Maker Pro
Maker Pro

Voltage Drop across NPN BJT C&E?

hevans1944

Hop - AC8NS
Jun 21, 2012
4,878
Joined
Jun 21, 2012
Messages
4,878
What I do find encouraging is that the engineers and scientists who actually design BJTs, and other semiconductor devices, do know what is going on. However, once the packaged device leaves the fab, it becomes a "black box" and it is open season for all sorts of "explanations" for what is going on inside.

You want to test your "knowledge" of BJTs? Try building a temperature-compensated logarithmic amplifier with six decades of response, say from one nanoampere to one milliampere of input current, using two BJTs and a few low bias-current op-amps. You might want to start with Linear Technology LT1008 op-amps or equivalent and carefully match, and temperature-control, your BJTs. Back-to-back epoxy bonding of the two BJTs and a small temperature-controlled oven will help. When I was building log amps in the 1970s, I could purchase matched NPN transistors built on an internally heated substrate, all in a nice TO-5 metal can. That was almost too easy, but I haven't been able to find any lately. The 24-bit delta-sigma ADC pretty much killed any further interest in precision log amps for data acquisition, although vanilla log-amps are still used for analog signal processing.

It is advisable to use ultra-clean ceramic circuit boards and Teflon insulation to control leakage currents. I had to go with FR4 boards, Teflon insulated solid wire, Teflon insulated stand-off terminals, and Teflon-insulated sockets. I don't think I ever quite reached six decades without substantial drift on the low end, but what I built was good enough to monitor a PIN photo-diode output.

It was an interesting one-off design experience with no follow-up. The effect they were looking for with the photo-diode (and also with a photo-multiplier tube or PMT) was a time delay in the Faraday rotation of linearly polarized light passing through a liquid sample, said delay to be characteristic of its chemical structure. Apparently the effect does not exist, or at least was not reproducible with our equipment.

If interested, you can Google the Allison Magneto-optical Effect. It is an example of pathological science. Fred Allison (July 4, 1882 - August 2, 1974) went to his grave still firmly believing he had discovered a unique and simple way to identify trace elements and compounds. AFAIK, the U.S. Air Force funded the last experiments trying to prove that Allison's apparatus (carefully replicated in a lab in Area B at Wright-Patterson AFB) worked and could be used to find trace contaminants in hydraulic fluid and engine lubricating oils. It would have been worth millions if that had panned out. The apparatus was dirt cheap compared to conventional spectroscopic analysis of fluid contaminants. I have a copy of the final report.

Part of that research funding payed my minuscule technician salary for a few months, along with a couple of PhD professors, a graduate-student engineer-employee, another full-time electrical engineer, and who knows how many civil servants and Air Force officers "supervising" the program over the course of a year or so. But that's the cost of basic research. Sometimes you spend big bux and it leads to a dead end.

Maybe someone will resurrect Allison's experiments with more modern electronics and better electro-optics than what we had available in the 1970s, which was not much more than what Allison used in the 1930s... some Lecher lines with trolleys to hold hand-wound coils containing the sample and a "reference" sample; a spark-gap discharge to send an electrical pulse down the Lecher lines; a pair of rotate-able, crossed, Nicol linear polarizing prisms to extinguish the light from a visible red-light HeNe laser equipped with a high-speed shutter (Allison used a spark gap illumination source); some photo detectors (Allison used his Mark I eyeball); and some Tektronix oscilloscopes with film cameras to record the traces. The laser shutter was synchronized to open with the spark-gap discharge that pulsed the Lecher lines.

You started out with two coaxial "reference" samples, typically carbon disulfide that has a known Verdet constant and "zeroed" the Nicol prisms, one in front of the light source and the other at the other end of the Lecher lines, for extinction of the laser beam. Then, pulsing the Lecher lines, the laser beam polarization is rotated when it passes through the sample in the first coil. After a time-of-flight delay the beam passes through the second coil which rotates the polarization back to its original orientation to extinguish the beam passing through both coils in succession. So far, so good. Just a complicated demonstration of Faraday rotation in a liquid.

