Differential Source Follower With Body Effect Compensation

1. Technical Field of the Invention

This invention generally pertains to analog signal processing circuits. In particular, this invention is directed to an improved input buffer circuit providing improved performance over a wider range of input signal levels.

2. Background Art

Analog signal processing circuits often require unity gain buffers at their inputs to minimize loading on the circuit which drives them. A method of implementing such a buffer is to use a single device unity-gain amplifier such as a source follower or common drain transistor. In such a configuration, the source is typically biased with a DC current of 10 to 100 .mu.A, the input signal is applied to the gate, and the output appears at the source, shifted by a voltage equal to the V.sub.GS of the device at the given bias level.

A problem occurs with this circuit because the shift voltage V.sub.GS is a function of the threshold voltage of the device. Since the device threshold is a function of its source-body voltage (i.e., is affected by the input signal), and the source voltage is the output signal, the level shift from input to output becomes a function of the input signal. Thus, this problem is inherent in the FET source-body structure and causes an attenuation of the output signal because V.sub.GS is proportional to the input signal, as shown in FIG. 1B (not drawn to scale). Ideally, the biasing means should compensate for the source-body voltage fluctuation and maintain V.sub.GS at a constant (DC) level without the variation seen in FIG. 1B.

Many analog circuits that use such conventional followers are differential circuits implemented to reject common-mode noise. In such a case the effect described above is doubled because the thresholds change in opposite directions. A typical differential source follower is illustrated in FIG. 1 and its transfer function shown in FIG. 2. Note the difference between the input signal voltage and the output signal voltage for signal magnitudes greater than about .+-.200 mV, which mismatch by about 5% in the .+-.500 mV range. For circuits that process the absolute amplitude of the input signal, such as peak detectors, this margin of error can be a significant drawback.

It is an object of the invention to provide a differential source follower with no body effect induced errors.

It is another object of the invention to provide an analog input buffer providing accurate performance over a wide range of input signal levels.

It is yet another object of the invention to provide a differential source follower with a low power mode.

As mentioned under Background Art, when an input signal is applied to the input gate of the differential source follower, the full input signal does not appear at the output because the voltage shift from gate to source is itself influenced by the input signal voltage, due to the inherent body effect described above. This results in an attenuation of the input signal from the gate to the source, and appears as a weakened output signal. However, it is essential to note that the voltage shift from gate to source (V.sub.GS) is a function of the drain current (I.sub.1 of FIG. 1A), heretofore assumed to be maintained constant by the current source.

Therefore, one preferred embodiment of the present invention comprises a method of modulating the drain current to cause a change in the gate to source voltage that exactly counteracts the source voltage swing caused by the body effect, thus, reducing or eliminating the overall attenuation.

The present invention achieves this goal by implementing a differential source follower biased by a pair of cross-coupled transistors of the same type as the followers. The sizes of the cross-coupled pair are selected to precisely cancel out the body effect in the followers. The circuit level-shifts both phases of the differential signal while not affecting the differential voltage.

Therefore, a second preferred embodiment of the present invention comprises an apparatus coupled to a voltage supply having inputs for receiving differential input signals, and providing unity gain output signals, in response to the differential input signals, having a much improved input-to-output mismatch. The improved unity gain is provided by a pair of FETs and current sources coupled to the outputs which counteract the inherent voltage fluctuation in the output FETs which attenuate the output signals. One of the outputs provides a modulation signal for an FET-current source pair controlling a bias current through the other output.

A third preferred embodiment comprises a differential source follower that counteracts the attenuation of its output signals and which also includes a low power control circuit for turning off bias currents flowing through the source followers in order to reduce power consumption.

Other features and advantages of this invention will become apparent from the following detailed description of the presently preferred embodiment of the invention, taken in conjunction with the accompanying drawings.

FIG. 1A illustrates a conventional differential source follower.

FIG. 1B (not drawn to scale) illustrates a typical V.sub.GS fluctuation in conventional differential source followers.

FIG. 2 illustrates the transfer function for the differential source follower of FIG. 1.

FIG. 3 illustrates the differential source follower of the present invention.

FIG. 4 illustrates the transfer function for the inventive differential source follower of FIG. 3 with a mismatch of essentially zero over the input signal level range.

FIG. 5 illustrates the inventive differential source follower with power disable feature.

A circuit configuration that achieves the objects of the present invention is illustrated in FIG. 3. This circuit includes a pair of FET's 37, 38 with their drains connected to the outputs of the source followers 31, 32, their gates cross-coupled to their drains, and their sources grounded (instead of ground a negative voltage supply would also work well). Current sources 35, 36 are connected to the followers 31, 32 in parallel with the FET's 37, 38, respectively, while the source followers 31, 32 receive differential inputs IN+ and IN- and are coupled to a supply terminal 39. Precision differential output signals are provided at the sources 33, 34 of the followers 31, 32.

This approach takes advantage of the fact that in a differential source follower, the signal at the opposing input changes in a way exactly opposite to that at the input under consideration. Hence, as the signal appearing at the source of one of the followers acts to increase that follower's V.sub.GS by increasing its threshold (or vice versa), the cross-coupled FET with its gate connected to the opposing output acts to decrease the follower's V.sub.GS by reducing its current (or increasing its current to counteract an excessive decrease in V.sub.GS). By proper tuning of the sizes of the cross-coupled devices 37, 38, the attenuation in the source followers 31, 32 can be eliminated. The selection of the cross-coupled devices must be precisely fitted to the overall circuit to achieve this desired performance improvement, i.e., an accurate size of devices is necessary to exactly cancel out the body effect of a given circuit. A general range of sizes for these devices likely will not achieve the precise fit necessary. One method for selecting properly sized FETs includes iterative simulation using computer models with varying FET sizes until an optimum transfer function is realized. The transfer function of FIG. 4 illustrates the improved performance of a circuit with properly selected devices. The mismatch error is negligible (essentially zero) over the entire signal range which involves a magnitude of about .+-.600 mV.

Note that a mathematical solution to the sizing of the cross-coupled devices is extremely difficult to develop. It first requires an accurate model of the relationship between the threshold voltage and the source voltage. This appears to be largely a square-law relationship, which complicates the computation tremendously. Setting the sizes precisely and empirically using the device models seems to work quite well as evidenced by the hardware built taking advantage of this technique, and the results shown in FIG. 4.

Referring to FIG. 5, a low (zero) power disable state was added to the circuit design of FIG. 3. In the power disable state, the DC bias currents can be turned off and the paths between the followers 51, 52 and the bias/cross coupled pair 59, 60 opened (or closed) by switching the newly added transistors 57, 58 via control input 63, and the outputs of the bias/cross-coupled pair 59, 60 are pulled to ground by switching the newly added transistors 61, 62 via control input 64. This circuit has the same functional characteristics of the circuit in FIG. 3 but does not draw any DC current when selectively disabled.

The matter contained in the above description or shown in the accompanying drawings have been described for purposes of illustration and shall not be interpreted in a limiting sense. It will be appreciated that various modifications may be made in the above structure and method without departing from the scope of the invention described herein. Thus, changes and alternatives will now become apparent to those skilled in the art without departing from the spirit and scope of the invention as set forth in the following claims. Accordingly, the scope of protection of this invention is limited only by the following claims and their equivalents.