I have already answered some of the questions you asked.
This schematic is in general use. I am also thinking of a possibility to use 2 operational amplifiers in differential mode each driving a half of the transistors.
Even as is the schematic performs very well. This schematic has bee n used a lot. For sound too. There is no logical problem. The logical problem with the transistors and the 0 has been logically overcome by the operational amplifier and the feedback. This is why I did not explain the principle on the first place neither looked into the problem. I just said something alike: "the feedback takes care of this..."
Obviously, practically, there will always be something. Another example is the transitors are noisier than the tubes. So why do they use them at all. Same applies to the operational amplifiers.
Free or paid, simulators are not a wrench but a pair of stockings. One can put a pair of stockings instead of a belt. But one may run into a lot of trouble unless one uses the new belt in specific circumstances: caution, slow revs, good cooling, hood off, temporarily until the next gas station.
You can run this circuit in a simulator but the best way is to build the circuit and test in a laboratory. Or there is a better way: pass through the simple mathematics of the circuit.
The only thing you need to do is to pass through the logic of the circuit, at least through the logic which is significant and ignore the insignificant logic. When you want mathematical analysis, go ahead: simple arithmetics or calculus. What do you need a similator for?
Simulators help in another way: human psychology works in a better way when there is certain type of visualisation. Things become more mechanical and easier for the people. This visualisation may make the people think of other things. As an example, when you run your circuits through a simulator, you get a lot of OTHER ideas or ideas for improvement not connected with what the simulator talks.
HOWEVER, SIMULATORS MAY BE MISLEADING. The only thing which will not be misleading is the logical way through the circuit.
Simulators may have been designed to give suggestions too, for example, change this OA with this, etcetera. But then, computer software may exist to do the whole design, so why do we tlak at this forum at all.
Here is what I would suggest: read everything I have written. Disregard the mistakes. Make a schematic. Do the same way as I have done. Write up the same style of explanations. Go through every single components. Be a simulator yourself. Then do something I don't. Once you are ready, go back and read everything you have written. Have a few goes.
Then go to the simulator. Run the schematics. Is there anything the simulator can tell you which you haven't known?
***
Here is what happens:
The input voltage adjusts the output voltage to be the same just amplified. This is done by the operational amplifier driving the push pull common collectors. The voltage at the output of the operational amplifier is 1.4V higher / lower than the voltage of the output for the positive / negative wave. As the input voltage goes from high to low, so does the output. When the input voltage is very close to 0, the input voltage requires the output voltage to also go close to zero. This will happen to the output voltage of the circuit but not to the output voltage of the operational amplifier which will remain higher than the voltage necessary for the transistors to stay open which is around 1.4V. This situation stays on until the input voltage goes to a bit lower than 0. Then the operational amplifier has to close one of the Darlingtons and open the other and adjust the output voltage to a value a bit lower than zero. So, the operational amplifier must quickly go from around 1.4V to around -1.4V. Then, as the input voltage goes lower, the operational amplifier output will go lower too and will be approximately 1.4V lower than what is necessary to equalise the divided by the feedback resistors output of the circuit to the input voltage.
During the “jump” of the operational amplifier output from around 1.4V to around -1.4V, the two transistors are closed and no output is displayed. However, the transistors are not perfect and have some inertia caused by the 5pF BE junction capacitor. So, the transistors will retain the around zero value for a while. The higher the current through the transistors the lower the delay the faster the switch.
The problems which have caused so many comments are related to the past. In the past, push pulls were used in transistors in order to boost the efficiency. These were not connected with an overall ( general ) feedback and were not run by operational amplifiers. Then, the first operational amplifiers were very slow, although they have improved the performance to a usable level. These days, the operational amplifiers are so fast, I would not be suprised they switch faster than the transistors.
The lower the closed loop gain the faster the switch. The output will not be divided before being presented to the input. A tiny change in the output would make a bigger change at the input subtractor ( differential input ). This is also why the operational amplifiers are faster at unity gain.
The higher the open loop gain the faster the switch.
