Compare commits

..

7 Commits

Author SHA1 Message Date
Jason e979363730 fix(mcu): volatile emergency state + AGC holdoff zero-guard (closes #83)
Bug 1 (main.cpp:630): system_emergency_state lacked volatile. Under -O1+
the compiler is permitted to hoist the read outside the blink loop, making
while (system_emergency_state) unconditionally infinite. Once entered, the
only escape was the 4 s IWDG timeout — which resets the MCU and
re-energizes the PA rails that Emergency_Stop() explicitly cut. Marking the
variable volatile forces a memory read on every iteration so an external
clear (ISR, USB command, manual reset) can break the loop correctly.

Bug 2 (ADAR1000_AGC.cpp:59): holdoff_frames is a public uint8_t; if a
caller sets it to 0, the condition holdoff_counter >= holdoff_frames is
always true (any uint8_t >= 0), causing the AGC outer loop to increase gain
on every non-saturated frame with no holdoff delay. With alternating
sat/no-sat frames this produces a ±step oscillation that prevents the
receiver from settling. Fix: clamp holdoff_frames to a minimum of 1 in the
constructor, preserving all existing test assertions (none use 0; default
remains 4).

Reported-by: shaun0927 (Junghwan) <https://github.com/shaun0927>
2026-04-21 03:35:48 +05:45
Jason 3366ac6417 fix(fpga): widen AGC gain arithmetic to 6-bit to prevent wraparound
5-bit signed subtraction in clamp_gain wrapped for agc_attack >= 10 or
agc_decay >= 9 when |agc_gain| + step > 16, inverting gain polarity
instead of clamping — e.g. gain=-7, attack=10 produced +7 (max amplify)
rather than -7 (max attenuate), causing ADC saturation on strong returns.

Widen clamp_gain input to [5:0] and sign-extend both operands to 6 bits
({agc_gain[3],agc_gain[3],agc_gain} and {2'b00,agc_attack/decay}),
covering the full [-22,+22] range before clamping. Default attack/decay
values (1-4) are unaffected; behaviour changes only for values >= 10/9.
2026-04-21 03:06:32 +05:45
Jason db80baf34d Merge remote-tracking branch 'origin/main' into develop 2026-04-21 01:33:27 +05:45
NawfalMotii79 33d21da7f2 Remove radar system image from README
Removed the AERIS-10 Radar System image from the README.
2026-04-20 19:04:08 +01:00
NawfalMotii79 18901be04a Fix image link and update mixer model in README
Updated image link and corrected mixer model in specifications.
2026-04-19 19:06:44 +01:00
NawfalMotii79 9f899b96e9 Add files via upload 2026-04-19 19:04:48 +01:00
NawfalMotii79 88ca1910ec Merge pull request #109 from NawfalMotii79/develop
Release: merge develop into main
2026-04-19 01:27:15 +01:00
7 changed files with 44 additions and 186 deletions
Binary file not shown.

After

Width:  |  Height:  |  Size: 378 KiB

@@ -24,6 +24,7 @@ ADAR1000_AGC::ADAR1000_AGC()
, saturation_event_count(0)
{
memset(cal_offset, 0, sizeof(cal_offset));
if (holdoff_frames == 0) holdoff_frames = 1;
}
// ---------------------------------------------------------------------------
@@ -627,7 +627,7 @@ typedef enum {
static SystemError_t last_error = ERROR_NONE;
static uint32_t error_count = 0;
static bool system_emergency_state = false;
static volatile bool system_emergency_state = false;
