di 1 incre

main.cpp

@@ -114,6 +114,8 @@ struct Config {
     std::optional<std::string> tty_path;
     bool di1_group_average = false;
     bool do1_toggle_per_frame = false;
+    bool do1_noise_subtract = false;
+    std::optional<uint32_t> noise_avg_steps;
 };
 
 [[noreturn]] void fail(const std::string& message) {
@@ -471,7 +473,7 @@ void print_help(const char* exe_name) {
         << "                [di1:zero|trace|ignore]\n"
         << "                [duration_ms:100] [packet_limit:0] [csv:capture.csv] [svg:capture.svg]\n"
         << "                [live_html:live_plot.html] [live_json:live_plot.json] [tty:/tmp/ttyADC_data] [di1_group_avg]\n"
-        << "                [do1_toggle_per_frame]\n"
+        << "                [do1_toggle_per_frame] [do1_noise_subtract] [noise_avg_steps:N]\n"
         << "                [recv_block:32768] [stats_period_ms:1000] [live_update_period_ms:1000] [svg_history_packets:50] [start_wait_ms:10000]\n"
         << "                [buffer_words:8388608] [step_words:32768]\n"
         << "                [pullup_syn1] [pullup_syn2] [pulldown_conv_in] [pulldown_start_in]\n"
@@ -496,19 +498,26 @@ void print_help(const char* exe_name) {
         << "  stats_period_ms:1000 -> print online transfer/capture statistics every 1000 ms (0 disables)\n"
         << "  recv_block:32768 -> request up to 32768 raw 32-bit words per X502_Recv() call\n"
         << "  live_update_period_ms:1000 -> refresh live HTML/JSON no more than once per second"
-        << " (ignored in tty+di1_group_avg fast mode)\n"
+        << " (ignored in tty fast stream-only modes)\n"
         << "  svg_history_packets:50 -> keep last 50 packets in RAM for final SVG (0 keeps all, risky)"
-        << " (ignored in tty+di1_group_avg fast mode)\n"
+        << " (ignored in tty fast stream-only modes)\n"
         << "  duration_ms:100   -> max packet length if stop edge does not arrive\n"
         << "  packet_limit:0    -> 0 means continuous until Ctrl+C, N means stop after N packets\n"
         << "  buffer_words:8388608 -> input stream buffer size in 32-bit words\n"
         << "  step_words:32768 -> input stream transfer step in 32-bit words\n"
-        << "  live_html/live_json -> live graph files updated as packets arrive outside tty+di1_group_avg fast mode\n"
+        << "  live_html/live_json -> live graph files updated as packets arrive outside tty fast stream-only modes\n"
         << "  tty:/tmp/ttyADC_data -> write a continuous legacy 4-word CH1/CH2 stream; with channels:1, CH2 is 0\n"
         << "  di1_group_avg -> with tty + di1:trace, emit one averaged 4-word step per constant DI1 run\n"
-        << "  do1_toggle_per_frame -> packet-gated DO1 output: LOW,HIGH,LOW,HIGH per 2-channel frame inside packet\n"
+        << "  do1_toggle_per_frame -> hardware cyclic DO1 pattern in module memory:\n"
-        << "                         outside packet windows DO1 is forced LOW\n"
+        << "                         DO1 outputs 00110011... continuously (toggle every 2 ADC ticks)\n"
-        << "  tty+di1_group_avg -> fast stream-only mode: skip CSV/SVG/live outputs and send only averaged tty data\n"
+        << "                         without per-tick DOUT updates from PC\n"
+        << "  do1_noise_subtract -> with tty + do1_toggle_per_frame, use DI1 state from synchronous DIN\n"
+        << "                        (expected DO1->DI1 loopback) to average recent noise steps and subtract it\n"
+        << "                        from useful steps before tty output\n"
+        << "                        (only corrected DO1=LOW steps are sent to tty)\n"
+        << "  noise_avg_steps:N -> number of recent DO1=HIGH noise steps per channel used as subtraction baseline\n"
+        << "                       (required when do1_noise_subtract is enabled)\n"
+        << "  tty fast stream-only modes -> di1_group_avg or do1_noise_subtract; skip CSV/SVG/live outputs\n"
         << "  If sample_clock_hz is omitted together with clock:internal, the maximum ADC speed is used\n"
         << "\n"
         << "Profiles:\n"
@@ -531,10 +540,12 @@ void print_help(const char* exe_name) {
         << "This build enables synchronous DIN together with ADC. DI_SYN2 stop edges are detected\n"
         << "inside the same input stream, packets are split continuously by DI_SYN2 edges, and DI1\n"
         << "can either zero ADC samples on change, be exported as a separate synchronous trace, or be ignored.\n"
-        << "Outside tty+di1_group_avg fast mode, open live_plot.html in a browser to see the live graph update over time.\n"
+        << "Outside tty fast stream-only modes, open live_plot.html in a browser to see the live graph update over time.\n"
         << "The live HTML supports X/Y zoom buttons, mouse-wheel zoom, and reset.\n"
         << "For tty output, use di1_group_avg together with di1:trace to emit one averaged 4-word step\n"
         << "per constant DI1 run while keeping the current ring-buffered binary frame format.\n"
+        << "For DO1 subtraction with loopback, use do1_toggle_per_frame + do1_noise_subtract + noise_avg_steps:N;\n"
+        << "DI1 from synchronous DIN is used as the actual DO1 state marker.\n"
         << "Amplitude mode keeps the raw AD8317 voltage as captured on X1; no inversion is applied.\n"
         << "\n"
         << "Phase profile example:\n"
@@ -551,7 +562,13 @@ void print_help(const char* exe_name) {
         << "  " << exe_name
         << " profile:amplitude clock:internal internal_ref_hz:2000000 start:di_syn2_rise stop:di_syn2_fall"
         << " sample_clock_hz:max range:0.2 di1:trace di1_group_avg duration_ms:100 packet_limit:0"
-        << " tty:/tmp/ttyADC_data\n";
+        << " tty:/tmp/ttyADC_data\n"
+        << "\n"
+        << "Amplitude TTY DO1 noise subtract example:\n"
+        << "  " << exe_name
+        << " profile:amplitude clock:internal internal_ref_hz:2000000 start:di_syn2_rise stop:di_syn2_fall"
+        << " sample_clock_hz:max range:0.2 do1_toggle_per_frame do1_noise_subtract noise_avg_steps:16"
+        << " duration_ms:100 packet_limit:0 tty:/tmp/ttyADC_data\n";
 }
 