The faster the operational amplifier and the faster the transistors the lower the imperfect effect. A major concern has been the operational amplifier speed and not the speed of the transistors. Therefore, one can have a close look at the specifications. Even a simple, general purpose operational amplifier as TL084 would have a slew rate of: 13 V/µs typical, which would give 200ns to the operational amplifier to jump. Fast operational amplifiers could go faster in the nanosecond range. This is nothing as compared to the other problems which exist with electronic circuits.
Again: the operational amplifier will make the output be the same as the input, just amplified, as long as this is physically possible. With this schematic, to do so is physically possible. And impossible without a problem. The problem appears when the input crosses 0 volts. The problem will make a very tiny voltage which is supposed to be displayed at the output to look 0. The faster the frequency the less tiny this voltage is ( the more the input would go before the output is on ( reacts ) ). HOWEVER, SOUND SIGNALS ARE INCREDIBLY SLOW SIGNALS AND THE VERY MAXIMAL FREQUENCY IS LESS THAN 30KHz.
Compare the problem with the switch with the digitisation which also does not display exactly the same voltage as well as does not get the zero exactly perfectly due to digitisation of the period as well as digitisation of the level ( voltage ). Yet, digital systems are the systems with the best sound.
Anyways, whoever wants to do calculus, you may wish to calculate how much a pure sine voltage changes after the 0 point when the frequency is 25KHz for 200ns.
Again, your concerns are concerns for push pull transistors without operational amplifier and a common feedback.
This schematic is in general use. I am also thinking of a possibility to use 2 operational amplifiers in differential mode each driving a half of the transistors.
Even as is the schematic performs very well. This schematic has bee n used a lot. For sound too. There is no logical problem. The logical problem with the transistors and the 0 has been logically overcome by the operational amplifier and the feedback. This is why I did not explain the principle on the first place neither looked into the problem. I just said something alike: "the feedback takes care of this..."
Obviously, practically, there will always be something. Another example is the transitors are noisier than the tubes. So why do they use them at all. Same applies to the operational amplifiers.
Free or paid, simulators are not a wrench but a pair of stockings. One can put a pair of stockings instead of a belt. But one may run into a lot of trouble unless one uses the new belt in specific circumstances: caution, slow revs, good cooling, hood off, temporarily until the next gas station.
You can run this circuit in a simulator but the best way is to build the circuit and test in a laboratory. Or there is a better way: pass through the simple mathematics of the circuit.
The only thing you need to do is to pass through the logic of the circuit, at least through the logic which is significant and ignore the insignificant logic. When you want mathematical analysis, go ahead: simple arithmetics or calculus. What do you need a similator for?
Simulators help in another way: human psychology works in a better way when there is certain type of visualisation. Things become more mechanical and easier for the people. This visualisation may make the people think of other things. As an example, when you run your circuits through a simulator, you get a lot of OTHER ideas or ideas for improvement not connected with what the simulator talks.
HOWEVER, SIMULATORS MAY BE MISLEADING. The only thing which will not be misleading is the logical way through the circuit.
Simulators may have been designed to give suggestions too, for example, change this OA with this, etcetera. But then, computer software may exist to do the whole design, so why do we tlak at this forum at all.
Here is what I would suggest: read everything I have written. Disregard the mistakes. Make a schematic. Do the same way as I have done. Write up the same style of explanations. Go through every single components. Be a simulator yourself. Then do something I don't. Once you are ready, go back and read everything you have written. Have a few goes.
Then go to the simulator. Run the schematics. Is there anything the simulator can tell you which you haven't known?
***
Here is what happens:
The input voltage adjusts the output voltage to be the same just amplified. This is done by the operational amplifier driving the push pull common collectors. The voltage at the output of the operational amplifier is 1.4V higher / lower than the voltage of the output for the positive / negative wave. As the input voltage goes from high to low, so does the output. When the input voltage is very close to 0, the input voltage requires the output voltage to also go close to zero. This will happen to the output voltage of the circuit but not to the output voltage of the operational amplifier which will remain higher than the voltage necessary for the transistors to stay open which is around 1.4V. This situation stays on until the input voltage goes to a bit lower than 0. Then the operational amplifier has to close one of the Darlingtons and open the other and adjust the output voltage to a value a bit lower than zero. So, the operational amplifier must quickly go from around 1.4V to around -1.4V. Then, as the input voltage goes lower, the operational amplifier output will go lower too and will be approximately 1.4V lower than what is necessary to equalise the divided by the feedback resistors output of the circuit to the input voltage.
