Your circuit has no voltage gain. I already said your circuit has unity voltage gain.
You still haven't told us what you're going to use for V bias. The reason I emphasize this is that your circuit has no practical value. It is a Thevenin equivalent circuit used for calculating component values etc. The circuit will actually work for illustrative purposes, so go ahead and build it and measure it. You can use a 1.5 volt battery for V bias as a start. You have to determine RE1 and RE2. What values do you propose?
Here's a more practical circuit from your favorite website Push-Pull Output Stage . You need to use emitter resistors on Q11 and Q12. Here is where you can actually control the quiescent current, input impedance, and current gain of your unity gain amplifier. The ratio of Q11/Q12 emitter resistors to RB1/RB2 is the approximate current gain of your amplifier, if their ratio is much less than the beta of the transistors.
So go ahead and select some values and simulate them. You can scale the circuit so you can use it with smaller transistors. You get real world linearity gains and thermal stability improvements. You control quiescent current and input impedance. And you can design something that you can actually build or use as a building block.
You still haven't told us what you're going to use for V bias. The reason I emphasize this is that your circuit has no practical value. It is a Thevenin equivalent circuit used for calculating component values etc. The circuit will actually work for illustrative purposes, so go ahead and build it and measure it. You can use a 1.5 volt battery for V bias as a start. You have to determine RE1 and RE2. What values do you propose?
Here's a more practical circuit from your favorite website Push-Pull Output Stage . You need to use emitter resistors on Q11 and Q12. Here is where you can actually control the quiescent current, input impedance, and current gain of your unity gain amplifier. The ratio of Q11/Q12 emitter resistors to RB1/RB2 is the approximate current gain of your amplifier, if their ratio is much less than the beta of the transistors.
So go ahead and select some values and simulate them. You can scale the circuit so you can use it with smaller transistors. You get real world linearity gains and thermal stability improvements. You control quiescent current and input impedance. And you can design something that you can actually build or use as a building block.
My friend has just come to me and said he can use a STP16NF06 single mosfet current drain as our output stage and still have extremely low output impedance.
And I'm kind of thrown of as to how this is possible as it wouldn't require feedback, or anything else complicated. There's got to be a catch here somewhere, right?
And I'm kind of thrown of as to how this is possible as it wouldn't require feedback, or anything else complicated. There's got to be a catch here somewhere, right?
Last edited:
Transistors (heck, all amplifying devices) are inherently non-linear. So there is always distortion along for the ride. (Sometimes it seems as though nature is just intrinsically in a bad mood, and fights all our human attempts to do things to our satisfaction...there is always some problem that comes along with every solution!)
BJTs are actually much more non-linear than FETs - but in a compound pair, there is 100% series voltage feedback built-in. That lowers distortion a lot.
A single common-drain FET also has 100% series voltage feedback built-in. But it has less inherent voltage gain to start with (FETs have much lower transconductance than BJTs). So it will probably have more distortion. So (oversimplifying a bit) there's less overall feedback, and therefore more distortion.
-Gnobuddy
BJTs are actually much more non-linear than FETs - but in a compound pair, there is 100% series voltage feedback built-in. That lowers distortion a lot.
A single common-drain FET also has 100% series voltage feedback built-in. But it has less inherent voltage gain to start with (FETs have much lower transconductance than BJTs). So it will probably have more distortion. So (oversimplifying a bit) there's less overall feedback, and therefore more distortion.
-Gnobuddy
Patience, Mr. Fahey! In this rushed and crazy world of today, most people now find it very hard to slow down and focus on one single thing long enough to learn anything difficult.
My nephew kills at least 100,000 invading aliens on his computer most days, usually several of them per second. How will he ever slow his racing mind down long enough to study the complexities of, say, a Sziklai pair? He has trained his mind to jump from one idea to another every tenth of a second...slowing down is impossible for him.
This is what we have done to ourselves, almost everyone has symptoms of ADD now. Our brains hop and bounce constantly from one thing to the next, and it's becoming harder to find anyone who can focus one one difficult problem long enough to think deeply about it, and solve it.
Gorge is sticking with his original intention to learn more, which is commendable. Let's allow him to do it in his own way!
-Gnobuddy
My nephew kills at least 100,000 invading aliens on his computer most days, usually several of them per second. How will he ever slow his racing mind down long enough to study the complexities of, say, a Sziklai pair? He has trained his mind to jump from one idea to another every tenth of a second...slowing down is impossible for him.
This is what we have done to ourselves, almost everyone has symptoms of ADD now. Our brains hop and bounce constantly from one thing to the next, and it's becoming harder to find anyone who can focus one one difficult problem long enough to think deeply about it, and solve it.
Gorge is sticking with his original intention to learn more, which is commendable. Let's allow him to do it in his own way!
-Gnobuddy
Why ask US?, ask your friend
There's a lot of old wives tales, dogma, and voodoo that is propagated about hi-fi; both about speakers (like simpler crossovers are better) and also about the electronics (like less transistors are better). This hype replaces actual knowledge (which very few people have) of the issues.
