For a lab class we were assigned a final project where we have to design an amplifier using BJTs, diodes, and capacitors (MOSFETs are not allowed) in order to drive an 8-ohm speaker. The requirements given are as follows:

  • that we were to be able to have gain adjust control for a certain range
  • use 9 volts VCC and must not use voltage dividers, instead use current mirrors
  • we must attain at least a certain voltage for a certain peak to peak sine wave input for a certain frequency range when gain adjust control is set to maximum, therefore there must be a minimum gain for the range of frequencies that our amplifier will support
  • we are not allowed to use any op-amps, except to invert signals should we choose to use differential amplifiers
  • there must be no clipping for a certain peak-to-peak sine wave input when gain adjust control is set to maximumm

I didn't include the exact specs for our final project in this question, since I think that isn't allowed isn't it? But our group's problem is how to approach this from a design perspective.

Our idea is to use a CE-CC configuration, but obviously this is a too basic thought.

I think it just so happened that all three of us always just thought of it from an analysis perspective. Or we all need to study more haha, which is actually true too.

We don't even know about A-B power supply configurations (is my terminology right)? Since it hasn't been discussed yet. Maybe in another class. I'm not sure if we were supposed to come up with a similar thing on our own using diode connected transistors or something, since we didn't cover using diodes in this class. This class is mainly about analyzing BJT and MOS configurations by the way, and their frequency responses.

So a guide on how to approach doing this would be great! So that we would at least know how to start doing this. Maybe we're also lacking insight on how does one actually drive an 8 ohm speaker... I mean we've already done this in lab, but how does it actually manage to emit sound? Is it because of the varying voltage due to the small-signal voltage introduced to the circuit? What would happen if a DC voltage is what's passing through the speaker? (We tried simulating a configuration in SIMetrix, and we made a mistake somewhere where the output voltage (the one across the speaker), is a constant DC level, so obviously we made a mistake somewhere). A clarification on how this actually works would be nice too :D Thanks to anyone who'll help!

[EDIT: Since people are commenting that I should include the requirements from the specs, I'll do so. I was hesitant to include these because it's classwork. But I suppose I really am being vague, and you guys are good judges on how to help me since you know what to do about this. Sorry for the oversight :(]


  • must drive an 8 ohm speaker
  • BJTs allowed: 2n3904, 2n3906, 2n4401, 2n4403
  • gain adjust control must be 50 to 250
  • 9V VCC and don't use voltage dividers
  • peak to peak voltage across the speaker must be at least 1v for a 20mV peak-to-peak sine wave input, for frequencies ranging 500Hz to 20kHz when gain adjust control is set to maximum. So the amp must have a gain of at least 50 for frequencies ranging from 500Hz to 20kHz
  • the amplified signal must be free of clipping for a 20mV peak-to-peak sine wave input when gain adjust control is set to maximum]
  • \$\begingroup\$ A push-pull output is probably best BUT a class A transformer output amplifier can provide bigger peak-to-peak output swings. \$\endgroup\$
    – Andy aka
    Commented May 9, 2015 at 10:28
  • \$\begingroup\$ Multi-stage BJT amplifiers are covered in most undergraduate textbooks and courses on microelectronics. \$\endgroup\$ Commented Dec 4, 2015 at 18:32
  • \$\begingroup\$ As for "how does it actually manage to emit sound?" That part does suggest you might be in the wrong line of business. \$\endgroup\$ Commented Dec 4, 2015 at 18:50

3 Answers 3


You are asking way too many basic questions for this stage in your course. This means you didn't understand some previous material, but worse, it's clear you don't have any intuition and feel for circuits. With some hard work you can possibly muddle your way thru and eventually receive a degree, but that will be just a hollow piece of paper. I've encountered such "engineers" a few times, and they are outed quickly in the real world. If you haven't been tinkering with this stuff since at least high school, then you don't have a passion for it and should seriously re-evaluate your career objectives.