During the “jump” of the operational amplifier output from around 1.4V to around -1.4V, the two transistors are closed and no output is displayed. However, the transistors are not perfect and have some inertia caused by the 5pF BE junction capacitor. So, the transistors will retain the around zero value for a while. The higher the current through the transistors the lower the delay the faster the switch.
The problems which have caused so many comments are related to the past. In the past, push pulls were used in transistors in order to boost the efficiency. These were not connected with an overall ( general ) feedback and were not run by operational amplifiers. Then, the first operational amplifiers were very slow, although they have improved the performance to a usable level. These days, the operational amplifiers are so fast, I would not be suprised they switch faster than the transistors.
The lower the closed loop gain the faster the switch. The output will not be divided before being presented to the input. A tiny change in the output would make a bigger change at the input subtractor ( differential input ). This is also why the operational amplifiers are faster at unity gain.
The higher the open loop gain the faster the switch.
The faster the operational amplifier and the faster the transistors the lower the imperfect effect. A major concern has been the operational amplifier speed and not the speed of the transistors. Therefore, one can have a close look at the specifications. Even a simple, general purpose operational amplifier as TL084 would have a slew rate of: 13 V/µs typical, which would give 200ns to the operational amplifier to jump. Fast operational amplifiers could go faster in the nanosecond range. This is nothing as compared to the other problems which exist with electronic circuits.
Again: the operational amplifier will make the output be the same as the input, just amplified, as long as this is physically possible. With this schematic, to do so is physically possible. And impossible without a problem. The problem appears when the input crosses 0 volts. The problem will make a very tiny voltage which is supposed to be displayed at the output to look 0. The faster the frequency the less tiny this voltage is ( the more the input would go before the output is on ( reacts ) ). HOWEVER, SOUND SIGNALS ARE INCREDIBLY SLOW SIGNALS AND THE VERY MAXIMAL FREQUENCY IS LESS THAN 30KHz.
Compare the problem with the switch with the digitisation which also does not display exactly the same voltage as well as does not get the zero exactly perfectly due to digitisation of the period as well as digitisation of the level ( voltage ). Yet, digital systems are the systems with the best sound.
Anyways, whoever wants to do calculus, you may wish to calculate how much a pure sine voltage changes after the 0 point when the frequency is 25KHz for 200ns.
Again, your concerns are concerns for push pull transistors without operational amplifier and a common feedback.
Ask yourself this question.
Why does nobody else use the method you are suggesting? (using a driving opamp to try to also bias the output devices) I do not know for sure. Perhaps you are the first person ever to think up this idea. Maybe no one ever tried it before and it works fine?
Whatever your objection to 100% free simulators might be, I can not imagine.
They are just a FREE tool to add to all other tools.
That is like saying "I will never use a wrench no matter what, even if you give me one for free". Ok?
It's either that or you have to build a prototype.
Have you built any of your designs as prototypes or finished units to date?
Many who do find that things often do not work exactly as theory or they expected them to.
Regards,
_-_-bear
Terrific
Is that you????
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Agreed. However you refuse to do either, which is perhaps why you still don't see the problems.
I have done that. You have not. When presented with my "Dummy's guide" to the math, you refused to believe it.
Well here are two possible and somewhat related ideas to toy with:
1) turn the opamp section into an "instrumentation" opamp, (single ended in, balanced out?) one with outputs in opposite phase, but bias the output stage of each output opamp into class A by some means, thereby placing DC bias on the ouput darlingtons - so, gaining a benefit in two regards.
2) use a "pull-up" (or pull-down) circuit at the output of your opamp to obtain bias voltage for the darlington stage. This also takes the opamp output out of Class B.
Why not eliminate a problem before having to "fix it" with feedback? This way the feedback is more effective and has to "fix less"?
What do you think?
No good?
