SPI vs I2C: Understanding the Two Most Popular Serial Protocols
SPI (Serial Peripheral Interface) and I2C (Inter-Integrated Circuit) are the two most widely used serial communication protocols in embedded systems. Choosing between them is a fundamental decision that affects system performance, complexity, and cost. This comprehensive guide provides the technical depth needed to make informed design decisions.
Quick Comparison Summary
| SPI | Faster (up to 100+ MHz), more pins, point-to-point optimized |
| I2C | Slower (up to 3.4 MHz), only 2 wires, true multi-master bus |
SPI Protocol Deep Dive
Signal Description
SPI uses four primary signals for full-duplex communication:
- SCLK (Serial Clock): Clock signal generated by the master
- MOSI (Master Out Slave In): Data from master to slave
- MISO (Master In Slave Out): Data from slave to master
- SS/CS (Slave Select/Chip Select): Active-low signal to select slave device
SPI Clock Modes
SPI defines four clock modes based on Clock Polarity (CPOL) and Clock Phase (CPHA):
| Mode | CPOL | CPHA | Clock Idle | Data Sampled | Data Shifted |
|---|---|---|---|---|---|
| Mode 0 | 0 | 0 | Low | Rising Edge | Falling Edge |
| Mode 1 | 0 | 1 | Low | Falling Edge | Rising Edge |
| Mode 2 | 1 | 0 | High | Falling Edge | Rising Edge |
| Mode 3 | 1 | 1 | High | Rising Edge | Falling Edge |
SPI Advantages
- High Speed: Commonly 10-50 MHz, some devices support 100+ MHz
- Full Duplex: Simultaneous bidirectional data transfer
- Simple Protocol: No addressing overhead, no arbitration needed
- Flexible Data Width: Not limited to 8-bit transfers
- No Pull-ups Required: Push-pull outputs provide faster edges
SPI Disadvantages
- Pin Count: Requires 3 + N pins (N = number of slaves)
- No Acknowledgment: No built-in error detection
- Short Distance: Limited by signal integrity at high speeds
- Single Master: No standard multi-master support
I2C Protocol Deep Dive
Signal Description
I2C uses only two bidirectional signals:
- SDA (Serial Data): Bidirectional data line
- SCL (Serial Clock): Clock line (typically driven by master)
I2C Speed Modes
| Mode | Maximum Speed | Typical Use Case |
|---|---|---|
| Standard Mode | 100 kbps | Basic sensors, EEPROMs |
| Fast Mode | 400 kbps | Most modern I2C devices |
| Fast Mode Plus (Fm+) | 1 Mbps | High-performance sensors |
| High Speed Mode | 3.4 Mbps | Specialized applications |
I2C Addressing
I2C supports two addressing modes:
- 7-bit Addressing: Up to 128 devices (112 usable, 16 reserved)
- 10-bit Addressing: Up to 1024 devices
I2C Protocol Features
- START Condition: SDA falls while SCL is high
- STOP Condition: SDA rises while SCL is high
- ACK/NACK: Receiver pulls SDA low (ACK) or leaves high (NACK)
- Clock Stretching: Slave can hold SCL low to slow communication
- Arbitration: Multiple masters can coexist with collision detection
I2C Advantages
- Minimal Wiring: Only 2 wires for any number of devices
- Addressing: Each device has unique address, no chip select needed
- Multi-Master: Built-in arbitration for multiple masters
- Acknowledgment: Built-in ACK/NACK for error detection
- Hot-Plugging: Devices can be added/removed dynamically
I2C Disadvantages
- Speed Limited: Maximum 3.4 Mbps (typically 400 kbps used)
- Half Duplex: Cannot send and receive simultaneously
- Pull-up Resistors: Required, affect rise time and power
- Addressing Conflicts: Fixed addresses can cause conflicts
- Bus Capacitance: Limited to 400 pF total bus capacitance
Detailed Technical Comparison
| Parameter | SPI | I2C |
|---|---|---|
| Maximum Speed | 100+ MHz | 3.4 MHz |
| Minimum Wires | 4 (3 + 1 per slave) | 2 |
| Duplex | Full Duplex | Half Duplex |
| Addressing | Chip Select lines | 7-bit or 10-bit address |
| Multi-Master | Not standard | Built-in support |
| Acknowledgment | None | ACK/NACK after each byte |
| Clock | Master only | Master (with stretching) |
| External Components | None required | Pull-up resistors |
| Power Consumption | Higher (push-pull) | Lower (open-drain) |
| Protocol Overhead | None | Start, Address, ACK, Stop |
When to Use SPI vs I2C
Choose SPI When:
- Speed is Critical: Applications requiring >1 Mbps throughput
- Streaming Data: ADCs, DACs, displays, flash memory
- Full Duplex Needed: Simultaneous read/write operations
- Few Devices: Small number of peripherals (1-4)
- No Multi-Master: Single controller systems
Choose I2C When:
- Pin Count Matters: Limited GPIO availability
- Many Devices: Connecting numerous sensors/peripherals
- Multi-Master Required: Multiple controllers on same bus
- Hot-Plugging Needed: Dynamic device connection
- Low Speed Acceptable: Configuration, sensor reading
Implementation Best Practices
SPI Implementation Tips
- Match Clock Modes: Verify CPOL/CPHA settings match device requirements
- CS Timing: Ensure proper setup/hold times for chip select
- Signal Integrity: Keep traces short at high frequencies, use ground planes
- MSB/LSB First: Confirm bit order matches device expectations
- FIFO Depth: Use FIFOs to reduce CPU overhead in IP cores
I2C Implementation Tips
- Pull-up Selection: Calculate based on bus capacitance and speed mode
- Address Verification: Check for conflicts before system integration
- Clock Stretching Timeout: Implement timeout to prevent bus lockup
- Bus Recovery: Implement stuck bus recovery (9 clock pulses)
- Glitch Filtering: Use filtering on SDA/SCL for noise immunity
I2C Pull-up Resistor Calculation
Minimum: Rmin = (VCC - VOL) / IOL
Maximum: Rmax = tr / (0.8473 x Cb)
Where: VOL = 0.4V, IOL = 3mA (standard mode), tr = rise time, Cb = bus capacitance
Conclusion
Both SPI and I2C have their place in embedded system design. SPI excels in high-speed, streaming applications where throughput is paramount, while I2C shines in multi-device, pin-constrained systems requiring flexibility and simplicity.
Many modern designs use both protocols: SPI for high-bandwidth peripherals like displays and flash memory, and I2C for low-speed sensors and configuration interfaces. Understanding the strengths and limitations of each protocol enables optimal system architecture decisions.
Vcores offers silicon-proven SPI and I2C IP cores with configurable features, comprehensive verification, and easy integration for your FPGA and ASIC designs.
