Please, you two, a bipolar transistor is BOTH. It just depends on the drive impedance, which is dominant. Who cares about some professor's soapbox, when it interferes with the actual efficient use of the device?
bjt operation & heresies
References from semiconductor OEMs describing FET as voltage controlled and bjt as current controlled.
On Semiconductor AN-913: under the heading "Comparing and Contrasting Bipolars and Power MOSFETs", top of page 2 - "The most marked difference is that the gate of the MOSFET is voltage driven whereas the base of the bipolar is current driven."
National Semiconductor AN 558: introduction on page 1 - "The high voltage power MOSFETs that are available today are N-channel, enhancement-mode, double diffused, Metal- Oxide-Silicon, Field Effect Transistors. They perform the same function as NPN, bipolar junction transistors except the former are voltage controlled in contrast to the current controlled bi-polar devices. Today MOSFETs owe their ever-increasing popularity to their high input impedance and to the fact that being a majority carrier device, they do not suffer from minority carrier storage time effects, thermal runaway, or second breakdown."
Fairchild Semiconductor AN-7500: under "general characteristics" on page 1 - "A conventional n-p-n bipolar power transistor is a current-driven device whose three terminals (base, emitter, and collector) are connected to the silicon by alloyed metal contacts. Bipolar transistors are described as minority-carrier devices in which injected minority carriers recombine with majority carriers. A drawback of recombination is that it limits the device's operating speed. And because of its current-driven base-emitter input, a bipolar transistor presents a low-impedance load to its driving circuit. In most power circuits, this low-impedance input requires somewhat complex drive circuitry.
By contrast, a power MOSFET is a voltage-driven device whose gate terminal, Figure 1(a), is electrically isolated from its silicon body by a thin layer of silicon dioxide (SiO2). As a majority-carrier semiconductor, the MOSFET operates at much higher speed than its bipolar counterpart because there is no charge-storage mechanism. A positive voltage applied to the gate of an n-type MOSFET creates an electric field in the channel region beneath the gate; that is, the electric charge on the gate causes the p-region beneath the gate to convert to an n-type region, as shown in Figure 1(b). This conversion, called the surface-inversion phenomenon, allows current to flow between the drain and source through an n-type material. In effect, the MOSFET ceases to be an n-p-n device when in this state. The region between the drain and source can be represented as a resistor, although it does not behave linearly, as a conventional resistor would. Because of this surface-inversion phenomenon, then, the operation of a MOSFET is entirely different from that of a bipolar transistor, which always retain its n-p-n characteristic."
Fairchild Semiconductor AN-9010: under "advantages of a MOSFET" on page 7 - "1. High input impedance - voltage controlled device - easy to drive. To maintain the on-state, a base drive current which is 1/5th or 1/10th of collector current is required for the current controlled device (BJT). And also a larger reverse base drive current is needed for the high speed turn-off of the current controlled device (BJT). Due to these characteristics base drive circuit design becomes complicated and expensive. On the other hand, a voltage controlled MOSFET is a switching device which is driven by a channel at the semiconductor’s surface due to the field effect produced by the voltage applied to the gate electrode, which is isolated from the semiconductor surface. As the required gate current during switching transient as well as the on and off states is small, the drive circuit design is simple and less expensive."
These are just a few off the top of my head. You can download any of the above app notes and confirm the above. I've been an EE for 30 years, and every OEM, FAE, app note, and data sheet describes FETs as voltage controlled and bjt's as current controlled. Of course, this is a model based on viewing the device as a 3-terminal black box. This macro view point does not consider internal fields, quantum mechanics, energy bands, doping, traps, crystal bonds and defects, stored charges, etc. It is a "big picture / external" view of the device. At the micro level, both are classified as "charge controlled". At the micro level the fundamental difference between the two is that FET's are "majority carrier" devices, whereas bjt's are "minority carrier devices". The terms "majority" and "minority" refer to *charge* carriers, not currents or voltages. At the micro level, for all of my 30 years as an EE, they are both considered "charge controlled".