// Error handler function
SystemError_t checkSystemHealth(void) {
@@ -18,23 +18,8 @@
# Bank 35: VCCO = 3.3V (FT2232H USB 2.0 FIFO — 15 signals)
#
# DRC Fix History:
# - PLIO-9 (REVERTED): Previously moved clk_120m_dac from C13 (N-type) to
# D13 (P-type MRCC) to satisfy the MRCC preference. However, a schematic
# audit (KiCad netlist export from the Eagle schematic, U42 pad->net map)
# revealed that D13 is UNCONNECTED on the physical PCB. The real
# /FPGA_DAC_CLOCK net from AD9523 OUT11 lands on C13 (IO_L11N_T1_SRCC_15,
# N-type). Moved back to C13 and added CLOCK_DEDICATED_ROUTE FALSE,
# matching the ft_clkout treatment on C4 (N-type MRCC).
# - Schematic audit added pin constraints for previously-unconstrained
# signals connected to the FPGA in hardware: ADC_OR_P/N (M6/N6, AD9484
# overflow flag), /FPGA_ADC_CLOCK_P/N (N11/N12, 400 MHz observation tap
# of the AD9523->AD9484 sample clock). Added to 50T wrapper as
# anchored-but-unused inputs to secure pin assignment and prevent
# accidental future contention; full RTL consumers are a follow-up.
# - PLIO-9 (original, historical): FT2232H CLKOUT routed to C4
# (IO_L12N_T1_MRCC_35, N-type). Clock inputs normally use P-type MRCC
# pins, but IBUFG works correctly on N-type. Demote PLIO-9 to warning
# in build script.
# - PLIO-9: Moved clk_120m_dac from C13 (N-type) to D13 (P-type MRCC).
# Clock inputs must use the P-type pin of a Multi-Region Clock-Capable pair.
# - BIVC-1 / Place 30-372: Bank 14 must have a single VCCO. LVDS_25 forces
# VCCO=2.5V, so adc_pwdn was changed from LVCMOS33 to LVCMOS25 to match.
# IBUFDS input buffers are VCCO-independent. BIVC-1 also waived via
@@ -43,6 +28,9 @@
# - UCIO/NSTD: Unconstrained ports (FT601 ports inactive with USB_MODE=1,
# status/debug outputs have no physical pins). Handled with SEVERITY
# demotion + default IOSTANDARD.
# - PLIO-9: FT2232H CLKOUT routed to C4 (IO_L12N_T1_MRCC_35, N-type).
# Clock inputs normally use P-type MRCC pins, but IBUFG works correctly
# on N-type. Demote PLIO-9 to warning in build script.
# ============================================================================
# ============================================================================
@@ -78,7 +66,7 @@ set_property IOSTANDARD LVCMOS33 [get_ports {clk_100m}]
create_clock -name clk_100m -period 10.0 [get_ports {clk_100m}]
set_input_jitter [get_clocks clk_100m] 0.1
# 120MHz DAC Clock (AD9523 OUT11 → /FPGA_DAC_CLOCK → Bank 15 pin C13)
# 120MHz DAC Clock (AD9523 OUT11 → FPGA_DAC_CLOCK → Bank 15 MRCC pin D13)
# NOTE: The physical DAC (U3, AD9708) receives its clock directly from the
# AD9523 via a separate net (DAC_CLOCK), NOT from the FPGA. The FPGA
# uses this clock input for internal DAC data timing only. The RTL port
@@ -86,19 +74,12 @@ set_input_jitter [get_clocks clk_100m] 0.1
# physical pin on the 50T board and is left unconnected here. The port
# CANNOT be removed from the RTL because the 200T board uses it with
# ODDR clock forwarding (pin H17, see xc7a200t_fbg484.xdc).
#
# PIN: C13 is IO_L11N_T1_SRCC_15 (N-type SRCC). A prior commit attempted to
# move this to D13 (MRCC P-type) to satisfy PLIO-9, but the schematic audit
# showed D13 is UNCONNECTED on the PCB — the /FPGA_DAC_CLOCK net physically
# lands on C13. Moving to D13 made the DAC clock input float. Restored to
# C13 and forced CLOCK_DEDICATED_ROUTE FALSE (same mechanism as ft_clkout on
# C4), which routes the IBUFG output through general fabric to a BUFG.
set_property PACKAGE_PIN C13 [get_ports {clk_120m_dac}]
