nsqdriver 0.12.16__cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

nsqdriver/NS_DDS_v4.py

@@ -0,0 +1,778 @@
+import copy
+import time
+from enum import Enum
+from math import ceil
+from collections import namedtuple
+from waveforms import Waveform, wave_eval, WaveVStack
+from waveforms.waveform import _zero
+# from waveforms.math.signal import getFTMatrix, shift
+import nsqdriver.nswave as nw
+
+import numpy as np
+
+try:
+    import waveforms
+
+    HAS_WAVEFORMS = True
+except ImportError as e:
+    HAS_WAVEFORMS = False
+
+try:
+    from .common import BaseDriver, Quantity, get_coef
+except ImportError as e:
+
+    class BaseDriver:
+
+        def __init__(self, addr, timeout, **kw):
+            self.addr = addr
+            self.timeout = timeout
+
+
+    class Quantity(object):
+
+        def __init__(self, name: str, value=None, ch: int = 1, unit: str = ''):
+            self.name = name
+            self.default = dict(value=value, ch=ch, unit=unit)
+
+    # def get_coef(*args):
+    #     return '', '', '', ''
+
+DEBUG_PRINT = False
+
+
+def get_coef(coef_info, sampleRate):
+    start, stop = coef_info['start'], coef_info['stop']
+    numberOfPoints = int(
+        (stop - start) * sampleRate)
+    if numberOfPoints % 64 != 0:
+        numberOfPoints = numberOfPoints + 64 - numberOfPoints % 64
+    t = np.arange(numberOfPoints) / sampleRate + start
+
+    fList = []
+    wList = []
+    phases = []
+
+    for kw in coef_info['wList']:
+        Delta, t0, weight, w, phase = kw['Delta'], kw['t0'], kw['weight'], kw['w'], kw['phase']
+        fList.append(Delta)
+
+        if w is not None:
+            w = np.zeros(numberOfPoints, dtype=complex)
+            w[:len(w)] = w
+            w = shift(w, t0 - start)
+            phases.append(np.mod(phase + 2 * np.pi * Delta * start, 2 * np.pi))
+        else:
+            weight = weight
+            if isinstance(weight, np.ndarray):
+                pass
+            else:
+                if isinstance(weight, str):
+                    fun = wave_eval(weight) >> t0
+                elif isinstance(weight, Waveform):
+                    fun = weight >> t0
+                else:
+                    raise TypeError(f'Unsupported type {weight}')
+                weight = fun(t)
+            phase += 2 * np.pi * Delta * start
+            w = getFTMatrix([Delta],
+                            numberOfPoints,
+                            phaseList=[phase],
+                            weight=weight,
+                            sampleRate=sampleRate)[:, 0]
+            phases.append(np.mod(phase, 2 * np.pi))
+        wList.append(w)
+    return np.asarray(wList), fList, numberOfPoints, phases, round((stop - t0) * sampleRate), t
+
+
+def get_demod_envelope(coef_info, demod_map, freq_map, sampleRate):
+    start, stop = coef_info['start'], coef_info['stop']
+    # t0 = coef_info['wList']['t0']
+    # numberOfPoints = int(
+    #     (stop - start) * sampleRate)
+    # if numberOfPoints % 64 != 0:
+    #     numberOfPoints = numberOfPoints + 64 - numberOfPoints % 64
+    # t = np.arange(numberOfPoints) / sampleRate
+    demod_width = 2.048e-6
+    t_p = int(demod_width * sampleRate)
+    t = np.linspace(0, demod_width, round(demod_width*sampleRate), endpoint=False)
+    demod_map_list = demod_map
+    weight_sum = np.zeros((len(freq_map), t_p))
+
+    for idx, weight in enumerate(demod_map_list):
+        if isinstance(weight, np.ndarray):
+            weight_sum[idx] = weight
+        else:
+            if isinstance(weight, str):
+                fun = wave_eval(weight)
+            elif isinstance(weight, Waveform):
+                fun = weight
+            else:
+                raise TypeError(f'Unsupported type {weight}')
+            weight_sum[idx] = fun(t)
+    print(f'{freq_map=}, {len(freq_map)=}')
+    combined_wave = []
+    for idx, freq in enumerate(freq_map):
+        wave = (np.exp(2 * np.pi * freq * t * 1j)).reshape((1, -1))
+        print(f"wave {wave} {weight_sum[idx, :]}")
+        wave = weight_sum[idx, :] * wave
