I have possibly found an issue which others may have taken a rather askance notice of or my have actually determined to be a source of distortion in amplifiers. It involves the use of capacitors close to the base terminals of the differential pair input transistors.
I suspect that they may be a cause of considerable distortion as the bias current through the base-emitter diode adds charge to them. Series input capacitors as well as feedback high pass ones would be the main culprits. The solution I propose would be to omit such capacitors from the circuit and use an opamp input stage with the needed high pass filter in its feedback circuit along with the DC blocking input capacitor. Another possible solution would be to use mosfets instead of BJT's. I think using JFET's could also solve the problem since their diodes should not have to conduct. But they, like mosfets, may not be as stable in maintaining DC offset since they may have less voltage gain. Maybe my ideas are wrong here. I would like to see some thoughts concerning this issue.
This problem may not be severe given the high current gain of the input transistors. But I still think I have noticed trouble on many occasions.
The best way yet may just to drive the output driver stage with the opamp, as has been done in some circuits. The problem with this method though is that more voltage gain is still needed since most opamps can only handle 36 volts or less. Even complete chip amps are limited to only 86 volts max. The problem may be best solved with the opamp driving either one conventional VAS transistor or two, one on each rail complementary style. Then, suitable base series resistors would be needed to convert the voltage of the opamp to current somewhat proportional to its output voltage. Alternately, an opamp would need a differential open collector output which could be cascoded to the traditional VAS transistor.
Another possible solution would be to use a complementary input stage. Then the base currents though the base-emitter diodes would cancel out. But then the choice of possible input transistors is reduced and the amount of circuitry is doubled.
One other solution may be to simply try to balance the capacitance present at both bases. So, if the high pass capacitor in the feedback loop is 68uF through a 1k resistor, then the input resistor and series capacitor should be the same. The problem here is the rather low 1k input resistor. The global feedback resistor does not factor in much since it does not lead to a capacitor. I would simply recommend that the base ground reference resistor be the same as the NFB resistor. Also, an adjustble DC offset may be needed somewhere to compensate for a high value input base grounding resistor, but maybe not.
I suspect that they may be a cause of considerable distortion as the bias current through the base-emitter diode adds charge to them. Series input capacitors as well as feedback high pass ones would be the main culprits. The solution I propose would be to omit such capacitors from the circuit and use an opamp input stage with the needed high pass filter in its feedback circuit along with the DC blocking input capacitor. Another possible solution would be to use mosfets instead of BJT's. I think using JFET's could also solve the problem since their diodes should not have to conduct. But they, like mosfets, may not be as stable in maintaining DC offset since they may have less voltage gain. Maybe my ideas are wrong here. I would like to see some thoughts concerning this issue.
This problem may not be severe given the high current gain of the input transistors. But I still think I have noticed trouble on many occasions.
The best way yet may just to drive the output driver stage with the opamp, as has been done in some circuits. The problem with this method though is that more voltage gain is still needed since most opamps can only handle 36 volts or less. Even complete chip amps are limited to only 86 volts max. The problem may be best solved with the opamp driving either one conventional VAS transistor or two, one on each rail complementary style. Then, suitable base series resistors would be needed to convert the voltage of the opamp to current somewhat proportional to its output voltage. Alternately, an opamp would need a differential open collector output which could be cascoded to the traditional VAS transistor.
Another possible solution would be to use a complementary input stage. Then the base currents though the base-emitter diodes would cancel out. But then the choice of possible input transistors is reduced and the amount of circuitry is doubled.
One other solution may be to simply try to balance the capacitance present at both bases. So, if the high pass capacitor in the feedback loop is 68uF through a 1k resistor, then the input resistor and series capacitor should be the same. The problem here is the rather low 1k input resistor. The global feedback resistor does not factor in much since it does not lead to a capacitor. I would simply recommend that the base ground reference resistor be the same as the NFB resistor. Also, an adjustble DC offset may be needed somewhere to compensate for a high value input base grounding resistor, but maybe not.
And what about non-linear input junction capacitance of semiconductors? (junction capacitance is a non-linear function of applied voltage). See Walt Jung's articles for Electronic Design.
- Status
- Not open for further replies.