Video Tutorial: Design and Characterization of a Single Transistor Amplifier

[Hugoneus]

In this episode Shahriar presents a tutorial on the design and characterization of a single-stage low-noise bipolar amplifier suitable for audio applications. Given a set of specifications, a common-emitter topology is investigated. The circuit employs a beta-insensitive biasing scheme which is simultaneously optimized for  maximum output swing. The small-signal gain of the circuit is calculated and the bandwidth is set for audio frequencies. A non-inverting operational amplifier is used as a second stage to achieve the desired overall gain. The circuit is assembled on a breadboard where the gain and bandwidth are measured and compared with design specifications. As the final experiment, the circuit is used to amplify signals from a microphone. All documents can be downloaded from TheSignalPath.

Video Link:
http://www.TheSignalPath.com/blogs/2012/09/23/tutorial-on-the-theory-design-and-characterization-of-a-single-transistor-bipolar-amplifier/

More Videos:
http://www.TheSignalPath.com/

[Mechatrommer]

long lived to you human.

[Hugoneus]

long lived to you human.

As to you.

[IanB]

I was watching carefully, but the video seemed to pass over the base biasing and the base resistors? There was much discussion of the collector and emitter resistors, but then RB1 and RB2 somehow seemed to just appear in the circuit. Was a segment missed when editing it together? (There also seems to be a duplicated segment around about 9m and about 11m20s, FYI.)

But those quibbles aside, a very nice tutorial video.

Is it a bit of a cheat using the op amp instead of a second transistor stage? It seems a bit odd using a single transistor to do most of the work and then throw in something built with dozens of transistors as a final buffer stage...it almost seems to go against the tutorial spirit of the piece. I'm sure the makers of my 1970's pocket transistor radios didn't have that luxury 

[poptones]

It would seem the entire point of the transistor is to "imbue" the audio with some of that rich second and third harmonic distortion. What sort of idiocy is this putting an op amp after a common collector "amplifier?" Why not include the transistor in the feedback path so as to minimize the distortion contribution? I can see putting the transistor in there to feed a power output stage, but not a simple op amp.

[Hugoneus]

I was watching carefully, but the video seemed to pass over the base biasing and the base resistors? There was much discussion of the collector and emitter resistors, but then RB1 and RB2 somehow seemed to just appear in the circuit. Was a segment missed when editing it together? (There also seems to be a duplicated segment around about 9m and about 11m20s, FYI.)

But those quibbles aside, a very nice tutorial video.

Is it a bit of a cheat using the op amp instead of a second transistor stage? It seems a bit odd using a single transistor to do most of the work and then throw in something built with dozens of transistors as a final buffer stage...it almost seems to go against the tutorial spirit of the piece. I'm sure the makers of my 1970's pocket transistor radios didn't have that luxury 

Thank you for the kind comments. I am in the process of fixing the duplicate section.

The VB and RB biasing section is essentially the thevenin equivalent of the RB1, RB2 and the +/- 2.5V supplies.

As for using the opamp, this circuit is intended for use in an analog electronics course lab work. I agree that a two stage amplifier with two transistors would be better. It is just for educational purposes to get the students to wire up an opamp as well.

[Hugoneus]

.. What sort of idiocy is this putting an op amp after a common collector "amplifier?" ...

There is no reason to insult.

[HLA-27b]

Great video! Seeing it as it is done teaches a lot.





On the other hand I am most ashamed that the makers of these very good videos should face derision and insult here. The obnoxious audiofool will not learn, neither electronics nor civilized behavior.

[robrenz]

+1

[ElektroQuark]

Great work!
I like your videos very much.
Thank you for sharing.