Even people in the know have their pet prejudices. My buddy's dad started to teach me electronics when I was 10. He was the real deal; he had built his own preamp and amplifier in the 50s, and his own speakers too. It was all tubes (of course) and it was actually very good. Anyway, he was very biased against transistors, and especially op amps (which were new at the time). By the time I was 15 or 16, my knowledge of electronics was starting to eclipse his and since it was my goal to be an engineer, I was going to be dealing with transistors and op amps. I designed and built an ultralinear amplifier that employed transistors in the power supply and also in the biasing. He said flat out "you can't mix tubes and transistors" and he considered the idea heresy. But when we hooked it up to his equipment, it all went out the window. It was far superior to what he had built and he pointed out the improvements. Then I designed and built an op amp based preamp, using the CA3140 devices (not much choice back then). He was appalled; he said "op amps belong in computers." But again, that all went out the window when we hooked it up to his system. It was obviously superior to his 1950s hand built equipment, and he pointed that out.
The moral of my story is that the best way to separate the wheat from the chaff is to built, test, and evaluate. So to our newbie friend, I say do it. Build, test, and evaluate. Then try another configuration. You'll find out what's real.
Funny, but it's true.
Some people make a lot of money by building amps, that look more like crafts preschoolers or infants.
This is what I love about the scientific method. An honest confrontation with reality will do more to clear superstition out of the brain than a thousand years of talking about it!
My hat is off to your friend, he is (was?) a true scientist. It isn't easy to give up one's cherished beliefs, even when they turn out to be wrong. But when the instruments told him he was wrong, he believed them. Bravo!
I also believe a tendency towards some degree of superstition is literally wired into the human brain. We are all susceptible to it, there is no escape as long as we are human. There is increasing scientific evidence from brain researchers showing that the left brain simply makes up an explanation for any data that comes along. The "explanation" may be complete nonsense ("I won $5 at the Lotto because I wore my lucky green sweater"), but that is all that is needed to satisfy the mind.
We all believe so much that is illogical: for instance, most people on this forum (myself included) would automatically prefer a piece of test equipment made by HP or Tektronix over some unknown Asian brand - but without actually testing the equipment in question, there really isn't an objective basis for that decision. It's simply a prejudice, until proven one way or the other by actual data.
I agree 100%. Throw your ideas at the wall, and see what actually sticks, as one person put it!
-Gnobuddy
Was my friend. This is ancient history we're talking about. I'm at the age where my peers are dropping like flies. He's been dead over 30 years.
Your stated belief is a scientific fact. There is no way that we could possibly get through life without a whole lot of prejudices; some of them hard wired. Our brains would be on constant overload without them, and we would never get anything done.
Why does a lazy housecat bother to chase a string? It's hard wired. Why does a dog bark when it hears a noise? Why does it assume there's something going on outside? Why do you and I assume it, for that matter? These are hard wired prejudices that have made it possible for populations to survive and evolve for millennia. They have served survival of species well.
Imagine what driving would be like if you paid attention to everything. You'd never make it down the block. We don't even notice people sitting on the porch because it's not relevant to the task at hand; but if a ball rolls into the street, we jam on the brakes without consciously thinking about it.
Different animals have different prejudices. These cognitive biases make it possible to survive without expending hardly any intellect at all. Look at the lower species (like insects); they have virtually no intellect yet manage to survive and propagate without a thought.
Apparently one of my prejudices is the belief that flies don't actually drop like flies. Instead, particularly when I lived on a farm for two years, they buzz in annoying irregular loops around me all summer long, simultaneously demonstrating both the existence of perpetual motion, and the existence of eternal life.
-Gnobuddy
-Gnobuddy
Nelson Pass has a useful hands on tutorial on DIY op.amps https://www.passdiy.com/pdf/diyopamp.pdf
The circuits he describes run from plus and minus 32 volt supplies whereas 15 volts would better suit the lesser needs a headphone amplifier.
In a low power application like this running the output stage in Class A is not wasteful of power. For low impedance headphones you might need an output standing current of 100 m.a. a small heat sink and a supply able to meet the current needs. The current requirement can be reduced for high impedance headphones.
Looking at the circuit on page 6 in figure 14B the output transistor Q4 sees a 10 m.a. constant current source at it's emitter lead.
The circuit for the latter is shown on page 7 in figure 15B where the 160R resistor provides a current of 4 m.a. You would need to replace this with a 5 watt 6R8 wire- wound type.
A 4 - 5 m.a. constant current should be OK for Q3 (100R in lieu of 160R) and 0.6 m.a. (300R) for the tail of input transistor pair. To get zero volts at the output that value may need to be changed).
With constant current sources in the output the output current cannot increase so if you get the design right in terms of transistor specifications heat sinks etc there will not be any thermal runaway.
Using +/- 15V on a headphone amp is potentially a way to cause severe loudness-induced hearing damage. That much supply voltage can put out well over 2 watts into 32 ohm headphones.
Since most contemporary headphones will generate over 90 dB SPL at the ears from one single milliwatt, with 2 - 3 watts input, we now have a system capable of around 160 dB maximum SPL.
