PicoScope 7 Software
Available on Windows, Mac and Linux
Noise floor of Pico 4262
Noise floor of Pico 4262
Hi, I am considering purchasing the Pico 4264 as a scope/ FFT signal analyzer for a low frequency application. One thing I would like to use it for is checking the noise level of my device before my amplifiers, and I'm trying to understand if the Pico will be sensitive enough to do this. I saw the Pico 4262 noise is specified at 8.5 uV rms-- is that across the whole 5 MHz bandwidth or only with the 200 kHz bandwidth limiter on? I read topic39910.html which was very helpful but focused on the max achievable signal to noise ratio. I would do this test with no actual signal, just trying to see the noise coming from my device . So I am trying to figure out what is the noise floor of the Picoscope in terms of nV/root Hz, on the minimum vertical scale.
To be very practical: if I terminate the scope input and look at an FFT ranging from 3 to 5 kHz with 1 Hz wide bins, how tall will the average bin be: 4nV tall, 20 nV tall, or will there be more noise in this frequency range?
Thanks for your help,
Kathryn
To be very practical: if I terminate the scope input and look at an FFT ranging from 3 to 5 kHz with 1 Hz wide bins, how tall will the average bin be: 4nV tall, 20 nV tall, or will there be more noise in this frequency range?
Thanks for your help,
Kathryn
Re: Noise floor of Pico 4262
Hi Kathryn,
Our Spectrum Mode only plots the Amplitude Spectrum of the converted Voltages (so we apply coherent gain before the FFT calculation, in order to Normalize the levels for voltages). Unfortunately, we don't plot a Power Spectrum of V^(2)/Hz or V/√(Hz). This means that a Spectrum plot won't be able to measure noise.
Technically you could convert the Spectrum by exporting the Spectrum data, but this would involve delving into the DSP details of our implementation and the method of conversion you would use (which is not trivial). A more straightforward alternative would be to export the Time Domain data, and import it into a Programming environment with a rich DSP library, such as Matlab or Python, and then use a Library function to perform the FFT.
Also, 8.5uV is the total rms noise level across the bandwidth of the Scope. So, this means that in order to achieve a meaningful measurement in the Time Domain, you would only be able to measure down to 34uV of noise (depending upon which rule of thumb you use, the measuring instrument should have at least 3 or 4 times the precision of the measurement that you are attempting to make).
Regards,
Gerry
Our Spectrum Mode only plots the Amplitude Spectrum of the converted Voltages (so we apply coherent gain before the FFT calculation, in order to Normalize the levels for voltages). Unfortunately, we don't plot a Power Spectrum of V^(2)/Hz or V/√(Hz). This means that a Spectrum plot won't be able to measure noise.
Technically you could convert the Spectrum by exporting the Spectrum data, but this would involve delving into the DSP details of our implementation and the method of conversion you would use (which is not trivial). A more straightforward alternative would be to export the Time Domain data, and import it into a Programming environment with a rich DSP library, such as Matlab or Python, and then use a Library function to perform the FFT.
Also, 8.5uV is the total rms noise level across the bandwidth of the Scope. So, this means that in order to achieve a meaningful measurement in the Time Domain, you would only be able to measure down to 34uV of noise (depending upon which rule of thumb you use, the measuring instrument should have at least 3 or 4 times the precision of the measurement that you are attempting to make).
Regards,
Gerry
Gerry
Technical Specialist
Technical Specialist