max sampling rate

Hi, I wanted to know, if there´s some experience and maybe examples of evaluating the possible speed to get out of the MCP3208 with python.

I´m totally aware about pythons overhead but it´s required for my project. As I saw for the ADC DAC board it states that with python 12ksps would be possible. Those boards are comparable, or am I misstaking?

I did some basic testing of sampling one channel with 512 samples and timed it. If I break it down to one sample it results in ~21ksps.

Just wanted to double check if some experienced simimlar performance or can give me some guidance.

Thanks & greetings

N=512
test = np.zeros(N)
adc = ExpanderPi.ADC()
adc.set_adc_refvoltage(4.096)

while True:
    T_start = datetime.datetime.now()
    for n in np.arange(0, N):
        test[n] = adc.read_adc_voltage(1, 0)
    T_end = datetime.datetime.now()
    T_sample_all = (T_end - T_start).total_seconds()
    T_sample_one = T_sample_all/N
    f_sample= 1/T_sample_all
    f_sample2=1/T_sample_one

Hi

21Ksps will be about the average for the Expander Pi ADC when using a Raspberry Pi 3. I have uploaded a speed test demo to our GitHub repository which works similarly to your code. You can find it at demo_adcspeed.py

The sample rate will vary depending on several factors like the Raspberry Pi model, whether you are using Python 2 or 3 and what other processes are running at the same time. You will probably see about a 10% speed gain when using Python 2 instead of 3 as 2 seems to be better optimised.

The get the maximum sample rate out of the ADC you will need to use a compiled language like C or C++. I have uploaded a new speed test to our Expander Pi C library so I could compare the results and the sample rate of the C demo is around 54Ksps, so more than twice the speed of the Python library.

Hi Andrew,



thanks for your quick and helpful response!

Good to know about the better performance of python2. Yeah I'm aware about the variety of timing related on the by that comes with the Linux kernel scheduling. For the start I'll try to deal with it and hope isolation of cpu cores and the preempt_rt patch will provide me with satisfying performance.

Maximum sampling rate is not the main goal rather to have it somehow stable at a certain sampling rate and 20ksps would absolutely fit my needs.

For now I'm kind of forced to stick to python because the project is meant to be a proof of concept to design a radar system(adc, dac and spectrum analysis). We'll see where it ends up :)

Thanks for your help I really appreciate!



Greetings