# FIX: Moved from C13 (IO_L12N = N-type) to D13 (IO_L12P = P-type MRCC).
# Clock inputs must use the P-type pin of an MRCC pair (PLIO-9 DRC).
set_property PACKAGE_PIN D13 [get_ports {clk_120m_dac}]
set_property IOSTANDARD LVCMOS33 [get_ports {clk_120m_dac}]
create_clock -name clk_120m_dac -period 8.333 [get_ports {clk_120m_dac}]
set_input_jitter [get_clocks clk_120m_dac] 0.1
# C13 is N-type SRCC (not dedicated-clock-capable); override the DRC check.
set_property CLOCK_DEDICATED_ROUTE FALSE [get_nets {clk_120m_dac_IBUF}]
# ADC DCO Clock (400MHz LVDS — AD9523 OUT5 → AD9484 → FPGA, Bank 14 MRCC)
# NOTE: LVDS_25 is the only valid differential input standard on 7-series HR
@@ -302,45 +283,6 @@ set_input_delay -clock [get_clocks adc_dco_p] -min 0.2 [get_ports {adc_d_p[*]}]
set_input_delay -clock [get_clocks adc_dco_p] -max 1.0 -clock_fall [get_ports {adc_d_p[*]}] -add_delay
set_input_delay -clock [get_clocks adc_dco_p] -min 0.2 -clock_fall [get_ports {adc_d_p[*]}] -add_delay
# --------------------------------------------------------------------------
# AD9484 Overflow / Out-Of-Range flag (schematic nets ADC_OR_P / ADC_OR_N)
# --------------------------------------------------------------------------
# AD9484 differential OR output on FPGA pads M6 (OR_P) / N6 (OR_N), Bank 14.
# This is the AD9484's full-scale overflow indicator, useful for AGC /
# gain-ranging feedback. The 50T RTL wrapper anchors this with an IBUFDS
# (DONT_TOUCH) so the pads cannot be accidentally driven as outputs (which
# would cause contention with the AD9484 driver). A future PR should wire
# the buffered signal into the receive-path status flags.
set_property PACKAGE_PIN M6 [get_ports {adc_or_p}]
set_property PACKAGE_PIN N6 [get_ports {adc_or_n}]
set_property IOSTANDARD LVDS_25 [get_ports {adc_or_p}]
set_property IOSTANDARD LVDS_25 [get_ports {adc_or_n}]
set_property DIFF_TERM TRUE [get_ports {adc_or_p}]
# --------------------------------------------------------------------------
# FPGA observation of AD9523->AD9484 sample clock (/FPGA_ADC_CLOCK_P/N)
# --------------------------------------------------------------------------
# AD9523 drives the AD9484 sample clock directly; the same differential
# pair is tapped to FPGA pads N11 (P) / N12 (N), Bank 14, MRCC-capable.
# This is an INPUT-ONLY tap (FPGA must never drive these pads — that would
# contend with the AD9523 driver feeding the ADC). The 50T wrapper anchors
# with IBUFDS + DONT_TOUCH so the pad assignment is preserved across all
# synthesis/optimization stages. The buffered net is unconsumed for now;
# create_clock and clock_groups are deferred until an RTL consumer exists
# (see commented template below).
set_property PACKAGE_PIN N11 [get_ports {fpga_adc_clock_p}]
set_property PACKAGE_PIN N12 [get_ports {fpga_adc_clock_n}]
set_property IOSTANDARD LVDS_25 [get_ports {fpga_adc_clock_p}]
set_property IOSTANDARD LVDS_25 [get_ports {fpga_adc_clock_n}]
set_property DIFF_TERM TRUE [get_ports {fpga_adc_clock_p}]
# No create_clock here on purpose: the IBUFDS output is unconsumed (anchored
# via DONT_TOUCH only), so declaring it as a clock would only generate
# "clock has no registered destinations" warnings. When a follow-up PR adds
# an actual consumer, add:
# create_clock -name fpga_adc_clock -period 2.5 [get_ports {fpga_adc_clock_p}]
# set_input_jitter [get_clocks fpga_adc_clock] 0.05
# set_clock_groups -asynchronous -group [get_clocks fpga_adc_clock] ...
# ============================================================================
# FT2232H USB 2.0 INTERFACE (Bank 35, VCCO=3.3V)
# ============================================================================
@@ -405,49 +347,29 @@ set_property DRIVE 8 [get_ports {ft_data[*]}]
