Skip to content
GitLab
Commit acf68183 authored by Guillaume Blanchy's avatar Guillaume Blanchy
Browse files

Merge branch 'code_refactor' of https://gitlab.irstea.fr/reversaal/OhmPi into code_refactor

parents d45b5938 4cf39502
master 144-create-testing-framework-for-tests-automation code_refactor dev/dummy v2024_rc v2025_alpha v2024.0.35 v2024.0.34 v2024.0.33 v2024.0.32 v2024.0.31 v2024.0.30 v2024.0.29 v2024.0.28 v2024.0.27 v2024.0.26 v2024.0.25 v2024.0.24 v2024.0.23 v2024.0.22 v2024.0.21 v2024.0.20 v2024.0.19 v2024.0.18 v2024.0.17 v2024.0.16 v2024.0.15 v2024.0.14 v2024.0.13 v2024.0.12 v2024.0.11 v2024.0.10 v2024.0.9 v2024.0.8 v2024.0.7 v2024.0.0
No related merge requests found
Hide whitespace changes
Inline Side-by-side
Showing
with 249 additions and 69 deletions
+249 -69
configs/config_mb_2024_0_2__4_mux_2023_dps5005.py
+ 1
− 1
View file @ acf68183
......@@ -50,7 +50,7 @@ HARDWARE_CONFIG = {
'roles': {'M': 'X'},
'cabling': {(i, j): ('mux_M', i) for j in ['M'] for i in range(1, 65)},
'voltage_max': 12.},
'mux_N':
'mux_N':
{'model': 'mux_2023_0_X',
'mux_tca_address': 0x73,
'roles': {'N': 'X'},
......
dev/test_mb_2024_4_mux_2024.py
+ 7
− 7
View file @ acf68183
......@@ -8,8 +8,8 @@ import logging
from ohmpi.config import HARDWARE_CONFIG
stand_alone = False
part_of_hardware_system = True
within_ohmpi = False
part_of_hardware_system = False
within_ohmpi = True
# Stand alone
if stand_alone:
......@@ -105,14 +105,14 @@ if within_ohmpi:
# k._hw.switch_mux([A, B, M, N], state='off')
# print(f'OhmPiHardware Resistance: {k._hw.last_rho :.2f} ohm, dev. {k._hw.last_dev:.2f} %, rx bias: {k._hw.rx._bias:.2f} mV')
# k._hw._plot_readings()
k.load_sequence('sequences/9991_GRAD_16_s1_a1.txt')
k.run_sequence(tx_volt=5, injection_duration=1., nb_stack=2, duty_cycle=0.5)
# k.load_sequence('sequences/9991_GRAD_16_s1_a1.txt')
# k.run_sequence(tx_volt=5., injection_duration=1., nb_stack=2, duty_cycle=0.5)
print('using OhmPi')
#d = k.run_measurement([A, B, M, N], injection_duration=1., nb_stack=2, duty_cycle=0.5)
d = k.run_measurement([A, B, M, N], injection_duration=1., nb_stack=2, duty_cycle=0.5)
# print(d)
# k._hw._plot_readings()
k._hw._plot_readings()
print(f'OhmPiHardware: Resistance: {k._hw.last_resistance() :.2f} ohm, dev. {k._hw.last_dev():.2f} %, sp: {k._hw.sp:.2f} mV, rx bias: {k._hw.rx._bias:.2f} mV')
print(f'OhmPi: Resistance: {d["R [ohm]"] :.2f} ohm, dev. {d["R_std [%]"]:.2f} %, rx bias: {k._hw.rx._bias:.2f} mV')
print(f'OhmPi: Resistance: {d["R [Ohm]"] :.2f} ohm, dev. {d["R_std [%]"]:.2f} %, rx bias: {k._hw.rx._bias:.2f} mV')
# k._hw._plot_readings(save_fig=False)
# plot_exec_log('ohmpi/logs/exec.log')
change_config('../configs/config_default.py', verbose=False)
ohmpi/hardware_components/abstract_hardware_components.py
+ 26
− 21
View file @ acf68183
......@@ -122,6 +122,7 @@ class PwrAbstract(ABC):
else:
self.voltage = self._voltage_min
class MuxAbstract(ABC):
def __init__(self, **kwargs):
self.model = kwargs.pop('model', 'unknown MUX hardware')
......@@ -279,7 +280,7 @@ class TxAbstract(ABC):
self.pwr = kwargs.pop('pwr', None)
self._polarity = 0
self._injection_duration = None
self._adc_gain = 1.