As far as the following goes "currents do not control currents. Voltages do. " That is the paradigm used by CMD's (current mode deniers) in justifying all of their contrarian positions. A cmd cannot fathom or conceptualize how a current can be controlled by another current. A transformer immediately comes to mind. If the xfmr is a voltage xfmr, or VT, the secondary voltage Vs, is determined or "controlled" if you will, by the primary voltage Vp. Vp is controlled by the constant voltage power source across the primary. Vs = Vp*(Ns/Np).
But the primary current Ip, is controlled or determined by the secondary current Is. When the CVS across the primary establishes Vs, the load resistance Rs establishes Is = Vs/Rs in accordance with Ohm's law. Then the primary current Ip = (Ns/Np)*Is. Except, of course, that the exciting current in the primary is there regardless of loading. So, Iexc exists in the primary and the balance of Ip is controlled by Is. The value of Ip is directly controlled by Is regardless of Vp, or Vs. Not too difficult.
The most common misconception in electrical science in general is that currents are "caused" by voltages. It's not true. The light bulb that illuminates your room can be used for example. Does the voltage across the filament cause the current? Or does the current through the filament cause the voltage? The only rational answer is that neither can exist without the other. It is a circular relation. For the bulb to output heat and light, which is power, power must be inputted. The input power is the product of current and voltage. Both must be non-zero to light up the bulb. The light requires both current and voltage. Because the power source is constant voltage, people can rush to judgement that voltage comes first, then current. But the constant voltage source is that way because the power company forces it to be that way. They could provide constant current, but the losses in the lines would increase.
Also, beta is all-important. Yet, variations in beta have to be neutralized. These 2 concepts are not at odds. The wcm beta value (worst case minimum) is all important. But the wf beta (windfall) value must be neutralized through proper topology and component value selection. If a device has a wcm beta of 100, vs. a different device with a 50 wcm beta, the 100 value gives us more flexibility if all other parameters are equal. If I'm designing a 1 stage amp, I need a high wcm beta value. Otherwise, my input impedance is too low, loading down the input source. If I increase Re to provide a high input Z, Zin, then my stage gain, Av = -Rc/Re, goes down. So a high Zin requires a high (hfe+1)*Re, but a high gain Av requires a high Rc/Re which means a LOW Re. If hfe is very large I can have both.
But, if my wcm beta is 100, and it can vary up to 600, then the overall performance must hardly change at all. In other words, I design my amps so that if the worst case min is 100, and the max is 600, the difference in overall gain is too small to notice. This is a good design because it possesses "beta immunity". But if the wcm beta is only 50, then the gain and Zin cannot both be as good as in the case with a 100 wcm beta. That is why beta is all important.
The wcm should be as high as you can obtain w/o comprimising other important parameters. A good topology will work very consistently with any beta value at or above the wcm value with negligible performance difference. This is beta immunity, a good thing.
Enough for now. BR.
Lumba Ogir said:
Lumba Ogir said:
References from semiconductor OEMs describing FET as voltage controlled and bjt as current controlled.
On Semiconductor AN-913: under the heading "Comparing and Contrasting Bipolars and Power MOSFETs", top of page 2 - "The most marked difference is that the gate of the MOSFET is voltage driven whereas the base of the bipolar is current driven."
National Semiconductor AN 558: introduction on page 1 - "The high voltage power MOSFETs that are available today are N-channel, enhancement-mode, double diffused, Metal- Oxide-Silicon, Field Effect Transistors. They perform the same function as NPN, bipolar junction transistors except the former are voltage controlled in contrast to the current controlled bi-polar devices. Today MOSFETs owe their ever-increasing popularity to their high input impedance and to the fact that being a majority carrier device, they do not suffer from minority carrier storage time effects, thermal runaway, or second breakdown."
Fairchild Semiconductor AN-7500: under "general characteristics" on page 1 - "A conventional n-p-n bipolar power transistor is a current-driven device whose three terminals (base, emitter, and collector) are connected to the silicon by alloyed metal contacts. Bipolar transistors are described as minority-carrier devices in which injected minority carriers recombine with majority carriers. A drawback of recombination is that it limits the device's operating speed. And because of its current-driven base-emitter input, a bipolar transistor presents a low-impedance load to its driving circuit. In most power circuits, this low-impedance input requires somewhat complex drive circuitry.
