Need of matching transistors & advanced circuits
As we analyzed in the thread “ matching transistors ”, became comprehensible that transistor of same type they have difference between them as for ac or dc gain (hfe and hFE respectively) in common emitter or common collector configuration (i.e. emitter follower) something that makes finally to be not so much similar between them. The mismatching of two or more transistor it is critical in stages where these collaborate. The most common stages of such type in amplifiers are two.
The first is the input difference amplifier or long tailed pair (LTP) where two transistors extract the error of whole circuit presented in output via the feedback loop from the input. Thus the amplifier remains stable. If the two transistors have different hfe or hFE under static condition (with no signal in input) this translated in output as dc offset negative or positive. Under dynamic condition (with signal in input) this also translated as more positive or more negative going amplified output signal around the zero volt cross point and thus increased distortion during instant peaks. There are three solutions to resolve this problem. First is the matching of transistors, second the addition of a trimmer between the emitters or at the base of input transistor to share manually the same current at collectors, and the third is the addition of a current mirror as an active load in the collectors which automatically balances the collector currents. The third solution seems to be the better because eliminates the extra work and the additional components needed in the input side which increases the fundamental Johnson noise or the transconductance factor occasionally (remembering that the input difference amplifier translates the voltage of input signal in an equivalent current in his output) and due to the dynamic characteristics of the two transistors of the current mirror (they have the beta factor) it is positive under all circumstances (static and dynamic). In conclusion, in the input stage there is no need to match the difference amplifier transistors because the existence of the two above solutions.
The second stage and the most difficult it is the output of amplifier. In this stage there are usually two and more paralleled pairs of complementary transistors. Suppose that we have three pairs of transistors and that are not matched between them. Under static condition (without input signal) the only current that flows through collectors it is the idle current usually between 50 to 100 mA for a pure class B amplifier and there is no problem due to unbalanced collector currents owed to the different hfe or hFE of the big output transistors characterized by the high SOA. What happens under dynamic conditions (signal in input) is that the transistor with the higher hFE from the other output transistors sources the output load or sinks from the supply rail the greater amount of the total current flows to the output load. Thus exceeds his limits of SOA constantly and it is under the danger of destruction at any time. To protect the output devices from the early years was added a special circuit the well known VI limiter. In the past used only two current sensing resistors around the emitter resistors of one pair of output transistors and the matching of all was critical as it is easily understood. After this, was added sensing resistors in each emitter resistor, thus caused a summing network which eliminates the drawback of the first approach, but and in this case the matching is necessary (not critical) to ensure that one transistor not operates against the others (it exists continuously the VI limiter circuit without reason). Further improvement in the present days given by the use of extra circuits (named analog computers) which with the aid of optocouplers mounted on one or two emitter resistors (complementary) and comparators which compares the input and output signal continuously activates the output disconnection relays, or an appropriate input signal compression circuit to decrease the drive signal instantly or continuously, offers positive protection of output devices yet under the worse output condition such as heavy or direct short circuited output. But yet under these types of output protection, the matching of output transistors remains necessary and a +/- 10% tolerance in hFE it is acceptable and there is no significant reason to examine the transfer characteristic curve of each transistor i.e. hfe, the simple hFE measurement is enough with all these auxiliary protection circuits.
This first article offered to our colleague in diyaudio forum who suffers with paleolithic amplifier circuits (simplicity??!! it is simply justification).
Fotios Anagnostou
As we analyzed in the thread “ matching transistors ”, became comprehensible that transistor of same type they have difference between them as for ac or dc gain (hfe and hFE respectively) in common emitter or common collector configuration (i.e. emitter follower) something that makes finally to be not so much similar between them. The mismatching of two or more transistor it is critical in stages where these collaborate. The most common stages of such type in amplifiers are two.
The first is the input difference amplifier or long tailed pair (LTP) where two transistors extract the error of whole circuit presented in output via the feedback loop from the input. Thus the amplifier remains stable. If the two transistors have different hfe or hFE under static condition (with no signal in input) this translated in output as dc offset negative or positive. Under dynamic condition (with signal in input) this also translated as more positive or more negative going amplified output signal around the zero volt cross point and thus increased distortion during instant peaks. There are three solutions to resolve this problem. First is the matching of transistors, second the addition of a trimmer between the emitters or at the base of input transistor to share manually the same current at collectors, and the third is the addition of a current mirror as an active load in the collectors which automatically balances the collector currents. The third solution seems to be the better because eliminates the extra work and the additional components needed in the input side which increases the fundamental Johnson noise or the transconductance factor occasionally (remembering that the input difference amplifier translates the voltage of input signal in an equivalent current in his output) and due to the dynamic characteristics of the two transistors of the current mirror (they have the beta factor) it is positive under all circumstances (static and dynamic). In conclusion, in the input stage there is no need to match the difference amplifier transistors because the existence of the two above solutions.
The second stage and the most difficult it is the output of amplifier. In this stage there are usually two and more paralleled pairs of complementary transistors. Suppose that we have three pairs of transistors and that are not matched between them. Under static condition (without input signal) the only current that flows through collectors it is the idle current usually between 50 to 100 mA for a pure class B amplifier and there is no problem due to unbalanced collector currents owed to the different hfe or hFE of the big output transistors characterized by the high SOA. What happens under dynamic conditions (signal in input) is that the transistor with the higher hFE from the other output transistors sources the output load or sinks from the supply rail the greater amount of the total current flows to the output load. Thus exceeds his limits of SOA constantly and it is under the danger of destruction at any time. To protect the output devices from the early years was added a special circuit the well known VI limiter. In the past used only two current sensing resistors around the emitter resistors of one pair of output transistors and the matching of all was critical as it is easily understood. After this, was added sensing resistors in each emitter resistor, thus caused a summing network which eliminates the drawback of the first approach, but and in this case the matching is necessary (not critical) to ensure that one transistor not operates against the others (it exists continuously the VI limiter circuit without reason). Further improvement in the present days given by the use of extra circuits (named analog computers) which with the aid of optocouplers mounted on one or two emitter resistors (complementary) and comparators which compares the input and output signal continuously activates the output disconnection relays, or an appropriate input signal compression circuit to decrease the drive signal instantly or continuously, offers positive protection of output devices yet under the worse output condition such as heavy or direct short circuited output. But yet under these types of output protection, the matching of output transistors remains necessary and a +/- 10% tolerance in hFE it is acceptable and there is no significant reason to examine the transfer characteristic curve of each transistor i.e. hfe, the simple hFE measurement is enough with all these auxiliary protection circuits.
This first article offered to our colleague in diyaudio forum who suffers with paleolithic amplifier circuits (simplicity??!! it is simply justification).
Fotios Anagnostou
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.