160 dB SPL is so unbelievably loud that it can literally cause permanent hearing damage almost instantaneously. One single full-power transient - say from a loose headphone cable - and you could be deaf for life.
I suggest headphone amps be designed to limit maximum SPL to no more than 100 dB. That is already way beyond the threshold where hearing damage starts to occur.
-Gnobuddy
Since most contemporary headphones will generate over 90 dB SPL at the ears from one single milliwatt, with 2 - 3 watts input, we now have a system capable of around 160 dB maximum SPL.
160 dB SPL is so unbelievably loud that it can literally cause permanent hearing damage almost instantaneously. One single full-power transient - say from a loose headphone cable - and you could be deaf for life.
I suggest headphone amps be designed to limit maximum SPL to no more than 100 dB. That is already way beyond the threshold where hearing damage starts to occur.
-Gnobuddy
Looking at what last I wrote, 100R for Q3 constant current source would allow 6 m.a. of operating current. A characteristic of transistors is that voltage gain increases (a) as Q3 collector load increases and (b) as the current Q3 passes is increased.
The formula is Av (voltage gain)=gm (in m.Siemens - roughly 40 per m.a of operating current) times RL ( Resistance of the transistor collector load in k.Ohms relative to the signal frequency).
A few caveats.
The form of Q3 is an inverting amplifier and there is some internal capacitance within the transistor directly to do with internal geometry. There is also a changeable element involved - in that the internal capacitance is inversely proportionate to the square root of the voltage seen at the collector.
Reducing the supply rails from 32 to 15 will increase the value of the internal capacitance and that will be amplified somewhat more due to increasing the Av previously mentioned. A 10 pf external capacitor in parallel is shown in Mr Pass' circuit the value of this will also be increased due to Miller Effect.
There is quite a lot you can tinker with if you want to give it a go.
The formula is Av (voltage gain)=gm (in m.Siemens - roughly 40 per m.a of operating current) times RL ( Resistance of the transistor collector load in k.Ohms relative to the signal frequency).
A few caveats.
The form of Q3 is an inverting amplifier and there is some internal capacitance within the transistor directly to do with internal geometry. There is also a changeable element involved - in that the internal capacitance is inversely proportionate to the square root of the voltage seen at the collector.
Reducing the supply rails from 32 to 15 will increase the value of the internal capacitance and that will be amplified somewhat more due to increasing the Av previously mentioned. A 10 pf external capacitor in parallel is shown in Mr Pass' circuit the value of this will also be increased due to Miller Effect.
There is quite a lot you can tinker with if you want to give it a go.
With due respect, you are off by about 1000:1 ratio.That much supply voltage can put out well over 2 watts into 32 ohm headphones.
Since most contemporary headphones will generate over 90 dB SPL at the ears from one single milliwatt, with 2 - 3 watts input, we now have a system capable of around 160 dB maximum SPL.
160 dB SPL is so unbelievably loud
-Gnobuddy
2000mW:1mW amounts to 33 dB so theoretical max SPL would be 90+33=123dB
Not that I would like to get even 123dB sound at my eardrums
Thank you for the correction! I used 20 log(P2/P1) when I should have used 10 log(P2/P1).With due respect, you are off by about 1000:1 ratio.
Sorry for the mistake!But, as you say, the original point is still valid. 120 dB is still ear-destroying SPL. One loud pop from a bad headphone socket, and you could have permanent hearing damage. That's simply not worth the risk, IMO.
-Gnobuddy
Thank you for the correction! I used 20 log(P2/P1) when I should have used 10 log(P2/P1).
But, as you say, the original point is still valid. 120 dB is still ear-destroying SPL. One loud pop from a bad headphone socket, and you could have permanent hearing damage. That's simply not worth the risk, IMO.
-Gnobuddy
The output stage I discussed has a fixed level of current delivery. With 100 ma this is not going to output 2 watts (RMS) into an impedance of 32 Ohms. There seems to have been some confusion with push-pull output stage capabilities here.
The volume of headphone amplifiers can be set by the closed loop gain and the volume control. For lower impedance phones a 4R7 resistor connecting the output to the socket is not a bad idea.
I have built headphone amplifiers running from 15 -0 -15 volts supplies.
3W into a 90dB/mW headphone results in ~125dB, not 160dB (edit: I see that this has been addressed).Using +/- 15V on a headphone amp is potentially a way to cause severe loudness-induced hearing damage. That much supply voltage can put out well over 2 watts into 32 ohm headphones.
Since most contemporary headphones will generate over 90 dB SPL at the ears from one single milliwatt, with 2 - 3 watts input, we now have a system capable of around 160 dB maximum SPL.
160 dB SPL is so unbelievably loud that it can literally cause permanent hearing damage almost instantaneously. One single full-power transient - say from a loose headphone cable - and you could be deaf for life.
I suggest headphone amps be designed to limit maximum SPL to no more than 100 dB. That is already way beyond the threshold where hearing damage starts to occur.
-Gnobuddy
Most headphone users would find a 100dB absolute limit on SPL to be completely inadequate.
Many will accept 110dB and quite a few would demand 120dB capability. and yes, continuous "too loud" does make one deaf.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.