 Jason
|
c6103b37de
|
Gap 7 MMCM jitter cleaner + CIC comb CREG pipeline + XDC clock-name fix
MMCM (Gap 7):
- Add adc_clk_mmcm.v: MMCME2_ADV wrapper (VCO=800MHz, CLKOUT0=400MHz)
- Modify ad9484_interface_400m.v: replace BUFG with MMCM path, gate reset on mmcm_locked
- Add adc_clk_mmcm.xdc: CDC false paths for clk_mmcm_out0 <-> clk_100m
XDC Fix (Build 19 WNS=-0.011 root cause):
- Remove conflicting create_generated_clock -name clk_400m_mmcm
- Replace all clk_400m_mmcm references with Vivado auto-generated clk_mmcm_out0
- CDC false paths now correctly apply to actual timing paths
CIC CREG Pipeline (Build 18 critical path fix):
- Explicit DSP48E1 for comb[0] with CREG=1/AREG=1/BREG=1/PREG=1
- Absorbs integrator_sampled_comb fabric FDRE into DSP48 C-port register
- Eliminates 0.643ns fabric->DSP routing delay (Build 18 tightest path)
- +1 cycle comb latency via data_valid_comb_0_out pipeline
- Move shared register declarations above ifndef SIMULATION (iverilog fix)
- Update golden data for +1 cycle CIC pipeline shift
Build scripts: build19_mmcm.tcl, build20_mmcm_creg.tcl
Regression: 18/18 FPGA pass, 20/20 MCU pass
Build 20 launched on remote Vivado (pending results)
|
2026-03-19 22:59:46 +02:00 |
|
|
Jason
|
1acedf494c
|
Migrate hardware platform from XC7A50T to XC7A200T-2FBG484I
Production FPGA: Artix-7 XC7A200T-2FBG484I (33,650 slices, 740 DSP48E1,
365 BRAM, -2 speed grade). Pin-mapped across 6 banks with proper VCCO
assignment (3.3V/2.5V/1.8V).
RTL timing primitives added for clean timing closure:
- ad9484_interface_400m.v: BUFIO for IDDR capture at 400MHz DDR,
BUFG for fabric logic, reset synchronizer (P1-7)
- dac_interface_single.v: ODDR for dac_clk forwarding + dac_data[7:0]
output registration, eliminates clock-forwarding insertion delay
- usb_data_interface.v: ODDR for ft601_clk_out forwarding, FSM runs
on ft601_clk_in domain with CDC synchronizers
Constraints:
- New production XDC (xc7a200t_fbg484.xdc): 182 pins, generated clocks
for ODDR outputs, BUFIO/DDR input delays, fixed false_path strategy
(from reset source, not to CLR pins), IOB packing on cells not ports
- Preserved upstream XDC as xc7a50t_ftg256.xdc for reference
- Updated cntrt.xdc with DRC fixes (I/O standards, missing constraints)
|
2026-03-16 22:24:22 +02:00 |
|
|
Jason
|
eefaf94e9e
|
Fix 6 RTL files + add xfft_32 stub for successful Vivado synthesis
Resolves all synthesis errors across attempts 3-11, achieving clean
Vivado 2025.2 synthesis on XC7A100T (0 errors, 831 LUTs, 320 FFs, 2 DSPs).
radar_receiver_final.v:
- reg clk_400m -> wire; output reg -> output wire (x4)
- Replace ad9484_lvds_to_cmos_400m with ad9484_interface_400m
- Remove duplicate IBUFDS lvds_to_cmos_400m instantiation
- Remove non-existent ref_i/ref_q port connections on matched filter
- Connect adc_dco_bufg as 400MHz clock source
ad9484_interface_400m.v:
- Add adc_dco_bufg output port with BUFG instance
- Route all internal logic through buffered DCO clock
cic_decimator_4x_enhanced.v:
- Move reset_monitors handling inside else branch (fixes Vivado
ambiguous clock error in both integrator and comb always blocks)
- Add separate comb_overflow_latched/comb_saturation_detected regs
to eliminate multi-driven nets between integrator and comb blocks
- Remove standalone always @(posedge reset_monitors) block
- Add output_counter to async reset branch
matched_filter_processing_chain.v:
- Wrap behavioral FFT body (uses $cos/$sin/$rtoi) in ifdef SIMULATION
- Add synthesis stub tying outputs to safe defaults
chirp_memory_loader_param.v:
- Replace hardcoded Windows paths with relative filenames for all
10 $readmem default parameters
latency_buffer_2159.v:
- Split single always block into separate BRAM write (synchronous
only) and control logic (with async reset) blocks
- Fixes Vivado Synth 8-3391: BRAM cannot infer with async reset
xfft_32.v (NEW):
- Synthesis stub for Xilinx 32-point FFT IP core
- AXI-Stream interface with pass-through and 1-cycle latency
- Placeholder until real xfft IP is generated
|
2026-03-15 17:37:59 +02:00 |
|
NawfalMotii79