# FPGA Write Path (FPGA drives data, FT2232H samples):
# - Data setup before next CLKOUT rising: t_su = 5.0 ns
# - Data hold after CLKOUT rising: t_hd = 0.0 ns
# - Board trace skew budget: ~0.5 ns
# - Output delay max = t_su + trace_max = 5.0 + 0.5 = 5.5 ns
# - Output delay min = t_hd - trace_min = 0.0 - 0.0 = 0.0 ns
#
# NOTE: Historical XDC used 'period - t_su = 11.667 ns' for output_delay -max,
# which is the wrong interpretation: set_output_delay takes the external setup
# requirement (+trace), not the remaining timing budget. The old value forced
# Vivado to close a path assuming FT2232H requires 11.667 ns of setup, which
# it does not, and caused WNS=-5.350 ns failures on ft_data/ft_rd_n/ft_wr_n/
# ft_oe_n/ft_siwu paths given the 5.513 ns clock insertion delay on the
# non-dedicated C4 routing.
# - Output delay max = period - t_su = 16.667 - 5.0 = 11.667 ns
# - Output delay min = t_hd = 0.0 ns
# --------------------------------------------------------------------------
# Input delays: FT2232H → FPGA (data bus and status signals)
#
# -min revision (Build N+1): was 0.0 ns, now 1.0 ns.
# Rationale: set_input_delay -min is the EARLIEST time data can change at the
# FPGA pin after the launch clock edge, i.e. FT2232H Tco_min + trace_min.
# Setting -min 0.0 claimed data could change simultaneously with the clock
# edge, which is pessimistically tight for hold analysis and caused a
# -0.079 ns hold violation on ft_rxf_n → FSM_sequential_wr_state in Build N
# (due to 2.895 ns clock insertion delay on non-dedicated C4 routing).
# FT2232H Sync FIFO Tco is spec'd 14 ns; using 1.0 ns is conservative and
# still covers worst-case silicon. Invariant preserved: hold_margin =
# Tco_min + trace_min - clk_insertion_delay - Th_fpga ≥ 0.
set_input_delay -clock [get_clocks ft_clkout] -max 9.667 [get_ports {ft_data[*]}]
set_input_delay -clock [get_clocks ft_clkout] -min 1.0 [get_ports {ft_data[*]}]
set_input_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_data[*]}]
set_input_delay -clock [get_clocks ft_clkout] -max 9.667 [get_ports {ft_rxf_n}]
set_input_delay -clock [get_clocks ft_clkout] -min 1.0 [get_ports {ft_rxf_n}]
set_input_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_rxf_n}]
set_input_delay -clock [get_clocks ft_clkout] -max 9.667 [get_ports {ft_txe_n}]
set_input_delay -clock [get_clocks ft_clkout] -min 1.0 [get_ports {ft_txe_n}]
set_input_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_txe_n}]
# Output delays: FPGA → FT2232H (control strobes and data bus when writing)
set_output_delay -clock [get_clocks ft_clkout] -max 5.5 [get_ports {ft_data[*]}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_data[*]}]
set_output_delay -clock [get_clocks ft_clkout] -max 5.5 [get_ports {ft_rd_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_rd_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 5.5 [get_ports {ft_wr_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_wr_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 5.5 [get_ports {ft_oe_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_oe_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 5.5 [get_ports {ft_siwu}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_siwu}]
set_output_delay -clock [get_clocks ft_clkout] -max 11.667 [get_ports {ft_data[*]}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_data[*]}]
set_output_delay -clock [get_clocks ft_clkout] -max 11.667 [get_ports {ft_rd_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_rd_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 11.667 [get_ports {ft_wr_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_wr_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 11.667 [get_ports {ft_oe_n}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_oe_n}]
set_output_delay -clock [get_clocks ft_clkout] -max 11.667 [get_ports {ft_siwu}]
set_output_delay -clock [get_clocks ft_clkout] -min 0.0 [get_ports {ft_siwu}]
# ============================================================================
# STATUS / DEBUG OUTPUTS — NO PHYSICAL CONNECTIONS
@@ -489,42 +411,24 @@ set_false_path -from [get_ports {stm32_mixers_enable}]
set_false_path -from [get_cells reset_sync_reg[*]] -to [get_pins -filter {REF_PIN_NAME == CLR} -of_objects [get_cells -hierarchical -filter {PRIMITIVE_TYPE =~ REGISTER.*.*}]]
