current operating vs voltage operating electronic devices

[MrAl]

Hello everyone,

Here the OP. I thank you for the various inputs which I think are all useful in different ways. As a beginner, I am still not comfortable with design and need some physics explanations too.
I will read through your answers carefully.

From a quick reading, it seems that voltage is what causes current, in general. A current operated device seems to be one where the flow of current in a specific part of the device enables the device to function the way it is supposed to function, fulfilling the purpose it was designed for. For instance, if a transistor is current operated, that current will make the transistor acts as a switch and block or pass current in some other area...Same reasoning goes for voltage controlled devices.
A LED is defined as a current operated device in the sense that the light brightness is dependent on the current in the LED, in a more correlated way than the voltage. That is the functional definition of why the LED is called current operated. But that current is surely generated by tweaking a voltage, even if the relation current-voltage is not linear...

Thanks,
Kavan

Hello again,


I dont want to confuse you, but the current and voltage relationship come sort of at the same time. So it is arguable whether or not the voltage came first or the current came first. But to simplify, it easier to think of EITHER the current or the voltage is what comes first depending on the application. Let me give a couple examples...

The inductor. When we apply a voltage, at the instant of application there is zero current. But an infinitesimally short time later current appears.
The capacitor. When we apply a current, at the instant of application there is zero voltage. But an infinitesimally short time later voltage appears.

But there is no such thing as an infinitesimally short time period, except in theory. A human being can never experience an infinitesimally short time period. So the question comes up as to whether or not nature actually allows infinitesimally short time periods or time is quantized like everything else. And if you try to get an infinitesimally short time period, you end up with zero time, which is no time at all.
So if we apply a signal at t=0 we havent actually done anything yet. This is evident also in noting that at t=0 there can be no power because it's impossible to start a current flowing at t=0 because again no time has yet passed, and every distance in nature inherently contains inductance and so no matter what the device is there is at least some small inductance ever present, and some capacitance as well, and some resistance.

So lets look at an application...
We have a switch and battery to power a resistor wired up. When we turn on the switch, the resistor gets current and voltage from the battery. That's the simplified view.
But what really happens is that when we press the switch, the contacts come closer and closer together, which of course means for a tiny amount of time we have a capacitor connected at one end of the switch, and it is getting larger and larger in value as the contacts come closer and closer to each other. Since the cap has some voltage across it (from the battery through the resistor and wires) that means the pseudo cap will start to conduct. So before the switch actually closes we have a tiny conduction current. It's not zero, but some small value.
So it appears that the voltage was there first, then the current started to flow. But that's not really true either is it? Because the switch contacts were always there, they didnt just suddenly burst into existence. So there must have been capacitance there too (again distance comes into the picture).

So before we even start this experiment, we had voltage AND current. But it's safe to say that the current in the switch had gone to zero some time after the battery was first connected into the circuit, and that had happened some time in the past. So the actual start of the experiment then is when we first touch the switch, and at least one of the switch contacts begins to come closer to the other. That increases the capacitance and so current starts to flow again.

So we started with a set of charges that were arranged in a certain distribution in space, which must have had an electric field associated with it. Then we applied a force to the switch, and that changed the capacitance of the switch. The changing capacitance of the switch caused a current to flow again. So it was our finger that caused the imbalanced charge distribution to change.

But we're not done breaking this down yet. When we push the button, we dont immediately get a change in the distribution of charge. We first have to overcome the parallel capacitance of the wires and the series inductance of the wires. That means initially we're back to square one where the change we cause doesnt have any effect until after some inductance was finished impeding the current flow and the capacitance has stopped preventing the voltage from rising.
So we find we're back to the same situation, where the current getting to the device is impeded by the inductance and the voltage cant appear immediately because of the capacitance (of the device or just the wires). Analyzing this electrically, we see that as the voltage application appears the current begins to rise as well as the voltage to the device, but it doesnt seem possible to determine which came first (as seen by the device, which is what this thread is all about).
To see this in light of past work in physics, there is nothing in Maxwells equations to suggest that one comes before the other. In spite of this there are other views that say that there might be a delay from one to the other.

I hope i've explained this well enough and not confused you at all. This is a subject that requires a very detailed look at nature, where we might be concerned with behavior down in the attoseconds of time to fully (hopefully) understand this.

 

[kavan]

thanks Mr. Al.

I see your point(s). Very good.

In essence, we are dealing with electric charge. Charge is that there, accumulated and separated by charge of the opposite sign can provide voltage and can cause a current (if the material allows it).
In the capacitor case, the capacitor is not charged (no voltage) until a current brings charge on its plates. The current diminishes with time as the voltage goes up to its max value. If we look at the capacitor only, then the current caused the voltage on the capacitor. But that current was causes by a voltage, i.e. the voltage of the battery. It may not be useful, from a design standpoint, to backtrack what causes what, but in the capacitor example it still seems that it is the voltage causing the current.
After all, can we agree on what voltage is? Using informal language, I would say that voltage derives from charges of opposite sign being separated from each other. These charge create an electric field between them. Voltage is a derived quantity proportional to that E field...