By contrast, a power MOSFET is a voltage-driven device whose gate terminal, Figure 1(a), is electrically isolated from its silicon body by a thin layer of silicon dioxide (SiO2). As a majority-carrier semiconductor, the MOSFET operates at much higher speed than its bipolar counterpart because there is no charge-storage mechanism. A positive voltage applied to the gate of an n-type MOSFET creates an electric field in the channel region beneath the gate; that is, the electric charge on the gate causes the p-region beneath the gate to convert to an n-type region, as shown in Figure 1(b). This conversion, called the surface-inversion phenomenon, allows current to flow between the drain and source through an n-type material. In effect, the MOSFET ceases to be an n-p-n device when in this state. The region between the drain and source can be represented as a resistor, although it does not behave linearly, as a conventional resistor would. Because of this surface-inversion phenomenon, then, the operation of a MOSFET is entirely different from that of a bipolar transistor, which always retain its n-p-n characteristic."
Fairchild Semiconductor AN-9010: under "advantages of a MOSFET" on page 7 - "1. High input impedance - voltage controlled device - easy to drive. To maintain the on-state, a base drive current which is 1/5th or 1/10th of collector current is required for the current controlled device (BJT). And also a larger reverse base drive current is needed for the high speed turn-off of the current controlled device (BJT). Due to these characteristics base drive circuit design becomes complicated and expensive. On the other hand, a voltage controlled MOSFET is a switching device which is driven by a channel at the semiconductor’s surface due to the field effect produced by the voltage applied to the gate electrode, which is isolated from the semiconductor surface. As the required gate current during switching transient as well as the on and off states is small, the drive circuit design is simple and less expensive."
These are just a few off the top of my head. You can download any of the above app notes and confirm the above. I've been an EE for 30 years, and every OEM, FAE, app note, and data sheet describes FETs as voltage controlled and bjt's as current controlled. Of course, this is a model based on viewing the device as a 3-terminal black box. This macro view point does not consider internal fields, quantum mechanics, energy bands, doping, traps, crystal bonds and defects, stored charges, etc. It is a "big picture / external" view of the device. At the micro level, both are classified as "charge controlled". At the micro level the fundamental difference between the two is that FET's are "majority carrier" devices, whereas bjt's are "minority carrier devices". The terms "majority" and "minority" refer to *charge* carriers, not currents or voltages. At the micro level, for all of my 30 years as an EE, they are both considered "charge controlled".
As far as the following goes "currents do not control currents. Voltages do. " That is the paradigm used by CMD's (current mode deniers) in justifying all of their contrarian positions. A cmd cannot fathom or conceptualize how a current can be controlled by another current. A transformer immediately comes to mind. If the xfmr is a voltage xfmr, or VT, the secondary voltage Vs, is determined or "controlled" if you will, by the primary voltage Vp. Vp is controlled by the constant voltage power source across the primary. Vs = Vp*(Ns/Np).
But the primary current Ip, is controlled or determined by the secondary current Is. When the CVS across the primary establishes Vs, the load resistance Rs establishes Is = Vs/Rs in accordance with Ohm's law. Then the primary current Ip = (Ns/Np)*Is. Except, of course, that the exciting current in the primary is there regardless of loading. So, Iexc exists in the primary and the balance of Ip is controlled by Is. The value of Ip is directly controlled by Is regardless of Vp, or Vs. Not too difficult.
The most common misconception in electrical science in general is that currents are "caused" by voltages. It's not true. The light bulb that illuminates your room can be used for example. Does the voltage across the filament cause the current? Or does the current through the filament cause the voltage? The only rational answer is that neither can exist without the other. It is a circular relation. For the bulb to output heat and light, which is power, power must be inputted. The input power is the product of current and voltage. Both must be non-zero to light up the bulb. The light requires both current and voltage. Because the power source is constant voltage, people can rush to judgement that voltage comes first, then current. But the constant voltage source is that way because the power company forces it to be that way. They could provide constant current, but the losses in the lines would increase.