# --------------------------------------------------------------------------
# Clock Domain Crossing — asynchronous clock groups
#
# Rationale: prefer `set_clock_groups -asynchronous` over pairwise
# `set_false_path -from CLK -to CLK`. The latter is an STA antipattern:
# it disables *all* paths between the two domains, including the
# synchronizer paths themselves and any future inadvertent crossings,
# which can mask real CDC bugs that only show up at temperature/voltage
# corners. Clock-groups is the idiomatic way to declare domains async
# while still letting STA flag newly-introduced unrelated paths.
#
# Register-level false_paths (e.g. reset_sync_reg above) remain
# appropriate — those restrict the waiver to specific, audited endpoints.
#
# Groups declared here mirror the pairwise false_paths that existed
# previously; no new pair is declared async.
# Clock Domain Crossing false paths
# --------------------------------------------------------------------------
# clk_100m ↔ adc_dco_p (400 MHz): DDC has internal CDC synchronizers
set_clock_groups -asynchronous \
-group [get_clocks clk_100m] \
-group [get_clocks adc_dco_p]
set_false_path -from [get_clocks clk_100m] -to [get_clocks adc_dco_p]
set_false_path -from [get_clocks adc_dco_p] -to [get_clocks clk_100m]
# clk_100m ↔ clk_120m_dac: CDC via synchronizers in radar_system_top
set_clock_groups -asynchronous \
-group [get_clocks clk_100m] \
-group [get_clocks clk_120m_dac]
set_false_path -from [get_clocks clk_100m] -to [get_clocks clk_120m_dac]
set_false_path -from [get_clocks clk_120m_dac] -to [get_clocks clk_100m]
# FT2232H CDC: clk_100m ↔ ft_clkout (60 MHz), toggle CDC in RTL
set_clock_groups -asynchronous \
-group [get_clocks clk_100m] \
-group [get_clocks ft_clkout]
set_false_path -from [get_clocks clk_100m] -to [get_clocks ft_clkout]
set_false_path -from [get_clocks ft_clkout] -to [get_clocks clk_100m]
# FT2232H CDC: clk_120m_dac ↔ ft_clkout (no direct crossing, but belt-and-suspenders)
set_clock_groups -asynchronous \
-group [get_clocks clk_120m_dac] \
-group [get_clocks ft_clkout]
set_false_path -from [get_clocks clk_120m_dac] -to [get_clocks ft_clkout]
set_false_path -from [get_clocks ft_clkout] -to [get_clocks clk_120m_dac]
# ============================================================================
# PHYSICAL CONSTRAINTS
@@ -62,20 +62,6 @@ module radar_system_top_50t (
input wire adc_dco_n,
output wire adc_pwdn,
// ----- AD9484 overflow flag (differential) -----
// Schematic pads M6 (OR_P) / N6 (OR_N). Anchored-only for now; a future
// PR will wire this into the receive-path status flags for AGC feedback.
input wire adc_or_p,
input wire adc_or_n,
// ----- Tap of AD9523 -> AD9484 sample clock (differential) -----
// Schematic pads N11 (P) / N12 (N). Must remain input-only driving
// these pads as outputs would contend with the AD9523 driver feeding
// the ADC. Anchored with an IBUFDS (DONT_TOUCH) below; buffered net is
// unconsumed pending a follow-up PR.
input wire fpga_adc_clock_p,
input wire fpga_adc_clock_n,
// ===== STM32 Control (Bank 15: 3.3V) =====
input wire stm32_new_chirp,
input wire stm32_new_elevation,
@@ -98,38 +84,6 @@ module radar_system_top_50t (
output wire gpio_dig7 // DIG_7 (H12PD15): reserved
);
// =====================================================================
// Anchored-but-unused schematic inputs (secured via IBUFDS + DONT_TOUCH)
// =====================================================================
// Without these buffer instantiations, synthesis would remove the
// orphan input ports (UCIO / NSTD warnings) and the XDC pin constraints
// would fail to bind. DONT_TOUCH forces Vivado to retain the buffer
// primitives and their package-pin connections across all optimization
// stages. The buffered nets are intentionally left unconsumed here;
// they will be wired into the RTL in a follow-up PR once the ADC
// status-flag and sample-clock-tap features are implemented.