I just took physics 2 and this the sense I made out of it...

Thanks,
Kavan

 

[Ratchit]

Claude Abraham,

Ratchit you keep re-iterating that voltage is needed to sustain LED current, but that point is not in contention. "Lowering barrier voltage" is a crutch you can't let go of. If an LED requires 1.80V at 10 MA forward current, then we know that a 10 mA current source w/ only 1.0 V of compliance will not work. The fact that the junction barrier potential must be overcome is not under discussion. Without the 1.80V, the 10 mA does not happen.

Good! I am glad that is settled. Now, let's see where that leads us.

Your whole case is built on the fact that w/o overcoming barrier voltage, no current can be sustained. We already know that. But the current is controlled by the source powering the LED, & to a very small extent, the difference between source voltage Vg & LED forward drop Vd. A voltage source plus resistor approximates current drive, but Vd still has an influence. But once the Vd barrier is overcome, Vd is not in control of Id, the input source is. Id is basically (Vg-Vd)/R, so that if Vg >> Vd, then Vd has only a minor influence on Id.

You are introducing external components with (Vg-Vd)/R. Naturally, Vd is going to be a function of Vg. But it is Vd that controls Id according to S's equation.

The mere fact that voltage is needed to overcome the junction barrier does not make a device "voltage controlled". Using that logic every electrical device would be voltage controlled & no such thing as current controlled would exist. A resistor has barriers formed by charge carriers colliding with the lattice structure forming barriers & releasing photons in the 5 um IR wavelength (heat). Even mag amps which have current conduction has internal barriers due to winding resistance. To generate the magnetic fields for the mag amp, current must exist. But resistive wires in the windings result in barriers due to lattice resistive collisions. So to sustain current, a potential barrier must be overcome. That would make a mag amp voltage controlled using your narrow self-styled definition of said term.

You are pettifogging the issue. Their is no barrier voltage present in resistors. A magnetic amp does not care what the input voltage is, only the input current.

Junction devices need voltage. W/o voltage no steady current could exist. The fact that the current which controls the LED brightness could not exist w/o voltage is accepted w/o argument. But the value of Vd under forward bias is literally determined by Id. Just as you cannot have Id w/o Vd, it is equally true in reverse. We already explained that Vd is determined by internal device geometry & doping, as well as Id & temp.

I am not saying that need determines control. I am saying that control is determined by the way one parameter influences another parameter with regard to the physics of the device. In this case, Vd controls Id according to S's equation.

A FET is universally described as VC. But in order to charge up the gate-source a CURRENT is needed. This gate current Ig precedes Vgs. So a voltage controlled FET, requires that Ig charges the gate-source giving rise to Vgs which activates the FET change of state. The fact that Vgs cannot change w/o Ig does not nake FETs current controlled. Ig is indispensable but not directly driven. We supply Ig from a CVS letting Ig be incidental, Vgs directly controlled.

I can just as well say that Ig needs Vgs to be applied in order to exist. That is not the point. A FET controls its current by an electrostatic field established by a voltage, so it a voltage controlled device. The FET current is not controlled by the Ig because Ig is not a voltage needed to establish an electrostatic field.

All electrical devices need I & V both. Which variable is the control variable is determined by which variable is directly forced on the device & which is incidental. But in all cases, the indirect incidental variable is indispensable. FETs cannot work w/o current, LED cannot work w/o voltage. But a FET driven from a CCS is destroyed. Gate-source is high resistance capacitor. A continuous current source charges the gate indefinitely until punch through occurs. But a CVS w/ low gate resistance provides high current at first, then current tapers off as Vgs approaches source value. Again Vgs cannot be attained w/o gate current Ig. That does not determine how FETs are classified.

Control is determined by the physics of the device, and the way one parameter controls another parameter. It is not determined by equations, graphs or need.

All devices cannot work w/o one of the variables missing. Even the indirect uncontrolled variable is indispensable.

Irrelevant with respect to control.

Kavan - voltage is NOT the CAUSE of current. Until you discard that heresy you will be stuck forever. The idea that voltage causes current is beyond a doubt the most common of all electrical heresies, so much so that I call it "heresy no. 1" of simply "H1".

Voltage is the energy density of the charge (joules/coulomb). Charge, and thereby current, will move from a higher energy density (voltage) to a lower energy density. Without this energy density difference, charge will not move and current will not exist. So from what I said, you can figure out whether H1 is true or not.

Ratch

 

[Ratchit]

kavan,

Now, I thought I was pretty solid in thinking that a voltage causes current though. Let me explain why.
current is the flow of charges (either positive or negative, electrons, ions, etc....).
Voltage (potential difference) is more the expression of charge separation and the ability to cause current (if the material allows ,i.e. it is not an insulator).

consider the parallel plate capacitor: there is a voltage between the plates. a charge positioned between the plate will move due to that voltage and represent a current....

I don't like heresies so please help me deconstruct this one if it is one.

