Take a peak hold circuit, reverse the direction of the diode, and bias the cap to + with a very high value resistor.capslock said:
Re: Temperature compensation
A good argument can be made for thermal lag in the temperature sensing element - in such a feedback loop stability is achieved through a rolloff, and the lag between sensing element and chip is a perfectly fine place to put it.nemestra said:
Nelson Pass
... and then you end up with the circuit I described earlier, which I found in Audio Amateur some five years ago or so. The problem with this circuit is that it is very fast when measuring the lowest value of the voltage which means that it is good when it comes to ensuring that the output stage does not run out of bias. But when the bias increases, as with higher temperatures, the speed relies entirely on the resistor that charges the cap. This resistor must be very high ohmic or the capacitor very large, otherwise you end up with a sawtooth bias. That is why I do not think that this circuit is worth the effort. You end up with a time constant that is longer, than with traditional thermal compensation.
I therefore would like to see a zero crossing measurement device instead, I will start working on it today🙂
... and then you end up with the circuit I described earlier, which I found in Audio Amateur some five years ago or so. The problem with this circuit is that it is very fast when measuring the lowest value of the voltage which means that it is good when it comes to ensuring that the output stage does not run out of bias. But when the bias increases, as with higher temperatures, the speed relies entirely on the resistor that charges the cap. This resistor must be very high ohmic or the capacitor very large, otherwise you end up with a sawtooth bias. That is why I do not think that this circuit is worth the effort. You end up with a time constant that is longer, than with traditional thermal compensation.
I therefore would like to see a zero crossing measurement device instead, I will start working on it today🙂
peak circuit
The details are here (21th years anno):
The Autobias Amplifier: A New Topology for Automatically Biased Audio Amplifiers Using Power MOSFETs
Vol. 30, Number 4 pp. 208 (1982)
Author: Bill Roehr
Abstract: An obstacle blocking wide acceptance of power MOSFETs in audio amplifiers is the lack of an automatic bias technique. A unique circuit topology senses and maintains quiescent current despite the half-wave pulses inherent in class AB operation.
The details are here (21th years anno):
The Autobias Amplifier: A New Topology for Automatically Biased Audio Amplifiers Using Power MOSFETs
Vol. 30, Number 4 pp. 208 (1982)
Author: Bill Roehr
Abstract: An obstacle blocking wide acceptance of power MOSFETs in audio amplifiers is the lack of an automatic bias technique. A unique circuit topology senses and maintains quiescent current despite the half-wave pulses inherent in class AB operation.
to pabo
I therefore would like to see a zero crossing measurement device instead, I will start working on it today ------- that might be overcomplicated, better have a look on our circuits
I therefore would like to see a zero crossing measurement device instead, I will start working on it today ------- that might be overcomplicated, better have a look on our circuits
Pabo said:
What is the reason you chose BC860A transistors for the current mirror in the Self input stage?
Ampman, is this the thread you were referring to? It has a schematic scan posted by Peter Daniel?
http://www.diyaudio.com/forums/showthread.php?s=&threadid=4320&highlight=bias
http://www.diyaudio.com/forums/showthread.php?s=&threadid=4320&highlight=bias
Quote from AMPMAN:
"this is a quote from the authers of the class AB amp.An important problem concerns the bias control loop . often acomplementary common-collecter output stage is used and the power transistors are included in the bias control loop. This can easily cause thermal instability due to the larte temperature variations in the ouput transistors.
Thermal coupling of all diodes and transistors in the class-AB control loop can improve the thermal stability of the quiescent current in the output stag, bu this is in most cases too slow to react to burst signals. As a result, emitter resistors are usually added to the power transistors to improve thermal stability. However, the voltage drop across the emitter resistors can switch off the transistor that is conducting the resisdual current."
To make things confusing, the authors of this article use the phrase "class-AB control loop" in a completely different sense than I did. Apparently they talk about the translinear loop (loop of forward-biased diode or base-emitter junction voltages) defining the way the current is split between the output devices, not about a non-linear common-mode feedback loop as invented by Huijsing and Tol.
In any case, if you use a non-linear common-mode loop for class AB control, you can also automatically get rid of all quiescent current stability issues and make your output stage non-switching at the same time. In principle, you can apply it to common source, common emitter, common drain and common collector as well as quasi-complementary output stages.
I don't know if the basic principle has ever been patented, but if it has, the patent has long passed its due by date anyway. Some specific implementations might be patent protected, but as far as I know, not the exp(-I1*R*q/(kT))+exp(-I2*R*q/(kT)) version I proposed in 1996.
About the exponential law for bipolar transistors: I cannot explain the exact device physics behind it, because I was never any good at that myself. But basically, the reason why the equation for an ideal transistor looks the same as for an ideal diode is simply that the base-emitter junction of a bipolar transistor IS a diode. The number of electrons injected into the base by the emitter (assuming an NPN) is governed by the same equations as the number of electrons injected into the anode by the cathode of a semiconductor diode. The difference is that the collector sucks most of them out of the base if the collector voltage is high enough, that is, if the transistor is in the forward active region.
Due to all kinds of second-order effects, in practice, transistors tend to follow the theoretical exponential law more accurately than diodes. It only goes wrong at high currents, due to the ohmic resistance of the base and emitter material and due to the so-called high injection effects.