_-_-bear
1) turn the opamp section into an "instrumentation" opamp, (single ended in, balanced out?) one with outputs in opposite phase, but bias the output stage of each output opamp into class A by some means, thereby placing DC bias on the ouput darlingtons - so, gaining a benefit in two regards.
2) use a "pull-up" (or pull-down) circuit at the output of your opamp to obtain bias voltage for the darlington stage. This also takes the opamp output out of Class B.
Why not eliminate a problem before having to "fix it" with feedback? This way the feedback is more effective and has to "fix less"?
What do you think?
No good?
_-_-bear
And your simulator has said the delay was 1 microsecond.
Yes. Looks very unbeliavable. Sounds like uA709.
OK. In case you say so, I believe you. Would you like to run the schematic with a fast operational amplifier? Probably not.
Then the circuit can be very well used for garage bands. Do you think they care how much the non linear distortions are? I do NOT. I have to ask some.
Also, the arangement can be redesigned. I hate to use non feedbacked transistors, though. They are nasty.
Also, your answer was not dummy's guide but the best possible, This is the language one must use because there is nothing else. The rest is useless talk of people who make themselves doing a non dummy work whereas the whole electronics is dummy. At least, your answer showed some logic. Most of the rwst are just empty talks.
I will continue to refuse to accept what you say is true until I do the math and I will not do this very soon. Instead, I can simply use a fast operational amplifier when I want higher quality. Thus I will use the same circuit which is what is important.
Yes. Looks very unbeliavable. Sounds like uA709.
OK. In case you say so, I believe you. Would you like to run the schematic with a fast operational amplifier? Probably not.
Then the circuit can be very well used for garage bands. Do you think they care how much the non linear distortions are? I do NOT. I have to ask some.
Also, the arangement can be redesigned. I hate to use non feedbacked transistors, though. They are nasty.
Also, your answer was not dummy's guide but the best possible, This is the language one must use because there is nothing else. The rest is useless talk of people who make themselves doing a non dummy work whereas the whole electronics is dummy. At least, your answer showed some logic. Most of the rwst are just empty talks.
I will continue to refuse to accept what you say is true until I do the math and I will not do this very soon. Instead, I can simply use a fast operational amplifier when I want higher quality. Thus I will use the same circuit which is what is important.
I had done #2. As a gross generalisation I kind of like what happened. I seem to like the original better because of the purer logic.
As a gross generalisation, I do not like #1 because of the logic of the schematic. The solution is not elegant enough. But OK.
What I like the most is your answer. Much better than before. Shows logical approach. You continue to use terms but OK. Many people do this.
As a gross generalisation, I do not like #1 because of the logic of the schematic. The solution is not elegant enough. But OK.
What I like the most is your answer. Much better than before. Shows logical approach. You continue to use terms but OK. Many people do this.
No, I hadn't tried a simulation, but just did a rough calculation based on a random guess at slew rate (without checking). I should be more careful.
Anyway, I did try a simulation now with TL084. According to the simulator, I was way out. The switching is actually much slower than I estimated.
You werent that off...
Make a try at 4V , 2V , 1V PK outputs...
Distorsions components will remain at about the same absolute level
when output level is reduced , that is , switching distorsion ratio
will increase when deacreasing output power.
Yes, agreed. I tested with 2V pk-pk @ 1KHz, and the voltage gain set to 11. It was a nice reminder how nasty crossover distortion is - all the odd order harmonics from 3'rd to 19'th were almost the same level (about -50dB).
What surprised me was that the visible "glitch" in the waveform at crossover lasted over 10uS. Thinking about that, I suppose it's mostly because 13V/uS is the maximum slew rate with a large error voltage between the input pins of the opamp. With a smaller error voltage, the actual slew rate is much lower.
I do agree with the OP that distortion can be reduced by using a faster opamp, or less voltage gain. With the circuit changed to unity gain, distortion was about 10 times lower, as expected.