Let's start with the basics. Voltage is the energy density of the charge (joules/coulomb). Two ways you can make a voltage are by cutting magnetic lines with a conductor at right angles, or gathering charged carriers of the same polarity together into one spot. Same polarity charges repel each other, so it takes energy to get them together. The more charges in one spot, the higher the energy density (voltage) will be. The closer the charges are to each other, the higher the voltage will be. In the case of a capacitor in a conductive circuit, charge is deposited on one plate and depleted on the opposite plate, causing a net charge change of zero. It takes electrical energy to do this, and this energy is stored in an electrostatic field between the plates. The capacitor is charged with energy, or energized, but its net stored charge is zero. No charge moves, and no current exists through the capacitor itself, unless leakage is involved. The current in the cap circuit is due only to the accumulation and depletion of charge on its two plates. It takes voltage to put and take away electrons from a capacitor's plates. As the charge accumulates and depletes, it sets up a back-voltage that opposes the energizing voltage, so the energy accumulation rate becomes smaller. Any questions?

Ratch

 

[Claude Abraham]

Claude Abraham,



Yes, there will be a current at small forward voltages, albeit small. Only when the forward voltage approaches 0.7 volts does the current become significant. This is revealed in Schockley's equation. Real world? You are starting to talk about design proceedures, as opposed to diode theory.



Again the voltage across the diode will sustain the diode currect according to Schockley's equation. Inductance has nothing to do with anything.



The carriers are at the junction since the diode was manufactured. So is the depletion zone. So is the accumulation of the ionic charge and E-field. We have been over the rest of this before. The applied voltage reduces the barrier voltage and allows more diode current to exist.




The value of I is determined by the value of the voltage that the CCS puts across the diode according to Schockley's equation. There will a one to one correspondence between the voltage and current, with the voltage determining the current according to the physics of the diode.



No matter what the V is in the external circuit, the voltage across the diode determines the diode current.



No matter if you drive the diode circuit with a voltage or current source, Schockley's equation will hold even if the diode is embedded within the circuit. Of course we don't try to regulate the current by directly controlling the Vd, but Vd and Id will automatically follow Schockley's equation no matter how many resistors you insert in the circuit containing the diode.



No, Vd will have a specific value determined by S's equation, but it will be Vd that controls Id. Within the diode itself, S's equation holds.



Yes, temperature is a part of S's equation.



Not so. I am avering that the current is related to the voltage across the junction according to S's equation, and the voltage is not necessarily constant. There will always be a voltage across the diode even if the external voltage is applied to the circuit containing the diode and not directly to the diode itself.



You have everything wrong in the above paragraph. When a electron diffuses from the N-type into the P-type, it bonds with an atom with only 3 covalent bonds to make an atom with 4 covalent bonds. So that atom has the 4 covalent bonds it wants, but now the extra electron makes it into a negative ion. And the extra electron from the 5 electron atom in the N-type is gone, so now that atom is a positive ion. So the time it takes to make the transfer is unimportant. It is the transfer itself, not how fast it is done that determines the ion count and thereby the barrier voltage. As I said before, this ion generation happens as soon as the P-type and N-type are bonded together during manufacturing.



I disagree. I think you should review what really happens when two different semiconductor types are placed in intimate contact with each other.



No, I do not consider him a scholarly source per se. It is a shame he does not defend his assertions.



Very good.

Ratch

I am at a loss to make it clearer. Regarding ionization, when e- from the n region enter the p region & combine with an atom having 3 bonds, ionizing it negatively, in a short time another e- exits the p region. Else the barrier would continue to ionize & with a constant current the barrier voltage rises indefinitely which we know does not happen. I don't know why you insist that external voltage source "lower the barrier voltage". Only a reduction in current can lower Vd. The external source gives e- enrgy to conduct. That source has a constant output voltage as an example. This voltage Vg is merely the ratio of energy imparted per unit charge. Vd OTOH is the energy lost per unit charge. The "voltage" Vg represents energy gained, vs. Vd which is energy lost. The diode current Id is not controlled by the lost energy per unit charge.

The diode forward drop Vd settles into equilibrium when constant Id exists. Vd is not controlling Id. Every e- crossing the barrier incurs loss via Vd, which gets replenished by input power source. To illustrate that Vd cannot be what controls Id, please search under key words "diode forward reverse recovery". When the polarity on a diode (p-n junction not Schottky), recovery takes place. If terminal voltage changes from positive to negative, a large reverse current exists until excess minority carriers are cleared from junction. For a brief time, the diode can have -15 volts across it with +1.0 amp through it, totally at odds with SE forms a) & b) both.

The above is reverse recovery phenomenon. A forward recovery can be described as well. If polarity suddenly changes from reverse to forward, the diode can be conducting +1.0 amp with a forward drop of +15 V. After equilibrium is reached the normal SE I-V relations settle down. The barrier potential affects Id only slightly. The external source is literally what controls Id by giving the e- energy. A portion of said energy is lost crossing the junction.