 Config parse_args(int argc, char** argv) {
@@ -635,10 +652,18 @@ Config parse_args(int argc, char** argv) {
             cfg.di1_group_average = true;
             continue;
         }
+        if (arg == "do1_noise_subtract") {
+            cfg.do1_noise_subtract = true;
+            continue;
+        }
         if (arg == "do1_toggle_per_frame") {
             cfg.do1_toggle_per_frame = true;
             continue;
         }
+        if (starts_with(arg, "noise_avg_steps:")) {
+            cfg.noise_avg_steps = parse_u32(arg.substr(16), "noise_avg_steps");
+            continue;
+        }
         if (starts_with(arg, "sample_clock_hz:")) {
             const std::string value = trim_copy(arg.substr(16));
             cfg.sample_clock_specified = true;
@@ -750,7 +775,7 @@ Config parse_args(int argc, char** argv) {
             cfg.live_update_period_ms = std::max<uint32_t>(cfg.live_update_period_ms, 1000U);
         }
     }
-    if (cfg.tty_path && cfg.di1_group_average) {
+    if (cfg.tty_path && (cfg.di1_group_average || cfg.do1_noise_subtract)) {
         if (!cfg.recv_block_specified) {
             cfg.recv_block_words = std::max<uint32_t>(cfg.recv_block_words, 65536U);
         }
@@ -773,8 +798,24 @@ Config parse_args(int argc, char** argv) {
     if (cfg.di1_group_average && (cfg.di1_mode != Di1Mode::Trace)) {
         fail("di1_group_avg requires di1:trace");
     }
-    if (cfg.do1_toggle_per_frame && (cfg.channel_count != 2U)) {
+    if (cfg.do1_noise_subtract) {
-        fail("do1_toggle_per_frame requires channels:2");
+        if (!cfg.tty_path.has_value()) {
+            fail("do1_noise_subtract requires tty:<path>");
+        }
+        if (!cfg.do1_toggle_per_frame) {
+            fail("do1_noise_subtract requires do1_toggle_per_frame");
+        }
+        if (cfg.di1_group_average) {
+            fail("do1_noise_subtract is incompatible with di1_group_avg");
+        }
+        if (!cfg.noise_avg_steps.has_value()) {
+            fail("do1_noise_subtract requires noise_avg_steps:N");
+        }
+        if (*cfg.noise_avg_steps == 0U) {
+            fail("noise_avg_steps must be > 0");
+        }
+    } else if (cfg.noise_avg_steps.has_value()) {
+        fail("noise_avg_steps:N can only be used together with do1_noise_subtract");
     }
     if (sync_uses_di_syn1(cfg.sync_mode) && sync_uses_di_syn1(cfg.sync_start_mode)) {
         fail("clock and start cannot both use DI_SYN1; use start_in or immediate start");
@@ -926,6 +967,7 @@ struct Api {
     decltype(&X502_GetAdcFreq) GetAdcFreq = nullptr;
     decltype(&X502_SetDinFreq) SetDinFreq = nullptr;
     decltype(&X502_SetOutFreq) SetOutFreq = nullptr;
+    decltype(&X502_AsyncOutDig) AsyncOutDig = nullptr;
     decltype(&X502_SetRefFreq) SetRefFreq = nullptr;
     decltype(&X502_SetStreamBufSize) SetStreamBufSize = nullptr;
     decltype(&X502_SetStreamStep) SetStreamStep = nullptr;
@@ -936,7 +978,9 @@ struct Api {
     decltype(&X502_StreamsStart) StreamsStart = nullptr;
     decltype(&X502_GetRecvReadyCount) GetRecvReadyCount = nullptr;
     decltype(&X502_Recv) Recv = nullptr;
-    decltype(&X502_PreloadStart) PreloadStart = nullptr;
+    decltype(&X502_OutCycleLoadStart) OutCycleLoadStart = nullptr;
+    decltype(&X502_OutCycleSetup) OutCycleSetup = nullptr;
+    decltype(&X502_OutCycleStop) OutCycleStop = nullptr;
     decltype(&X502_Send) Send = nullptr;
     decltype(&X502_PrepareData) PrepareData = nullptr;
     decltype(&X502_ProcessData) ProcessData = nullptr;
@@ -983,6 +1027,7 @@ struct Api {
         GetAdcFreq = load_symbol<decltype(GetAdcFreq)>(x502_module, "X502_GetAdcFreq");
         SetDinFreq = load_symbol<decltype(SetDinFreq)>(x502_module, "X502_SetDinFreq");
         SetOutFreq = load_symbol<decltype(SetOutFreq)>(x502_module, "X502_SetOutFreq");
+        AsyncOutDig = load_symbol<decltype(AsyncOutDig)>(x502_module, "X502_AsyncOutDig");
         SetRefFreq = load_symbol<decltype(SetRefFreq)>(x502_module, "X502_SetRefFreq");
         SetStreamBufSize = load_symbol<decltype(SetStreamBufSize)>(x502_module, "X502_SetStreamBufSize");
         SetStreamStep = load_symbol<decltype(SetStreamStep)>(x502_module, "X502_SetStreamStep");
@@ -993,7 +1038,9 @@ struct Api {
         StreamsStart = load_symbol<decltype(StreamsStart)>(x502_module, "X502_StreamsStart");
         GetRecvReadyCount = load_symbol<decltype(GetRecvReadyCount)>(x502_module, "X502_GetRecvReadyCount");
         Recv = load_symbol<decltype(Recv)>(x502_module, "X502_Recv");
-        PreloadStart = load_symbol<decltype(PreloadStart)>(x502_module, "X502_PreloadStart");
+        OutCycleLoadStart = try_load_symbol<decltype(OutCycleLoadStart)>(x502_module, "X502_OutCycleLoadStart");
+        OutCycleSetup = try_load_symbol<decltype(OutCycleSetup)>(x502_module, "X502_OutCycleSetup");
+        OutCycleStop = try_load_symbol<decltype(OutCycleStop)>(x502_module, "X502_OutCycleStop");
         Send = load_symbol<decltype(Send)>(x502_module, "X502_Send");
         PrepareData = load_symbol<decltype(PrepareData)>(x502_module, "X502_PrepareData");
         ProcessData = load_symbol<decltype(ProcessData)>(x502_module, "X502_ProcessData");
@@ -1029,6 +1076,8 @@ constexpr uint32_t kE502DiSyn2Mask =
 constexpr uint32_t kE502Digital1Mask = (static_cast<uint32_t>(1U) << 0U);
 constexpr uint32_t kE502Digital2Mask = (static_cast<uint32_t>(1U) << 1U);
 constexpr uint32_t kE502Do1Mask = (static_cast<uint32_t>(1U) << 0U);
+constexpr uint32_t kDo1TogglePeriodTicks = 2U;
+constexpr uint32_t kDo1CyclePatternWords = kDo1TogglePeriodTicks * 2U;
 constexpr uint32_t kStreamInputAdcFlag = 0x80000000U;
 constexpr uint32_t kStreamInputCalibratedAdcFlag = 0x40000000U;
 