MOSFET's are a completely different story. They behave more or less exponentially at low gate-source voltages (weak inversion region, few hundreds of millivolts below threshold), then gradually become more or less quadratic (strong inversion, a few hundred millivolts above threshold) and finally more or less linear (velocity saturation).
"this is a quote from the authers of the class AB amp.An important problem concerns the bias control loop . often acomplementary common-collecter output stage is used and the power transistors are included in the bias control loop. This can easily cause thermal instability due to the larte temperature variations in the ouput transistors.
Thermal coupling of all diodes and transistors in the class-AB control loop can improve the thermal stability of the quiescent current in the output stag, bu this is in most cases too slow to react to burst signals. As a result, emitter resistors are usually added to the power transistors to improve thermal stability. However, the voltage drop across the emitter resistors can switch off the transistor that is conducting the resisdual current."
To make things confusing, the authors of this article use the phrase "class-AB control loop" in a completely different sense than I did. Apparently they talk about the translinear loop (loop of forward-biased diode or base-emitter junction voltages) defining the way the current is split between the output devices, not about a non-linear common-mode feedback loop as invented by Huijsing and Tol.
In any case, if you use a non-linear common-mode loop for class AB control, you can also automatically get rid of all quiescent current stability issues and make your output stage non-switching at the same time. In principle, you can apply it to common source, common emitter, common drain and common collector as well as quasi-complementary output stages.
I don't know if the basic principle has ever been patented, but if it has, the patent has long passed its due by date anyway. Some specific implementations might be patent protected, but as far as I know, not the exp(-I1*R*q/(kT))+exp(-I2*R*q/(kT)) version I proposed in 1996.
About the exponential law for bipolar transistors: I cannot explain the exact device physics behind it, because I was never any good at that myself. But basically, the reason why the equation for an ideal transistor looks the same as for an ideal diode is simply that the base-emitter junction of a bipolar transistor IS a diode. The number of electrons injected into the base by the emitter (assuming an NPN) is governed by the same equations as the number of electrons injected into the anode by the cathode of a semiconductor diode. The difference is that the collector sucks most of them out of the base if the collector voltage is high enough, that is, if the transistor is in the forward active region.
Due to all kinds of second-order effects, in practice, transistors tend to follow the theoretical exponential law more accurately than diodes. It only goes wrong at high currents, due to the ohmic resistance of the base and emitter material and due to the so-called high injection effects.
MOSFET's are a completely different story. They behave more or less exponentially at low gate-source voltages (weak inversion region, few hundreds of millivolts below threshold), then gradually become more or less quadratic (strong inversion, a few hundred millivolts above threshold) and finally more or less linear (velocity saturation).
to capslock
I cannot sent a paper to you until will get your answer on my private message with your email
I cannot sent a paper to you until will get your answer on my private message with your email
sam jaywara
yes I did build the amp but I could only get one channel working inthe meantime I have concentrated on the aleph x.re the out put transistors,the authers have this to say,A problem with power transistors driven by a current source isthat there is no turn-off resistor for them.Under high-frequency,high-amplitude drive there will be a tendency for the effective bias current to rise dynamically.By useing HF power transistors with an ft of 80mhz, This bias current rise is reduced to 60% at 20khz at full drive.
yes I did build the amp but I could only get one channel working inthe meantime I have concentrated on the aleph x.re the out put transistors,the authers have this to say,A problem with power transistors driven by a current source isthat there is no turn-off resistor for them.Under high-frequency,high-amplitude drive there will be a tendency for the effective bias current to rise dynamically.By useing HF power transistors with an ft of 80mhz, This bias current rise is reduced to 60% at 20khz at full drive.
Ampman,
I am aware of that comment, ie., the requirement for 80MHz transistors in order to keep the bias stable. But the 2SC3281(5200) and 2SA1302(1943) pair are only 30MHz devices. Hence, my question - was the amplifier stable in long term operation and what about the sonics?
Thanks,
I am aware of that comment, ie., the requirement for 80MHz transistors in order to keep the bias stable. But the 2SC3281(5200) and 2SA1302(1943) pair are only 30MHz devices. Hence, my question - was the amplifier stable in long term operation and what about the sonics?
Thanks,
Ampman,
How big is this effect? Say the dc bias current is 50mA. If you then output 20KHz into 8 ohms, how much does the bias increase with, say, 30MHz ft transistors? How do you differentiate the bias increase due to slow turn-off effect from thermal changes? How much %change in effective bias matters?
Bam
How big is this effect? Say the dc bias current is 50mA. If you then output 20KHz into 8 ohms, how much does the bias increase with, say, 30MHz ft transistors? How do you differentiate the bias increase due to slow turn-off effect from thermal changes? How much %change in effective bias matters?
Bam
Bam and Sam
I'm afraid I cannot anwser your question's as I have'nt got astereo pair working
and secondly I'm not an EE. anybody care to put mosfets on the output.
I'm afraid I cannot anwser your question's as I have'nt got astereo pair working
and secondly I'm not an EE. anybody care to put mosfets on the output.Ampman,
Don't worry about a STEREO version of the amp. That one channel that you have working, that is the one in question. Just let us know if the amp works stable for a relatively long time driving a nominal speaker load.
Thanks,
Don't worry about a STEREO version of the amp. That one channel that you have working, that is the one in question. Just let us know if the amp works stable for a relatively long time driving a nominal speaker load.
Thanks,
- Status
- Not open for further replies.