Analysis and Design of VCO-Based Phase-Domain $\Sigma \Delta $ Modulators
read more
Citations
SNDR Limits of Oscillator-Based Sensor Readout Circuits
A Highly Digital 2210μm 2 Resistor-Based Temperature Sensor with a 1-Point Trimmed Inaccuracy of ± 1.3 ° C (3 σ) from -55 ° C to 125 ° C in 65nm CMOS
A 0.9-V 28-MHz Highly Digital CMOS Dual-RC Frequency Reference With ±200 ppm Inaccuracy From −40 °C to 85 °C
Linearity Analysis of Phase-Domain Delta-Sigma Modulator
Linearity Analysis of Phase-Domain Delta-Sigma Modulator
References
Understanding Delta-Sigma Data Converters
All-digital PLL and transmitter for mobile phones
Oversampled Sigma-Delta Modulation
A 12-Bit, 10-MHz Bandwidth, Continuous-Time $\Sigma\Delta$ ADC With a 5-Bit, 950-MS/s VCO-Based Quantizer
Related Papers (5)
Frequently Asked Questions (17)
Q2. How can a VCO-based PDM be used?
Quantization noise, counter wrap-around and settling time can add additional errors and design constraints in VCO-based PDΣΔMs with respect to fully analog PDΣΔMs.
Q3. What is the cost of a low-noise amplifier?
If the system allows it, the carrier frequency FIN can also be increased to avoid flicker noise, at the cost of additional quantization noise.
Q4. What is the main noise source of the PDM?
The VCO-based PDΣΔM has three major noise sources: ΣΔ quantization noise, up/down counter’s quantization error andVCO’s phase noise.
Q5. How do the authors make the analysis consistent with the model in Fig. (6)?
In order to make the analysis consistent with the model in Fig. (6), the authors first derive the effect of phase noise in terms of fractional counts and then refer the error back to phase.
Q6. How can the authors determine the KVCO value?
In practice, since KVCO can change dramatically due to process spread and temperature, the authors chose KVCO = 200 MHz/mV in the nominal case with worst-case of KVCO =160 MHz/mV.
Q7. What is the way to degrade the resolution of the temperature sensor?
In order not to degrade the sensor’s resolution, the authors need to derive the KVCO value that sufficiently suppresses the counter’squantization noise.
Q8. Why do low frequency idle tones not appear in the simulation shown in Fig. 8?
Although low-frequency idle tones are a typical issue for 1st-order modulator, they do not appear in the simulation shown in Fig. 8 because of the dithering action of the thermal noise superimposed on the input signal.
Q9. What is the amplitude of the N-th harmonic component?
Considering only second and third harmonic of FIN, the VCO output frequency can be expressed as:l = a/ cos 2π 4l + + a) cos 4π 4l + +2 + a cos 6π 4l + 3 + (24)where AN is the amplitude of the N-th harmonic component.
Q10. How many counter bits can be abridged?
with enough VCO gain and a sufficient number of counter bits, the performance gap between analog and VCO-based modulators can be abridged.
Q11. What is the cost of the improvement in SNR?
This improvement in SNR comes at a cost of larger area, higher VCO frequency, and requirement for higher VA and better VCO phase noise and linearity specs.
Q12. How much error does the phase range of the VCO cause?
When the phase range, i.e. the maximum − & , is changed to 11.25°, as shown in Fig. 14 (b), the error then reduces to less than 2 m°.
Q13. what is the minimum size of a non-wrapping counter?
From (2) and (3), assuming equal up and down periods, the minimum size of a non-wrapping counter ( b cd,efeghMij) is:b cd,efeghMij > l dt (17)Note that b cd,efeghMij in this case must be at least larger than VWX, which is large (~8 bits) for typical values (FNOM > 500 MHz, τUP > 100 ns).
Q14. What is the maximum count for a multi-bit PDM?
In this case, maximum value of | − & | ≤ Δ/2, and from (5), the authors find the maximum count (CMAX) to be:m = ± 2 sin 1/2 (21)To avoid wrap around, the following condition must hold:J > tuv) {2 sin 1/2 | + 1 (22)Note that (21) and (22) hold for a multi-bit PDΣΔM where | − & | ≤ Δ/2.
Q15. What is the trade-off between the counter size and quantization noise?
For low quantization noise, VIN and KVCO need to be high [from (13)], which means a larger counter is necessary to avoid wrap-around.
Q16. What is the difference between the phase noise of the input and the output?
For KVCO=140 MHz/V and VA = 0.5 mV, the phase-noiseinduced output noise will be below the quantization-noiseinduced output noise if the VCO phase noise is below -65 dBc/Hz for offset frequencies above 1.17 MHz, which is easily attainable by low-power VCO’s [20].
Q17. What is the difference between a wrapping counter and a DAC?
Since a wrapping counter can be of smaller length and does not need any additional logic for overload detection, it is simpler and hence occupies less silicon area.