Also, beta is all-important. Yet, variations in beta have to be neutralized. These 2 concepts are not at odds. The wcm beta value (worst case minimum) is all important. But the wf beta (windfall) value must be neutralized through proper topology and component value selection. If a device has a wcm beta of 100, vs. a different device with a 50 wcm beta, the 100 value gives us more flexibility if all other parameters are equal. If I'm designing a 1 stage amp, I need a high wcm beta value. Otherwise, my input impedance is too low, loading down the input source. If I increase Re to provide a high input Z, Zin, then my stage gain, Av = -Rc/Re, goes down. So a high Zin requires a high (hfe+1)*Re, but a high gain Av requires a high Rc/Re which means a LOW Re. If hfe is very large I can have both.
But, if my wcm beta is 100, and it can vary up to 600, then the overall performance must hardly change at all. In other words, I design my amps so that if the worst case min is 100, and the max is 600, the difference in overall gain is too small to notice. This is a good design because it possesses "beta immunity". But if the wcm beta is only 50, then the gain and Zin cannot both be as good as in the case with a 100 wcm beta. That is why beta is all important.
The wcm should be as high as you can obtain w/o comprimising other important parameters. A good topology will work very consistently with any beta value at or above the wcm value with negligible performance difference. This is beta immunity, a good thing.
Enough for now. BR.
Well said, Claude. I have not completely analyzed what you put down, but it is certainly in the right direction.
Young people forget that at one time BETA was almost everything! However, sometimes with low impedance drive, VOLTAGE became important. Often, people tried to optimize the voltage distortion with the current distortion to get the best linearity.
Young people forget that at one time BETA was almost everything! However, sometimes with low impedance drive, VOLTAGE became important. Often, people tried to optimize the voltage distortion with the current distortion to get the best linearity.
Thank you Claude!
That fanfares played by vendors forgot to play one tune: in order to make their devices easier drivable they have to make them easier crargable.
That fanfares played by vendors forgot to play one tune: in order to make their devices easier drivable they have to make them easier crargable.
Hi Edmond,
Thanks for the links.
Considering error correction, it was the Jan Didden's Pax amplifier which opened my eyes. In it, the implementation of EC is very simple to understand. I have been so fascinated by it that I then studied what I had about current conveyors and then on error correction and feedforward techniques, all became much more clear than before (still some difficulties with the Quad 405). This amp has real didactic virtues.
Thanks for the links.
Considering error correction, it was the Jan Didden's Pax amplifier which opened my eyes. In it, the implementation of EC is very simple to understand. I have been so fascinated by it that I then studied what I had about current conveyors and then on error correction and feedforward techniques, all became much more clear than before (still some difficulties with the Quad 405). This amp has real didactic virtues.
john curl said:
Greetings John. You also posted a very good treatise earlier on this subject. You display the experience of someone who has designed a lot of products. You really know your stuff.
I was only trying to emphasize to young folks (or anyone as well) that one cannot ignore the importance of beta simply because beta variations should be nulled out. The difference between "windfall" and "worst case minimum" beta value must be fully explained and dealt with.
I believe that back in the '50's all the way until today, the one parameter that still is all important is *alpha*. In order to make amp performance predictable and precise, we must lock our design around parameters that are very stable wrt temperature, speciman, & loading. All bjt amps, discrete or IC, hinge on 2 things, alpha, and resistor tolerance.
By establishing a value of emitter current, Ie, we know the quiescent Ic from the ever familiar Ic = alpha*Ie. But if we attempt to control Ic with Vbe, it is virtually impossible. The equation:
Ic = alpha*Ies*exp((Vbe/Vt)-1)) is too unpredictable. Ies, the saturation current of the b-e junction, varies non-linearly with temperature. The Vbe vs. Ic curve is wilder and crazier than beta. From -40 deg C, to +125 C, for a given Vbe of 0.65V, Ic can vary in the millions.