Again, your whole case is built on the fact that when source voltage is just below 0.65V, little current exists, then right above 0.65V, current becomes significant, which we all know. Vd is not controlling Id but rather it is as follows. A CCS unable to provide 0.65V cannot overcome Vd barrier hence current is very low. It is understood that the 0.65V loss is incurred & the source must compensate. This is not what defines current control. The 0.65V will never occur until the external source provides current. WHen switch first closes, Id can momentarily exist w/o 0.65V. But when e- cross junction (holes as well on other side), ionization occurs, DZ is formed, then if the CCS cannot overcome 0.65V Vd barrier, current ceases.

The fact that I cannot be sustained w/o V merely says that we do not have a superconductor. A semiconductor cannot have I w/o V, nor V w/o I. You always place V ahead of I but physics lends this idea no support. H1 is indeed heresy. V does not "control" I. BR.

 

[Ratchit]

Claude Abraham,

I am at a loss to make it clearer. Regarding ionization, when e- from the n region enter the p region & combine with an atom having 3 bonds, ionizing it negatively, in a short time another e- exits the p region. Else the barrier would continue to ionize & with a constant current the barrier voltage rises indefinitely which we know does not happen. I don't know why you insist that external voltage source "lower the barrier voltage". Only a reduction in current can lower Vd. The external source gives e- enrgy to conduct. That source has a constant output voltage as an example. This voltage Vg is merely the ratio of energy imparted per unit charge. Vd OTOH is the energy lost per unit charge. The "voltage" Vg represents energy gained, vs. Vd which is energy lost. The diode current Id is not controlled by the lost energy per unit charge.

I read the above paragraph over several times carefully, and I still don't know what you are talking about. When there is no connection to either the N-type or P-type region, as when a BJT in its package, how does any electron exit the P region? Yet there is still a barrier voltage formed by ions and a depletion region. When N and P are connected together, still nothing happens because the barrier voltage is equal to diffusion voltage. When a forward Vg is applied between N and P, then the barrier voltage is lowered and electrons can exit the P region and holes are annihilated by the incoming electrons at the N terminal. This produces a current. I have idea what you mean by "a reduction in current can lower Vd". The voltage determines the current, not the other way around. Your description of voltage ratios and energy gained is also confusing and not needed.

The diode forward drop Vd settles into equilibrium when constant Id exists. Vd is not controlling Id. Every e- crossing the barrier incurs loss via Vd, which gets replenished by input power source. To illustrate that Vd cannot be what controls Id, please search under key words "diode forward reverse recovery". When the polarity on a diode (p-n junction not Schottky), recovery takes place. If terminal voltage changes from positive to negative, a large reverse current exists until excess minority carriers are cleared from junction. For a brief time, the diode can have -15 volts across it with +1.0 amp through it, totally at odds with SE forms a) & b) both.

No matter how many times you say the Id controls the Vd, the physics of the BJT say otherwise. If a storage charge is present at the junction, that has to be cleared away first before the BJT operates again according S's equation. That storage charge is a temporary, and not to be confused with long term steady state forward voltage operaton.

The above is reverse recovery phenomenon. A forward recovery can be described as well. If polarity suddenly changes from reverse to forward, the diode can be conducting +1.0 amp with a forward drop of +15 V. After equilibrium is reached the normal SE I-V relations settle down. The barrier potential affects Id only slightly. The external source is literally what controls Id by giving the e- energy. A portion of said energy is lost crossing the junction.

Again you want to introduce storage charge. That transient phenomena will not prove that Id controls Vd. If we don't reverse the voltage, what then?

Again, your whole case is built on the fact that when source voltage is just below 0.65V, little current exists, then right above 0.65V, current becomes significant, which we all know. Vd is not controlling Id but rather it is as follows. A CCS unable to provide 0.65V cannot overcome Vd barrier hence current is very low. It is understood that the 0.65V loss is incurred & the source must compensate. This is not what defines current control. The 0.65V will never occur until the external source provides current. WHen switch first closes, Id can momentarily exist w/o 0.65V. But when e- cross junction (holes as well on other side), ionization occurs, DZ is formed, then if the CCS cannot overcome 0.65V Vd barrier, current ceases.

Ionization voltage is always present at the PN junction barrier. How effective it is depends on the applied forward voltage. I don't know what the 0.65 volts has to do with your assertion that Id controls Vd. If the CCS cannot put out 0.65 volts, then it is not a very good CCS, is it?

The fact that I cannot be sustained w/o V merely says that we do not have a superconductor. A semiconductor cannot have I w/o V, nor V w/o I. You always place V ahead of I but physics lends this idea no support. H1 is indeed heresy. V does not "control" I. BR.

What a strange conclusion! I don't know what you mean by placing V ahead of current. Current needs V to exist and current is sometimes a consequence of V. They both can exist together.

Ratch

 

[misterT]

I missed these debates.. So entertaining.

 

[MrAl]

Hi,


FET's are often called voltage controlled devices because in the most common circuit the gate voltage is used to control the drain current, period.

FET's can be used in other ways however, such that the drain to source voltage controls the drain current and the gate voltage is held constant. That's the so called 'constant current' mode. FET's are often used this way inside integrated circuits to provide the bias for the other circuitry.

