Electronic circuit linearization method

[0001] This application relates in general to the field of analog electronics circuit design. More particularly, the application relates to a design for transistor circuits that linearizes gain characteristics of the amplifier.

[0002] Modern transistors have an inherent non-linearity in operation. This is explained by the Ebers-Moll equations:

[0003] $I_E=I_{\mathrm{ES}}\left(e^{\frac{V_{\mathrm{BE}}}{V_T}}-1\right)$ $I_C={\alpha }_F{I}_E$ $I_B=\left(1-{\alpha }_F\right)I_E$

• • [0004] Where:
  • [0005] V_T is the thermal voltage (approximately 26 mV at 300 K≈room temperature).
  • [0006] I_E is the emitter current.
  • [0007] I_C is the collector current.
  • [0008] α_F is the common base forward short-circuit current gain (approximately 0.99 up to nearly 1.0).
  • [0009] I_ES is the reverse saturation current of the base-emitter diode (on the order of 10⁻¹⁵ to 10⁻¹² A).
  • [0010] V_BE is the base-emitter voltage.

[0011] The nonlinear component is termed “Shockley Resistance,” or R_S which invariably leads to distortion in the amplifier output signal. This effect is inherent in all transistor based amplifiers. R_S is also temperature sensitive. R_S≈0.026 V/I, at room temperature, 300 K, I in Amps. This gives approximately 26Ω at 0.001 A.

[0012] Prior art methods of distortion reduction of amplifier gains stages typically employ signal feedback from the output node to the input node of the gain stage, sometimes across multiple gains stages, thereby using feedback to compensate for the inherent nonlinearity of the stages. These methods have several drawbacks. Deriving the input feedback signal from the output node allows undesired distortion artifacts to be present on the output node, which are subsequently passed on to the following stages and ultimately, e.g., an amplifier output. The feedback imparts changes to the overall signal that would not otherwise be present in the signal path. These changes can be perceived as a coloration of the signal, which is particularly noticeable in audio amplifiers. The capacitance and resistance of the feedback network and associated components causes a degradation of transients present in the signal. Percussive sounds such as cymbal crashes and drums are particularly impacted due to the fast transients presented by these sounds. The risetime of the transients are colored by the response of the feedback electronics.

[0013] The invention concerns an electronic circuit which corrects for non-linearities present in transistor devices. The inventive circuit allows for the construction of amplifiers that are inherently low in distortion, eliminating the requirement for the use of feedback to linearize the amplifier. The invention is particularly useful for driving audio amplifiers, but can be used with other audio devices and amplifiers. The circuit has an input transistor connected in tandem with a second transistor. The transistor pair functions as a unit to cancel the log varying base-emitter resistance of the input transistor, while also canceling the base-emitter temperature drift of the input transistor.

[0014] It is a major object of the invention to provide an improved method and circuit design for correcting for the inherent non-linearities in transistor circuits.

[0015] It is another object of the invention to provide an improved method for reducing distortion in audio amplifiers.

[0016] It is another object of the invention to provide a linearity correction technique for electronic circuits requiring minimal electronic components.

[0017] It is another object of the invention to provide a linearity correction technique for electronic circuits that does not use feedback.

[0018] It is another object of the invention to provide a linearity correction technique for electronic circuits that cancels the base-emitter temperature drift of the input transistor.

[0023] Referring now to FIGS. 1-4, embodiments and applications of the inventive method and circuit design are shown. The inventive method linearizes the gain characteristics of transistor amplifiers so as to minimize distortion without requiring feedback. The method applies standard electronic circuit components to overcome inherent current dependent nonlinearities in transistor circuits and can be employed in various forms for amplifier circuits such as operational amplifiers, RF circuits, or amplifier drivers. As stated above, the method corrects the nonlinearity caused by the Shockley Resistance R_S, thereby reducing distortion at each stage of signal transmission or amplification. As is known in the art, R_S is temperature sensitive. The inventive method compensates for this temperature dependence, and also has an inherent noise reduction feature as will be explained in more detail below.