This is why we never directly use a low-Z voltage source to directly control Vbe. The bjt would be incinerated. As temp increases due to self heating, Ies increases dramatically. Voltage drive of Vbe is thermally unstable, resulting in runaway. But by controlling Ie, we know Ic within 1%. Alpha is around 0.99 give or take 0.01. The gm value directly varies with Ic, and also is temperature dependent. From -40 deg C to +125 deg C, gm increases by 71%. But a well designed amp would exhibit variation much much less. This is why we never design amps that rely on beta, gm, or Ies. They are unpredictable. Good designs are beta-proof, gm-proof, and Ies-proof, as well as Early-proof.
Instead we make use of the Gibraltar-like alpha value, and resistor tolerance. If resistors had a tolerance of -50%, plus 100%, we would be unable to precisely control gain. Resistors with tolerance under 1% are easily obtainable.
Alpha and resistor tolerance is what bjt amp stages rely on. Mine do, anyway.
Functional vs. causal
Voltage controlled vs. current controlled BJT, from where a circuit designer looks, it doesn't matter much. Both models should lead to the same results and I fully agree with the requirement to design circuits which are as much as possible independent to all intrinsic device parameters and leave the performance in the trusted hands of passive devices (BTW, negative feedback fits in this picture as well).
However, if we have to chose one, based on the intrinsic device physics, the voltage controlled model is the right one. The root cause of the "current controlled" misconception is the lack of distinguishing a functional relation from a causal relation. In a BJT the base current must always exist, but the transistor effect is not causally dependant on such. Base current is an effect that results from an application of a voltage at the base emitter junction, and here's how it goes.
The BE junction is obviously a diode and hence the exponential dependence of current vs. Vbe holds. By applying Vbe, the emitter injects this current in the thin base, where part of the associated charge recombines. The recombination charge/current is what defines the base current, together with a diffusion component! The rest of the injected charge/current reaches the CB junction, where it is transferred to the collector. Simplified, the thinner the base, the less charge recombines, therefore the higher the beta. However, the base can't be infinitely thin, as it has the support part of the CB space charge region, without breakdown or punch through.
As you see, strictly from a causal perspective, the BJT is a voltage controlled device. In any decent book on device physics you'll though find this dilemma (current vs. voltage controlled) as irrelevant, and this is for over 30 years.
Now, I could live with the current controlled model without any problems. However, current controlled models have (to me) the big problem of not relating the small signal parameters to the device bias point. I don't care about Ib=Ic/beta, but I do care about gm=Ic/Vt and I may care, depending on the application, of r(pi)=beta/gm and r0=Va/Ic as in the hybrid-pi model.
On the large signal side, what you described (alpha, etc...) pretty much fits the Ebers-Moll model. Which is correct but, as much the hybrid-pi model replaced the hij parameters mode, was superseded by the Gummel Poon charge-control model which again makes use of the causality in the physical device. If you follow the Gummel Poon model, you'll find out that beta forward and beta reverse are derived quantities, that can be expressed as a function of the device/material properties. And it can easily be shown that, in certain simplifying circumstances, the Gummel Poon model reduces to the Ebers Moll model.
Again, current controlled or voltage controlled, it doesn't really matter as long as the circuit results are identical. The voltage controlled model is though closer to the device physics and respects the causality principle.
Voltage controlled vs. current controlled BJT, from where a circuit designer looks, it doesn't matter much. Both models should lead to the same results and I fully agree with the requirement to design circuits which are as much as possible independent to all intrinsic device parameters and leave the performance in the trusted hands of passive devices (BTW, negative feedback fits in this picture as well).
However, if we have to chose one, based on the intrinsic device physics, the voltage controlled model is the right one. The root cause of the "current controlled" misconception is the lack of distinguishing a functional relation from a causal relation. In a BJT the base current must always exist, but the transistor effect is not causally dependant on such. Base current is an effect that results from an application of a voltage at the base emitter junction, and here's how it goes.
The BE junction is obviously a diode and hence the exponential dependence of current vs. Vbe holds. By applying Vbe, the emitter injects this current in the thin base, where part of the associated charge recombines. The recombination charge/current is what defines the base current, together with a diffusion component! The rest of the injected charge/current reaches the CB junction, where it is transferred to the collector. Simplified, the thinner the base, the less charge recombines, therefore the higher the beta. However, the base can't be infinitely thin, as it has the support part of the CB space charge region, without breakdown or punch through.
