Page 1 of 1

DSO noise floor - seems high?

PostPosted: Sat Aug 31, 2019 3:57 am
by wireless
I was taking some measurements of harmonic levels from a 7MHz sine wave carrier and was struggling to spot the harmonics amongst the general noise levels, so I thought I would investigate some more.

My setup is a Windows 10 laptop (latest software - build 1903), plus a Virtins DSO-2820E running MI v3.8.5.2. The laptop was running on battery. WiFi was switched off. No other equipment was running in the same room.

With only the 'A' channel enabled and the probe connected but short-circuited, the attached spectrum was seen using a 200MHz sample rate. The 'Y' scale was set to 'dbR'. Some smoothing was enabled (20 frames linear average) in order to achieve a stable display.

This noise floor seems very high and if correct is only allowing a very poor dynamic range. Would you regard this as any way normal? Is there something wrong with my setup. Is this noise being generated within the DSO?

I have tried using ferrite magnet cable clamps on the USB connecting cable - this made no difference.

I would appreciate your thoughts please.

Noise levels.PNG
Noise levels.PNG (50.01 KiB) Viewed 18236 times


Re: DSO noise floor - seems high?

PostPosted: Sat Aug 31, 2019 1:54 pm
by VirtinsTech
1. Normalized Amplitude Spectrum Mode
The Spectrum Analyzer is in Normalized Amplitude Spectrum mode when "dBr" is selected as the "Y Scale". Under this mode, the highest peak in the spectrum will be set to 0dB (the maximum) and the rest spectral components will be normalized by this peak, regardless of their absolute amplitudes. This mode helps the user to view spectral components even if they are very small in absolute amplitude. But it often gives a wrong impression that the noise level is very high when there is no signal present.

2. Interleaved Sampling at the Max. Sampling Rate
For VT DSO-2820E, when the sampling rate is set to 200MHz, the displayed waveform is assembled from two ADC channels through interleaved sampling. Admittedly the disparity between the two ADC channels often leads to a higher noise level. Therefore the sampling rate of 200MHz should be used only when necessary.

3. Probe CAL signal output
By default, the 1kHz squarewave probe CAL signal is enabled. It can be disabled via [Setting]>[DAC Device] to help to reduce some noises. This is especially important when measuring with low voltage ranges.

Re: DSO noise floor - seems high?

PostPosted: Tue Sep 03, 2019 1:39 am
by wireless
Thank you - my mistake!
When using a more appropriate scale, the levels are significantly lower. Using 'dbu' for example I am seeing noise peaks at up to - 50dbu with a 200MHz sample rate, up to -60dbu with a 100MHz sample rate and at a 50MHz sample rate the worst case is a 10MHz peak at -68dbu: All taken with the probe inputs shorted.

This is considerably better than I had thought, although there are some puzzling peaks in the noise distribution even with the inputs shorted and the cal square wave signal turned off as suggested. It would be helpful if you could indicate if this is roughly as expected under these conditions - or failing that, if you could provide a test setup that I can copy and some indicative results for me to compare. This will help me confirm that the unit is operational.

Many thanks again

Re: DSO noise floor - seems high?

PostPosted: Tue Sep 03, 2019 2:12 pm
by VirtinsTech
A flat or smooth noise spectrum at zero input is desirable, however, that ideal condition is disturbed by various digital signals especially digital clocks inside the unit. Despite that, as long as the absolute noise level is reasonably low, the unit is operational.