lot more tuned in than me but shouldn't a new physical phenomenon at least require a new derived unit?

Michael Collins

backboiler wrote: »

...shouldn't a new physical phenomenon at least require a new derived unit?

In actual fact, no! The fact you described above is actually why it's called a memristor. The new part about it is that the instantenous resistance depends on the previous charge through it, the physical property that changes is in effect a resistance.

Take an RC circuit for example, if you work out the units of RC and see you'll see the product of them comes out as seconds. We don't invent a new unit for it just because it does something other circuits don't.

JoeB-

This all sounds very interesting... we should be able to build fantastic new circuits.. I am most excited by the fact that this works (and becomes more efficent) on the nano scale and so has the potential to provide for incredible miniturisation. (See link given by Democrates,
http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=207403521
great stuff, 3 pages in total)

I don't know a whole lot about electronics but is the transistor not a fundamental 'element' of circuitry?
What I mean is that people say three elements, resistor, capacitor and inductor are fundamental.. what about the transistor?

Or is it that the crucial point to realise is that a transistors operation can be duplicated exactly by some combination of the three basic elements, i.e resistors, capacitors and inductors?

So any usual circuit (without memristor elements) can be re-drawn wth only three elements, resistors, capacitors and inductors? Is this right? Is it always possible to reduce any circuit using some existing technique?

Obviously the memristor itself can't be represented by any combination of the other three basic elements, is that right?

What configuration of the three basic elements produces the transistor behaviour, (i.e switching?)

Surely also light emitting diodes (or simply diodes themselves) are also basic elements??? It can't be possible to duplicate these using the 'three', surely???

Cheers
Joe

JoeB-

Hi

I was thinking more about this and I was thinking of super conductors...

Then I realised that a superconductor is simply a perfect wire... i.e in all circuit diagrams the elements like resistors etc are connected using wires.. however in the real world all wires have resistances themselves so the circuit diagram could actually be redrawn with no wires and very small resistances instead..

So should wires themselves be considered a basic element... not ordinary wires of course because they are really very small resistors.. but superconductors which are in fact the perfect 'wire' and thus the fundamental circuit element.

Actually thinking about it some more I realised that wires don't really provide any behaviour??? they just sit there and conduct... so why is a superconductor any use? it just sits there and circulates a current, it doesn't do anything useful....

Cheers

Thomas_S_Hunterson

JoeBallantine wrote: »

I don't know a whole lot about electronics but is the transistor not a fundamental 'element' of circuitry?
What I mean is that people say three elements, resistor, capacitor and inductor are fundamental.. what about the transistor?

As I understand it, the transistor isn't a fundamental passive component. Essentially, it's just a switch, and acts as infrastructure for the fundamental elements.

democrates

Some clarity on 'elements' and note 'element' refers to an idealised mathematical representation used for design and analysis rather than a physical component :
http://en.wikipedia.org/wiki/Electrical_element Excerpts:

Any electrical network can be modeled by decomposing it down to multiple, interconnected electrical elements in a schematic diagram or circuit diagram.
...
The four fundamental circuit variables are current, I; voltage, V, charge, Q; and magnetic flux, Φm. Only 5 elements are required to represent any component or network by manipulating these four variables:

• Two sources:
  • Current source, measured in amperes - produces a current in a conductor. Affects charge according to the relation dQ = − Idt.
  • Voltage source, measured in volts - produces a potential difference between two points. Affects magnetic flux according to the relation dΦm = Vdt.
• Three passive elements:
  • Resistance R, measured in ohms - produces a voltage proportional to the current flowing through it. Relates voltage and current according to the relation dV = RdI.
  • Capacitance C, measured in farads - produces a current proportional to the rate of change of voltage across it. Relates charge and voltage according to the relation dQ = CdV.
  • Inductance L, measured in henries - produces a voltage proportional to the rate of change of current through it. Relates flux and current according to the relation dΦm = LdI.

(the memristor is then mentioned)
...
Current sources are more often used when representing semiconductors. For example, on a first degree of approximation, a bipolar transistor may be represented by a variable current source that is controlled by the input voltage.