As you see, strictly from a causal perspective, the BJT is a voltage controlled device. In any decent book on device physics you'll though find this dilemma (current vs. voltage controlled) as irrelevant, and this is for over 30 years.
Now, I could live with the current controlled model without any problems. However, current controlled models have (to me) the big problem of not relating the small signal parameters to the device bias point. I don't care about Ib=Ic/beta, but I do care about gm=Ic/Vt and I may care, depending on the application, of r(pi)=beta/gm and r0=Va/Ic as in the hybrid-pi model.
On the large signal side, what you described (alpha, etc...) pretty much fits the Ebers-Moll model. Which is correct but, as much the hybrid-pi model replaced the hij parameters mode, was superseded by the Gummel Poon charge-control model which again makes use of the causality in the physical device. If you follow the Gummel Poon model, you'll find out that beta forward and beta reverse are derived quantities, that can be expressed as a function of the device/material properties. And it can easily be shown that, in certain simplifying circumstances, the Gummel Poon model reduces to the Ebers Moll model.
Again, current controlled or voltage controlled, it doesn't really matter as long as the circuit results are identical. The voltage controlled model is though closer to the device physics and respects the causality principle.
Claude Abraham said:
Sorry folks this is a pile of **** end of story. We are at a a crossroads either we move forward or mire in the muck.
Scott;
while you are here, can you answer please my question regarding Zetex VS Toshiba?
Toshiba has nice low noice devices with flat beta on Ic dependence.
Zetex has devices capable of amperes, with Icb of less than few picoamperes guaranteed, with milivolts when saturated, but they don't position their devices for low-noise inputs.
Why?
while you are here, can you answer please my question regarding Zetex VS Toshiba?
Toshiba has nice low noice devices with flat beta on Ic dependence.
Zetex has devices capable of amperes, with Icb of less than few picoamperes guaranteed, with milivolts when saturated, but they don't position their devices for low-noise inputs.
Why?
If bjts are voltage driven ....
Then shouldn't an inverting riaa amp be viable ?
That helps with 'noise gain' right ?
Say something like a 1/2 5532 ........
cartridge connected to inverting input
(series capacitor optional (suggested))
Positive input to ground (split supplies)
Or maybe a 072 type ?
Then shouldn't an inverting riaa amp be viable ?
That helps with 'noise gain' right ?
Say something like a 1/2 5532 ........
cartridge connected to inverting input
(series capacitor optional (suggested))
Positive input to ground (split supplies)
Or maybe a 072 type ?
Syn08, you have made one of the most sensible commentaries on this subject that I have seen yet!
Re: Functional vs. causal
Sorry, but I am well aware of the difference between a causal vs. functional relationship. You've been led astray by a web site I won't name. I know not all, but more than you give me credit for. "Misconception". I'm in good company. Every semi maker above describes bjt as current controlled. Do you produce bjt?
I'm informing you that if you examine all 3 equations for collector current:
1) Ic = beta*Ib.
2) Ic = alpha*Ies*exp((Vbe/Vt)-1).
3) Ic = alpha*Ie.
You will realize that all 3 of the above equations are functional relations, as none are causal. I don't think you really know semiconductor physics. You're just repeating what (KA) puts on his web site. Have you taken courses in semiconductor physics? KA hasn't. Yet you take him at his word and reject Fairchild, On Semi, TI, etc. They've forgotten more than KA ever knew. His material is not peer-reviewed and he's free to say anything w/o challenge.
As far as your "causal" argument goes, since I can remember, a handful of electronics buffs, techs, etc. have insisted that voltage is "causal" and current is an effect. The laws of e/m field theory lend this idea no support whatsoever. Ib & Vbe can never exist independently. In the ac domain, ib changes before vbe. The effect can never precede the cause. Before vbe can increase, the charge transport across the junctions must first increase. Then with a larger number of minority carriers in the depletion region, the b-e junction voltage will increase. You cannot increase vbe w/o first increasing ib. No causality here.
