Imaging system with shot-noise-matched and floating-point ramp analog-to-digital converters
An image sensor may include an array of image sensor pixels that are read out using analog-to-digital converters (ADCs). The ADC may be shot-noise-matched to reduce the number of decision cycles required. A ramp with limited resolution spanning only a small portion of the full scale voltage range may be used. For small analog input voltages, this limited ramp range is sufficient. For large analog input voltages, less resolution is needed due to the increasing shot noise in the photo signal. The larger input voltages may be successively divided by a selected attenuation factor until the analog input signal is within the range of the reduced ramp. The ADC keeps track of the number of divisions being performed to determine an exponent value for a floating-point output value and then convert the residual signal with the smaller ramp to determine a mantissa value for the floating-point output value.
1. An image sensor, comprising:
an image pixel;
a pixel output line that is coupled to the image pixel, wherein the image pixel is configured to output a signal onto the pixel output line; and
an analog-to-digital converter (ADC) circuit configured to receive the signal from the image pixel via the pixel output line, to divide the signal, and to use a voltage ramp with a ramp resolution that is less than the total resolution of the ADC circuit to determine the voltage of the divided signal.
2. The image sensor of claim 1 , wherein the ADC circuit is configured to divide the received signal by a factor of 2.
3. The image sensor of claim 1 , wherein the ADC circuit is configured to divide the received signal by a factor that is greater than 2 to provide margin for the voltage ramp.
4. The image sensor of claim 1 , wherein the ADC circuit is configured to divide the received signal by up to F times, and wherein F is an integer greater than one.
5. The image sensor of claim 4 , wherein the ADC circuit is configured to divide the received signal until the signal falls below a predetermined comparison level, and wherein the number of times the ADC circuit actually divides the received signal is equal to an exponent value E, and wherein E is less than or equal to integer F.
6. The image sensor of claim 5 , wherein the ADC circuit is configured to incrementally adjust the voltage ramp until the voltage ramp crosses the divided signal, and wherein the amount by which the voltage ramp is adjusted is equal to a mantissa value M.
7. The image sensor of claim 6 , wherein the ADC circuit is configured to compute a floating-point output based on the exponent value E and the mantissa value M.
8. The image sensor of claim 1 , wherein the ADC circuit has a full scale voltage range, and wherein the voltage ramp overlaps with only a subset of the full scale voltage range.
9. The image sensor of claim 1 , wherein the ADC circuit comprises a comparator having an offset, and wherein the voltage ramp is over-ranged to account for the offset of the comparator.
10. The image sensor of claim 1 , wherein the voltage ramp has a fixed step size that is independent of the voltage level of the signal received from the image pixel.
11. The image sensor of claim 1 , wherein the ADC circuit is a shot-noise-matched data converter.
12. A method of operating an image sensor having an image pixel and an analog-to-digital converter (ADC), the method comprising:
using the image pixel to output a pixel signal;
receiving the pixel signal at the ADC;
with the ADC, progressively attenuating the pixel signal to obtain a floating-point exponent value;
with the ADC, identifying the voltage level of the attenuated pixel signal by incrementally adjusting an analog ramp to obtain a floating-point mantissa value; and
computing a final output based on the floating-point exponent value and the floating-point mantissa value.
13. The method of claim 12 , wherein ADC has an overall resolution, and wherein the analog ramp has a reduced resolution that is less than the overall resolution of the ADC.
14. The method of claim 12 , wherein progressively attenuating the pixel signal comprises repeatedly dividing the pixel signal by a predetermined factor that is equal to or greater than two.
15. The method of claim 12 , wherein obtaining the floating-point exponent value takes a first amount of time, and wherein obtaining the floating-point mantissa value takes a second amount of time that is greater than the first amount of time.
16. The method of claim 12 , further comprising:
identifying a reset voltage level by incrementally adjusting the analog ramp; and
performing correlated double sampling based on the identified reset voltage level and the identified pixel signal level to remove column fixed-pattern noise.
17. An analog-to-digital converter (ADC) circuit having a total resolution, comprising:
a comparator;
a plurality of capacitors coupled to the comparator;
a first set of switches coupled to the plurality of capacitors; and
a second set of switches coupled to the plurality of capacitors, wherein the comparator has a input configured to receive an analog ramp having a resolution that is less than the total resolution of the ADC circuit, wherein the ADC circuit is further configured to receive a pixel signal, and wherein the first and second sets of switches are toggled in an alternating fashion to progressively attenuate the received pixel signal.
18. The analog-to-digital converter circuit of claim 17 , wherein the progressive attenuation scheme of the ADC circuit is a passive capacitive charge sharing division scheme.
19. The analog-to-digital converter circuit of claim 17 , wherein the ADC circuit has a non-flat quantization noise level that is shot-noise-matched, and wherein the analog ramp is capacitively coupled to the input of the comparator to provide a non-sampled ramp architecture.