[Mechatrommer]

On the other hand I am most ashamed that the makers of these very good videos should face derision and insult here. The obnoxious audiofool will not learn, neither electronics nor civilized behavior.

when you hang around a bit longer, you know this is normal. either the guy needs attention (just as i did recently) or he think build from a bunch of diodes or vacuum pumps is the only right thing to do. we or the OP should simply ignore it and avoid feeding it (opps did i just do that?  ) or try to explain the reason wisely.

i recalled in the video, shariar reasoned that using transistor is to avoid high noise property of opamp if used in the first stage. though i'm not sure the validity of this claim, yet a good reason (maybe?) for me to blame the opamp/design i used in one of my circuit lately, noisy, not a cheap opamp i tell ya, but still noisy not within my desired spec (lack of knowledge if you wish to say). i think i gotta try sometime, using bjt in the first stage followed by opamp for further gain, exactly like what he demonsrated. idiocy? i'm not sure either, never tried and will never know if i do not try.

edit: even though he claimed low noise idea of the design, but i sensed noisy circuit he built from his dso display albeit he's using average mode

[T4P]

The seemingly high noise exists only if you have been using quite high gains on a very low input signal, try using cascaded* gain stages

[Mechatrommer]

The seemingly high noise exists only if you have been using quite high gains on a very low input signal, try using cascaded* gain stages

example? do you mean cascaded transistors or cascaded opamps or cascaded both?

[Alana]

Tank you, some time ago i needed exacly that, now i know how to build one.

[Hugoneus]

Thank you for everyone's contribution. Allow me to clarify some things.

In a well designed system of cascaded amplifiers, the over all noise figure is dominated by the first stage. Intuitively, this is true since the noise contribution of the later stages is scaled by the gain of the first stage when referred to the input. In this application, the gain of the first stage is about 35dB and thus the noise of the opamp stage is almost irrelevant.

For the purposes of this educational circuit, the opamp is meant to be a cheap classic 741 opamp which is rather noisy. In the same technology, a properly biased single transistor will have lower noise than a full opamp circuit. The transistor that I am using has a NFmin of 4dB in the frequency range of interest. Furthermore, to get the most out of the first stage common-emitter, simultaneous signal and noise matching must be achieved. This is something that I did not discuss in the video. If used with a microphone, signal and noise matching can be achieved through biasing and component sizes. The minimum noise figure current density of the bipolar transistor in the audio range can be found from the datasheet.

Lastly, the noise that you see from the DSO is actually not from the circuit. There is a bug in the new firmware of the scope which is causing some issues. It will be resolved soon.

[SeanB]

The matched pairs did a lot of good as phono preamps as a first stage, and were the dominant noise source. I still have some 2N2223 matched pairs around, and a couple of MPQ6502's as well.

[hans]

Very interesting tutorial.

I've been busy with some (high) gain stages for a DC power analyzer a little while back (college project). I ended up using an instrumentation amplifier of +/-15V. The purpose was ofcourse to measure currents through a 0.1ohm high-side shunt with supply voltages up to 12V or so, and to measure current (transients) of devices. Should be able to measure uA's (in addition to detect in-rush currents of 1-2A)..
The first stage was a very high gain stage of 1000x. The next stages added an additional +/-1x to 50x (depending on the range- the 16-bit A/D didn't had enough resolution for full scale).
I did saw some appnotes about transistor amplifiers that would get very very low noise results, in the order of 3 times lower than I used (I abused a microphone amplifier, INA103). I believe it was like 0.33nV/sqrt(Hz). It was indeed a set of matched transistors..
I still wonder, how is the 1/f (DC) noise with transistor amplifiers compared to op-amps? I really wished we did more analysis on transistor amplifiers in college..

In the end the DC analyzer was 'not bad', but ofcourse not calibrated. With some oversampling (16x, 1kSPS samples sent out) and additional filtering I ended up with <0.5uA noise on a +/- 2.5mA range.

[Hugoneus]

Very interesting tutorial.

I've been busy with some (high) gain stages for a DC power analyzer a little while back (college project). I ended up using an instrumentation amplifier of +/-15V. The purpose was ofcourse to measure currents through a 0.1ohm high-side shunt with supply voltages up to 12V or so, and to measure current (transients) of devices. Should be able to measure uA's (in addition to detect in-rush currents of 1-2A)..
The first stage was a very high gain stage of 1000x. The next stages added an additional +/-1x to 50x (depending on the range- the 16-bit A/D didn't had enough resolution for full scale).
I did saw some appnotes about transistor amplifiers that would get very very low noise results, in the order of 3 times lower than I used (I abused a microphone amplifier, INA103). I believe it was like 0.33nV/sqrt(Hz). It was indeed a set of matched transistors..
I still wonder, how is the 1/f (DC) noise with transistor amplifiers compared to op-amps? I really wished we did more analysis on transistor amplifiers in college..