Another solid refutation of your causality theory is conservation of energy. The E field is related to the charge transport from b to e for h+ and vice-versa for e- from e to b. An increase in ic requires an increase in E field. To increase the E field requires an increase in energy since w = 0.5*D*E, where D = epsilon*E. But an increase in energy takes finite time, t. If w increases over time t, then dw/dt is non-zero. But dw/dt, the time rate of change of energy, is just the power, P. So dw/dt = P = non-zero = ib*vbe. Since power is the product of both current and voltage, and power is non-zero, then both i & v are non-zero as well.
It is impossible to change ic (collector current small signal value or ac) unless ib & vbe both change as well. It's impossible to change one without the other. Neither is "causal", but both participate.
Also, to say that vbe is "causal" while ib is not indicates a very limited understanding of semi physics. To create Ic, what is required? Well, we need to reverse bias the c-b junction. The voltage is Vcb. But to change Vcb from 0 to nonzero requires displacement current to move and separate charges across the b-c junction. Hence Icdisp is necessary. Of course we must forward bias the b-e junction as well. Only a superconductor can carry I w/o V. Only an insulator can carry V w/o I. A semiconductor is so named because it's neither insulator nor conductor but "semi". No Ib can exist w/o Vbe, and vice-versa. Also, Ie is included with Ib & Vbe.
You will never be able to show me a bjt with Ib, Vbe, & Ie independent. These 3 can only exist all in unison, or all 3 are zero. No exceptions at all.
So we need Vbc, Icdisp, Ib, Vbe, & Ie to set up conditions to "cause" bjt action. Also, if the base is not ultra-thin, it doesn't work. Hence Ic = alpha*Ie.
No single variable, Vbc, Ib, Vbe, Ie, alpha, etc. can "cause" bjt action. All of them in unison work together.
Nothing else even remotely makes sense.
FWIW, all 3 eqns. cannot be causal for yet another reason. There is no time involved. You cannot describe bjt in terms of device physics w/o time delays and speed limitations. Eqns 1), 2), & 3) do not hold when frequency f, approaches/exceeds ft the transition frequency. Tuesday I'll post the charge control eqtn including propogation time.
No disrespect, but I'd recommend doing serious research before lecturing me about misconceptions as if you are some kind of EE professor. I accept correction when I got it wrong, but your arguments are cut and pasted from a non-peer-reviewed web site of a contrarian who has very limited knowledge. Study the physics laws, and correct me if I erred. Peace and nothing personal.
syn08 said:
Sorry, but I am well aware of the difference between a causal vs. functional relationship. You've been led astray by a web site I won't name. I know not all, but more than you give me credit for. "Misconception". I'm in good company. Every semi maker above describes bjt as current controlled. Do you produce bjt?
I'm informing you that if you examine all 3 equations for collector current:
1) Ic = beta*Ib.
2) Ic = alpha*Ies*exp((Vbe/Vt)-1).
3) Ic = alpha*Ie.
You will realize that all 3 of the above equations are functional relations, as none are causal. I don't think you really know semiconductor physics. You're just repeating what (KA) puts on his web site. Have you taken courses in semiconductor physics? KA hasn't. Yet you take him at his word and reject Fairchild, On Semi, TI, etc. They've forgotten more than KA ever knew. His material is not peer-reviewed and he's free to say anything w/o challenge.
As far as your "causal" argument goes, since I can remember, a handful of electronics buffs, techs, etc. have insisted that voltage is "causal" and current is an effect. The laws of e/m field theory lend this idea no support whatsoever. Ib & Vbe can never exist independently. In the ac domain, ib changes before vbe. The effect can never precede the cause. Before vbe can increase, the charge transport across the junctions must first increase. Then with a larger number of minority carriers in the depletion region, the b-e junction voltage will increase. You cannot increase vbe w/o first increasing ib. No causality here.
Another solid refutation of your causality theory is conservation of energy. The E field is related to the charge transport from b to e for h+ and vice-versa for e- from e to b. An increase in ic requires an increase in E field. To increase the E field requires an increase in energy since w = 0.5*D*E, where D = epsilon*E. But an increase in energy takes finite time, t. If w increases over time t, then dw/dt is non-zero. But dw/dt, the time rate of change of energy, is just the power, P. So dw/dt = P = non-zero = ib*vbe. Since power is the product of both current and voltage, and power is non-zero, then both i & v are non-zero as well.