(* DONT_TOUCH = "TRUE" *) wire adc_or_buf;
(* DONT_TOUCH = "TRUE" *) IBUFDS #(
.DIFF_TERM ("TRUE"),
.IBUF_LOW_PWR("FALSE"),
.IOSTANDARD ("LVDS_25")
) u_ibufds_adc_or (
.O (adc_or_buf),
.I (adc_or_p),
.IB (adc_or_n)
);
(* DONT_TOUCH = "TRUE" *) wire fpga_adc_clock_buf;
(* DONT_TOUCH = "TRUE" *) IBUFDS #(
.DIFF_TERM ("TRUE"),
.IBUF_LOW_PWR("FALSE"),
.IOSTANDARD ("LVDS_25")
) u_ibufds_fpga_adc_clk (
.O (fpga_adc_clock_buf),
.I (fpga_adc_clock_p),
.IB (fpga_adc_clock_n)
);
// ===== Tie-off wires for unconstrained FT601 inputs (inactive with USB_MODE=1) =====
wire ft601_txe_tied = 1'b0;
wire ft601_rxf_tied = 1'b0;
+7 -7
View File
@@ -169,11 +169,11 @@ endfunction
// =========================================================================
// Clamp a wider signed value to [-7, +7]
function signed [3:0] clamp_gain;
input signed [4:0] val; // 5-bit to handle overflow from add
input signed [5:0] val; // 6-bit: covers [-22,+22] (max |gain|+step = 7+15)
begin
if (val > 5'sd7)
if (val > 6'sd7)
clamp_gain = 4'sd7;
else if (val < -5'sd7)
else if (val < -6'sd7)
clamp_gain = -4'sd7;
else
clamp_gain = val[3:0];
@@ -246,15 +246,15 @@ always @(posedge clk or negedge reset_n) begin
// Use inclusive counts/peaks (accounting for simultaneous valid_in)
if (wire_frame_sat_incr || frame_sat_count > 8'd0) begin
// Clipping detected: reduce gain immediately (attack)
agc_gain <= clamp_gain($signed({agc_gain[3], agc_gain}) -
$signed({1'b0, agc_attack}));
agc_gain <= clamp_gain($signed({agc_gain[3], agc_gain[3], agc_gain}) -
$signed({2'b00, agc_attack}));
holdoff_counter <= agc_holdoff; // Reset holdoff
end else if ((wire_frame_peak_update ? max_iq[14:7] : frame_peak[14:7])
< agc_target) begin
// Signal too weak: increase gain after holdoff expires
if (holdoff_counter == 4'd0) begin
agc_gain <= clamp_gain($signed({agc_gain[3], agc_gain}) +
$signed({1'b0, agc_decay}));
agc_gain <= clamp_gain($signed({agc_gain[3], agc_gain[3], agc_gain}) +
$signed({2'b00, agc_decay}));
end else begin
holdoff_counter <= holdoff_counter - 4'd1;
end
+2 -3
View File
@@ -7,7 +7,6 @@
[![Frequency: 10.5GHz](https://img.shields.io/badge/Frequency-10.5GHz-blue)](https://github.com/NawfalMotii79/PLFM_RADAR)
[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](https://github.com/NawfalMotii79/PLFM_RADAR/pulls)
![AERIS-10 Radar System](https://raw.githubusercontent.com/NawfalMotii79/PLFM_RADAR/main/8_Utils/3fb1dabf-2c6d-4b5d-b471-48bc461ce914.jpg)
AERIS-10 is an open-source, low-cost 10.5 GHz phased array radar system featuring Pulse Linear Frequency Modulated (LFM) modulation. Available in two versions (3km and 20km range), it's designed for researchers, drone developers, and serious SDR enthusiasts who want to explore and experiment with phased array radar technology.
@@ -47,7 +46,7 @@ The AERIS-10 main sub-systems are:
- **Main Board** containing:
- **DAC** - Generates the RADAR Chirps
- **2x Microwave Mixers (LT5552)** - For up-conversion and IF-down-conversion
- **2x Microwave Mixers (LTC5552)** - For up-conversion and IF-down-conversion
- **4x 4-Channel Phase Shifters (ADAR1000)** - For RX and TX chain beamforming
- **16x Front End Chips (ADTR1107)** - Used for both Low Noise Amplifying (RX) and Power Amplifying (TX) stages
- **XC7A50T FPGA** - Handles RADAR Signal Processing on the upstream FTG256 board:
@@ -92,7 +91,7 @@ The AERIS-10 main sub-systems are:
### Processing Pipeline
1. **Waveform Generation** - DAC creates LFM chirps
2. **Up/Down Conversion** - LT5552 mixers handle frequency translation
2. **Up/Down Conversion** - LTC5552 mixers handle frequency translation
3. **Beam Steering** - ADAR1000 phase shifters control 16 elements
4. **Signal Processing (FPGA)**:
- Raw ADC data capture