Maybe a circuit something along these lines might fit into your scheme?
http://www.analog.com/static/imported-files/data_sheets/SSM2142.pdf
SSB, one of the things I mentioned to you early on is that if your design is going to have relatively high distortion of the sort that spreads the spectra of distortion up in frequency and then of relatively high level compared to the 2nd and 3rd order, then one can simply buy for less $$ than it takes to build or design from scratch all sorts of commercially built amps and integrated amps that will be of higher quality and more usefulness.
SSB, you said: "Then the circuit can be very well used for garage bands. Do you think they care how much the non linear distortions are? I do NOT. I have to ask some."
That makes it not terribly "universal"?
Perhaps it would be a good idea to more carefully define your design goals.
As a thought exercise, then whatever you do is fine,
As a guide for others to follow, perhaps not so good?
The title of the thread strongly suggests that it is a guide or tutorial, or else a design to duplicate or follow.
Did you look around this site at all?
Have you looked at the other designs?
Perhaps there are some things to be discovered and delighted in?
Are you aware that the topic of "cross over distortion" - the off time of output transistors - is a point of significant concern and discussion? Many ideas and methods have been suggested, engineered, analyzed, simulated, tested and implemented to mediate and minimize this seemingly inconsequential part of a P-P output design.
_-_-bear
PS. some people are even concerned about permitting transistors in a P-P output stage TO switch "off" at all. Did you ever come across this concern?
http://www.analog.com/static/imported-files/data_sheets/SSM2142.pdf
SSB, one of the things I mentioned to you early on is that if your design is going to have relatively high distortion of the sort that spreads the spectra of distortion up in frequency and then of relatively high level compared to the 2nd and 3rd order, then one can simply buy for less $$ than it takes to build or design from scratch all sorts of commercially built amps and integrated amps that will be of higher quality and more usefulness.
SSB, you said: "Then the circuit can be very well used for garage bands. Do you think they care how much the non linear distortions are? I do NOT. I have to ask some."
That makes it not terribly "universal"?
Perhaps it would be a good idea to more carefully define your design goals.
As a thought exercise, then whatever you do is fine,
As a guide for others to follow, perhaps not so good?
The title of the thread strongly suggests that it is a guide or tutorial, or else a design to duplicate or follow.
Did you look around this site at all?
Have you looked at the other designs?
Perhaps there are some things to be discovered and delighted in?
Are you aware that the topic of "cross over distortion" - the off time of output transistors - is a point of significant concern and discussion? Many ideas and methods have been suggested, engineered, analyzed, simulated, tested and implemented to mediate and minimize this seemingly inconsequential part of a P-P output design.
_-_-bear
PS. some people are even concerned about permitting transistors in a P-P output stage TO switch "off" at all. Did you ever come across this concern?
Last edited:
This will fix your amplifier without the ditzy opamp kludges.
Class B amplifier has automatic bias : EDN Europe
http://www.edn-europe.com/cmsimages/1209/35f1.jpg
G²
SSB, you really should spend more time looking at other designs for amps on here, rather than trying to do your own design right now. Take a look? It's free. There is one amp on here that is ALL chips. Very neat. Up your alley I would think. Multiple chips that is. Ultralow distortion spec. Below what you can measure. There are also those who are specializing in "chipamp" designs (IC power opamps). Saves the trouble of a whole lot of design in the amp itself (or does it?). Take a look around...
Gotta tell you, that by way of analogy, you are a bit like someone saying that they are going to blaze a trail across the continent, when Lewis & Clark already did it, went, came back and retired...
_-_-bear
Gotta tell you, that by way of analogy, you are a bit like someone saying that they are going to blaze a trail across the continent, when Lewis & Clark already did it, went, came back and retired...
_-_-bear
Be specific. The idea is to have the two transistors work in the mid range and work differentially with the same voltage just inverted which is displayed at the load.
The obvious disadvantage has been noted: No common point between source and load. Other than this, please specify.
Generally, there have been mistakes before as far as I remember. There may be here too. In case you say what you have found, I would look at this. Otherwise, I do not want to do the effort just for fun.
The obvious disadvantage has been noted: No common point between source and load. Other than this, please specify.
Generally, there have been mistakes before as far as I remember. There may be here too. In case you say what you have found, I would look at this. Otherwise, I do not want to do the effort just for fun.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.