@@ -1182,6 +1231,103 @@ struct TtyGroupAverageState {
     }
 };
 
+struct Do1NoiseSubtractState {
+    uint32_t noise_avg_steps = 0;
+    std::array<std::vector<double>, 2> noise_ring;
+    std::array<std::size_t, 2> noise_ring_head {};
+    std::array<std::size_t, 2> noise_ring_size {};
+    std::array<double, 2> noise_ring_sum {};
+    std::array<double, 2> step_sum {};
+    std::array<uint32_t, 2> step_count {};
+    bool step_level_initialized = false;
+    bool step_di1_high = false;
+    uint16_t next_index = 1;
+
+    void configure(uint32_t history_steps) {
+        noise_avg_steps = history_steps;
+        for (auto& ring : noise_ring) {
+            ring.assign(noise_avg_steps, 0.0);
+        }
+        reset_packet();
+    }
+
+    void clear_step() {
+        step_sum = {};
+        step_count = {};
+    }
+
+    void reset_packet() {
+        clear_step();
+        step_level_initialized = false;
+        step_di1_high = false;
+        next_index = 1;
+        noise_ring_head = {};
+        noise_ring_size = {};
+        noise_ring_sum = {};
+    }
+
+    void start_new_step(bool di1_high) {
+        clear_step();
+        step_di1_high = di1_high;
+        step_level_initialized = true;
+    }
+
+    void add_sample(uint32_t lch, double raw_code) {
+        if (lch >= step_sum.size()) {
+            return;
+        }
+        step_sum[lch] += raw_code;
+        ++step_count[lch];
+    }
+
+    bool has_complete_step(uint32_t channel_count) const {
+        return (step_count[0] != 0U) && ((channel_count <= 1U) || (step_count[1] != 0U));
+    }
+
+    double step_average(uint32_t lch) const {
+        if ((lch >= step_sum.size()) || (step_count[lch] == 0U)) {
+            return 0.0;
+        }
+        return step_sum[lch] / static_cast<double>(step_count[lch]);
+    }
+
+    void push_noise_average(uint32_t lch, double avg_value) {
+        if ((noise_avg_steps == 0U) || (lch >= noise_ring.size())) {
+            return;
+        }
+
+        auto& ring = noise_ring[lch];
+        auto& head = noise_ring_head[lch];
+        auto& size = noise_ring_size[lch];
+        auto& sum = noise_ring_sum[lch];
+
+        if (size < noise_avg_steps) {
+            ring[size] = avg_value;
+            sum += avg_value;
+            ++size;
+            return;
+        }
+
+        sum -= ring[head];
+        ring[head] = avg_value;
+        sum += avg_value;
+        head = (head + 1U) % noise_avg_steps;
+    }
+
+    double noise_baseline(uint32_t lch) const {
+        if (lch >= noise_ring_size.size()) {
+            return 0.0;
+        }
+        const std::size_t size = noise_ring_size[lch];
+        if (size == 0U) {
+            return 0.0;
+        }
+        return noise_ring_sum[lch] / static_cast<double>(size);
+    }
+
+    void finish_step() { clear_step(); }
+};
+
 int16_t pack_raw_code_to_int16(double avg_raw_code) {
     const double scaled =
         avg_raw_code * 32767.0 / static_cast<double>(X502_ADC_SCALE_CODE_MAX);
@@ -1383,9 +1529,10 @@ int run(const Config& cfg) {
     }
     double actual_dout_freq_hz = actual_frame_freq_hz;
     if (cfg.do1_toggle_per_frame) {
-        if (actual_frame_freq_hz <= 0.0) {
+        if (actual_sample_clock_hz <= 0.0) {
-            fail("do1_toggle_per_frame requires a positive ADC frame rate");
+            fail("do1_toggle_per_frame requires a positive ADC sample rate");
         }
+        actual_dout_freq_hz = actual_sample_clock_hz;
         expect_ok(api, api.SetOutFreq(device.hnd, &actual_dout_freq_hz), "Set DOUT frequency");
     }
 