self._gain = 1.
self.injection_duration = injection_duration
self._latency = kwargs.pop('latency', 0.)
self.tx_sync = kwargs.pop('tx_sync', Event())
......@@ -287,26 +288,23 @@ class TxAbstract(ABC):
self._pwr_state = 'off'
@property
def adc_gain(self):
return self._adc_gain
def gain(self):
return self._gain
@adc_gain.setter
def adc_gain(self, value):
@gain.setter
def gain(self, value):
"""
Set gain on RX ADC
Set gain on RX
Parameters
----------
value: float
"""
self._adc_gain = value
self.exec_logger.debug(f'Setting TX ADC gain to {value}')
self._gain = value
self.exec_logger.debug(f'Setting TX gain to {value}')
@abstractmethod
def _adc_gain_auto(self):
pass
@abstractmethod
def current_pulse(self, **kurwargs):
def current_pulse(self, **kwargs):
pass
@abstractmethod
......@@ -403,6 +401,10 @@ class TxAbstract(ABC):
self._pwr_state = 'off'
self.exec_logger.debug(f'{self.model} cannot switch off power source')
def reset_gain(self):
self.gain = 1.
class RxAbstract(ABC):
def __init__(self, **kwargs):
self.exec_logger = kwargs.pop('exec_logger', None)
......@@ -416,31 +418,34 @@ class RxAbstract(ABC):
self._sampling_rate = kwargs.pop('sampling_rate', 1) # ms
self.exec_logger.debug(f'{self.model} RX initialization')
self._voltage_max = kwargs.pop('voltage_max', 0.)
self._adc_gain = 1.
self._gain = 1.
self._max_sampling_rate = np.inf
self._latency = kwargs.pop('latency', 0.)
self._bias = kwargs.pop('bias', 0.)
self._vmn_hardware_offset = kwargs.pop('vmn_hardware_offset', 0.)
@property
def adc_gain(self):
return self._adc_gain
def gain(self):
return self._gain
@adc_gain.setter
def adc_gain(self, value):
@gain.setter
def gain(self, value):
"""
Sets gain on RX ADC
Sets gain on RX
Parameters
----------
value: float
"""
self._adc_gain = value
self.exec_logger.debug(f'Setting RX ADC gain to {value}')
self._gain = value
self.exec_logger.debug(f'Setting RX gain to {value}')
@abstractmethod
def _adc_gain_auto(self):
def gain_auto(self):
pass
def reset_gain(self):
self.gain = 1.
@property
def sampling_rate(self):
return self._sampling_rate
......
ohmpi/hardware_components/mb_2023_0_X.py
+ 11
− 6
View file @ acf68183
......@@ -31,7 +31,7 @@ SPECS = {'rx': {'model': {'default': os.path.basename(__file__).rstrip('.py')},
'mcp_address': {'default': 0x20},
'ads_address': {'default': 0x48},
'compatible_power_sources': {'default': ['pwr_batt', 'dps5005']},
'r_shunt': {'min': 0., 'default': 2. },
'r_shunt': {'min': 0., 'default': 2.},
'activation_delay': {'default': 0.005}, # Max turn on time of 211EH relays = 5ms
'release_delay': {'default': 0.001}, # Max turn off time of 211EH relays = 1ms
}}
......@@ -91,7 +91,8 @@ class Tx(TxAbstract):
# ADS1115 for current measurement (AB)
self._ads_current_address = kwargs['ads_address']
self._ads_current_data_rate = kwargs['data_rate']
self._ads_current = ads.ADS1115(self.connection, gain=self.adc_gain, data_rate=self._ads_current_data_rate,
self._adc_gain = 2 / 3
self._ads_current = ads.ADS1115(self.connection, gain=self._adc_gain, data_rate=self._ads_current_data_rate,
address=self._ads_current_address)
self._ads_current.mode = Mode.CONTINUOUS
self.r_shunt = kwargs['r_shunt']
......@@ -116,7 +117,7 @@ class Tx(TxAbstract):
def gain(self, value):
assert value in [2/3, 2, 4, 8, 16]
self._adc_gain = value
self._ads_current = ads.ADS1115(self.connection, gain=self.adc_gain,
self._ads_current = ads.ADS1115(self.connection, gain=self._adc_gain,
data_rate=SPECS['tx']['data_rate']['default'],
address=self._ads_current_address)
self._ads_current.mode = Mode.CONTINUOUS
......@@ -182,12 +183,12 @@ class Tx(TxAbstract):
@property
def tx_bat(self):
if np.isnan(self.tx.pwr.battery_voltage):
if np.isnan(self.pwr.battery_voltage):
self.soh_logger.warning(f'Cannot get battery voltage on {self.model}')
self.exec_logger.debug(f'{self.model} cannot read battery voltage. Returning default battery voltage.')