+        # plt.figure()
+        # plt.plot(weight_sum[idx, :])
+        # plt.plot(wave.T)
+        # plt.show()
+        combined_wave.append(wave)
+    combined_wave = np.concatenate(combined_wave, axis=0)
+    return weight_sum, combined_wave
+
+
+@nw.kernel
+def program_cap(param: nw.Var, indelay: nw.Var):
+
+    nw.wait_for_trigger()
+    i: nw.Var
+    # param: [[100e-9, 1e-6], [200e-9, 1e-6]]
+    nw.wait(indelay)
+    for i in param:
+        nw.wait(i[0])
+        nw.capture(i[1], 0, i[1])
+
+
+@nw.kernel
+def program_da(p: nw.Var):
+    i: nw.Var
+    e: nw.Var
+    nw.init_frame(0, 0)
+    nw.wait_for_trigger()
+    # nw.reset_frame()
+    e = nw.ins_envelope(p[0][1])
+    for i in p:
+        nw.wait(i[0])
+        # e = nw.ins_envelope(i[1])
+        nw.play_wave(e, 1, 0, 0)
+
+
+ProbeSegment = namedtuple('ProbeSegment', ['start', 'stop', 'freq', 'demod', 'idx'])
+
+CaptureCmd = namedtuple('CaptureCmd', [
+    'start', 'ad_duration', 'delay', 'da_duration', 'freqs', 'delays', 'demod_wave_list', 'idx_list'
+])
+
+
+class DemodulateMode(str, Enum):
+    MORE_QUBIT = 'more_qubit'
+    COMPLEX_SEQ = 'complex_seq'
+
+
+class Driver(BaseDriver):
+    CHs = list(range(1, 25))
+    segment = ('ns', '111|112|113|114|115')
+    res_map = []
+
+    quants = [
+        Quantity('ReInit', value={}, ch=1),  # set, 设备重新初始化
+        Quantity('Instruction', value=None, ch=1),  # set   参数化波形指令队列配置
+        # 采集运行参数
+        Quantity('Shot', value=1024, ch=1),  # set,运行次数
+        Quantity('PointNumber', value=16384, unit='point'),  # set/get,AD采样点数
+        Quantity('TriggerDelay', value=0, ch=1, unit='s'),  # set/get,AD采样延时
+        Quantity('FrequencyList', value=[], ch=1,
+                 unit='Hz'),  # set/get,解调频率列表，list，单位Hz
+        Quantity('PhaseList', value=[], ch=1,
+                 unit='Hz'),  # set/get,解调频率列表，list，单位Hz
+        Quantity('Coefficient', value=None, ch=1),
+        Quantity('DemodulationParam', value=None, ch=1),
+        Quantity('CaptureMode'),
+        Quantity('StartCapture'),  # set,开启采集（执行前复位）
+        Quantity('TraceIQ', ch=1),  # get,获取原始时域数据
+        # 返回：array(shot, point)
+        Quantity('IQ', ch=1),  # get,获取解调后数据,默认复数返回
+        # 系统参数，宏定义修改，open时下发
+        # 复数返回：array(shot,frequency)
+        # 实数返回：array(IQ,shot,frequency)
+
+        # 任意波形发生器
+        Quantity('Waveform', value=np.array([]), ch=1),  # set/get,下发原始波形数据
+        Quantity('Delay', value=0, ch=1),  # set/get,播放延时
+        Quantity('KeepAmp', value=0
+                 ),  # set, 电平是否维持在波形最后一个值, 0：波形播放完成后归0，1：保持波形最后一个值，2:保持波形第一个值
+        Quantity('Biasing', value=0, ch=1),  # set, 播放延迟
+        Quantity('LinSpace', value=[0, 30e-6, 1000],
+                 ch=1),  # set/get, np.linspace函数，用于生成timeline
+        Quantity('Output', value=True, ch=1),  # set/get,播放通道开关设置
+        Quantity('GenWave', value=None,
+                 ch=1),  # set/get, 设备接收waveform对象，根据waveform对象直接生成波形
+        # set/get, 设备接收IQ分离的waveform对象列表，根据waveform对象列表直接生成波形
+        Quantity('GenWaveIQ', value=None, ch=1),
+        Quantity('MultiGenWave', value={1: np.ndarray([])}),  # 多通道波形同时下发
+        Quantity('EnableWaveCache', value=False),  # 是否开启waveform缓存
+        Quantity('PushWaveCache'),  # 使waveform缓存中的波形数据生效
+        # 混频相关配置
+        Quantity('EnableDAMixer', value=False, ch=1),  # DA通道混频模式开关
+        Quantity('MixingWave', ),  # 修改完混频相关参数后，运行混频器
+        Quantity('DAIQRate', value=1e9, ch=1),  # 基带信号采样率
+        Quantity('DALOFreq', value=100e6, ch=1),  # 中频信号频率
+        Quantity('DALOPhase', value=0, ch=1),  # 基带信号相位，弧度制
+        Quantity('DASideband', value='lower', ch=1),  # 混频后取的边带
+        Quantity('DAWindow', value=None, ch=1),
+        # 基带信号升采样率时所使用的窗函数，默认不使用任何窗，
+        # 可选：None、boxcar、triang、blackman、hamming、hann、bartlett、flattop、parzen、bohman、blackmanharris、nuttall、
+        # barthann、cosine、exponential、tukey、taylor
+
+        # 内触发
+        Quantity('GenerateTrig', value=1e7,
+                 unit='ns'),  # set/get,触发周期单位ns，触发数量=shot
+        Quantity('UpdateFirmware', value='', ch=1),  # qsync固件更新