@@ -1449,7 +1596,8 @@ int run(const Config& cfg) {
               << ((cfg.channel_count >= 2U) ? phy_channel_name(cfg.mode, cfg.ch2) : std::string("disabled")) << "\n"
               << "  DI1 handling: " << di1_mode_to_string(cfg.di1_mode) << "\n"
               << "  DO1 toggle per frame: "
-              << (cfg.do1_toggle_per_frame ? std::string("enabled (packet-gated, LOW-first)") : std::string("disabled")) << "\n"
+              << (cfg.do1_toggle_per_frame ? std::string("enabled (hardware cyclic 00110011..., toggle every 2 ADC ticks)")
+                                           : std::string("disabled")) << "\n"
               << "  DOUT rate: "
               << (cfg.do1_toggle_per_frame ? std::to_string(actual_dout_freq_hz) + " Hz" : std::string("disabled")) << "\n"
               << "  ADC range: +/-" << range_to_volts(cfg.range) << " V\n"
@@ -1459,9 +1607,14 @@ int run(const Config& cfg) {
     std::size_t tty_ring_capacity_bytes = 0;
     std::unique_ptr<TtyProtocolWriter> tty_writer;
     const bool tty_di1_group_average = cfg.tty_path.has_value() && cfg.di1_group_average;
-    const bool fast_tty_avg_stream_mode = tty_di1_group_average;
+    const bool tty_do1_noise_subtract = cfg.tty_path.has_value() && cfg.do1_noise_subtract;
+    const bool fast_tty_avg_stream_mode = tty_di1_group_average || tty_do1_noise_subtract;
     const uint16_t tty_packet_start_marker =
         (cfg.profile == CaptureProfile::Amplitude) ? 0x001AU : 0x000AU;
+    const std::string fast_tty_mode_name =
+        tty_do1_noise_subtract ? std::string("do1_noise_subtract")
+                               : (tty_di1_group_average ? std::string("di1_group_avg")
+                                                        : std::string("none"));
     if (cfg.tty_path) {
         const uint64_t typical_tty_batch_frames = (static_cast<uint64_t>(read_capacity_words) + 1U) / 2U;
         const uint64_t typical_tty_batch_bytes = typical_tty_batch_frames * sizeof(uint16_t) * 4U;
@@ -1472,7 +1625,7 @@ int run(const Config& cfg) {
         }
         tty_ring_capacity_bytes = static_cast<std::size_t>(tty_ring_bytes_u64);
         tty_writer = std::make_unique<TtyProtocolWriter>(*cfg.tty_path, tty_ring_capacity_bytes);
-        if (!tty_di1_group_average) {
+        if (!tty_di1_group_average && !tty_do1_noise_subtract) {
             tty_writer->emit_packet_start(tty_packet_start_marker);
         }
     }
@@ -1483,13 +1636,18 @@ int run(const Config& cfg) {
         writer->initialize_csv(cfg.channel_count, cfg.di1_mode == Di1Mode::Trace);
     }
     if (fast_tty_avg_stream_mode) {
-        std::cout << "  tty avg stream-only mode: enabled\n"
+        std::cout << "  tty fast stream-only mode: enabled (" << fast_tty_mode_name << ")\n"
                   << "  tty stream output: " << *cfg.tty_path << "\n"
                   << "  tty ring buffer bytes: " << tty_ring_capacity_bytes << "\n"
                   << "  recv block words: " << cfg.recv_block_words << "\n"
                   << "  input step words: " << cfg.input_step_words << "\n"
                   << "  input buffer words: " << cfg.input_buffer_words << "\n"
-                  << "  tty di1_group_avg: enabled\n"
+                  << "  tty di1_group_avg: " << (tty_di1_group_average ? "enabled" : "disabled") << "\n"
+                  << "  tty do1_noise_subtract: " << (tty_do1_noise_subtract ? "enabled" : "disabled") << "\n"
+                  << "  do1_noise_subtract marker: "
+                  << (tty_do1_noise_subtract ? "DI1 from DIN (DO1->DI1 loopback expected)" : "n/a") << "\n"
+                  << "  noise_avg_steps: "
+                  << (tty_do1_noise_subtract ? std::to_string(*cfg.noise_avg_steps) : std::string("n/a")) << "\n"
                   << "  stream-only note: ignoring csv/svg/live_html/live_json/live_update_period_ms/svg_history_packets\n";
     } else {
         std::cout << "  live plot html: " << writer->live_html_path() << "\n"
@@ -1505,6 +1663,11 @@ int run(const Config& cfg) {
                                             : (cfg.di1_group_average ? std::string("requested, tty disabled")
                                                                      : std::string("disabled")))
                   << "\n"
+                  << "  tty do1_noise_subtract: "
+                  << (tty_do1_noise_subtract ? std::string("enabled")
+                                             : (cfg.do1_noise_subtract ? std::string("requested, tty disabled")
+                                                                       : std::string("disabled")))
+                  << "\n"
                   << "  SVG history packets kept in RAM: "
                   << ((cfg.svg_history_packets == 0U) ? std::string("all (unbounded)") : std::to_string(cfg.svg_history_packets))
                   << "\n";
@@ -1515,20 +1678,46 @@ int run(const Config& cfg) {
         std::cerr << "Warning: Ctrl+C handler could not be installed; continuous mode may stop abruptly.\n";
     }
 