You can demonstrate Faraday rotation with just one coil and DC excitation of the coil by observing how far the second (analyzing) prism has to be rotated to extinguish the beam again when there is current (and a magnetic field) in the coil. The Verdet "constant" is usually measured this way. What Allison was claiming was (1) a delay in the onset of the Faraday effect, (2) that his Lecher-line apparatus could measure this delay, and (3) the measured delay was characteristic of the elemental composition of the sample. It is said that Ernest O. Lawrence, the inventor of the cyclotron, visited Allison's lab one day, peered into the telescope, adjusted the Lecher lines, then turned around and said, "Gentlemen, this man is perpetrating a fraud!" That should have ended Allison's career, but he was adamant that his apparatus worked and spent the rest of his life trying to prove it.

Fast forward now to the 1970s. Fred is dying but he has the ear, and the funding, of someone willing to try his apparatus one more time. Now you build his apparatus and replace one of the samples with an unknown sample and measure the trolley positions where extinction occurs. Allison claimed these positions (there were supposed to be several) were indicative of sample composition... right down to parts per billion sensitivity. Problem was, it was difficult to find the extinction point while manipulating the sample trolley along the Lecher lines. You were looking for something that wasn't there, viz., the absence of light at each extinction position. Multiple observers reported different extinction points. Instrumentation with photo-diodes and PMTs yielded no conclusive results, although measurements were difficult because of the large RF field around the pulsed Lecher lines. Maybe better shielding was needed?

It is a simple experiment using simple technology, but I sure won't be involved in resurrecting it. I still have yet to build my cold-fusion design that will forever free me from the local power grid. Maybe the profits from that would allow me to revive a search for the Allison Effect. And I don't mean audio acoustics, which is the first thing a Google search turns up..

So, if construction of a log-amp is too much trouble, and five decades of input range is good enough for your application, purchase a Maxim MAX4206 with all the design done for you. But at least try to play around with log-amp circuits when you are ready for it. They would make a nice LDR interface to a microprocessor ADC even without temperature compensation. Your spiffy LDO will help you see what is going on. Enjoy your electronics adventure. Have fun!

Hop
 

LvW

Apr 12, 2014
604
Joined
Apr 12, 2014
Messages
604
Given a black box with these functions, you could not prove that the base current is not controlling the collector current. So, given that it is a linear function instead of an exponential function, we choose to use it in calculations. This is not wrong since it provides the same answer.
Bob
Bob - with all respect, this is another phenomenon to me: Some people believe they would use the current-control view in their calculations - but in reality they don`t!
They apply voltage feedback (using Re) for creating a voltage VE (caused by the desired current IE) and they design a low-resistive base voltage divider (as low-resistive as allowed with respect to other constraints like power consumption and input resistance) for supplying the base with a "stiff" voltage VB. Hence, they make a design based on VBE=VB-VE (assuming a good guess for VBE). Such a guess is sufficient because of the feedback effect.
And, of course, during calculations they consider the base current Ib=Ic/B as a small portion of the divider current (which is pretty uncertain due to heavy B-tolerances).
Nevertheless, they still think, that they have assumed the current-control view durung these design steps.
At which step???

Fazit: Not neglecting the base current does not mean that we have assumed any controlling function!
 
Last edited:

chopnhack

Apr 28, 2014
1,576
Joined
Apr 28, 2014
Messages
1,576
So, if construction of a log-amp is too much trouble, and five decades of input range is good enough for your application, purchase a Maxim MAX4206 with all the design done for you. But at least try to play around with log-amp circuits when you are ready for it. They would make a nice LDR interface to a microprocessor ADC even without temperature compensation.
Wow... The application note looks very interesting, I have only read the abstract and first paragraph, good history there.

I thought the comparator would be sufficient for the LDR setup, since we only need to know on or off and not really caring for a specific triggering point nor values at either end of the range. That is why I left the trimmer in so that the comparator could be calibrated to make up for the differences in LDR or in light intensity at the installation site. At least that is what I thought!

On the brighter side of things, delivery of the FET's should occur today according to the information from the good ole USPS. I guess I should go and drill out those holes in the adapter board and let the epoxy setup hard enough to withstand plugging into the breadboard.
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Everyone here knows that I have been playing Whack-a-Mole with every BJT base current control freak that pops up here. They just can't seem to understand that base current is an indicator of the emitter/collector current, not the control of it.

The physics of the BJT determine what is in control. And, Vbe controls the diffusion, which in turn controls the emitter/collector current. Also, because the BJT is a diffusion device, it has a nonlinear curve for Vbe vs Ie/Ic.

Ratch
Yes, and you cannot see that it does not matter whether it is base current or Vbe that controls the collector current. The two are precisely related, and using a black box model it is equally correct to consider either one controlling the collector current.

