Communication between a CPLD (Complex Programmable Logic Device) and an FPGA (Field-Programmable Gate Array) depends on the required speed, complexity, and available I/O resources. Below are common methods, along with their advantages and trade-offs.

1. Parallel Bus (Fast & Simple)
Method:
Use multiple I/O pins for data transfer (e.g., 8-bit, 16-bit, or 32-bit parallel bus).

Add control signals like:

• CLK (synchronization clock)
• WR (write enable)
• RD (read enable)
• CS (chip select)
• ACK (acknowledge handshake, if needed).

Pros:
✅ High-speed data transfer (good for burst transfers).
✅ Simple to implement (no complex protocols).

Cons:
❌ Requires many I/O pins (not ideal for pin-limited designs).
❌ Susceptible to noise (if long traces are used).

Example Connection:

FPGA (Master)          CPLD (Slave)
-------------------------
DATA[7:0]  <------->  DATA[7:0]
ADDR[3:0]  -------->  ADDR[3:0]
WR         -------->  WR
RD         -------->  RD
CS         -------->  CS
CLK        -------->  CLK

2. Serial Communication (Saves Pins)
Methods:
A. SPI (Serial Peripheral Interface)

• 4-wire protocol (SCLK, MOSI, MISO, SS).
• Supports full-duplex communication.
• Typical speeds: 1–100 MHz.

B. I²C (Inter-Integrated Circuit)

• 2-wire protocol (SCL, SDA).
• Supports multi-master mode.
• Slower (~400 kHz – 3.4 MHz).

C. UART (Asynchronous Serial)

• 2-wire (TX, RX) or 1-wire (half-duplex).
• No clock needed (baud rate must match).
• Simple but slower (typical speeds: 9600–115200 bps).

Pros:
✅ Fewer pins required (good for small CPLDs).
✅ Standardized protocols (easy to interface with MCUs).

Cons:
❌ Slower than parallel bus.
❌ SPI/I²C require clock synchronization.

Example (SPI Connection):

FPGA (Master)          CPLD (Slave)
-------------------------
SCLK       -------->  SCLK
MOSI       -------->  SDI
MISO       <-------  SDO
SS         -------->  CS

3. Dual-Port RAM / FIFO (High-Speed Buffering)
Method:

• Use a shared memory block (if FPGA has embedded RAM).
• FPGA writes data, CPLD reads (or vice versa).
• Handshake signals (FULL, EMPTY, WR_EN, RD_EN) synchronize access.

Pros:
✅ High throughput (good for streaming data).
✅ Reduces real-time synchronization issues.

Cons:
❌ Requires RAM resources (may not be available in small CPLDs).

Example (FPGA-to-CPLD FIFO):

verilog
// FPGA writes data
always @(posedge clk) begin
    if (wr_en && !fifo_full) begin
        fifo_data <= data_to_send;
    end
end

// CPLD reads data
always @(posedge clk) begin
    if (rd_en && !fifo_empty) begin
        received_data <= fifo_data;
    end
end

4. LVDS / Differential Signaling (Noise Immunity)
Method:

• Use LVDS (Low-Voltage Differential Signaling) for high-speed serial links.
• Example: Xilinx SelectIO, Intel LVDS.
• Requires serializer/deserializer (SerDes) if using high speeds (>1 Gbps).

Pros:
✅ High-speed & noise-resistant (good for long PCB traces).
✅ Fewer pins than parallel bus.

Cons:
❌ More complex to implement.
❌ CPLD may not support LVDS (check datasheet).

Example (LVDS Connection):

FPGA (TX)          CPLD (RX)
-------------------------
TX_P       -------->  RX_P
TX_N       -------->  RX_N

5. Custom Protocol (Flexible but Complex)
Method:
Define a custom handshake protocol (e.g., source-synchronous).

Example:

• FPGA sends data + strobe signal.
• CPLD latches data on strobe edge.

Pros:
✅ Optimized for specific needs.

Cons:
❌ Requires careful timing analysis.

Example (Custom Strobe-Based Transfer):

verilog
// FPGA sends data
always @(posedge clk) begin
    if (send_data) begin
        data_bus <= data;
        strobe <= ~strobe; // Toggle strobe
    end
end

// CPLD receives data
always @(posedge strobe) begin
    received_data <= data_bus;
end

Comparison of Methods

How to Choose?

1. Need high speed? → Parallel bus or LVDS.
2. Limited pins? → SPI, I²C, or UART.
3. Buffered data transfer? → Dual-port RAM/FIFO.
4. Noise immunity needed? → LVDS.