Analog-to-digital converter comprising dual oscillators for linearity compensation

[0001] Analog-to-digital converters (ADC) have been implemented using a voltage controlled oscillator (VCO) wherein an input voltage is converted to an output frequency. The output frequency may then be converted to a digital value by driving a counter over a sampling interval. The counter output may then represent an average of the input voltage over the sampling interval. The accuracy of this type of ADC is affected by the center frequency (gain) of the VCO which may drift over changing environmental conditions (e.g., temperature), non-linearity of the VCO transfer function, and quantization error of the digital counter.

[0008] FIG. 1A shows an analog-to-digital converter (ADC) operable to convert a sensor signal 2 to a digital value. A differential amplifier 4 responsive to the sensor signal 2 and a reference signal 6 generates a first analog signal (8A or 8B) representing a first offset above the reference signal 6 and a second analog signal (8A or 8B) representing a second offset below the reference signal 6 (FIG. 1B). A first oscillator 10A generates a first output frequency 12A dependent on the first analog signal 8A, and a second oscillator 10B generates a second output frequency 12B dependent on the second analog signal 8B. A difference 14 between the first output frequency 12A and the second output frequency 12B is generated, and the digital value representing the sensor signal is generated in response to the difference 14.

[0009] FIG. 1C illustrates a non-linear response that each oscillator 10A and 10B may exhibit, including a non-linear response about a center operating point (represented by the reference signal 6). As the sensor signal 2 exceeds the reference signal 6, a digital value is generated in response to a difference Δf between the output frequencies f1 and f2 of the oscillators. Similarly, as the sensor signal 2 falls below the reference signal 6, a digital value is generated in response to a difference Δf between the output frequencies f1 and f2 of the oscillators. Accordingly, the difference Δf between the output frequencies f1 and f2 will be the same for a sensor signal that is above the reference frequency by an offset and below the reference frequency by the same offset, thereby compensating for non-linearity in the curve shown in FIG. 1C.

[0010] FIG. 2A shows an embodiment of the present invention wherein the first oscillator comprises a ring oscillator 16A and the second oscillator comprises a ring oscillator 16B. In one embodiment, the ring oscillators 16A and 16B are fabricated so as to substantially match one another so they have essentially the same response such as shown in FIG. 1C. The output frequency 12A of the first ring oscillator 16A drives a first counter 18A, and the output frequency 12B of the second ring oscillator 16B drives a second counter 18B, wherein the outputs 20A and 20B of the counters are subtracted to generate the difference 14 between the output frequencies. In one embodiment, the counters 18A and 18B may be reset at the beginning of a sample interval, and then the outputs evaluated at the end of the sample interval.

[0011] FIG. 2B shows an embodiment of a ring oscillator 16A or 16B comprising a plurality of inverter elements 22₀-22_N connected in series, wherein the output of the last inverting element 22_N is fed back to the input of the first inverting element 22₀ through a NAND gate 24 which is used to enable the ring oscillator in response to an enable signal 26. In one embodiment, there is an odd number of inversions in the ring oscillator which ensures the ring oscillator will oscillate. The analog voltage 8A or 8B operates as a supply voltage to the inverter elements 22₀-22_N and NAND gate 24, such that the output frequency of the ring oscillator depends on the amplitude of the analog voltage 8A or 8B which determines the propagation delay of each element. That is, as the amplitude of the analog voltage 8A or 8B increases, the propagation delay of the elements will decrease thereby causing an increase in the output frequency, and vise versa. In one embodiment, the ring oscillators 16A and 16B are enabled synchronously by the enable signal 26 after resetting the counters 18A and 18B at the beginning of a sample interval.

[0012] FIG. 3 shows an embodiment of the present invention wherein the outputs 20A and 20B are summed, and the sum 28 processed by control circuitry 30 in order to adjust at least one of the reference signal 6 and a control signal 32 applied to the sensor 34. In one embodiment, the control circuitry 30 adjusts the reference signal 6 so that it remains substantially centered about the response curve of the oscillators (e.g., as shown in FIG. 1C). That is, the reference signal 6 may be adjusted until the sum 28 substantially matches a target value. In another embodiment, the control circuitry 30 adjusts the control signal 32 in order to adjust a gain of the sensor 34 (e.g., by adjusting a gain of an op-amp). For example, during a calibration phase the sensor 34 may be excited with a known signal and the gain adjusted until the sum 28 substantially matches a target value corresponding to the known signal.

[0013] Any suitable sensor 34 may be employed in the embodiments of the present invention, such as an optical sensor, piezoelectric sensor, electromagnetic sensor, temperature sensor, pressure sensor, vibration sensor, etc. In one embodiment, the sensor 34 may be employed in a storage device, such as a disk drive that may utilize one or more of the aforementioned sensors.