But my point though this discussion and the other one in this forum is that in order to change the gate voltage we have to have at least some current to back it up, or at least to get it to appear where we want it to appear (across the oxide layer for example) and i dont think it is possible to specify which comes first, the voltage or the current, because inside the wires we are dealing with something that is 'other worldly', in that we cant emulate it exactly with anything large that we see normally in life like a basketball. So it has properties that we are not used to seeing commonly in everyday life, and it's very hard to experience this phenomenon because we cant deal with it in our hands like most things we come across.

The potential exists sort of on it's own, but to get it to do anything we have to have some of the charge move, and that's what we call current
There's no way to "apply" a voltage without getting some charge to move.

We can turn on a water faucet half way to get a medium water flow. Thus we've "controlled" it, in a manner of speaking. But more exactly can we say that we are *still* controlling it, when we stop turning the knob? Control in electrical circuits usually means we are trying to get one variable to stay constant while we monitor that variable and forcibly cause another variable to change, sometimes drastically. So we might say that in order to control an FET with the gate we have to be able to change the gate voltage, not just set it to a certain level and leave it there. While the gate voltage is changing we can say we are "controlling" the drain current, but after we stop changing the gate voltage we have to say that we "have controlled" the drain current, not that we are still controlling it.

This doesnt change the typical statement that an FET is voltage controlled, but it does show us that the way we deem a device (voltage or current controlled) is still application dependent, and physically we always need both voltage and current except in the static case which is not exactly fundamental to the way view the device (dynamics are the definitive rulers).

 

[steveB]

Hello Forum,
I have heard of current operating devices, like LEDs, and voltage operated devices...

What is the difference?

It seems to me that all electronic devices, once provided a certain voltage, draw the current they need....so what is this distinction between current operating and voltage operating?

thanks
kavan

Putting aside the semantics debate for a moment, let's focus on the practical issues with how engineers design from the point of view of "controllability". After all, this is the reason for the terminology you are seeing. Engineers don't get paid to debate semantics, but do get paid to make things work. Sure, understanding physics is important, but if you want to understand device physics then study the device physics. But, clearly this is not what your question is about, as i read it.

If you have a high power LED and desire to control the light output, clearly you should consider that LED to be a current controlled device, because light output is uncontrollable when only voltage is used as the input control. This is because applying a voltage to an LED at room temperature results in a particular light output. However, then the device heats up and the VI curve changes. Hence, the light output changes despite the fact that you're controlling voltage. Why? Because you're not controlling (or at least measureing and adapting to) temperature at the same time. However, if you use current as the control, the light ouput is nearly constant because hole-electron pairs are converted to photons with an efficiency that is relatively stable with temperature.

If you have an incandescent light bulb and desire to control the light output, clearly you should consider that bulb to be a voltage controlled device, because light ouput is uncontrollable when current is used as the input control. This is because applying a current to a bulb at room temperautre results in a particular light output. However, then the device heats up and changes the resistance (R goes up with T for metals). This causes thermal runaway because P=I^2R, and resistance/temperature keeps going up as more and more power is added. With voltage constant, the current is initally high and the device heats up quickly which then increases the resistance and lowers the current. This provides a stable operation.

This isn't to say you can't control an LED with voltage or a light bulb with current, but just that those are harder ways. And, remember that classifications don't usually have clear dividing lines. Often there is ambiguity when we classify. For example, a resistor used as a heater is both a current controlled and voltage controlled device. Then, the consideration is which source is more practical.

I'll make a quick comment on the debate issue, to help you deal with the differences in opinion. Our view of the world is always based on models which always gives a biased point of view. So, when people have differing opinions, it is often because their models (and hence their viewpoint) is different. A model and "the mystery it emulates" are two very different things. The former we can understand very well, while the latter is always a riddle, wrapped in an enigma ,shrouded in a model, touted by an expert.

 

[ronv]

Yea Steve!

 

[MrAl]

Putting aside the semantics debate for a moment, let's focus on the practical issues with how engineers design from the point of view of "controllability". After all, this is the reason for the terminology you are seeing. Engineers don't get paid to debate semantics, but do get paid to make things work. Sure, understanding physics is important, but if you want to understand device physics then study the device physics. But, clearly this is not what your question is about, as i read it.

If you have a high power LED and desire to control the light output, clearly you should consider that LED to be a current controlled device, because light output is uncontrollable when only voltage is used as the input control. This is because applying a voltage to an LED at room temperature results in a particular light output. However, then the device heats up and the VI curve changes. Hence, the light output changes despite the fact that you're controlling voltage. Why? Because you're not controlling (or at least measureing and adapting to) temperature at the same time. However, if you use current as the control, the light ouput is nearly constant because hole-electron pairs are converted to photons with an efficiency that is relatively stable with temperature.