+        Quantity('PipInstall'),  # pip install in instance
+        Quantity('Timeout'),
+    ]
+
+    def __init__(self, addr: str = '', timeout: float = 20.0, **kw):
+        super().__init__(addr, timeout=timeout, **kw)
+        self.handle = None
+        self.model = 'NS_MCI'  # 默认为设备名字
+        self.srate = 8e9
+        self.ad_srate = 4e9
+        self.addr = addr
+        self.timeout = timeout
+        self.chs = set()  # 记录配置过的ch通道
+        self.IQ_cache = {}
+        self.coef_cache = {}
+        self.res_maps = {}
+        self.demod_maps = {}
+        self.probe_da_wave = {}
+        self.programout_para = {} # {ch : para}
+        self.programin_para = {}
+        self.programin_para_indelay = {i: 136e-9 for i in range(1, 13)}
+        # self.probe_delay = 32e-9
+        self.probe_delay = 0
+        self.capture_cmds: "dict[int, list[CaptureCmd]]" = {}
+        self.capture_cali_param: "dict[int, np.ndarray]" = {}
+        self.capture_points: "dict[int, np.ndarray]" = {}
+        self.demodulate_mode = DemodulateMode.MORE_QUBIT
+        self.demode_calculus: "dict[int, np.ndarray]" = {}
+
+    def open(self, **kw):
+        """
+        输入IP打开设备，配置默认超时时间为5秒
+        打开设备时配置RFSoC采样时钟，采样时钟以参数定义
+        """
+        from nsqdriver import MCIDriver
+
+        DArate = 8e9
+        ADrate = 4e9
+        sysparam = {
+            "MixMode": 2,
+            "RefClock": "out",
+            "DArate": DArate,
+            "ADrate": ADrate,
+            "CaptureMode": 0,
+            "INMixMode": 2,  # 4～6 GHz 取 1， 6 ～ 8 GHz 取 2
+        }
+        sysparam.update(kw.get('system_parameter', {}))
+        print(f"{self.timeout=}")
+        device = MCIDriver(self.addr, self.timeout)
+        device.open(system_parameter=sysparam)
+        self.handle = device
+
+    def granularity4ns(self, delay):
+        # points_4ns = 16  # self.ad_srate*4e-6
+        return delay // 4 * 4
+
+    @staticmethod
+    def _delay2_phase(delay, freq):
+        return 2 * np.pi * freq * (delay * 1e-9)
+
+    def in_sequence_in_time(self, coef_info: dict) -> list[CaptureCmd]:
+        """
+        合并重叠项，取并集，记录合并延迟时间，合并频点，合并包络
+        """
+
+        w_list = coef_info.get('wList', [])
+        time_segments: "list[ProbeSegment]" = []
+
+        for idx, wave in enumerate(w_list):
+            t0 = int(round(wave['t0'] * 1e9))
+            weight_expr = wave['weight']
+
+            # 假设 weight 表达式格式为 "square(X) >> Y"，我们提取实际时间宽度
+            # duration = float(weight_expr.split('>>')[1].strip())
+            _start, _stop, _ = wave_eval(weight_expr).bounds
+            _start, _stop = int(round(_start * 1e9)), int(round(_stop * 1e9))
+
+            # 将区间加入列表
+            seg = ProbeSegment(t0 + _start, t0 + _stop, wave['Delta'], weight_expr, idx)
+            time_segments.append(seg)
+
+        # 按起始时间排序
+        time_segments.sort()
+
+        # 结果存储
+        non_overlapping_segments: list[CaptureCmd] = []
+        current_start, current_end = time_segments[0].start, time_segments[0].stop
+        current_cmd = CaptureCmd(0, 0, 0, 0, [time_segments[0].freq], [0.], [time_segments[0].demod],
+                                 [time_segments[0].idx])
+        pointer = 0
+        for seg in time_segments[1:]:
+            if seg.start > current_end:
+                # 如果不重叠，保存当前段并移动到下一段
+                if pointer == 0:
+                    current_cmd = current_cmd._replace(start=current_start)
+                else:
+                    current_cmd = current_cmd._replace(start=current_start - self.probe_delay)
+                current_cmd = current_cmd._replace(ad_duration=current_end - current_start)
+                current_cmd = current_cmd._replace(delay=self.probe_delay)
+                current_cmd = current_cmd._replace(da_duration=current_end - current_start)
+                non_overlapping_segments.append(current_cmd)
+
+                current_cmd = CaptureCmd(0, 0, 0, 0, [seg.freq], [0.], [seg.demod], [seg.idx])
+                pointer = current_end
+                current_start, current_end = seg.start, seg.stop