In the end the DC analyzer was 'not bad', but ofcourse not calibrated. With some oversampling (16x, 1kSPS samples sent out) and additional filtering I ended up with <0.5uA noise on a +/- 2.5mA range.

The 1/f noise of a bipolar transistor is quiet low. A fully bipolar opamp would also have low 1/f noise. If you consult a good analog electronics book, you should be able to find the optimum noise figure current density of a bipolar transistor as a function of the technology parameters.

[T4P]

The seemingly high noise exists only if you have been using quite high gains on a very low input signal, try using cascaded* gain stages

example? do you mean cascaded transistors or cascaded opamps or cascaded both?

Opamps

[poptones]

Idiocy, of course, means idiocy not that the person is an idiot. So, why not use an emitter follower for the second stage?

[Hugoneus]

Idiocy, of course, means idiocy not that the person is an idiot. So, why not use an emitter follower for the second stage?

The maximum available gain from an emitter-follower is 1. So using an emitter-follower as a second stage would not provide an overall gain of  > 100 V/V. It would provide a low impedance output if that was needed. Regardless of this fact, the purpose of this video is educational and is meant to encourage the students to also play around with an opamp. It also highlights that the NF of the overall system is dominated by the NF of the common-emitter front-end.

A heavily degenerated second stage common-emitter with a gain of 2 V/V would do the trick if one doesn't wish to use the opamp.

[Hugoneus]

Idiocy, of course, means idiocy not that the person is an idiot. So, why not use an emitter follower for the second stage?

Does the above answer your question poptones?

[miceuz]

Shahriar,

I've built the circuit as you suggested, but I didn't quite understand, why do you divide voltage drop on collector resistor, transistor and emitter resistor in equal parts? Another thing - base resistors and their values come out of nowhere somehow, but are very important really as they set the operating point of transistor too.

Then i've tried to build circuit by this tutorial: http://www.sentex.ca/~mec1995/tutorial/xtor/xtor8/xtor8.html which explains everything in more praxic but single flow manner.

Could you comment of difference between aproach descirbed by you and by this tutorial? One thing to note is that it relies on gain set by Rc/Re rather than intristic gain that you descirbe, but what about dividing voltage in three equal parts and concentrating on output voltage to be Vcc/2?

[Hugoneus]

Shahriar,

I've built the circuit as you suggested, but I didn't quite understand, why do you divide voltage drop on collector resistor, transistor and emitter resistor in equal parts? Another thing - base resistors and their values come out of nowhere somehow, but are very important really as they set the operating point of transistor too.

Then i've tried to build circuit by this tutorial: http://www.sentex.ca/~mec1995/tutorial/xtor/xtor8/xtor8.html which explains everything in more praxic but single flow manner.

Could you comment of difference between aproach descirbed by you and by this tutorial? One thing to note is that it relies on gain set by Rc/Re rather than intristic gain that you descirbe, but what about dividing voltage in three equal parts and concentrating on output voltage to be Vcc/2?

I divided the voltage equally between the resistors and transistor to to simultaneously optimize for maximum output swing and beta-insensitive biasing.  The splitting of the voltage equally between the Rc and Vce is important to achieve a symmetric output swing. The voltage on Re will help achieve beta-insensitive biasing. You could spend less voltage on Re and give it to Rc and Vce to get more voltage swing for example.

I had a quick look at the link you have provided. It is essentially the same design. The gain is -Rc/Re ONLY if no capacitor is used in the emitter. With the capacitor, the emitter node is an AC ground and the gain becomes -gm* (Rc || ro) which is approximately equal to -gm * Rc since ro >> Rc.

The voltage on the base of the device is determined based on the emitter voltage that is required. Then a voltage divider is created to achieve the necessary base voltage. If (and this is important) you are willing to burn more current in the input resistor divider, then you are able to ignore the base current (Ib) since it will be much less than the current in the divider.