If you have an incandescent light bulb and desire to control the light output, clearly you should consider that bulb to be a voltage controlled device, because light ouput is uncontrollable when current is used as the input control. This is because applying a current to a bulb at room temperautre results in a particular light output. However, then the device heats up and changes the resistance (R goes up with T for metals). This causes thermal runaway because P=I^2R, and resistance/temperature keeps going up as more and more power is added. With voltage constant, the current is initally high and the device heats up quickly which then increases the resistance and lowers the current. This provides a stable operation.

This isn't to say you can't control an LED with voltage or a light bulb with current, but just that those are harder ways. And, remember that classifications don't usually have clear dividing lines. Often there is ambiguity when we classify. For example, a resistor used as a heater is both a current controlled and voltage controlled device. Then, the consideration is which source is more practical.

I'll make a quick comment on the debate issue, to help you deal with the differences in opinion. Our view of the world is always based on models which always gives a biased point of view. So, when people have differing opinions, it is often because their models (and hence their viewpoint) is different. A model and "the mystery it emulates" are two very different things. The former we can understand very well, while the latter is always a riddle, wrapped in an enigma ,shrouded in a model, touted by an expert.

Hi there Steve,


Thanks for joining the discussion here.

I agree with most of what you had said, except i have to disagree a little with the "uncontrollable" part. I'll explain in a second.

I also agree with the 'viewpoint' idea which is really the same thing as the "application" which i have been trying to get across to readers.

So ok back to the uncontrollable part...

The view that something is uncontrollable because one type of control seems better than the other could be a little too strict. It appears that one type of control 'better' controls the device so that often leads to us trying to label it one way or the other. It's not entirely wrong to do this, but it can be misleading. That's because it is almost always based on a given application (we're back to the application specific problem again). When the application changes, it could change the 'view' entirely. Not just a little, but the entire outlook about the device. This is what causes confusion among newcomers into the field when they see a new device and read that it is "voltage controlled" they think they never have to worry or think about current at all. The typical error comes in the form of a 100k resistor to drive the MOSFET gate when the MOSFET is used in a switching converter. Since it is a "voltage controlled" device, why worry about current? And then later they find that the MOSFET takes so darn long to switch that it overheats or just doesnt work at all at the required frequency. Ive seen this happen over and over on forums and that's because the MOSFET (or just FET) is always written about as being voltage controlled (via the gate source).

It's harder to argue about the LED being current controlled, because that's the way it should be viewed in most applications. It's a little hard to argue about the filament bulb because voltage is the most important aspect of it. But i think uncontrollable is not the right way to deal with it. If we were to monitor the temperature of the LED we could control it via it's terminal voltage. Not that we want to have to do it that way, but it is possible. With other devices (which are more confusing) it's not as apparent though.

More to the point are the devices which cause the most confusion, like the MOSFET and BJT. To illustrate how important the application is to the 'view' of the device, the BJT is current controlled when we want to design a current amplifier, but it is voltage controlled when we want to design a voltage reference diode.
So how do we call the BJT then, current controlled, or voltage controlled? We cant really call it either even though we got paid to design both circuits. What we can do perhaps is simply explain that:
"In this given application, the BJT is thought of as being current controlled", or
"In this other application, the BJT is thought of as being voltage controlled".
And note that the equations used for these two cases are most likely different equations, or at least they can be.

So we not only declare the BJT to be either, we also use different equations so that we can determine what the circuit is going to do, and those equations reflect how we view the BJT.

Most application design intends to accomplish it's goals by hoping to ignore one parameter over the other, hence the preference to call a device current or voltage controlled, because if we call it one we can often (almost) ignore the other. But that's not really the way to characterize a device, and that is most noted on any data sheet which gives many specifications for both current and voltage for almost every device under the sun. We choose to concentrate on some specs while almost ignoring others, and that leads us to want to take the short cut and call it one or the other. But looking more broadly, we see that we have to consider everything before we can be sure the circuit will work or not.

 

[steveB]

I agree with most of what you had said, except i have to disagree a little with the "uncontrollable" part. I'll explain in a second.

I understand what you are saying and pretty much agree. I was using the term uncontrollable a little too loosely. We have a rigid definition of "controllability" from feedback theory, but I was using the term more colloquially.

In the context of what I'm saying, with LEDs and bulbs, there are varying degrees of "uncontrollable" if only one control variable is used. In both cases, we could have thermal runaway, which really is a case of losing control or instability. At the other extreme, we may have stability and some degree of control, but we might lack precision of control. In these examples, we might see light power vary much more if the wrong method is used, and in that case we would need to improve the control by monitoring other variables (in this case either light power or temperature) and compensating either with feedback or with feedforward control.

Essentially, at the end of my discussion, I hinted at this point by saying. "This isn't to say you can't control an LED with voltage or a light bulb with current, but just that those are harder ways."

 

[Mr RB]

The OP said "operated" not "controlled".

Although the current produces the actual result inside the component, is that the best definition of "operated"?

If I said "How is a tractor operated?" the correct answer would be that it is operated by the user, who is operating the levers. The tractor is NOT operated by the diesel fuel.

So although its almost always current that produces the results inside a component or circuit, the "procedure of operation" or "how it is operated" is via voltage.

