Thank you for producing this material. It is considerable work. The question I have is - What is the design process to optimally bias the circuit? (Before current mirrors and the like are introduced). The (Distorted triangle wave) scenario was introduced, but I'm not sure it was ever answered or closed off. It's the design process that particularly interests me.
The best answer I can give is somewhere between 'it depends' and 'don't do that;' Those who look for excessively high gain in a single-stage open-loop amplifier will be punished with distortion. If you stick to a stage gain of, say, 20-40, you can use a big enough emitter resistor that you won't see the pointed-arch distortion that was in the video. A good rule of thumb is that the external emitter resistor (the one carrying the ac signal, not the one setting the bias point) needs to be at least 5-10x the intrinsic emitter resistance at the desired collector current. You can use a high-gain stage, distortion be damned, if you wrap it in a feedback loop - which I haven't discussed yet, but I'm working up to. Later in the series, I plan a couple of 'putting it all together' videos that demonstrate higher-performing multi-transistor amplifiers. Each of those amps will have several stages, usually including current sources (or mirrors), differential pairs, common-emitter gain blocks, and emitter followers. They all work together so that each stage compensates for the failings of the others (a follower to provide low output impedance, for instance, or a current source to give the appearance of a nearly-infinite tail resistor.) All of them use feedback to stabilise gain and minimise offsets. For small-signal circuits, pretty much everyone just uses an op-amp rather than a single-transistor circuit, unless you need to do something particular - for example, use the exponential relation between base voltage and collector current, or use the transistor to regulate a current, or use the transistor to handle more power than the op-amp can put out. But understanding the basics of what's going on _inside_ the op amp - which this series is working up to - is vital to understanding the limitations of what an op amp can do. That will come up in discussing op-amp integrators (input offset voltages), using an op amp to drive a power amp (crossover distortion and op-amp slew rate), making a radiofrequency amplifier (Miller effect) and so on... just as the discussion of current mirrors led us to the Early effect.
@@KludgesFromKevinsCave Thank you and understand. I guess my question was finding the optimal bias point.in the most linear region. Say for high-quality audio, where you wish to use discrete transistors over Op Amps, what is the design process to find that bias point? I understand that bigger picture objective gets way more complicated and requires far more sophisticated approaches, plus negative feedback, etc. I'm just trying get the design process clear in my mind. My original background was electronics, but my career moved away from it many years ago.
@tonykirkham1663 I wish I could give you a single recipe. In a way, this entire series is working toward the Holy Grail of linearity and dynamic range. Once feedback is in play, as it always should be, the linearity is nearly assured, and biasing becomes a question of 'how far can the signal swing without clipping? Center the bias point in that range.' Each presentation of a given circuit topology (follower, common emitter, etc) has at least an example of finding the bias point. I'm generally skeptical of the "no op-amps in high-quality audio" camp. Sometimes you need discretes in a low-noise preamp, where signal levels are minuscule, or in a power stage where you have to dump a lot of heat, but with reasonable signal levels, op-amps are awfully good. With that said, I surely don't have a 'golden ear' - I grew up very near a busy airport, and have had some minimal hearing loss all my life. I trust my instruments rather than my senses.
The dual transistors all seem to come in strange pinouts. DMMT5401 and its complementary partner, DMMT5551, even have more than one choice for them, both peculiar. www.diodes.com/assets/Datasheets/ds30436.pdf It's actually not horrible for layout. Common emitter (as for a differential pair or exponential converter) and common base (as for a current mirror) both put the connected pads right next to one another.
@@KludgesFromKevinsCave Mmm. True. That makes sense for a current mirror layout. I see onsemi has a different layout that is EBC EBC, but the EBC on each side do not all belong to the same transistor (which is also confusing to me in a different way). For me, it would make sense to just have each side be EBC or CBE for each transistor, both for most layouts and conceptually, but that's just me. No doubt there's a reason not to do this on the die.
Excellent tutorial. It is may be worthwhile to recognise that the difference of diode forward voltage drops is of the order of 2.5%, i.e. neither very significant nor insignificant.
Hmm. I suppose. I think the important point is that the difference is always there, so you can tell emitter from collector that way. Sometimes at the bench it's quicker to use a multimeter than look up the data sheet for the transistor.
Thank you professor ◜◯◯◝. I'm not sure the girls will like "me" more but I know I will. All this is going onto my resume! Cheers from So.Ca.USA 3rd house on the left.