[0024] Referring now particularly to FIG. 1, two transistors in tandem Q1 and Q2 are shown. Q1 is the input transistor, with its base as the input and emitter attached to the collector of Q2. Q2 functions as a saturated constant current source with a forward biased collector base diode, thus cancelling the log varying emitter Shockley resistance of Q1. This means that the exponential nonlinearity of Q1 will be offset to a good approximation by the inverse nonlinearity of Q2. This is the basis of the present invention, the method providing a substantially linear (about 95%) element within each gain stage. Three embodiments of a circuit designed with this principle are presented.

[0025] Referring again to the first embodiment of FIG. 1, the output of the tandem is the Q1 collector. A decreasing signal on Q1 will decrease the current in Q1 in a linear manner. An increasing signal on Q1 increases Q1 current in a linear manner until Q2 comes out of saturation at which point the current in Q1 is limited to the current provided by constant current source Q2. This tandem configuration also cancels the base-emitter temperature drift of Q1. The circuit of FIG. 1 can be used in many applications such as a differential amplifiers, opamps, power amplifier drivers, RF stage circuits and other analog circuits. In the inventive arrangement, Q1 and Q2 function essentially as a linear device which is why feedback is not required to linearize the overall circuit.

[0026] Referring now to FIG. 2, a second embodiment of the inventive method is shown, this embodiment using two complementary transistors in a pair, where Q1 is the input transistor, with its base as the input and its emitter attached to the collector of Q2. Q2 functions as a saturated current source for Q1. In this embodiment, the input bias voltage is adjusted for 1 mA through Q1 and Q2, approximately 1.4V. The saturated collector-base diode of Q2 thus cancels the log varying emitter resistance R_S of Q1. The output of the pair is the Q1 collector. A decreasing signal on Q1 will decrease the current in Q1 in a linear manner, while an increasing signal on Q1 increases the Q1 current in a linear manner until Q2s collector base diode fully saturates and all the current flows to ground, the current can go up limited by the collector load resistor when Q1 saturates. A gain limiting resistor from Q2's base to ground can be utilized. The complementary pair Q1, Q2 also cancels the base-emitter temperature drift of Q1.

[0027] Referring now to FIGS. 3A-3I, a series of prior art circuit configurations are shown illustrating the effects of current dependent non-linearities, while FIGS. 3J and 3K show the application of the inventive method to correct for the current dependent nonlinearities. The currents and resistor loads are merely exemplary and operations are not limited to these values. FIGS. 3A-3F show common emitter amplifier circuits using NPN transistors. FIGS. 3G-3I show common emitter amplifier circuits using PNP transistors.

[0028] For all of the circuits, the base inputs are referenced at ground (zero Volts) and the chosen collector load resistor is 10 KΩ. For a nominal gain of 10, RC/RE is set to 10. The collector load bias voltage is set at +15 volts to avoid collector saturation. The emitter load biased for 1 ma emitter current. (−1.7 volts).

[0029] Ib1 is the base current which has electron flow out of the base towards ground in the NPN transistor, while Ib2 is the base current which has electron flow into the base from ground in the PNP transistor. If the transistors have a Beta of 100 then the base currents in these examples will be 10 ua.

[0030] FIGS. 3A and 3G show common emitter amplifiers for both NPN and PNP transistors showing classic static (no input signal applied) NPN and PNP configurations, with RE at −1.7 and +1.7, respectively. FIG. 3C shows a classic static NPN common emitter amplifier with real world Shockley resistance {rs=0.026/Emitter current. Rs=0.026/1 ma=26 ohms}. RE (−1.7) Gain=~9.75. FIG. 3H shows a classic static PNP common emitter amplifier with real world Shockley resistance {rs=0.026/Emitter current. Rs=0.026/1 ma=26 ohms}. RE (+1.7) Gain=~9.75.

[0031] FIGS. 3E-3I show various common emitter amplifiers (PNP and NPN), illustrating operation of the amplifiers with an applied input signal. Referring now to FIG. 3E, when the input signal goes in a positive direction the current goes up. This also makes R_S go down, creating an increase in gain with varying positive signal. Gain varying with the input signal creates distortion in the collector load ‘RC’.

[0032] For FIG. 3F, when the input signal goes in a negative direction this makes the current go down. This also makes R_S go up, creating a decrease in gain with varying positive signal. Gain varying with the input signal creates distortion in the collector load ‘RC’.

