AC Voltammetry and sample rate

BlairH

I've been attempting to use the rodeostat to perform AC voltammetry as per the example and using the 50k dummy cell which gives a nice output at very low sample rates however at higher sample rates (though still quite low) of 115 Hz I've found the output is heavily affected by the sample rate.
I also implement the firmware changes as mentioned in this post however noticed no effect.

Any advice on how to overcome this would be greatly appreciated

2 Hz
2 Hz

115 Hz
115 Hz

I've used the following parameters

Run parameters

    t0 = 0.0  # Transition start time (s)
    t1 = 8.0  # Transition stop time (s)

    v0 = -0.1  # Initial voltage (V)
    v1 =  -0.9  # Final voltage (V)

    amp = 0.1  # Sinusoid ampliude (V)
    per = 0.5  # Sinusoid period (s)

    dt = 0.005  # Time step for setting voltage and measurements
    t_total = t0 + t1 # Total experiment duration 

    volt_func = create_sin_linear_func(t0,t1,v0,v1,amp,per)

    # Create device object, set voltage/current ranges and run test
    pstat = Potentiostat('COM5')
    pstat.set_volt_range('2V')
    pstat.set_curr_range('100uA')
    t, volt, curr = run_manual_test(pstat, volt_func, dt, t_total)

Will Dickson

@BlairH

I don't think the method in post is going to help in the case of example as it is run in manual/direct mode rather than in firmware. In manual/direct mode the every time the output voltage is changed and the current is sampled, as the test is run, requires USB/serial communication with the host PC. In this manner the schedule for the waveform output is running PC side and the voltage is set using set_volt and the current is sample using the get_curr methods. So each sample requires two USB/Serial command/response pairs - one to set_volt and one to get_curr.. Each of these commands take some time to complete. Because of this tests run in manual /direct mode cannot achieve as high a sample rate as those implemented in firmware and run using the 'run_test' method. Unfortunately, at this time, there isn't an AC voltammetry test implement in the firmware. That said even if it was implemented in firmware the maximum sample rate would be about 1000Hz with the stock firmware - and you might be able to achieve up to around 10kHz for very short bursts with the firmware modifications in post.

Regarding the manual/direct mode AC voltammetry test you are running I'm not quite sure what is happening. I'm not able to replicate this myself - I'm guessing it might be something specific to your system. That said this manual/direct mode test can't achieve very high sample rates as the 'set_volt' and 'get_curr' method calls take some time to complete. So as you lower dt more and more eventually the actually time taken per sample will plateau as it is dominated by the time required for the 'set_volt' and 'get_curr' method calls.

One thing that is useful to look to help diagnose the timing of manual/direct mode test is the time difference between samples. You can get this by converting the list of samples returned to an array an then looking at the consecutive differences e.g.

t = scipy.array(t)
dt = scipy.diff(t)
plt.plot(t[1:], dt)

This is what I get for dt = 0.05 and dt = 0.005
dt=0.05
and
dt=0.005

Note, that in when dt parameter is set to 0.005 the measured "actual' dt between time steps much more erratic and has a median value of only 0.0143. this is due to the fact that the sampling time is being dominated by the time required for the 'set_volt' and 'get_curr' calls.

I've also attached the modified AC voltammetry script I used for generating these plots.

from __future__ import print_function
from potentiostat import Potentiostat
import time
import sched
import math
import scipy
import matplotlib.pyplot as plt


def run_manual_test(pstat, volt_func, dt, t_stop):
    """
    Run a voltammetric test in maunal/direct mode. 

    pstat      = potentiostat
    volt_func  = output voltage function
    dt         = sample time step
    t_stop     = duration of the trial
    """
    t = 0 
    cnt = 0
    t_start = time.time()
    time_list, volt_list, curr_list = [], [], []
    scheduler = sched.scheduler(time.time, time.sleep)

    while t < t_stop:
        # Set potentiostat output voltage and samle current 
        volt = volt_func(t)
        pstat.set_volt(volt)
        curr = pstat.get_curr()
        print('{0:1.2f}, {1:1.2f}, {2:1.2f}'.format(t, volt, curr))
        time_list.append(t)
        volt_list.append(volt)
        curr_list.append(curr)
        # Run scheduler to until time for the next sample (dt seconds)
        t_next = t_start + (cnt+1)*dt
        scheduler.enterabs(t_next, 1, lambda:None, ()) 
        scheduler.run()
        t = time.time() - t_start
        cnt+=1
    return time_list, volt_list, curr_list


def create_sin_linear_func(t0, t1, v0, v1, amp, per):
    """
    Returns a function which in the interval [t0,t1] is a sum of linear
    function and a sine wave.

    t0   = time at which linear transition, from v0 to v1, begins
    t1   = time at which linear transition, from v0 to v1, ends
    v0   = initial value
    v1   = final value
    amp  = amplitude of superimposed sinewave
    per  = period on superimposed sinewave

    """
    def func(t):
        if t < t0:
            return v0 
        elif t < t1:
            dt_trans = t1-t0
            v_lin = (v1-v0)*(t-t0)/dt_trans + v0
            v_sin = amp*math.sin(2*math.pi*(t-t1)/per)
            return v_lin + v_sin
        else:
            return v1
    return func 


if __name__ == '__main__':

    # Run parameters
    t0 = 0.0  # Transition start time (s)
    t1 = 8.0  # Transition stop time (s)

    v0 = -0.1  # Initial voltage (V)
    v1 =  -0.9  # Final voltage (V)

    amp = 0.1  # Sinusoid ampliude (V)
    per = 0.5  # Sinusoid period (s)

    dt = 0.05  # Time step for setting voltage and measurements
    t_total = t0 + t1 # Total experiment duration 

    volt_func = create_sin_linear_func(t0,t1,v0,v1,amp,per)

    # Create device object, set voltage/current ranges and run test
    pstat = Potentiostat('/dev/ttyACM0')
    pstat.set_volt_range('2V')
    pstat.set_curr_range('100uA')
    t, volt, curr = run_manual_test(pstat, volt_func, dt, t_total)

    t_array = scipy.array(t)
    dt_array = scipy.diff(t)
    dt_median = scipy.median(dt_array)
    print('median dt: {}'.format(dt_median))

    # Plot results
    plt.figure()
    plt.subplot(311)
    plt.plot(t,volt)
    plt.ylabel('potential (V)')
    plt.title('param dt = {:0.3},  median actual dt = {:0.4}'.format(dt,dt_median))
    plt.grid(True)

    plt.subplot(312)
    plt.plot(t,curr)
    plt.xlabel('time (s)')
    plt.ylabel('current (uA)')
    plt.grid(True)

    plt.subplot(313)
    plt.plot(t,curr)
    plt.plot(t_array[1:], dt_array)
    plt.grid(True)
    plt.xlabel('time (s)')
    plt.ylabel(r'$\Delta t$ (s)')
    plt.ylim(0.0, dt_array.max()*(1 + 0.1))
    plt.show()

BlairH

Hi Will,

Thanks for your feeback. You're right about my system specific differences, even for param dt =0.05 I get a median of 0.058.

I've attempted to implent the firmware changes outlined as I only need a sample rate of 115Hz however I haven't noticed any difference. I'm not very familar with arduino programming and I suspect I haven't implmented the changes in ps_system_state.cpp correctly. I've attached my changes in hopes that you might notice some sort of obvious error.

Regards

Blair

void SystemState::serviceDataBuffer()
    {
        // Check for last sample flag to see if done
        bool run_complete = false;
        if (lastSampleFlag_)
        {
            if ((run_complete = true));
        }

       		 // Empty data buffer
        	size_t buffer_size;
        	ATOMIC_BLOCK(ATOMIC_RESTORESTATE)
        	{
            	buffer_size = dataBuffer_.size();
        	}

        	while (buffer_size > 0)
        	{
            	Sample sample;
            	ATOMIC_BLOCK(ATOMIC_RESTORESTATE)
            	{
              	  	sample = dataBuffer_.front();
                	dataBuffer_.pop_front();
                	buffer_size = dataBuffer_.size();
            	}
            		messageSender_.sendSample(sample);
        	}

        	// Send indication that the run is complete 
        	if (run_complete)
        	{
            	messageSender_.sendSampleEnd();
            	lastSampleFlag_ = false;
        	}
    	}