However, to answer some of your question, break the problem into three stages.

The first stage should have a reasonably high input impedance with the ability to receive the feedback signal from the output. This could be, as only one example, a differential pair as found on the input of BJT opamps. The main purpose of this stage is to take the input signal, subtract the feedback signal, and multiply by some gain. It should also lower the impedance of the signal that it passes to the next stage.

The second stage is largely for gain and to make the signal even lower impedance. At this point it's just a single ended signal. There are many ways to do this.

The third stage is to provide the relatively large current needed to drive the speaker. This stage won't have much voltage gain. In fact in some configurations it actually has a voltage gain less than 1. However, it has significant power gain due to transforming the impedance down to something small compared to the 8 Ω of the speaker. Hint 1: A BJT in emitter follower configuration has voltage gain a little less than 1, but large current gain. One problem is that the output voltage is the input voltage minus the B-E drop. Hint 2: Ordinary silicon diodes have about the same voltage drop across them when forward biased as the B-E junction of a transistor. With a little cleverness, you can use a diode to offset the input of a emitter follower to get roughly the same output level as the input level, but still with the large current gain. Hint 3: Both NPN and PNP can be used as emitter followers, each working in the opposite polarity. Both together can be used to make a emitter follower that works symmetrically in both directions.


Firstly, can I just say you haven't really provided enough information, and are also asking too much in one question.

I would definitely suggest getting a book. A really good one is "the art of electronics". It has been the "bible" for a long time, and will give you a range of circuits from the most basic to more elaborate, with all the explanations, for ideas and technical guidance.

Then: Step 1: establish requirements. (Features, cost etc). Then design, and test. Various methodologies exist. Try http://engineering.stackexchange.com for that. Perhaps adapt a methodology used in computer science, such as the waterfall method. The simplest methodology might look like this:

  1. Establish requirements
  2. Design
  3. Prototype / Testing
  4. Production

The waterfall method has loops which take you back to the design stage if there are problems in the latter stages.

As for design: come up with a few designs, evaluate each in a table with cost, performance, and so on. Use simulation tools and/or calculations to establish performance, such as efficiency, THD, etc.

Come back when you want help with a specific task such as help with a particular design, or calculations for one aspect, and so on. Split up the task.

I believe the reason you feel stuck is you haven't actually completed the first stage; requirements. You don't know what it is you actually are required to produce. Either do research, and/or go through the iterative cycle of development until you are satisfied with a design.

I will give you a few steps in the right direction below. But I can't emphasise highly enough that you play around with designs; while one method might seem like the best way, you won't really know until you have investigated others. I.e., don't take the circuit I give as the only or best way!!

Firstly, your question about speakers and sound. Sound is a wave. It is a longitudinal wave. Imagine a "slinky" spring toy. Stretch it out over the ground, then repeatedly push/pull one end (towards and away from the other end). That is a longitudinal wave. Where the spring's spiral is compressed, it is like the compressed air in sound, and where the spring's spiral is expanded, it is like the lower pressure part of the sound wave.

All the speaker needs to do (or anything in air, for that matter) to make sound, is vibrate. As it moves into the air, it creates high pressure. As it moves away from the air, it creates low pressure; both of which travel at the speed of sound, like the waves in the slinky. (Although the slinky waves probably only move at a few metres per second.)

The way a speaker vibrates is quite straight forward. It's just a coil and a magnet. When current flows in one direction in the coil, it generates magnetism, which makes it move in one direction with respect to the permanent magnet. And when the current moves in the other direction, the coil's poles reverse, and so is pulled in the other direction with respect to the permanent magnet.