+    std::array<uint32_t, kDo1CyclePatternWords> do1_cycle_pattern {};
+    std::array<uint32_t, kDo1CyclePatternWords> do1_cycle_encoded {};
     if (cfg.do1_toggle_per_frame) {
-        expect_ok(api, api.PreloadStart(device.hnd), "Start DOUT preload");
+        if ((api.OutCycleLoadStart == nullptr) || (api.OutCycleSetup == nullptr) || (api.OutCycleStop == nullptr)) {
-        const uint32_t preload_low_word = 0U;
+            fail("do1_toggle_per_frame requires X502_OutCycle* API support (cyclic output mode)");
-        uint32_t preload_encoded_word = 0U;
+        }
+
+        for (std::size_t i = 0; i < do1_cycle_pattern.size(); ++i) {
+            do1_cycle_pattern[i] = (i < kDo1TogglePeriodTicks) ? 0U : kE502Do1Mask;
+        }
+
         expect_ok(api,
-                  api.PrepareData(device.hnd, nullptr, nullptr, &preload_low_word, 1U, 0, &preload_encoded_word),
+                  api.OutCycleLoadStart(device.hnd, static_cast<uint32_t>(do1_cycle_pattern.size())),
-                  "Prepare initial DOUT LOW");
+                  "Allocate cyclic DOUT buffer");
-        const int32_t preload_sent = api.Send(device.hnd, &preload_encoded_word, 1U, cfg.recv_timeout_ms);
+        expect_ok(api,
-        if (preload_sent < 0) {
+                  api.PrepareData(device.hnd,
-            fail("Send initial DOUT LOW: " + x502_error(api, preload_sent));
+                                  nullptr,
+                                  nullptr,
+                                  do1_cycle_pattern.data(),
+                                  static_cast<uint32_t>(do1_cycle_pattern.size()),
+                                  0,
+                                  do1_cycle_encoded.data()),
+                  "Prepare cyclic DOUT pattern");
+
+        std::size_t sent_cycle_words = 0;
+        while (sent_cycle_words < do1_cycle_encoded.size()) {
+            const int32_t sent = api.Send(device.hnd,
+                                          do1_cycle_encoded.data() + sent_cycle_words,
+                                          static_cast<uint32_t>(do1_cycle_encoded.size() - sent_cycle_words),
+                                          cfg.recv_timeout_ms);
+            if (sent < 0) {
+                fail("Send cyclic DOUT pattern: " + x502_error(api, sent));
             }
-        if (preload_sent != 1) {
+            if (sent == 0) {
-            fail("Send initial DOUT LOW did not transmit exactly one word");
+                fail("Send cyclic DOUT pattern timed out before all words were queued");
             }
+            sent_cycle_words += static_cast<std::size_t>(sent);
+        }
+
+        expect_ok(api, api.OutCycleSetup(device.hnd, X502_OUT_CYCLE_FLAGS_WAIT_DONE), "Activate cyclic DOUT pattern");
     }
 