Bob
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Bob - with all respect, this is another phenomenon to me: Some people believe they would use the current-control view in their calculations - but in reality they don`t!
They apply voltage feedback (using Re) for creating a voltage VE (caused by the desired current IE) and they design a low-resistive base voltage divider (as low-resistive as allowed with respect to other constraints like power consumption and input resistance) for supplying the base with a "stiff" voltage VB. Hence, they make a design based on VBE=VB-VE (assuming a good guess for VBE). Such a guess is sufficient because of the feedback effect.
And, of course, during calculations they consider the base current Ib=Ic/B as a small portion of the divider current (which is pretty uncertain due to heavy B-tolerances).
Nevertheless, they still think, that they have assumed the current-control view durung these design steps.
At which step???

Fazit: Not neglecting the base current does not mean that we have assumed any controlling function!
You are making some assumptions there about how I calculate bias, which are incorrect. First of all, one does not want a stiff divider, this will lower the input impedance too much. One wants the biasing resistors to be as large as possible while still supplying the required base current, which is the way I calculate them.

Bob
 

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
Yes, and you cannot see that it does not matter whether it is base current or Vbe that controls the collector current. The two are precisely related, and using a black box model it is equally correct to consider either one controlling the collector current.

Bob
I don't have to mention that the relationship between base and emitter/collector current (Beta) is not precise between two different BJTs, do I? Beta can have quite a range. Designs that depend on beta either use an excessive amount of feedback or are prone to unreliable performance. See http://cr4.globalspec.com/thread/68055#comment720374

Ratch
 

chopnhack

Apr 28, 2014
1,576
Joined
Apr 28, 2014
Messages
1,576
The physics of the BJT determine what is in control. And, Vbe controls the diffusion, which in turn controls the emitter/collector current. Also, because the BJT is a diffusion device, it has a nonlinear curve for Vbe vs Ie/Ic.
This is what makes it difficult for a hobbyist to use these devices as a simple switch, there is a lot more going on than a simple on/off action!
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
I don't have to mention that the relationship between base and emitter/collector current (Beta) is not precise between two different BJTs, do I? Beta can have quite a range. Designs that depend on beta either use an excessive amount of feedback or are prone to unreliable performance. See http://cr4.globalspec.com/thread/68055#comment720374

Ratch
I have never seen a datasheet that gives any relationship between Vbe and Ic much less a precise one. So how is trying to control the current by controlling Vbe any easier?

We all know that the BJT transistors must be designed to be insensitive to the exact beta.

Bob
 

LvW

Apr 12, 2014
604
Joined
Apr 12, 2014
Messages
604
Yes, and you cannot see that it does not matter whether it is base current or Vbe that controls the collector current. The two are precisely related, and using a black box model it is equally correct to consider either one controlling the collector current.
Bob
Forgetting the black box model (that has - for my opinion - no meaning in a technical/physical discussion like this) I have one simple question to you:
When a newcomer comes to you asking for the difference between a class-A and a class AB (or class-B) amplifier - what would be your explanation?

Another question: Seriously - do you really think that "it does not matter".....because the two are "precisely related"?
Is "correlation" the same as "causation"?
 
Last edited:

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
This is what makes it difficult for a hobbyist to use these devices as a simple switch, there is a lot more going on than a simple on/off action!
A switch is easy. Zero volts on the base, and no current will be present in the emitter/collector. Applying some forward voltage on the base to saturate the BJT (not much is needed), will turn on the switch.

Ratch
 

LvW

Apr 12, 2014
604
Joined
Apr 12, 2014
Messages
604
First of all, one does not want a stiff divider, this will lower the input impedance too much. One wants the biasing resistors to be as large as possible while still supplying the required base current, which is the way I calculate them.
Bob
....as large as possible? Are you aware that (a) this contradicts all the common design strategies and (b) such a design degrades the desired feedback effect caused by RE?
 

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
I have never seen a datasheet that gives any relationship between Vbe and Ic much less a precise one. So how is trying to control the current by controlling Vbe any easier?

We all know that the BJT transistors must be designed to be insensitive to the exact beta.

Bob

Controlling a BJT by Vbe is difficult, if not impossible. I was not advocating designing that way. But by knowing what controls a BJT, one can apply alternate design techniques. Did you look at the link I sent?