Also, the more current that flows through the coil, the further the speaker cone will extend (or retract). So if you wanted to, you could actually create sound in a speaker with a sine wave that has a DC offset, for example so that the voltage across the coil is only ever in one direction, just with varying current. So the sine wave might have a voltage from +2v to +10v. However this would be a waste of energy; for example, if you have a sine wave that goes from +5.9v to +6.1v, the speaker will be moving very little, yet will have an average of 6v across it, and for an 8 ohm speaker, that would mean I=V/R, a current of 0.75 amps, so P=IV, power of 4.5 Watts, almost all wasted (as heat in the loudspeaker).

So, now onto hints to a possible circuit...

As you probably know, a bipolar transistor has a voltage drop between the base and emitter of approximately 0.7v. This happens to be the same as a forward-biased diode; no coincidence since parts of the structure of bipolar transistors resemble a diode.

That is what this circuit uses to work:


See how you get on with understanding how that works, and take it from there. This question/answer has reached its limit, it really needs to be split into more questions, so please do ask further questions separately, trying your best to isolate your questions into distinct atomic problems.

  • \$\begingroup\$ @Jodse Sorry I didn't include the exact requirements before, I now included it. I was hesitant to do so since it's class work, I didn't realize that I was being too vague. I'll check out the book :D Thanks for the recommendation! \$\endgroup\$
    – Val Croft
    Commented May 9, 2015 at 19:47
  • \$\begingroup\$ @ValCroft I second the Art or Electronics! Get your hands on a copy, its years old so should be available used from the internet, Craigslist etc. Read it cover to cover, just to absorb the range and depth of the subject. \$\endgroup\$
    – tomnexus
    Commented May 10, 2015 at 13:02

Since it is class work, I am hesitant to say too much. Also, I am not an audio amp designer. But I would suggest Douglas Self's book on audio power amplifier design, since he also does not use MOS or IC's (such as op-amps). It may be the intent of your teacher that you all struggle through the trade-offs of different amplifier architectures. But there is kind of a standard way that these amps are designed. They use a long tailed pair for input stage, a common emitter amplifier for the Voltage gain stage, and class B emitter follower output stage (push-pull with NPN on top and PNP on bottom). I would avoid class A output as it turns into a thermal design problem rather than strictly an electronics design problem. If the required power output is very low, you could consider class A.

There are also many other practical things you have to tend to to make sure it will be stable. Thermal design and thermal stability is also really important.

As a final note, since you opted to not include most of the specifications, it is hard to say how difficult this problem will be. The topology I listed above cannot drive power from rail to rail, for example. You cannot drive the NPN on top high enough to get VCC output. Likewise, you cannot drive the PNP on bottom low enough to get VEE output. The devil is in the details.

Good luck.

  • \$\begingroup\$ Thank you for the tips and the book recommendation! We indeed didn't consider thermal design, and used a class A output :)) Though as for now I don't really get push-pull configurations. Someone just hinted that we should use it. I updated the specs on my post. We asked for help from higher years, and they asked about the required power output too, however it's not included in the specs. We looked at similar problems and power is also the requirement. I'm not really sure how to derive it from the requirements given to us from a power output perspective. \$\endgroup\$
    – Val Croft
    Commented May 9, 2015 at 19:46
  • 1
    \$\begingroup\$ Is it your interpretation that the maximum required peak-to-peak output is 1V? If so, the maximum instantaneous power output is 0.5V ^2 / 8 Ohms (This is V^2/R formula). This is a total power output of 1/32 Watt. Another way to look at it is that the maximum current is 0.5/8 Ohms which is 1/16 Ampere = around 65 mA. Class A should be no problem for the transistors you are using (double-check the datasheet). You may consider Darlington configuration for the input pair and/or the voltage amplifier stage since your required gain is fairly high. \$\endgroup\$
    – user57037
    Commented May 9, 2015 at 22:40
  • \$\begingroup\$ Self also made a power amp entirely of NE5532... And his audio amp books also cover ICs... perhaps to some silly extent, like making an inamp with NE5532. I tried that for amusement. \$\endgroup\$ Commented Dec 4, 2015 at 18:52

Not the answer you're looking for? Browse other questions tagged or ask your own question.