     expect_ok(api, api.StreamsStart(device.hnd), "Start streams");
@@ -1541,13 +1730,16 @@ int run(const Config& cfg) {
     std::deque<double> pending_adc;
     std::deque<double> pending_adc_raw;
     std::deque<uint32_t> pending_din;
-    std::deque<uint32_t> pending_dout;
     std::deque<CapturePacket> packets;
     PacketAccumulator current_packet;
     TtyContinuousState tty_state;
     TtyGroupAverageState tty_group_state;
+    Do1NoiseSubtractState tty_do1_noise_state;
     std::vector<uint16_t> tty_frame_words;
     tty_frame_words.reserve(static_cast<std::size_t>(read_capacity_words) * 2U + 16U);
+    if (tty_do1_noise_subtract) {
+        tty_do1_noise_state.configure(*cfg.noise_avg_steps);
+    }
     if (!fast_tty_avg_stream_mode) {
         current_packet.reset(target_frames, cfg.channel_count);
     }
@@ -1572,14 +1764,6 @@ int run(const Config& cfg) {
     TickMs stats_window_start = capture_loop_start;
     TickMs last_stats_print = capture_loop_start;
     TickMs last_live_update = 0;
-    bool do1_next_state_high = false;
-    const std::size_t max_pending_dout_words = 1000000U;
-    std::vector<uint32_t> dout_prepare_input;
-    std::vector<uint32_t> dout_send_words;
-    if (cfg.do1_toggle_per_frame) {
-        dout_prepare_input.resize(read_capacity_words);
-        dout_send_words.resize(read_capacity_words);
-    }
 
     uint64_t total_raw_words = 0;
     uint64_t total_adc_samples = 0;
@@ -1622,12 +1806,14 @@ int run(const Config& cfg) {
                   << ", DIN samples/s=" << din_samples_per_s
                   << ", frames/s per channel=" << frames_per_ch_per_s
                   << ", packets/s=" << packets_per_s;
+        if (!fast_tty_avg_stream_mode) {
             if (cfg.di1_mode == Di1Mode::ZeroOnChange) {
                 std::cout << ", zeroed on DI1 change=" << zeroed_fraction << "% ("
                           << stats_zeroed_samples << "/" << stats_stored_adc_samples << ")";
             } else if (cfg.di1_mode == Di1Mode::Trace) {
                 std::cout << ", DI1 trace=enabled";
             }
+        }
         if (tty_writer) {
             std::cout << ", tty_frames_written=" << tty_frames_written
                       << ", tty_frames_dropped=" << tty_frames_dropped
@@ -1681,6 +1867,49 @@ int run(const Config& cfg) {
         ++packet_avg_steps;
     };
 
+    auto append_tty_do1_subtracted_step = [&]() {
+        if (!tty_do1_noise_subtract) {
+            return;
+        }
+
+        if (!tty_do1_noise_state.step_level_initialized) {
+            return;
+        }
+
+        if (!tty_do1_noise_state.has_complete_step(cfg.channel_count)) {
+            tty_do1_noise_state.finish_step();
+            return;
+        }
+
+        const double ch1_avg = tty_do1_noise_state.step_average(0U);
+        const double ch2_avg = (cfg.channel_count <= 1U) ? 0.0 : tty_do1_noise_state.step_average(1U);
+
+        if (tty_do1_noise_state.step_di1_high) {
+            tty_do1_noise_state.push_noise_average(0U, ch1_avg);
+            if (cfg.channel_count > 1U) {
+                tty_do1_noise_state.push_noise_average(1U, ch2_avg);
+            }
+            tty_do1_noise_state.finish_step();
+            return;
+        }
+
+        if (tty_do1_noise_state.next_index >= 0xFFFFU) {
+            fail("TTY protocol step index overflow within packet");
+        }
+
+        const double ch1_corrected = ch1_avg - tty_do1_noise_state.noise_baseline(0U);
+        const double ch2_corrected =
+            (cfg.channel_count <= 1U) ? 0.0 : (ch2_avg - tty_do1_noise_state.noise_baseline(1U));
+
+        append_tty_frame((cfg.profile == CaptureProfile::Amplitude) ? 0x001AU : 0x000AU,
+                         tty_do1_noise_state.next_index,
+                         static_cast<uint16_t>(pack_raw_code_to_int16(ch1_corrected)),
+                         static_cast<uint16_t>(pack_raw_code_to_int16(ch2_corrected)));
+        ++tty_do1_noise_state.next_index;
+        ++packet_avg_steps;
+        tty_do1_noise_state.finish_step();
+    };
+
     auto flush_tty_frames = [&]() {
         if (!tty_writer || tty_frame_words.empty()) {
             return;
@@ -1696,66 +1925,6 @@ int run(const Config& cfg) {
         }
     };
 