Also see
http://en.wikipedia.org/wiki/Passive_component
http://en.wikipedia.org/wiki/Linear_element

HTH.

CathalMc

democrates wrote: »

The moral dilemma with AI hinges on whether it is regarded as a sentient being.

I like that! Interesting, should we consider ourselves Gods - or even Intelligent Designers to these beings. Chuckle chuckle.

Chua is more popularly famous for his work on chaotic circuits - Chua's Circuit, which is simple enough to be explored by undergraduates.

CathalMc

JoeBallantine wrote: »

I don't know a whole lot about electronics but is the transistor not a fundamental 'element' of circuitry?

Or is it that the crucial point to realise is that a transistors operation can be duplicated exactly by some combination of the three basic elements, i.e resistors, capacitors and inductors?

Those are good questions Joe, and really get at the root of the issue. Transistors are not fundamental, because, among other criteria, they are non-linear devices. Neither are they unique: if I told you I had a transistor, I could be talking about a whole host of various flavours - of which "acting like a switch" is a gross simplification of their shared behaviors. For instance, there's bipolar junction, and MOSFET transistors - which share (vaguely) similar behavior between two terminals, but very different behavior in their third.

Now the confusing thing about all of this, is that we frequently discuss transistors in terms of (grossly simplified & linearized) models that are based on notions of elementary (RLC) components. For instance, in a resistor, we take a voltage across two nodes, and it causes a current to flow between those two nodes - we term this resistance or conductance. In our abstract transistor model, we use a abstract, yet fundamental entity called a transconductance. This takes a voltage across two nodes, and produces a current across some other two nodes. (Alternatively, we can use a transresistance which is current to voltage).

Another intuitive, more practical helper here, is to consider that RLC's are all constructed from wires - or lets call them conductors.

A resistor - well that's could be just a very thin conductor, intuitively satisfying.
An inductor is a composite of loops of wire and the surrounding medium (say air).
A capacitor is a composite of two close, non-touching flat conductors and the medium between them (say air).

There is something very fundamental about these constructions. A transistor is a far more complex construction, especially in operation, as it relies on the composition of layers of chemically treated semiconductor, and the interaction of this with various electron densities, current flows, electric and magnetic fields. The result is, unsurprisingly, nonlinear, as our end-game is a switching device.

JoeBallantine wrote: »

So any usual circuit (without memristor elements) can be re-drawn wth only three elements, resistors, capacitors and inductors? Is this right? Is it always possible to reduce any circuit using some existing technique?

Surely also light emitting diodes (or simply diodes themselves) are also basic elements??? It can't be possible to duplicate these using the 'three', surely???

Sure, a circuit could be re-drawn in terms of RLC, if the original circuit was real and linear - but that definition is almost circular, as that system would already be a network of linear, real elements.
To model a linearized (ie: very incorrectly, but useful to designers) transistor, we need to introduce our transconductance - an abstract device. To correctly model a transistor, we need vastly complicated equations - which, interestingly, we sometimes use notions of non-linear resistances and non-linear capacitances.

Finally, diodes, as you well spotted - are exactly like transistors, non-linear and thus non-fundamental to circuit and network theorists.

CathalMc

JoeBallantine wrote: »

... Then I realised that a superconductor is simply a perfect wire... i.e in all circuit diagrams the elements like resistors etc are connected using wires.. however in the real world all wires have resistances themselves so the circuit diagram could actually be redrawn with no wires and very small resistances instead..

Actually thinking about it some more I realised that wires don't really provide any behaviour??? they just sit there and conduct... so why is a superconductor any use? it just sits there and circulates a current, it doesn't do anything useful....

It's a terminology issue - a wire is a wire as long as it's resistance is "low enough" that it isn't a significant in the context of the signal on it. If the resistance is large, there's a few practical effects to tend to manifest: your system may get slower as a result, and the signal might be attenutated in size. If it's wire transferring power, then considerable power may be lost as heat (hence we would like superconductors).
At that point, you would say, okay, I now going to model that wire as an ideal wire plus some ideal resistance. Then you're back to fundamentals.