 

[MrAl]

Hi there MrRB (Roman),


I think the analogy you brought up is good, and i think we can use that to further expound on what we all have been trying to say. But one point i want to bring up though is that what you said does show a bias toward what aspect of the operation we are looking at, or what context the discussion falls under. I'll explain...

In the context of human vs automatic operation, your statement is entirely accurate and to the point. But in the context of diesel vs propane fuels it's entirely irrelevant. That's simple to see that when we talk about human vs automatic operation we want to know one or the other, but when we talk about which type of fuel we dont care if a human is operating it we just want to know what kind of fuel it uses.

So being under the category of voltage vs current, our only choices are voltage, current, both, or neither. Since we are talking about electronic or electrical devices, we narrow it down to voltage, current, or both.

Now the most strict view (and actually most accurate) is to say "both", and that's because physics tells us that we cant tell what came first, the voltage or the current. But when we try to narrow the range of possibilities to make the design challenge more simple, we start to assign one or the other depending on what we are designing *for*.

So to sum up i like to call this "application specific" because how we view the device depends highly on what we are designing in the first place. The clear example is the BJT where we view it one way for one application and another way for another application. But that's just to simply the design procedure. In reality it's always going to be *both*.

I see a lot of physics errors in this thread and the other though, where we would like to believe that one is favored over the other because of the way the device appears to work internally. But that extreme view is another example of theory only, and we know that as soon as we try to apply that theory to another application we find suddenly it doesnt work anymore, and that's because often theory narrows down the playing field even more to show a given aspect of the device, not how it works in it's entirety. So we cant say that the underlying physics says this or that, unless we consider all of the physics at the same time. All we can do is explain the theory with that assumption in mind.
Case in point: A resistor powered by a voltage. We know I=E/R, but we cant always just end it there. We have to talk about a lot more yet.
Case 2: The FET works via the 'field' that's why it's called a Field Effect Transistor. But the field doesnt get there unless we apply a voltage, and that voltage has low enough impedance to cause the internal voltage to rise. Also, we find a limit in the frequency of the device.

 

[atferrari]

Our view of the world is always based on models which always gives a biased point of view. So, when people have differing opinions, it is often because their models (and hence their viewpoint) is different. A model and "the mystery it emulates" are two very different things. The former we can understand very well, while the latter is always a riddle, wrapped in an enigma ,shrouded in a model, touted by an expert.

I am afraid that in the future I will be quoting this often, blatantly free of charge .

 

[steveB]

I see a lot of physics errors in this thread and the other though, where we would like to believe that one is favored over the other because of the way the device appears to work internally. But that extreme view is another example of theory only, and we know that as soon as we try to apply that theory to another application we find suddenly it doesnt work anymore, and that's because often theory narrows down the playing field even more to show a given aspect of the device, not how it works in it's entirety. So we cant say that the underlying physics says this or that, unless we consider all of the physics at the same time.

I agree, and I also see a number of physics errors. Here you are noting a very big misconception about physics in general, and, quite often, good physicists slip into this mindset as easily as the rest of us. In a number of places in these threads we see the idea that there are models that are used for design, and then there is the "underlying physics" which is somehow a clearly percieved reality. In fact, all of the physics we use is some type of a model, whether it is a conceptual model or a mathematical model.

Let's be very specific about the models in question with semiconductor devices, because it provides a very interesting example of just how distinct the device-entity is from the theory we use to understand device-physics (leaving design models aside for now).

The specific theoretical tool we use to develop the concepts and the calculations for semiconductor devices is a forced joining of two disparate physical models. Specifically, we take classical electrodynamics (i.e. Maxwell's equations or EM theory) and join them with basic quantum mechanics QM. Let's look closely at this strange marriage of an odd-couple. We have EM theory which is a classical theory that has special-relativity embedded intimately within it. Then, we have QM which is a non-classical theory that does not include special-relativistic effects. Despite the clear incompatibility of these theories, we force them to coexist in one theoretical model that we use to understand semiconductor devices. And, fortunately this works accurately enough because the quantization effects don't become important on the electrical side for the types of devices we deal with right now (note, this could change in the future). And, the relativistic effects don't become important on the mechanical side. This strikes me as a clear example of using approximate models to develop a workable understanding of devices to describe and predict their behavior, and to have a mental picture that allows human intuition to be used practically.

The other aspect of this that seems relevant is that we are talking about the voltages and currents here, which means that the subject of this thread should emphasize the EM theory aspect of the model. If we do this, then an obvious fact has the spotlight shone on it. That is, in EM theory it is charges and currents (or, more accurately charge density and current density) that are the "sources" in the model. In other words, charges and currents are the causes or inputs and the fields are the responses to those sources. So, if we really did want to use the underlying physics as a basis (which I don't recommend by the way) to identify what "controls" a device, why would we place voltage (a potential derived from electric field) as the "control" above the true sources in the theory? My answer is that we do that when it makes sense to do it, in order to have a better mental picture of what is going on.