Ratch
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
A switch is easy. Zero volts on the base, and no current will be present in the emitter/collector. Applying some forward voltage on the base to saturate the BJT (not much is needed), will turn on the switch.

Ratch
Really? So next time I design a BJT switch I should just apply a stiff 3V to the base with the emitter grounded?

I just looked at the PN2222A datasheet. The Vbe at saturation is anywhere from 0.6V to 1.2V. Please tell me what voltage I should put on the base to saturate it. "Some forward volage" does not work well.

Bob
....as large as possible? Are you aware that (a) this contradicts all the common design strategies and (b) such a design degrades the desired feedback effect caused by RE?
As large as possible was a bad choice of words. Just a "stiff" was in you post. If you want a still voltage, the resistors will lower your input impedance to to the point that the amplifier is useless. In reality it is somewhere in between.


Bob
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Controlling a BJT by Vbe is difficult, if not impossible. I was not advocating designing that way. But by knowing what controls a BJT, one can apply alternate design techniques. Did you look at the link I sent?

Ratch
And I started my response in this thread by saying that I do know that it is actually controlled by Vbe, which I have never disputed. So I guess we are in agreement.

Bob
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Forgetting the black box model (that has - for my opinion - no meaning in a technical/physical discussion like this) I have one simple question to you:
When a newcomer comes to you asking for the difference between a class-A and a class AB (or class-B) amplifier - what would be your explanation?

Another question: Seriously - do you really think that "it does not matter".....because the two are "precisely related"?
Is "correlation" the same as "causation"?
I originally stated that I know that is is Vbe that controls the current, so no, I have never claimed that Ib controls Ic, only that pretending it does produces the correct results.

And I have no idea what you are talking about with the class A and class AB example. I would not tell anyone that their operation is the same any more than I would tell someone that the operation of a MOSFET is the same as that of a BJT

Bob
 

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
Really? So next time I design a BJT switch I should just apply a stiff 3V to the base with the emitter grounded?

I just looked at the PN2222A datasheet. The Vbe at saturation is anywhere from 0.6V to 1.2V. Please tell me what voltage I should put on the base to saturate it. "Some forward volage" does not work well.

Bob

Of course not. But, just enough forward voltage to saturate the transistor reliably. You will have to calculate it. It will depend of Vc, Rc,Rb, and Re.

Ratch
 

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
There is another problem with having a high base resistance. It is thermal instability. Whenever the reverse bias voltage on the base collector circuit is applied, the Icbo generator fires up and sends its current into the base terminal. Then the current splits, with some of it going into the base emitter junction, and the remainder going out to the base circuit. If it goes into the base emitter circuit, it becomes "betatized" and greatly increases the emitter/collector current. There is no way to stop this current from being generated. It is thermally powered, and doubles with every 6° C increase. It can be diverted harmlessly out to the base circuit if the base resistance is low and the emitter resistance is high. You probably will not have any problems if you just use silicon BJTs for signal applications at room temperatures. But for power transistors, good heat sinking and attention to the instability factor is a wise precaution. Also, silicon transistors with their inherently lower Icbo help a lot. Thermal runaway can occur if the increase in collector current causes a higher temperature to cause the Icbo to increase further to form a positive feedback. It becomes a race between the thermal generation and the dissipation of the heat sink.

Ratch
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
Of course not. But, just enough forward voltage to saturate the transistor reliably. You will have to calculate it. It will depend of Vc, Rc,Rb, and Re.

Ratch
Show me how I can calculate that from parameters I find in the datasheet.

Bob
 

Ratch

Mar 10, 2013
1,098
Joined
Mar 10, 2013
Messages
1,098
Show me how I can calculate that from parameters I find in the datasheet.

Bob
You can't. You have to know the component values of your circuit. A BJT is not an inherent switch, anymore than it is a inherent current amplifier. But, it can be made into either by the right circuit and component values.

Ratch
 

chopnhack

Apr 28, 2014
1,576
Joined
Apr 28, 2014
Messages
1,576
A switch is easy. Zero volts on the base, and no current will be present in the emitter/collector. Applying some forward voltage on the base to saturate the BJT (not much is needed), will turn on the switch.

Ratch

Perhaps you find it easy, but I don't find it to be trivial. Especially trying to design a circuit to have a clean on/off controlled by the presence/absence of 2-3v. The difficulty I allude to is selecting an appropriate transistor to do the job based on the datasheet.
 
Top