Happy Holidays folks! It's good to relax for a couple of weeks, huh?
I don't remember exactly what I read, but I recall a device that stated that its output is dependent on the output current, so essentially it was labeled current output device. Assuming if the current in the device is very small and requires amplification, do I avoid using FET opamps?
I was just reading the differences between BJTs and FETs, and it says that FETs are voltage controlled, so the gate is essentially isolated from the source and drain.
In JFETs, there is a gate-channel diode so you can get current. In MOSFETs, there is an SiO2 insulator between the gate and the channel so you will not get any gate current unless the gate is overvoltaged (that destroys the device).
FETs are voltage controlled current sources and BJTs are current controlled.
The choice of an OP AMP is often trivial, but it can be quite challenging ... entire books on this subject!
No, the idea of an opamp is to have a low (zero is ideal) output impedance (can drive heavy loads) and infinite input impedance - that's were FETs come in.
A FET is close to a vacuum toobe - bias the grid (gate) with a signal and virtually no current flows from the signal source. Very handy when you need a close to ideal op amp - input current in the small nanoamps. Outstanding for electret mikes, analog sample and hold, and anywhere a really high input impedance is needed.
Downside? mainly cost.
----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==----
The #1 Newsgroup Service in the World! 120,000+ Newsgroups
----= East and West-Coast Server Farms - Total Privacy via Encryption =----
He he, I know one off the major ATE makers that unintentionally _did_ it on a precision measurement muxing JFETs circuit (they, well the guy in charge, thought that with a 10M gate resistor it didn't matter) and got unexpected fluctuations in their calibration sequences. I had to go by length to make them admit the gate current just made for some VDS voltage buildup (you know, all that trusted design review staff couldn't have been so wrong for so long).
The Ebers-Moll equations say no such thing. The emitter current in a bjt may be expressed as:
1) Ie = (beta + 1) * Ib, or as 2) Ie = Ies * (e^(Vbe/Vt) - 1). Likewise the collector current may be expressed three ways. 3) Ic = alpha * Ie, or alternately, 4) Ie = beta * Ib, or we may combine equations 2) and 3) to obtain 5) Ic = alpha * Ies * (e^(Vbe/Vt) - 1). Equation 5) is the one being referred to above. It is derived from equations 2) and 3). Many critics when quoting Ebers-Moll, omit the "alpha" factor, which is all important. "Alpha" is the ratio of Ic to Ie, and measures how effective the bjt functions as a current amplifier, aka common base forward current gain.
In a bjt, there are but 3 terminals, 2 of which constitute the control electrodes, namely base and emitter. Since current by its very nature is a *through* quantity, whereas voltage is an *across* quantity, there are 3 input quantities associated with the base and emitter terminals. They are base current Ib, base-emitter voltage Vbe, and emitter current Ie. Neither of these 3 quantitiea could exist without the other two, nor can either two exist without the other one. There are no exceptions, under static or dynamic conditions. Thus Ib, Vbe, and Ie are mutually and intimately related. The collector current may be expressed as a function of either of the three, Ib, Vbe, or Ie.
To "control" the collector current, we must control one of the input variables, and let the others and the collector current be defined by the device characteristics and laws of physics. If we control Ib directly, then 3) Ic = beta * Ib, and 6) Vbe = Vt * ln((((beta +
1)*Ib)/Ies) + 1). The problem with this approach is known as "beta dependency", as beta varies widely with specimen and temperature. This approach is only used when the bjt is used as a switch, toggling betwen cutoff and saturated states. By driving the base with current greater than Ic/beta, using minimum worst case beta value, assures saturation.
If we attempt to control Ic with Vbe, another problem exists. "Ies", the base-emitter junction reverse saturation current in the Ebers-Moll equation, is a very strong function of temperature. In fact, 7) Ies = Ieso * e^(a*(T-To)), where Ieso is the reverse saturation at a reference temperature (usually 25C or 298K, or 300K), and To is the ref temp. I've designed logarithmic amps using diodes and bjt's. For a
1N914 axial diode, Ieso is around 3 to 5 nanoamp, and "a", the temperature coefficient is about 0.35 neper / degree K. If a voltage source, let's say 0.65V dc, is impressed across the base-emitter junction, the current is given by Ie = Ies * (e^(Vbe/Vt) - 1), and of course, Ic = alpha * Ie. But Ies = Ieso * e^(a(T-To)). Observe as follows. The voltage Vbe is held fixed (voltage drive or control ), and Ic and Ie are established. The product of Vbe and Ie is *power*, which is dissipated by the bjt device. The temperature can only
*increase* due to non-zero power. This temp increase results in an
*increase* in Ies, the b-e junction reverse saturation current. Since Ie = Ieso * e^(a*(T-To)) * ((e^(Vbe/Vt)) - 1), Ie must increase due to the increase in Ies. The temperature will increase even further, so that Ies increases further, increasing temp further, etc. This is clearly a runaway condition. Any p-n junction, such as a diode, LED SCR gate to cathode, bjt base to emitter, cannot survive and must never be *voltage-driven*. Controlloing Ie and/or Ic with Vbe is futile, and results in thermal runaway. Under no circumstances are bjt's ever voltage-driven or controlled.