return self.pwr.voltage
else:
return self.tx.pwr.battery_voltage
return self.pwr.battery_voltage
def voltage_pulse(self, voltage=None, length=None, polarity=1):
......@@ -248,7 +249,7 @@ class Rx(RxAbstract):
def gain(self, value):
assert value in [2/3, 2, 4, 8, 16]
self._adc_gain = value
self._ads_voltage = ads.ADS1115(self.connection, gain=self.adc_gain,
self._ads_voltage = ads.ADS1115(self.connection, gain=self._adc_gain,
data_rate=SPECS['rx']['data_rate']['default'],
address=self._ads_voltage_address)
self._ads_voltage.mode = Mode.CONTINUOUS
......@@ -263,6 +264,10 @@ class Rx(RxAbstract):
def gain_auto(self):
self._adc_gain_auto()
def reset_gain(self):
self.gain = 2/3
@property
def voltage(self):
""" Gets the voltage VMN in Volts
......
ohmpi/hardware_components/mb_2024_0_2.py
+ 3
− 0
View file @ acf68183
......@@ -192,6 +192,9 @@ class Rx(Rx_mb_2023):
self._dg411_gain_auto()
self.exec_logger.debug(f'Setting RX gain automatically to {self.gain}')
def reset_gain(self):
self.gain = 1/3
@property
def voltage(self):
""" Gets the voltage VMN in Volts
......
ohmpi/hardware_components/raspberry_pi.py
+ 2
− 2
View file @ acf68183
......@@ -37,12 +37,13 @@ class Ctl(CtlAbstract):
# None interface for battery
self.interfaces['none'] = None
# warnings.filterwarnings("error") # to filter out adafruit warning about setting I2C frequency
warnings.filterwarnings("error") # to filter out adafruit warning about setting I2C frequency
# I2C
try:
self.interfaces['i2c'] = busio.I2C(board.SCL, board.SDA) # noqa
except RuntimeWarning:
pass
warnings.resetwarnings()
# Extended I2C
try:
......@@ -51,7 +52,6 @@ class Ctl(CtlAbstract):
pass
except Exception as e:
self.exec_logger.warning(f'Could not initialize Extended I2C:\n{e}')
warnings.resetwarnings()
# modbus
try:
......
ohmpi/hardware_system.py
+ 198
− 32
View file @ acf68183
......@@ -317,8 +317,94 @@ class OhmPiHardware:
sp = np.mean(mean_vmn[np.ix_(polarity == 1)] + mean_vmn[np.ix_(polarity == -1)]) / 2
return sp
# def _compute_tx_volt(self, pulse_duration=0.1, strategy='vmax', tx_volt=5.,
# vab_max=voltage_max, vmn_min=voltage_min):
# """Estimates best Tx voltage based on different strategies.
# At first a half-cycle is made for a short duration with a fixed
# known voltage. This gives us Iab and Rab. We also measure Vmn.
# A constant c = vmn/iab is computed (only depends on geometric
# factor and ground resistivity, that doesn't change during a
# quadrupole). Then depending on the strategy, we compute which
# vab to inject to reach the minimum/maximum Iab current or
# min/max Vmn.
# This function also compute the polarity on Vmn (on which pin
# of the ADS1115 we need to measure Vmn to get the positive value).
#
# Parameters
# ----------
# pulse_duration : float, optional
# Time in seconds for the pulse used to compute Rab.
# strategy : str, optional
# Either:
# - vmax : compute Vab to reach a maximum Iab without exceeding vab_max
# - vmin : compute Vab to reach at least vmn_min
# - constant : apply given Vab
# tx_volt : float, optional
# Voltage to apply for guessing the best voltage. 5 V applied
# by default. If strategy "constant" is chosen, constant voltage
# to applied is "tx_volt".
# vab_max : float, optional
# Maximum injection voltage to apply to tx (used by all strategies)
# vmn_min : float, optional
# Minimum voltage target for rx (used by vmin strategy)
#
# Returns
# -------
# vab : float
# Proposed Vab according to the given strategy.
# polarity:
# Polarity of VMN relative to polarity of VAB
# rab : float
# Resistance between injection electrodes
# """
#
# vab_max = np.abs(vab_max)
# vmn_min = np.abs(vmn_min)
# vab = np.min([np.abs(tx_volt), vab_max])
# # self.tx.turn_on()
# switch_pwr_off, switch_tx_pwr_off = False, False #TODO: check if these should be moved in kwargs
# if self.tx.pwr_state == 'off':
# self.tx.pwr_state = 'on'
# switch_tx_pwr_off = True
# if self.tx.pwr.pwr_state == 'off':
# self.tx.pwr.pwr_state = 'on'
# switch_pwr_off = True
# if 1. / self.rx.sampling_rate > pulse_duration:
# sampling_rate = 1. / pulse_duration # TODO: check this...
# else:
# sampling_rate = self.tx.sampling_rate
# self._vab_pulse(vab=vab, duration=pulse_duration, sampling_rate=sampling_rate) # TODO: use a square wave pulse?
# vmn = np.mean(self.readings[:, 4])
# iab = np.mean(self.readings[:, 3])
# # if np.abs(vmn) is too small (smaller than voltage_min), strategy is not constant and vab < vab_max ,
# # then we could call _compute_tx_volt with a tx_volt increased to np.min([vab_max, tx_volt*2.]) for example
# if strategy == 'vmax':
# # implement different strategies
# if vab < vab_max and iab < current_max:
# vab = vab * np.min([0.9 * vab_max / vab, 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
# self.tx.exec_logger.debug(f'vmax strategy: setting VAB to {vab} V.')
# elif strategy == 'vmin':
# if vab <= vab_max and iab < current_max:
# vab = vab * np.min([0.9 * vab_max / vab, vmn_min / np.abs(vmn), 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
# elif strategy != 'constant':
# self.tx.exec_logger.warning(f'Unknown strategy {strategy} for setting VAB! Using {vab} V')
# else:
# self.tx.exec_logger.debug(f'Constant strategy for setting VAB, using {vab} V')
# # self.tx.turn_off()
# if switch_pwr_off:
# self.tx.pwr.pwr_state = 'off'
# if switch_tx_pwr_off:
# self.tx.pwr_state = 'off'
# rab = (np.abs(vab) * 1000.) / iab
# self.exec_logger.debug(f'RAB = {rab:.2f} Ohms')
# if vmn < 0:
# polarity = -1 # TODO: check if we really need to return polarity
# else:
# polarity = 1
# return vab, polarity, rab
def _compute_tx_volt(self, pulse_duration=0.1, strategy='vmax', tx_volt=5.,
vab_max=voltage_max, vmn_min=voltage_min):
vab_max=voltage_max, vmn_min=voltage_min, polarities=(1, -1), delay=0.050):
"""Estimates best Tx voltage based on different strategies.
At first a half-cycle is made for a short duration with a fixed
known voltage. This gives us Iab and Rab. We also measure Vmn.
......@@ -358,11 +444,26 @@ class OhmPiHardware:
Resistance between injection electrodes
"""
# TODO: Get those values from components
p_max = 2.5
vmn_max = 5.
vab_max = 50.
# define a sill
diff_vab_lim = 2.5
n_steps = 4
# Set gain at min
self.rx.reset_gain()
vab_max = np.abs(vab_max)
vmn_min = np.abs(vmn_min)
k = 0
vab_list = np.zeros(n_steps + 1) * np.nan
vab = np.min([np.abs(tx_volt), vab_max])
vab_list[k] = vab
# self.tx.turn_on()
switch_pwr_off, switch_tx_pwr_off = False, False #TODO: check if these should be moved in kwargs
switch_pwr_off, switch_tx_pwr_off = False, False # TODO: check if these should be moved in kwargs
if self.tx.pwr_state == 'off':
self.tx.pwr_state = 'on'
switch_tx_pwr_off = True
......@@ -372,36 +473,100 @@ class OhmPiHardware:
if 1. / self.rx.sampling_rate > pulse_duration:
sampling_rate = 1. / pulse_duration # TODO: check this...
else:
sampling_rate = self.tx.sampling_rate
self._vab_pulse(vab=vab, duration=pulse_duration, sampling_rate=sampling_rate) # TODO: use a square wave pulse?
vmn = np.mean(self.readings[:, 4])
iab = np.mean(self.readings[:, 3])
# if np.abs(vmn) is too small (smaller than voltage_min), strategy is not constant and vab < vab_max ,
# then we could call _compute_tx_volt with a tx_volt increased to np.min([vab_max, tx_volt*2.]) for example
if strategy == 'vmax':
# implement different strategies
if vab < vab_max and iab < current_max:
vab = vab * np.min([0.9 * vab_max / vab, 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
self.tx.exec_logger.debug(f'vmax strategy: setting VAB to {vab} V.')
elif strategy == 'vmin':
if vab <= vab_max and iab < current_max:
vab = vab * np.min([0.9 * vab_max / vab, vmn_min / np.abs(vmn), 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
elif strategy != 'constant':
self.tx.exec_logger.warning(f'Unknown strategy {strategy} for setting VAB! Using {vab} V')
else:
self.tx.exec_logger.debug(f'Constant strategy for setting VAB, using {vab} V')
# self.tx.turn_off()
if switch_pwr_off:
self.tx.pwr.pwr_state = 'off'
if switch_tx_pwr_off:
self.tx.pwr_state = 'off'
rab = (np.abs(vab) * 1000.) / iab
self.exec_logger.debug(f'RAB = {rab:.2f} Ohms')
if vmn < 0:
polarity = -1 # TODO: check if we really need to return polarity
else:
polarity = 1
return vab, polarity, rab
sampling_rate = self.rx.sampling_rate
current, voltage = 0., 0.
if self.tx.pwr.voltage_adjustable:
self.tx.pwr.voltage = vab
if self.tx.pwr.pwr_state == 'off':
self.tx.pwr.pwr_state = 'on'
switch_pwr_off = True
diff_vab = np.inf
while (k < n_steps) and (diff_vab > diff_vab_lim):
vabs = []
for pol in polarities:
# self.tx.polarity = pol
# set gains automatically
injection = Thread(target=self._inject, kwargs={'injection_duration': 0.2, 'polarity': pol})
readings = Thread(target=self._read_values)
injection.start()
self.tx_sync.wait()
readings.start()
readings.join()
injection.join()
v = np.where((self.readings[:, 0] > delay) & (self.readings[:, 2] != 0))[0] # NOTE : discard data aquired in the first x ms
iab = self.readings[v, 3]/1000.
vmn = np.abs(self.readings[v, 4]/1000. * self.readings[v, 2])
iab_mean = np.mean(iab)
iab_std = np.std(iab)
vmn_mean = np.mean(vmn)
vmn_std = np.std(vmn)
print(f'iab: ({iab_mean:.5f}, {iab_std:5f}), vmn: ({vmn_mean:.4f}, {vmn_std:.4f})')
# bounds on iab
iab_upper_bound = iab_mean + 2 * iab_std
iab_lower_bound = np.max([0.00001, iab_mean - 2 * iab_std])
# bounds on vmn
vmn_upper_bound = vmn_mean + 2 * vmn_std
vmn_lower_bound = np.max([0.000001, vmn_mean - 2 * vmn_std])
# ax.plot([vab[k]] * 2, [vmn_lower_bound, vmn_upper_bound], '-b')
# ax.plot([0, vab_max], [0, vmn_upper_bound * vab_max / vab[k]], '-r', alpha=(k + 1) / n_steps)
# ax.plot([0, vab_max], [0, vmn_lower_bound * vab_max / vab[k]], '-g', alpha=(k + 1) / n_steps)
# bounds on rab
print(f'rab_lb: {vab_list[k] / iab_upper_bound}')
rab_lower_bound = np.min([0.1, np.abs(vab_list[k] / iab_upper_bound)])
rab_upper_bound = np.min([0.1, np.abs(vab_list[k] / iab_lower_bound)])
# bounds on r
r_lower_bound = np.min([0.01, np.abs(vmn_lower_bound / iab_upper_bound)])
r_upper_bound = np.min([0.01, np.abs(vmn_upper_bound / iab_lower_bound)])
# conditions for vab update
cond_vmn_max = rab_lower_bound / r_upper_bound * vmn_max
cond_p_max = np.sqrt(p_max * rab_lower_bound)
new_vab = np.min([vab_max, cond_vmn_max, cond_p_max])
diff_vab = np.abs(new_vab - vab_list[k])
print(f'Rab: [{rab_lower_bound:.1f}, {rab_upper_bound:.1f}], R: [{r_lower_bound:.1f},{r_upper_bound:.1f}]')
print(
f'{k}: [{vab_max:.1f}, {cond_vmn_max:.1f}, {cond_p_max:.1f}], new_vab: {new_vab}, diff_vab: {diff_vab}\n')
if new_vab == vab_max:
print('Vab bounded by Vab max')
elif new_vab == cond_p_max:
print('Vab bounded by Pmax')
elif new_vab == cond_vmn_max:
print('Vab bounded by Vmn max')
else:
print('Next step')
vabs.append(new_vab)
if diff_vab < diff_vab_lim:
print('stopped on vab increase too small')
else:
print('stopped on maximum number of steps reached')
k = k + 1
vab_list[k] = np.min(vabs)
vab_opt = vab_list[k]
print(f'Selected Vab: {vab_opt:.2f}')
# if strategy == 'vmax':
# # implement different strategies
# if vab < vab_max and iab < current_max:
# vab = vab * np.min([0.9 * vab_max / vab, 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
# self.tx.exec_logger.debug(f'vmax strategy: setting VAB to {vab} V.')
# elif strategy == 'vmin':
# if vab <= vab_max and iab < current_max:
# vab = vab * np.min([0.9 * vab_max / vab, vmn_min / np.abs(vmn), 0.9 * current_max / iab]) # TODO: check if setting at 90% of max as a safety margin is OK
# elif strategy != 'constant':
# self.tx.exec_logger.warning(f'Unknown strategy {strategy} for setting VAB! Using {vab} V')
# else:
# self.tx.exec_logger.debug(f'Constant strategy for setting VAB, using {vab} V')
# # self.tx.turn_off()
# if switch_pwr_off:
# self.tx.pwr.pwr_state = 'off'
# if switch_tx_pwr_off:
# self.tx.pwr_state = 'off'
# rab = (np.abs(vab) * 1000.) / iab
# self.exec_logger.debug(f'RAB = {rab:.2f} Ohms')
# if vmn < 0:
# polarity = -1 # TODO: check if we really need to return polarity
# else:
# polarity = 1
return vab_opt, None, None
def _plot_readings(self, save_fig=False):
# Plot graphs
......@@ -440,6 +605,7 @@ class OhmPiHardware:
# if self.tx.pwr.pwr_state == 'off':
# self.tx.pwr.pwr_state = 'on'
# switch_pwr_off = True
self._gain_auto(vab=vab)
assert 0. <= duty_cycle <= 1.
if duty_cycle < 1.:
......
ohmpi/ohmpi.py
+ 1
− 0
View file @ acf68183
......@@ -500,6 +500,7 @@ class OhmPi(object):
bypass_check = kwargs['bypass_check'] if 'bypass_check' in kwargs.keys() else False
d = {}
if self.switch_mux_on(quad, bypass_check=bypass_check, cmd_id=cmd_id):
tx_volt,_ ,_ = self._hw._compute_tx_volt(tx_volt=tx_volt, strategy=strategy)
self._hw.vab_square_wave(tx_volt, cycle_duration=injection_duration*2/duty_cycle, cycles=nb_stack, duty_cycle=duty_cycle)
if 'delay' in kwargs.keys():
delay = kwargs['delay']
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment