sar adc design

**SAR ADC (Successive Approximation Register Analog-to-Digital Converter)** is the **most widely used ADC architecture that converts analog voltages to digital codes through a binary search algorithm** — offering the best combination of moderate speed (1-100 MSPS), medium-to-high resolution (8-18 bits), low power consumption, and compact area that makes it the default choice for SoC-embedded data conversion. **How SAR ADC Works** 1. **Sample**: Track-and-hold circuit captures the input voltage (Vin). 2. **Compare MSB**: Internal DAC set to Vref/2. Comparator checks: Is Vin > Vref/2? - Yes → MSB = 1, keep Vref/2. No → MSB = 0, remove Vref/2. 3. **Compare MSB-1**: DAC adds/subtracts Vref/4. Compare again. 4. **Repeat**: N comparisons for N-bit resolution. 5. **Output**: N-bit digital code after N clock cycles. **Key Components** | Component | Function | Critical Parameter | |-----------|----------|-----------------| | Capacitor DAC | Generates comparison voltages | Matching (< 0.1% for 10-bit) | | Comparator | Compares Vin vs DAC output | Offset, noise, speed | | SAR Logic | Binary search controller | Switching sequence | | Sample/Hold | Captures input voltage | Bandwidth, settling | **Capacitive DAC (CDAC)** - Binary-weighted capacitor array: C, C/2, C/4, ... C/2^N. - Charge redistribution: Switch capacitor plates between Vin, Vref, and GND. - **Advantage**: Capacitors in CMOS are more linear and match better than resistors. - **Bottom-plate sampling**: Reduces charge injection error. **SAR ADC Advantages** - **Low Power**: Only 1 comparator decision per bit per sample → minimal switching. - Power scales with: $P \propto C_{total} \times V_{ref}^2 \times f_s$. - State-of-art: < 10 fJ/conversion-step (Walden FOM). - **Compact Area**: No op-amps needed (unlike pipeline ADC). - **Scalable with CMOS**: Better performance at smaller nodes (smaller caps = less power). **SAR ADC vs. Other Architectures** | Architecture | Speed | Resolution | Power | Area | |-------------|-------|-----------|-------|------| | SAR | 1-100 MSPS | 8-18 bit | Very Low | Small | | Pipeline | 100 MSPS-1 GSPS | 8-14 bit | Medium | Large | | Flash | 1-10 GSPS | 4-8 bit | High | Very Large | | Sigma-Delta | < 10 MSPS | 16-24 bit | Low | Medium | **Advanced SAR Techniques** - **Time-Interleaved SAR**: Multiple SAR channels sampling at offset times → aggregate bandwidth multiplied. - **Noise-Shaping SAR**: Embed sigma-delta noise shaping in SAR loop → higher ENOB without oversampling penalty. - **Redundant Bit SAR**: Extra comparison bits relax comparator speed requirements. SAR ADC is **the workhorse data converter of the semiconductor industry** — its elegant binary search algorithm delivers the optimal power-resolution-speed tradeoff that has made it the most prevalent ADC architecture in modern SoCs, from IoT sensors to 5G transceivers.

Go deeper with CFSGPT

Get AI-powered deep-dives, save terms, and run advanced simulations — free account.

Create Free Account