Thank you for producing this material. It is considerable work. The question I have is - What is the design process to optimally bias the circuit? (Before current mirrors and the like are introduced). The (Distorted triangle wave) scenario was introduced, but I'm not sure it was ever answered or closed off. It's the design process that particularly interests me.
The best answer I can give is somewhere between 'it depends' and 'don't do that;' Those who look for excessively high gain in a single-stage open-loop amplifier will be punished with distortion. If you stick to a stage gain of, say, 20-40, you can use a big enough emitter resistor that you won't see the pointed-arch distortion that was in the video. A good rule of thumb is that the external emitter resistor (the one carrying the ac signal, not the one setting the bias point) needs to be at least 5-10x the intrinsic emitter resistance at the desired collector current.
You can use a high-gain stage, distortion be damned, if you wrap it in a feedback loop - which I haven't discussed yet, but I'm working up to.
Later in the series, I plan a couple of 'putting it all together' videos that demonstrate higher-performing multi-transistor amplifiers. Each of those amps will have several stages, usually including current sources (or mirrors), differential pairs, common-emitter gain blocks, and emitter followers. They all work together so that each stage compensates for the failings of the others (a follower to provide low output impedance, for instance, or a current source to give the appearance of a nearly-infinite tail resistor.) All of them use feedback to stabilise gain and minimise offsets.
For small-signal circuits, pretty much everyone just uses an op-amp rather than a single-transistor circuit, unless you need to do something particular - for example, use the exponential relation between base voltage and collector current, or use the transistor to regulate a current, or use the transistor to handle more power than the op-amp can put out. But understanding the basics of what's going on _inside_ the op amp - which this series is working up to - is vital to understanding the limitations of what an op amp can do. That will come up in discussing op-amp integrators (input offset voltages), using an op amp to drive a power amp (crossover distortion and op-amp slew rate), making a radiofrequency amplifier (Miller effect) and so on... just as the discussion of current mirrors led us to the Early effect.
@@KludgesFromKevinsCave Thank you and understand. I guess my question was finding the optimal bias point.in the most linear region. Say for high-quality audio, where you wish to use discrete transistors over Op Amps, what is the design process to find that bias point? I understand that bigger picture objective gets way more complicated and requires far more sophisticated approaches, plus negative feedback, etc. I'm just trying get the design process clear in my mind. My original background was electronics, but my career moved away from it many years ago.
@tonykirkham1663 I wish I could give you a single recipe. In a way, this entire series is working toward the Holy Grail of linearity and dynamic range. Once feedback is in play, as it always should be, the linearity is nearly assured, and biasing becomes a question of 'how far can the signal swing without clipping? Center the bias point in that range.' Each presentation of a given circuit topology (follower, common emitter, etc) has at least an example of finding the bias point.
I'm generally skeptical of the "no op-amps in high-quality audio" camp. Sometimes you need discretes in a low-noise preamp, where signal levels are minuscule, or in a power stage where you have to dump a lot of heat, but with reasonable signal levels, op-amps are awfully good. With that said, I surely don't have a 'golden ear' - I grew up very near a busy airport, and have had some minimal hearing loss all my life. I trust my instruments rather than my senses.
The engineer of that pinout woke up in the morning and chose violence.
The dual transistors all seem to come in strange pinouts. DMMT5401 and its complementary partner, DMMT5551, even have more than one choice for them, both peculiar. www.diodes.com/assets/Datasheets/ds30436.pdf
It's actually not horrible for layout. Common emitter (as for a differential pair or exponential converter) and common base (as for a current mirror) both put the connected pads right next to one another.
@@KludgesFromKevinsCave Mmm. True. That makes sense for a current mirror layout. I see onsemi has a different layout that is EBC EBC, but the EBC on each side do not all belong to the same transistor (which is also confusing to me in a different way). For me, it would make sense to just have each side be EBC or CBE for each transistor, both for most layouts and conceptually, but that's just me. No doubt there's a reason not to do this on the die.
Спасибо!
Очень плзнавательный ролик!
Excellent tutorial. It is may be worthwhile to recognise that the difference of diode forward voltage drops is of the order of 2.5%, i.e. neither very significant nor insignificant.
Hmm. I suppose. I think the important point is that the difference is always there, so you can tell emitter from collector that way. Sometimes at the bench it's quicker to use a multimeter than look up the data sheet for the transistor.
@@KludgesFromKevinsCave True
Very good
Glad you liked!
Thank you professor ◜◯◯◝. I'm not sure the girls will like "me" more but I know I will. All this is going onto my resume! Cheers from So.Ca.USA 3rd house on the left.
Bye Bye.(Just like me)
Ok, here's a like. Bye!