If we attempt to control Ic by controlling Ie, the relation is given by
3) Ic = alpha * Ie. Alpha varies for around 0.98 to 0.998 over specimen and temperature range. Alpha is very stable and predictable. The base-emitter voltage is given by 6) Vbe = Vt * (ln((Ie/Ies) + 1)). The temperature behavior is markedly different. If we fix the emitter current Ie, Vbe is given above, and the power dissipated by the bjt is non-zero. As a result, the temperature must increase. An increase in temperature incurs an increase in Ies. But observing equation 6), an increase in Ies incurs a *decrease* in Vbe, the opposite of the voltage-controlled case. This is very desirable, because the power in the p-n junction will now decrease. Thermal stability is insured. To be fair, the base driven scenario above, is also thermally stable. The problem with base current control is beta-dependency, but it is thermally stable. The thermal characteristics of forward biased p-n junctions mandate the they are always "current-driven" or "current-controlled". This is what the phrase "current-controlled" means. We fix the *current* to some value, and the voltage is determined by physics and device characteristics. Both are equally important and Ic won't exist without both of them. Ditto for a FET.
To summarize, it is impossible for collector current to exist in a bjt, without Ib, Vbe, and Ie, all together in unison. You can never have one without the other two. Neither of the 3 quantities Ib, Vbe, and/or Ie, is responsible for "causing" the other two, or Ic. The collector current may be expressed as a function of Ib, Vbe, or Ie. Just as Ic = beta * Ib, it is equally true that Ib = Ic / beta. Just as Ic = alpha
Ies * (e^(Vbe/Vt) - 1)), it is equally true that Vbe = Vt * ln((Ic/(alpha*Ies)) + 1). The order in which the variables appear does not imply a pecking order or "cause/effect" relation. Neither Ib, nor Vbe, nor Ie, can ever be the sole cause of Ic. All 3 work together. There is no pecking order. Does this help?
Just how does the E-M eqtn show that bjt's are "voltage controlled"? You say that a delta Vbe incurs a *predictable* delta Ic, which is preposterous. The temperature variation makes this impossible. If I tell you that a certain bjt, a 2222A just for example, has Vbe = 0.65V, and asked you what is the collector current, you could not possibly predict it. At 0 deg C, vs. 100 deg C junction temperature, the result could vary by a factor of a million. This is due to Ies being an exponential function of temperature. If I told you that Ib = 1.0 mA, you could make a rough approximation that Ic is between 50 and 500 mA, accounting for part to part (speciman) and temperature variations of beta. This is known as "beta dependency", which good circuit designers make it a point to avoid. Now, if I told you that Ie was 100 mA, you or anyone could easily tell me the collector current. A value of 99 mA would be within 1% under all conditions. When it comes to predictable behavior, neither Ib nor Vbe can be relied upon. Only alpha, Ic/Ie, is predictable and consistent. In other words, Ic should be controlled by the emitter current Ie, for linear amplification applications. The same scenario takes place for the ac variation, delta Ic.
I don't think we interpret the expressions "voltage-controlled, and "current-controlled" the same way. Just because the E-M equation is more often expressed as current being an exponential function of voltage, you seem to think that the current, particularly Ie/Ic, exists due to the voltage Vbe. The E-M equation could just as well be written so that voltage is a logarithmic function of current. If y = e^x, then x = ln y. In other words, delta Ic cannot take place unless delta Ib/Vbe/Ie all change simultaneously. Only by controlling delta Ie can we obtain consistent and predictable delta Ic. Controlling delta Ib results in a delta Ic that varies widely, a factor of 10 or more with specimen and temperature due to beta variation. Controlling delta Vbe is worse than controlling Ib, incurring variation of 10 000 to a million. Only by controlling delta Ie can we safely predict delta Ic. Any university level textbook will confirm this. No crack circuit designer relies on Ib or Vbe to control Ic. Best regards.