-    auto append_dout_word = [&](bool do1_high) {
-        if (!cfg.do1_toggle_per_frame) {
-            return;
-        }
-        pending_dout.push_back(do1_high ? kE502Do1Mask : 0U);
-        if (pending_dout.size() > max_pending_dout_words) {
-            fail("Internal DOUT backlog grew too large");
-        }
-    };
-
-    auto flush_dout_queue = [&](uint32_t send_timeout_ms, bool require_full_flush) {
-        if (!cfg.do1_toggle_per_frame || pending_dout.empty()) {
-            return;
-        }
-
-        while (!pending_dout.empty()) {
-            const std::size_t chunk_words = std::min<std::size_t>(pending_dout.size(), dout_prepare_input.size());
-            for (std::size_t i = 0; i < chunk_words; ++i) {
-                dout_prepare_input[i] = pending_dout[i];
-            }
-
-            expect_ok(api,
-                      api.PrepareData(device.hnd,
-                                      nullptr,
-                                      nullptr,
-                                      dout_prepare_input.data(),
-                                      static_cast<uint32_t>(chunk_words),
-                                      0,
-                                      dout_send_words.data()),
-                      "Prepare DOUT data");
-
-            std::size_t sent_words = 0;
-            while (sent_words < chunk_words) {
-                const int32_t sent = api.Send(device.hnd,
-                                              dout_send_words.data() + sent_words,
-                                              static_cast<uint32_t>(chunk_words - sent_words),
-                                              send_timeout_ms);
-                if (sent < 0) {
-                    fail("Send DOUT data: " + x502_error(api, sent));
-                }
-                if (sent == 0) {
-                    break;
-                }
-                sent_words += static_cast<std::size_t>(sent);
-            }
-
-            for (std::size_t i = 0; i < sent_words; ++i) {
-                pending_dout.pop_front();
-            }
-
-            if (sent_words < chunk_words) {
-                break;
-            }
-        }
-
-        if (require_full_flush && !pending_dout.empty()) {
-            fail("DOUT queue could not be fully flushed before shutdown");
-        }
-    };
-
     auto start_packet = [&]() {
         if (packet_active) {
             return;
@@ -1771,9 +1940,9 @@ int run(const Config& cfg) {
         if (tty_di1_group_average) {
             tty_group_state.reset_packet();
             append_tty_packet_start();
-        }
+        } else if (tty_do1_noise_subtract) {
-        if (cfg.do1_toggle_per_frame) {
+            tty_do1_noise_state.reset_packet();
-            do1_next_state_high = false;
+            append_tty_packet_start();
         }
         packet_active = true;
     };
@@ -1790,6 +1959,10 @@ int run(const Config& cfg) {
         if (tty_di1_group_average) {
             append_tty_group_step();
             tty_group_state.clear_step();
+        } else if (tty_do1_noise_subtract) {
+            // Finalize the last DI1 run in packet, drop only incomplete channel tuples.
+            append_tty_do1_subtracted_step();
+            tty_do1_noise_state.clear_step();
         }
 
         const std::size_t frames = fast_tty_avg_stream_mode
@@ -1908,6 +2081,8 @@ int run(const Config& cfg) {
         }
         if (tty_di1_group_average) {
             tty_group_state.clear_step();
+        } else if (tty_do1_noise_subtract) {
+            tty_do1_noise_state.clear_step();
         }
     };
 
@@ -2132,7 +2307,6 @@ int run(const Config& cfg) {
                 }
                 trigger_prev_level = di_syn2_level;
             }
-
             if (!packet_active && (cfg.sync_start_mode == X502_SYNC_INTERNAL)) {
                 start_packet();
             }
@@ -2150,24 +2324,12 @@ int run(const Config& cfg) {
 
             if (packet_active && stop_edge) {
                 finish_packet(PacketCloseReason::ExternalStopEdge);
-                if (cfg.do1_toggle_per_frame && (lch == (cfg.channel_count - 1U))) {
-                    append_dout_word(false);
-                }
                 if ((cfg.packet_limit != 0U) && (total_completed_packets >= cfg.packet_limit)) {
                     stop_loop_requested = true;
                 }
                 continue;
             }
 
-            if (cfg.do1_toggle_per_frame && (lch == (cfg.channel_count - 1U))) {
-                if (packet_active) {
-                    append_dout_word(do1_next_state_high);
-                    do1_next_state_high = !do1_next_state_high;
-                } else {
-                    append_dout_word(false);
-                }
-            }
-
             if (!packet_active) {
                 continue;
             }
@@ -2182,6 +2344,13 @@ int run(const Config& cfg) {
             if (tty_di1_group_average && di1_changed) {
                 append_tty_group_step();
                 tty_group_state.clear_step();
+            } else if (tty_do1_noise_subtract) {
+                if (!tty_do1_noise_state.step_level_initialized) {
+                    tty_do1_noise_state.start_new_step(di1_level);
+                } else if (di1_changed) {
+                    append_tty_do1_subtracted_step();
+                    tty_do1_noise_state.start_new_step(di1_level);
+                }
             }
 
             if (!fast_tty_avg_stream_mode &&
@@ -2193,7 +2362,11 @@ int run(const Config& cfg) {
 
             if (fast_tty_avg_stream_mode) {
                 if (fast_packet_frames < target_frames) {
+                    if (tty_do1_noise_subtract) {
+                        tty_do1_noise_state.add_sample(lch, adc_raw_value);
+                    } else {
                         tty_group_state.add_sample(lch, adc_raw_value);
+                    }
                     if (lch == (cfg.channel_count - 1U)) {
                         ++fast_packet_frames;
                         ++total_completed_frames;
@@ -2238,7 +2411,6 @@ int run(const Config& cfg) {
             }
         }
 
-        flush_dout_queue(0U, false);
         flush_tty_frames();
 
         if (console_stop_requested()) {
@@ -2254,11 +2426,15 @@ int run(const Config& cfg) {
         }
     }
 
-    append_dout_word(false);
+    if (cfg.do1_toggle_per_frame) {
-    flush_dout_queue(cfg.recv_timeout_ms, true);
+        expect_ok(api, api.OutCycleStop(device.hnd, X502_OUT_CYCLE_FLAGS_WAIT_DONE), "Stop cyclic DOUT pattern");
+    }
 
     expect_ok(api, api.StreamsStop(device.hnd), "Stop streams");
     device.streams_started = false;
+    if (cfg.do1_toggle_per_frame) {
+        expect_ok(api, api.AsyncOutDig(device.hnd, 0U, ~kE502Do1Mask), "Force DO1 LOW after cyclic DOUT stop");
+    }
 
     if ((cfg.stats_period_ms != 0U) && ((stats_raw_words != 0U) || (stats_adc_samples != 0U) ||
         (stats_din_samples != 0U) || (stats_stored_adc_samples != 0U) || (stats_zeroed_samples != 0U) ||
@@ -2274,26 +2450,15 @@ int run(const Config& cfg) {
     }
 
     if (!fast_tty_avg_stream_mode) {
-        const std::vector<CapturePacket> svg_packets(packets.begin(), packets.end());
+        std::cout << "Captured " << total_completed_packets << " packet(s) in capture mode\n";
-        writer->write(svg_packets, actual_frame_freq_hz, range_to_volts(cfg.range));
-
-        std::size_t total_packet_frames = 0;
-        for (const auto& packet : svg_packets) {
-            total_packet_frames += (packet.channel_count <= 1U)
-                ? packet.ch1.size()
-                : std::min(packet.ch1.size(), packet.ch2.size());
-        }
-
-        std::cout << "Captured " << total_completed_packets << " packet(s), "
-                  << total_packet_frames << " total frames per channel kept for final SVG\n";
         if (cfg.di1_mode == Di1Mode::ZeroOnChange) {
             std::cout << "ADC samples forced to 0 on DI1 change: " << total_zeroed_samples << "\n";
         } else if (cfg.di1_mode == Di1Mode::Trace) {
-            std::cout << "DI1 synchronous trace exported to CSV/live plot/SVG\n";
+            std::cout << "DI1 synchronous trace exported to live plot\n";
         }
     } else {
         std::cout << "Captured " << total_completed_packets
-                  << " packet(s) in tty avg stream-only mode\n";
+                  << " packet(s) in tty fast stream-only mode (" << fast_tty_mode_name << ")\n";
     }
     std::cout << "Average stats: "
               << "MB/s=" << std::fixed << std::setprecision(3)
@@ -2313,6 +2478,7 @@ int run(const Config& cfg) {
               << (static_cast<double>(total_completed_packets) /
                   std::max(1e-9, static_cast<double>(tick_count_ms() - capture_loop_start) / 1000.0))
               << ", packets captured=" << total_completed_packets;
+    if (!fast_tty_avg_stream_mode) {
         if (cfg.di1_mode == Di1Mode::ZeroOnChange) {
             std::cout << ", zeroed on DI1 change="
                       << ((total_stored_adc_samples == 0U)
@@ -2322,20 +2488,16 @@ int run(const Config& cfg) {
         } else if (cfg.di1_mode == Di1Mode::Trace) {
             std::cout << ", DI1 trace=enabled";
         }
+    }
     if (tty_writer) {
         std::cout << ", tty_frames_written=" << tty_final_stats.frames_written
                   << ", tty_frames_dropped=" << tty_final_stats.frames_dropped
                   << ", tty_ring_overflows=" << tty_final_stats.ring_overflows;
     }
     if (fast_tty_avg_stream_mode) {
-        std::cout << ", tty avg stream-only mode=enabled";
+        std::cout << ", tty fast stream-only mode=" << fast_tty_mode_name;
     }
     std::cout << "\n";
-    if (!fast_tty_avg_stream_mode) {
-        std::cout << "Final SVG packets retained in memory: " << packets.size() << "\n"
-                  << "CSV: " << cfg.csv_path << "\n"
-                  << "SVG: " << cfg.svg_path << "\n";
-    }
 
     return 0;
 }

run_do1_noise_subtract.sh

@@ -0,0 +1,45 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+cd "$SCRIPT_DIR"
+
+BIN="${BIN:-./main.exe}"
+TTY_PATH="${TTY_PATH:-/tmp/ttyADC_data}"
+LIB_DIR="${LIB_DIR:-$HOME/.local/lib}"
+NOISE_AVG_STEPS="${NOISE_AVG_STEPS:-}"
+
+if [[ ! -x "$BIN" ]]; then
+  echo "Binary '$BIN' not found or not executable. Run ./build_main.sh first." >&2
+  exit 1
+fi
+
+if [[ -z "$NOISE_AVG_STEPS" ]]; then
+  echo "Set NOISE_AVG_STEPS to a positive integer, for example: NOISE_AVG_STEPS=16" >&2
+  exit 1
+fi
+
+if ! [[ "$NOISE_AVG_STEPS" =~ ^[0-9]+$ ]] || [[ "$NOISE_AVG_STEPS" -eq 0 ]]; then
+  echo "NOISE_AVG_STEPS must be a positive integer, got '$NOISE_AVG_STEPS'" >&2
+  exit 1
+fi
+
+if [[ -d "$LIB_DIR" ]]; then
+  export LD_LIBRARY_PATH="${LIB_DIR}${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}"
+fi
+
+exec "$BIN" \
+  clock:internal \
+  internal_ref_hz:2000000 \
+  start:di_syn2_rise \
+  stop:di_syn2_fall \
+  sample_clock_hz:max \
+  range:5 \
+  duration_ms:100 \
+  packet_limit:0 \
+  do1_toggle_per_frame \
+  do1_noise_subtract \
+  "noise_avg_steps:${NOISE_AVG_STEPS}" \
+  profile:phase \
+  "tty:${TTY_PATH}" \
+  "$@"