[0033] Referring again to FIG. 3A, when an input signal applied to the circuit goes positive, this creates an increase in the collector current to the load to give gain, but also causes a decrease in R_S, distorting the collector current. When the input signal goes negative, this creates a decrease in the collector current to the load to give gain. But this also gives an increase in R_S, distorting the collector current.

[0034] Ideally, linearization is greatly improved when, as the input signal goes positive, the output collector current increases and R_S decreases. The present inventive method adds a parallel (to the input transistor) device so when the signal goes positive, the output collector current also increases but having R_S increase and not decrease. By summing these two currents, the resulting collector current of both devices gives an increase in current, but one has a decreasing R_S and the other has an increasing R_S of comparable amount, thereby canceling the R_S variation induced distortion.

[0035] Referring now to FIG. 3 J, the inventive method as applied to a classic common emitter amplifier is shown. This embodiment has the inputs of both the NPN and PNP transistors (QN and QP respectively) with their bases tied together, similar to combining the circuit of FIGS. 3E and 3F in a positive polarity output configuration. For a half cycle of the input signal, the emitter load REp and bias of the QP transistor is chosen for 1 ma. The QN emitter load which is also the QP collector load is chosen for 2 ma. This QP collector connection to the QN emitter is where the above summing of the signal currents occurs to provide the QN positive collector current output into the QN Collector load Rc (10 K). Since the QP current (1 ma) is decreasing and through the same 500 Ohm resistor (R500) the remainder of the R500s current increases with the QN Emitter current of 1 ma. This gives, in the collector current load of QN an increase in currents from both devices with a simultaneous current dependent Rs decrease which cancels this undesirable signal dependent varying load current, thus canceling distortion.

[0036] The base current of the QN flows into the base of QP. If the base currents are equal from matched NPN/PNP Betas then the input bias current can be effectively zero. Or at least only the difference of slightly different base currents. This makes the input impedance much higher than that of a single transistor with similar operating currents, another improvement provided by the inventive method. Another advantage of the inventive method is lower noise. The signal currents of QN and QP are magnified effectively by a factor of two. While the noise of two transistors in parallel goes up only by the square root of two. FIG. 3K shows the negative output signal version embodiment of the invention.

[0037] Referring now to FIG. 4, transistors Q1 and Q3 are set up as a classic differential amplifier of the type with two individual current sources utilizing a single emitter gain determining resistor. This example shows +15 volts to supply the 7.5 K collector loads. However, their emitter connection with the inventive distortion cancellation method is described as follows:

[0038] The left side of the circuit is set up in the common emitter configuration. Q1's base is used for the input signal. Q1's collector is the output, and Q1's emitter is connected to Q2's collector. Q2 is a constant current source however its collector base diode is run in saturation. The 2 resistors' values are for example only but one skilled in the art would select values, so that the saturated current source passes half the current out of Q2's base to ground and the other half out Q2's collector supplying Q1's emitter current. Selecting the static bias can be monitored, for example, by setting Q1's collector to static mid-point for symmetrical clipping, or some other voltage to accommodate the next stage of the circuit design. The significance of the forward biased collector base diode of Q2 in series with the forward biased BASE EMITTER diode of Q1 is the distortion cancellation characteristics from the interaction of the current dependent Rs of the devices.

[0039] In operation with an input signal applied, when the input signal goes positive, Q1's current will increase, and Rs will go down. Since more current is flowing through Q1 the available current will decrease through the forward biased collector base diode of Q2 causing its Rs to go up. This countervailing interaction cancels out the current and signal dependent nonlinearities of the amplifier element, and thus distortion.

[0040] When the input signal goes negative, Q1's current will decrease, and Rs will go up. Since less current is flowing through Q1 the available current will Increase through forward biased collector base diode of Q2 causing its Shockley resistance to go down. This countervailing interaction cancels out the distortion.

[0041] The same operation on the right side occurs for Q3 and Q4. This classic differential circuit topology can be implemented such that each side of the differential has its distortion reduced independently.

[0042] It is to be understood that the present invention is not limited to the sole embodiment described above, but encompasses any and all embodiments within the scope of the following claims: