Speeding up trace transfer?

[gsteele13]

Hi all,

I'm working on some code to create a "live update" plots of traces acquired by my STEMlab125-14. The code I'm using is here:

Code: Select all

def get_IV_t(timebase=-3):
    tn = [time.time()]
    
    # Configure timebase and input range settings
    rp.tx_txt("ACQ:DEC %d" % dec[timebase])
    #print("setting dec to %d " % dec[timebase])
    #print("t_total is %f" % t_total[timebase])
    rp.tx_txt('ACQ:SOUR1:GAIN HV')
    rp.tx_txt('ACQ:SOUR2:GAIN HV')
    
    # Start the acquisition
    rp.tx_txt('ACQ:START')
    time.sleep(t_total[timebase]+0.1) # to avoid getting old data in the buffer
    rp.tx_txt('ACQ:TRIG NOW')
    
    tn.append(time.time())
    
    # Wait until it's done
    while 1:
        rp.tx_txt('ACQ:TRIG:STAT?')
        if rp.rx_txt() == 'TD':
            break
    
    tn.append(time.time())
    
    # Download the traces
    rp.tx_txt('ACQ:SOUR1:DATA?')
    buff_string = rp.rx_txt()
    buff_string = buff_string.strip('{}\n\r').replace("  ", "").split(',')
    I = np.array(list(map(float, buff_string))) / 1e4
    
    tn.append(time.time())
    
    rp.tx_txt('ACQ:SOUR2:DATA?')
    buff_string = rp.rx_txt()
    buff_string = buff_string.strip('{}\n\r').replace("  ", "").split(',')
    V = np.array(list(map(float, buff_string))) / 1e4
    tn.append(time.time())
  
    # Create the appropriate time array
    t = np.linspace(0,t_total[timebase],len(I))
    
    if True:
        print("Total time: %f" % (tn[-1]-tn[0]))
        print("Splits:")
        for i in range(len(tn)-1):
            print("%f" % (tn[i+1]-tn[i]))

    # Return the traces
    return I,V,t

The full notebook can be found on github:

github / gsteele13 / gary-misc-notebooks / Prototype SQUID practicum software.ipynb

What I notice is that getting the traces off the board is very slow: for a 134 microsecond acquision, the code above takes nearly 1.8 seconds to get the data off!

Code: Select all

Total time: 1.767000
Splits:
0.236000
0.396000
0.585000
0.550000

Is there a way to speed this up somehow? An transfer overhead of nearly 1000% is not so handy...

Any suggestions (or working code even) highly welcome.

Thanks,
Gary