+            else:
+                # 如果有重叠，扩展当前段
+                current_end = max(current_end, seg.stop)
+                current_cmd.idx_list.append(seg.idx)
+                current_cmd.freqs.append(seg.freq)
+                current_cmd.demod_wave_list.append(seg.demod)
+                # 由delay换算解缠绕相位
+                current_cmd.delays.append(seg.start - current_start)
+            print(f'{current_cmd=}')
+        else:
+            # 添加最后一个段
+            current_cmd = current_cmd._replace(start=current_start - self.probe_delay)
+            current_cmd = current_cmd._replace(ad_duration=current_end - current_start)
+            current_cmd = current_cmd._replace(delay=self.probe_delay)
+            current_cmd = current_cmd._replace(da_duration=current_end - current_start)
+            non_overlapping_segments.append(current_cmd)
+        return non_overlapping_segments
+
+    def generate_in_program(self, coef_info, ch):
+        freq_map = []
+        demod_wave_map = []
+        seq_param = []
+
+        self.capture_cmds[ch] = seq = self.in_sequence_in_time(coef_info) # 得到合并重叠后的list
+        print(f'{seq=}')
+        # for segment in seq:
+        #     demod_wave_map.extend(segment.demod_wave_list)
+        #     demod_wave_map = list(set(demod_wave_map))
+        #     freq_map.extend(segment.freqs)
+        #     freq_map = list(set(freq_map))
+
+        for segment in seq:
+            for n, f in enumerate(segment.freqs):
+                if f not in freq_map:
+                    demod_wave_map.append(segment.demod_wave_list[n])
+                    freq_map.append(f)
+
+        _t_end = 0
+        res_map = [[]] * len(coef_info['wList'])
+        phase_map = [0] * len(coef_info['wList'])
+        points_map = [0] * len(coef_info['wList'])
+        for cap_num, segment in enumerate(seq):
+            _align_start = self.granularity4ns(segment.start)  # 向前取整
+            _start_diff = segment.start - _align_start
+            # _align_end = ceil((segment.start + segment.ad_duration) / 4) * 4  # 向上取整
+            _align_end = (segment.start + segment.ad_duration) // 4 * 4  # 向上取整
+            seq_param.append([
+                (_align_start - _t_end)  * 1e-9,
+                (_align_end - _align_start) * 1e-9,
+                segment.delay * 1e-9,
+                (_align_end - _align_start)  * 1e-9,
+            ])
+            print(f"{_align_start=} {_align_end=} {(_align_start - _t_end)} {_t_end=}")
+            _t_end = _align_end
+            for idx, delay, freq, demod_wave in zip(segment.idx_list, segment.delays, segment.freqs,
+                                                    segment.demod_wave_list):
+                res_map[idx] = [freq_map.index(freq), cap_num]
+                # print("下面 +  t0")
+                # phase_map[idx] = self._delay2_phase(delay + _start_diff, freq)  # 向前取整的缩进加上起始时间的差值来计算相位
+                # phase_map[idx] = self._delay2_phase(_align_start + _start_diff, freq)  # 向前取整的缩进加上起始时间的差值来计算相位
+                # phase_map[idx] = self._delay2_phase(0, freq)  # 向前取整的缩进加上起始时间的差值来计算相位
+                points_map[idx] = (_align_end - _align_start) * 1e-9 * self.ad_srate
+                # points_map[idx] = segment.ad_duration * 1e-9 * self.ad_srate
+
+        ad_abs_end = 0
+        da_abs_end = 0
+        # 根据ad 的延迟重新下发da program
+        delta_t = self.programout_para[ch][0][0] - seq_param[0][0]
+        for n, i in enumerate(self.programout_para[ch]):
+            if n == 0:
+                continue
+            ad_abs_end += seq_param[n-1][0] + seq_param[n-1][1]
+            da_abs_end += self.programout_para[ch][n-1][0] + self.programout_para[ch][n-1][1].shape[0] / self.srate
+            da_next_start = ad_abs_end + seq_param[n][0] + delta_t
+            da_wait = da_next_start - da_abs_end
+            i[0] = da_wait
+        # kernel_da = program_da(self.programout_para[ch])
+        # self.handle.set("ProgramOUT", kernel_da, ch)
+        print(f"重下da 程序 {self.programout_para[ch]=}")
+        
+        for idx, freq in zip(segment.idx_list, segment.freqs):
+            phase_map[idx] = self._delay2_phase(0 , freq)
+
+        self.res_maps[ch] = res_map
+        self.capture_cali_param[ch] = np.exp(-1j * np.array(phase_map)).reshape((-1, 1))
+        self.capture_points[ch] = np.array(points_map).reshape((-1, 1))
+        print(f"{seq_param=} para_angle {np.angle(self.capture_cali_param[ch], deg=True)} {self.capture_points}")
+        self.programin_para[ch] = seq_param
+        return program_cap(seq_param, self.programin_para_indelay[ch]), freq_map, demod_wave_map
+
+    def out_sequence_in_time(self, wave_list: list):
+        last_start = wave_list[0][0]
+        last_stop = wave_list[0][1]
+        temp_w = [wave_list[0][2]]
+        _res = []
+
+        for idx, (start, stop, seg) in enumerate(wave_list[1:]):
+            if start > last_stop:
+                _res.append([last_start, last_stop, np.hstack(temp_w)])
+                last_start = start
+                last_stop = stop
+                temp_w.clear()
+                temp_w.append(seg)
+            else:
+                last_stop = max(last_stop, stop)
+                temp_w.append(seg)
+        else:
+            _res.append([last_start, last_stop, np.hstack(temp_w)])
+        return _res
+
+    def gen_wave_frag(self, x, wave: "Waveform"):
+        range_list = np.searchsorted(x, wave.bounds)
+        # ret = np.zeros_like(x) 
+        ret = []
+        start, stop = 0, 0
+        for i, stop in enumerate(range_list):
+            if start < stop and wave.seq[i] != _zero:
+                _w = copy.deepcopy(wave)
+                _w.start = start / self.srate
+                _w.stop = stop / self.srate
+                part = _w.sample(self.srate)
+                part = part if part is None else part[:(stop - start)]
+                ret.append((start, stop, part))
+            start = stop
+        else:
+            if not ret:
+                ret.append((0, 128, np.zeros((128,))))
+        return ret
+
+    def generate_out_program(self, _wave, ch):
+        align_points = 32  # 4ns*8e9
+        if isinstance(_wave, WaveVStack):
+            _wave = _wave.simplify()
+        if len(_wave.seq) == 1 and _wave.seq[0] == _zero:
+            wave_list = [(0, 128, np.zeros((128,)))]
+        else:
+            _wave.stop = _wave.bounds[-2]
+            wave_list = self.gen_wave_frag(
+                np.linspace(_wave.start, _wave.stop, int((_wave.stop - _wave.start) * self.srate)), _wave)
+        print(f'generate_out_program: {_wave.start=}, {_wave.stop=}, {len(wave_list)=}, {ch=}')
+        _t_end = 0
+        para = []
+        wave = self.out_sequence_in_time(wave_list)  # 得到合并重叠后的list
+
+        for num, i in enumerate(wave):
+            wait = (i[0] - _t_end)
+            # if wait % 32 != 0:
+            #     # 若wait 不是4ns整倍数，根据ad的逻辑会往后多财季4ns 
+            align_wait = wait // align_points * align_points
+            zero_num = wait - align_wait
+            align_end = i[1] // align_points * align_points
+            align_wave = [i[2],]
+            para.append([align_wait / self.srate, np.hstack(align_wave)])
+            _t_end = align_end
+        print(f"out {para=}")
+        self.programout_para[ch] = para
+        # print(para[0][1].max())
+        # plt.plot(para[0][1])
+        # plt.show()
+        return program_da(para)
+
+    def get_coef_res(self, iq_res, ch):
+        res = []
+        print(f'{self.res_maps[ch]=}')
+        for (freq_num, cap_num) in self.res_maps[ch]:
+            res.append(iq_res[freq_num][cap_num::len(self.capture_cmds[ch])])
+        # 采样点归一化
+        res = np.array(res) / self.demode_calculus[ch]
+        # 校准相位
+        res *= self.capture_cali_param[ch]
+
+        return res
+
+    def close(self, **kw):
+        """
+        关闭设备
+        """
+        if getattr(self, 'handle', None) is not None:
+            self.handle.close()
+            self.handle = None
+
+    def set(self, *args, **kwargs):
+        return self.handle.set(*args, **kwargs)
+
+    def get(self, *args, **kwargs):
+
+        return self.handle.get(*args, **kwargs)
+
+    # def generate_demo(self, co):
+    #     _wave = wf.zero()
+    #     min_t0 = 10
+    #     for _w in co['wList']:
+    #         t0 = _w['t0']
+    #         min_t0 = min(min_t0, t0)
+    #         _wave += (wf.wave_eval(_w['weight']) * wf.cos(2 * np.pi * _w['Delta'])) >> t0
+    #     _wave /= 8
+    #     # _wave = _wave << 50e-9
+    #     _wave.start = 0
+    #     _wave.stop = co['stop']
+    #     bk = self.srate
+    #     self.srate = self.ad_srate
+    #     _, para = self.generate_out_program(co, 1)
+    #     self.srate = bk
+    #     demo = para[0][1]
+    #     return demo
+
+    def write(self, name: str, value, **kw):
+        channel = kw.get('ch', 1)
+        print(f'NS_DDS_v3 write: {name=}, {channel=}')
+        if name in {'Coefficient'}:
+            print("Coefficient" * 3)
+            coef_info = value
+            self.chs.add(channel)
+            kernel, freq_map, demod_wave_map = self.generate_in_program(coef_info, channel)
+            self.handle.set("ProgramIN", kernel, channel)
+            demode_weight, demode_wave = get_demod_envelope(coef_info, demod_wave_map, freq_map, sampleRate=4e9)
+            self.demode_calculus[channel] = np.sum(demode_weight[0])
+            self.handle.set("DemodulationParam", demode_wave, channel)
+            # print(f"demode_wave {demode_wave}")
+            # plt.figure()
+            # plt.plot(demode_wave[0].real)
+            # plt.plot(demode_wave[0].imag)
+            # plt.show()
+            self.handle.set('TimeWidth', self.capture_points[channel].max()/self.ad_srate, channel)
+            # self.handle.set('TimeWidth', 87 / self.ad_srate, channel)
+            # self.handle.set("FreqList", freq_map, channel)
+            self.coef_cache.update({channel: coef_info})
+        elif name in {"TriggerDelay", "INDelay"}:
+            print("INDelay" * 3)
+            self.programin_para_indelay[channel] = value
+            # kernel = program_cap(self.programin_para[channel], self.programin_para_indelay[channel])
+            # self.handle.set("ProgramIN", kernel, channel)
+        elif name in {
+            'CaptureMode', 'SystemSync', 'ResetTrig', 'TrigPeriod',
+            'TrigFrom'
+        }:
+            pass
+        elif name in {
+            'GenWave', 'Waveform'
+        } and isinstance(value, Waveform):
+            kernel_da = self.generate_out_program(value, channel)
+            # self.handle.set("ProgramOUT", kernel_da, channel)
+        elif name in {
+            'StartCapture', 'Capture'
+        }:
+            for channel, param in self.programin_para.items():
+                kernel = program_cap(self.programin_para[channel], self.programin_para_indelay[channel])
+                self.handle.set("ProgramIN", kernel, channel)
+            for channel in self.programout_para:
+                kernel = program_da(self.programout_para[channel])
+                self.handle.set("ProgramOUT", kernel, channel)
+            print(f"{self.programin_para=} {self.programin_para_indelay=} {self.programout_para=}")
+            return self.handle.set(name, value)
+        else:
+            if name in {"Shot"}:
+                self.shots = value
+            return self.handle.set(name, value, channel)
+
+    def read(self, name: str, **kw):
+        channel = kw.get('ch', 1)
+        if name in {"IQ"}:
+            iq_res = self.handle.get(
+                "IQ", channel, round(self.shots * len(self.capture_cmds[channel]))
+            )
+            result = self.get_coef_res(iq_res, channel).T
+            if len(self.chs) != 0 and channel in self.chs:
+                self.chs.remove(channel)
+            # self.IQ_cache.update({channel: result})
+            if len(self.chs) == 0:
+                self.write("TerminateUpload", 1)  # 实验的开始必须加此句话
+        elif name in {'TraceIQ'}:
+            print(f"{self.shots=} {len(self.capture_cmds[channel])=}")
+            result = self.handle.get(
+                "TraceIQ", channel, round(self.shots * len(self.capture_cmds[channel]))
+            )
+        else:
+            result = self.handle.get(name, channel)
+        return result
+
+
+if __name__ == '__main__':
+    # 7.052186177715091e9 1.418e-6 7.062146892655367e9 1.416e-6 6.191950464396285e9 1.615e-6
+    # 6.188118811881188e9 1.616e-6
+    # 6.184291898577612e9 1.617e-6
+    # 6.180469715698393e9 1.618e-6
+    # 6.176652254478073e9 1.619e-6
+    co = {'start': 0.0, 'stop': 70.605e-06, 'wList': [
+        {'Delta': 6967500000.0, 'phase': -0.0, 'weight': 'square(0.8e-06)>>(4e-07)', 'window': (0, 1024), 'w': None,
+         't0': 1.618e-6, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 4.176652254478073e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1.618e-6, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 6.180469715698393e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1.618e-6 * 20, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 4.176652254478073e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1.618e-6 * 20, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 6.180469715698393e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1.618e-6 * 40, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 6.176652254478073e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1.618e-6 * 40, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 5.12311e9, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1/5.12311 * 1e-4 * 2, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 5.2231e9, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 1/5.12311 * 1e-4 * 2, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 5.1e9, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 6.401e-06, 'phi': 2.1739656328752264, 'threshold': 20.36802101135254},
+        # {'Delta': 5.2e9, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 6.401e-06, 'phi': 1.851749364542847, 'threshold': 21.65827751159668},
+        # {'Delta': 1e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 5.5e-06, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 1.1e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 5.5e-06, 'phi': 2.1739656328752264, 'threshold': 20.36802101135254},
+        # {'Delta': 1.2e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 5.5e-06, 'phi': 1.851749364542847, 'threshold': 21.65827751159668},
+        # {'Delta': 1e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 7.805e-06, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 1e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 8.805e-06, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+        # {'Delta': 1.1e9, 'phase': -0.0, 'weight': 'square(8e-07) >> 4e-07', 'window': (0, 1024), 'w': None,
+        #  't0': 9.005e-06, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926}
+    ]}
+    co = {'start': 5.760000000000001e-07, 'stop': 1.581e-06, 'wList': [{'Delta': 6967500000.0, 'phase': -0.0, 'weight': '(sin(3141592.6535897935)**3)*(square(1e-06)>>(5e-07))',
+                                                                         'window': (0, 1024), 'w': None, 't0': 5.81e-07, 'phi': -0.16222877291938465, 'threshold': 0.4922424554824829}]}
+    co = {'start': 5.760000000000001e-07, 'stop': 9.310000000000001e-07, 'wList': [{'Delta': 6967500000.0, 'phase': -0.0, 'weight': '(sin(8975979.010256553)**3)*(square(3.5e-07)>>(1.75e-07))', 'window': (0, 1024), 'w': None, 't0': 5.81e-07,
+                                                                                     'phi': 2.4011441876721005, 'threshold': 3.5368497371673584}]}
+    co = {'start': 5.760000000000001e-07, 'stop': 9.310000000000001e-07, 'wList': [{'Delta': 6967500000.0, 'phase': -0.0, 'weight': '(sin(8975979.010256553)**3)*(square(3.5e-07)>>(1.74e-07))', 'window': (0, 1024), 'w': None, 't0': 5.81e-07,
+                                                                                     'phi': 2.4011441876721005, 'threshold': 3.5368497371673584}]}
+
+    import numpy as np
+    # from nsqdriver.NS_DDS_v3_2 import Driver, get_coef
+    from nsqdriver import QSYNCDriver
+    from nsqdriver.NS_MCI import SHARED_DEVICE_MEM
+    import matplotlib.pyplot as plt
+    import waveforms as wf
+
+    SHARED_DEVICE_MEM.clear_ip()
+    _d = Driver('192.168.0.229', 50)
+    _q = QSYNCDriver('192.168.0.229')
+    _q.open(system_parameter={'RefClock': 'in'})
+    _d.open(system_parameter={'MixMode': 2, 'CaptureMode': 0, "DArate": 8e9, "RefClock": "out"})
+    _q.sync_system()
+    time.sleep(2)
+    _wave = wf.zero()
+    min_t0 = 10
+    for _w in co['wList']:
+        t0 = _w['t0']
+        min_t0 = min(min_t0, t0)
+        _wave += (wf.wave_eval(_w['weight']) * wf.cos(2 * np.pi * _w['Delta'])) >> t0
+    _wave /= 8
+    # _wave = _wave << 50e-9
+    _wave.start = 0
+    _wave.stop = co['stop']
+
+    _wave(np.linspace(0, _wave.stop, int(_wave.stop * 8e9)), frag=True)
+
+    wave = _wave.sample(8e9)
+    # plt.figure()
+    # plt.plot(wave)
+    # plt.show()
+    ch = 1
+    _wave.start = 0
+    _wave.stop = co["stop"]
+    shots = 8192
+
+    _q.set('Shot', shots)
+    _d.write('Shot', shots)
+
+
+# 测试解模数据
+
+    # _d.write("INDelay", 136e-9, ch=ch) # INDelay 要在 Coefficient前面
+    # _d.write("Coefficient", co, ch=ch)
+    # _d.write("GenWave", _wave, ch=ch)
+
+    # _d.set('StartCapture')
+    # _q.set('GenerateTrig', 90e-6)
+    # data = _d.read("IQ", ch=ch)
+    # print(f"angle= {np.angle(data.mean(axis=0), deg=True)}")
+    # print(f"abs= {np.abs(data.mean(axis=0))}")
+
+# 测试原始数据
+#   将采集间隔改大
+    _d.set('CaptureMode', 1)
+    _d.write("GenWave", _wave, ch=ch)
+    _d.write("Coefficient", co, ch=ch) # 获取原始数据也要下发，用于配置indelay
+    _d.write("INDelay", 136e-9, ch=ch) # INDelay 要在 Coefficient前面
+
+    _d.write('StartCapture', 1)
+    _q.set('GenerateTrig', 500e-6)
+    data = _d.read("TraceIQ", ch=ch)
+    data = data.reshape((shots, -1))
+    plt.figure()
+    plt.plot(data.mean(axis=0))
+    plt.show()
+
+# 测试波形连续播放
+#   修改波形频率，在示波器可见
+
+    # _q.set("Shot", 0xFFFFFFFF)
+    # _q.set("GenerateTrig", 500e-6)
+    # time.sleep(10)
+    # _q.set("ResetTrig")
+
+# 带包络扫描S21
+    # freq_range = np.linspace(4.01e9, 6.01e9, 51)
+    # s21_res = []
+    # _d.write("INDelay", 136e-9, ch=ch) # INDelay 要在 Coefficient前面
+    # for f in freq_range:
+    #     co = {'start': 0.0, 'stop': 70.605e-06, 'wList': [
+    #         {'Delta': f, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+    #         't0': 1.618e-6, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+    #         {'Delta': 6.176652254478073e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+    #         't0': 1.618e-6, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+    #         {'Delta': f, 'phase': -0.0, 'weight': 'square(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+    #         't0': 1.618e-6 * 20, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+    #         {'Delta': 6.176652254478073e9, 'phase': -0.0, 'weight': 'gaussian(0.8e-06) >> 4e-07', 'window': (0, 1024), 'w': None,
+    #         't0': 1.618e-6 * 20, 'phi': 2.4311851282940524, 'threshold': 9.645718574523926},
+    #         ]}
+    #     _wave = wf.zero()
+    #     for _w in co['wList']:
+    #         t0 = _w['t0']
+    #         min_t0 = min(min_t0, t0)
+    #         _wave += (wf.wave_eval(_w['weight']) * wf.cos(2 * np.pi * _w['Delta'])) >> t0
+    #     _wave /= 8
+    #     # _wave = _wave << 50e-9
+    #     _wave.start = 0
+    #     _wave.stop = co['stop']
+    #     # plt.figure()
+    #     # plt.plot(_wave.sample(8e9))
+    #     # plt.show()
+    #     _d.write("Coefficient", co, ch=ch)
+    #     _d.write("GenWave", _wave, ch=ch)
+
+    #     _d.set('StartCapture')
+    #     _q.set('GenerateTrig', 90e-6)
+    #     data = _d.read("IQ", ch=ch)
+    #     s21_res.append(data)
+    # # 取第一次采集的第一个频点画图
+    # cap1 = np.array(s21_res)
+    # cap1 = cap1[:, :, 0].mean(axis=1)
+    # cap1 = 20 * np.log10(np.abs(cap1))
+    # plt.figure()
+    # plt.plot(freq_range, cap1)
+    # plt.show()
+    # # 取第二次采集的第一个频点画图
+    # cap2 = np.array(s21_res)
+    # cap2 = cap2[:, :, 2].mean(axis=1)
+    # cap2 = 20 * np.log10(np.abs(cap2))
+    # plt.figure()
+    # plt.plot(freq_range, cap2)
+    # plt.show()