I believe that right now the best theoretical tool, with the most logical consistency, that could be used to most accurately model semiconductor devices would be based on a quantum electrodynamics, or quantum field theory. The problem with these theories is that they are too unwieldy to use to do calculations, or to even develop a usable mental picture of what is happening. However, even if we did this, we would need to understand that this theory is also a model, and the future is likely to see this model replaced by an even more sophisticated and accurate (and probably even more unwieldy) model.

 

[MrAl]

Hi Steve,


I must say, nicely written post there. I enjoyed reading it and thinking about what it says.

What i was getting at might be a little different than what you were explaining there. I was trying to get at the *underlying physics* at the very core of nature, where i think we find less variance than in models of semiconductors and whatnot, and less subject to interpretation.

What i was talking about was energy itself. The way energy works doesnt usually change too much, at least from what we know now. Sure it might change in the future, but i seriously doubt it will change in the way it would have to change in order to make some of the statements made by some in this and other threads completely valid. The entire discussion hinges more or less on the difference between voltage and energy, or current and energy.

Since voltage vs energy was more prevalent, i'll mention that first.

We all know that voltage can not do anything itself. It can not force a single electron through even a small band gap (band gap model note below). It takes a certain minimum energy to push an electron through the band gap. That should really be all that has to be said. So that energy has to come from somewhere, and the only place it can come from is (closed circuit with no other external energy sources and excluding the thermal energy either in concept or by pure experimental set up) is the terminals of the device. And the only way energy can enter the terminals of a device is via current and voltage. It takes both, even though one may be much smaller than the other (and hence the 'voltage control' analogies).

So im taking such fundamental physics here that there should be no doubt that it must certainly be true, or we'd have to figure out how a voltage or current alone can do anything at all by itself without requiring energy, and that would lead us to a machine that is more than 100 percent efficient.

 

[Mr RB]

... So, when people have differing opinions, it is often because their models (and hence their viewpoint) is different.
...

Which maybe explains why people from a physics background tend to embrace physics modelling instead of a good solid practical generalisation?

The terms "controled", "operated" or "activated" must respect heirarchy of control, ie what generally controls what.

And generally in electronics, an applied voltage is the controlling factor that causes a current to be established. In an inductor voltage can be applied which starts "operating" the inductor straight away even though it will take time for current to rise.

I can't think of a single example where you "apply a current" to something to activate it without any voltage being required, because of course some voltage difference must be present or you could not cause a current. In a "Chicken and egg" argument then voltage is coming out well on top in a control heirarchy.

Personally I would describe a LED as "current controlled" because the desired operation more closely matches the current then the voltage (by a small margin), but in a more correct sense I think the LED is still "voltage controlled" because it requires the voltage to be applied for the desired result current to flow.

 

[steveB]

I can't think of a single example where you "apply a current" to something to activate it without any voltage being required, because of course some voltage difference must be present or you could not cause a current. In a "Chicken and egg" argument then voltage is coming out well on top in a control heirarchy.

I pretty much agree, but I can think of at least one example. By reversing your inductor example, consider an uncharged capacitor. We can (and should) charge the capacitor with a current source, and in that first instant, there is no voltage. The field (and voltage) is a result of the charges getting on the plates. Of course, in that instant when even those first electrons are moved, an electric field is present, but it is the charge movement that generates a net electric field which manifests the potential voltage. Really, electric fields exist already (as generated by subatomic particles - primarily electrons and protons) and we consider the net field to be zero because they cancel out (at least at the macroscopic level) when we have charge equilibrium.

We also have to consider that many devices have capacitance as an integral part of the structure of the device. For example, transistor models always include capacitance when we are interested in high frequency performance. I suppose this is why some people treat FETs as charge controlled devices, rather than voltage controlled devices.

 

[MrAl]

Hello again MrRB and Steve,


So i believe you are starting to see that voltage and current work together to transfer energy. And fundamental physics says that energy is the only thing that can move anything physical no matter how small. That i believe is the heart of the matter.

But we may have gotten too particular with the discussion of the PN junction anyway, as the OP in both threads wants to know about practical devices. Yes, PN junction theory might be interesting too, but it doesnt explain the entire character of a real life device.

BTW, i forgot the note about the band gap theory. We wanted to stay away from models of any kind, but i believe the band gap model is still elementary enough to be used for this discussion. If you dont agree you can add a note explaining why. So im using something so elementary to physics that it is not likely to be questioned for a long time (energy is the prime mover) and band gap theory which isnt too questionable if it is at all.

The band gap theory tells us that a minimum amount of energy is needed to move a charge from one place to another. In particular, across the band gap. It's called the band gap because of the statistical nature of the atoms, when working together, create a barrier for carriers which can be almost any energy level (not just some discrete quantum level like in a single atom). So it's actually an energy level that must be present, and energy comes from both current and voltage, and the energy level is dependent on the type of atoms used in the two or more semiconductor constituent materials.

I consider these concepts to be elementary enough to be considered indisputable. I also believe that future improvements in the theory will not void the theories entirely, but will simply improve the precision of the results and add to the applicability.