It is impossible to change ic (collector current small signal value or ac) unless ib & vbe both change as well. It's impossible to change one without the other. Neither is "causal", but both participate.
Also, to say that vbe is "causal" while ib is not indicates a very limited understanding of semi physics. To create Ic, what is required? Well, we need to reverse bias the c-b junction. The voltage is Vcb. But to change Vcb from 0 to nonzero requires displacement current to move and separate charges across the b-c junction. Hence Icdisp is necessary. Of course we must forward bias the b-e junction as well. Only a superconductor can carry I w/o V. Only an insulator can carry V w/o I. A semiconductor is so named because it's neither insulator nor conductor but "semi". No Ib can exist w/o Vbe, and vice-versa. Also, Ie is included with Ib & Vbe.
You will never be able to show me a bjt with Ib, Vbe, & Ie independent. These 3 can only exist all in unison, or all 3 are zero. No exceptions at all.
So we need Vbc, Icdisp, Ib, Vbe, & Ie to set up conditions to "cause" bjt action. Also, if the base is not ultra-thin, it doesn't work. Hence Ic = alpha*Ie.
No single variable, Vbc, Ib, Vbe, Ie, alpha, etc. can "cause" bjt action. All of them in unison work together.
Nothing else even remotely makes sense.
FWIW, all 3 eqns. cannot be causal for yet another reason. There is no time involved. You cannot describe bjt in terms of device physics w/o time delays and speed limitations. Eqns 1), 2), & 3) do not hold when frequency f, approaches/exceeds ft the transition frequency. Tuesday I'll post the charge control eqtn including propogation time.
No disrespect, but I'd recommend doing serious research before lecturing me about misconceptions as if you are some kind of EE professor. I accept correction when I got it wrong, but your arguments are cut and pasted from a non-peer-reviewed web site of a contrarian who has very limited knowledge. Study the physics laws, and correct me if I erred. Peace and nothing personal.
scott wurcer said:
If I've erred, I'll gladly accept correction. What issue do you have with my postings? Is it the physics, or my complementing John? I haven't insulted anyone, so please inform me what I've done that gives you heartburn.
No disrespect. I just want to know why you find my info so irritating. Fair enough? Peace to all.
Gentlemen;
Can you please stop fighting for a while?
Thank you.
Suppose, we have a NPN transistor.
We grounded a base and supplied it's emitter by current from + rail that has enough voltage for E-B junction to breakdown.
Now, what shall we measure on collector in respect to base?
Why? 😱
Can you please stop fighting for a while?
Thank you.
Suppose, we have a NPN transistor.
We grounded a base and supplied it's emitter by current from + rail that has enough voltage for E-B junction to breakdown.
Now, what shall we measure on collector in respect to base?
Why? 😱
Re: serendipity
Hey!
I modified your idea for a compensated nfb clamp to adapt it to a unipolar VAS. See the attachment to post 16 here to see it added to the improved version of Cherry's NDFL amp I came up with.
It fixed the clipping issue, but it doesn't stop the Rush current amplifier / intergrator stage throwing a fit when the amplifier is slew rate limiting, so I abandoned the idea.
The other issue is that the VAS must be capable of supplying enough current to drive the feedback network to reliably track an over-voltage input signal. This is a limitation when low value feedback resistors are used for low noise.
Cheers,
Glen
Edmond Stuart said:
Hey!
I modified your idea for a compensated nfb clamp to adapt it to a unipolar VAS. See the attachment to post 16 here to see it added to the improved version of Cherry's NDFL amp I came up with.
It fixed the clipping issue, but it doesn't stop the Rush current amplifier / intergrator stage throwing a fit when the amplifier is slew rate limiting, so I abandoned the idea.
The other issue is that the VAS must be capable of supplying enough current to drive the feedback network to reliably track an over-voltage input signal. This is a limitation when low value feedback resistors are used for low noise.
Cheers,
Glen
- Status
- Not open for further replies.