Measuring signals buried in noise with an Oscilloscope

[@roelandriemens]

You can also turn on bandwidth filter to eliminate a lot of the noise

[@ruhnet]

Very useful info!! Thanks for sharing!

[@optroncordian7863]

How we can detemine the frequency of the signal? Would the fft function help ... ?

[@paulpaulzadeh6172]

Very Nice

[@tarunarya1780]

Thanks for the video. I am new to oscilloscopes. I have a Hantek DSO2c15 which sat on the shelf for more than 1 year. The ch1 and ch2 get 200-300mv noise on low voltage settings (20-100mv/div) when attached to only the bnc alligator probe. This is regardless of whether the earth and signal are joined. Is this normal? Is the alligator clip bnc probe acting a radio antenna or sis it a faulty oscilloscope?

[@laurentiusmichaelgeorge1118]

What a cool poor man's lock in amplifier. Mathematically, does this have something have to do with sine orthogonality while integrating them (in order to acquire the average)? Is that why the noise just went to zero when we multiplying it? And another question, if lock in amplifier's concept is this well known, what's so special about the really expensive lock in amplifiers out there?

[@markschnittker4770]

A lock in does multiplying, but this is different but has some similarities. Each time the Oscope triggers, it makes a trace which has a phase relationship to the reference (CH2). All the noise has a random relationship to the reference and so with the average of multiple traces, the noise averages out (similar to a lock in). The buried signal has a fixed (not random) phase relationship to the reference and thus it is the only portion that survives the averaging of many traces. Lock-ins do this my mixing (multiplying) the signal by the reference to relate the two, but the Oscope method uses the repetitive triggering. In both methods though the noise is averaged out and only the signal that has a fixed phase relationship to the refence remains. A nice advantage to this method is that the phase relationship between signal and reference need only be fixed, but phase offset does not need to be known... just be a constant. I think a different Oscope would measure that phase offset better than mine, but in many applications it is not necessary to know. Better lock in amplifiers have lower noise front ends and larger dynamic range, but both do the same concept which is that the longer time averaging is used, the more narrow the bandwidth and thus improved SNR. Like a narrower band pass filter. This Oscope method is about as sensitive as an AD630 Lock in board, but a true lock in would have orders of magnitude more sensitivity and SNR.

[@WilianMarangoni]

Great video, how could you know there is a 1khz sine wave in channel 1?

[@markschnittker4770]

In this case, I knew the frequency of the signal buried in the noise because of experimental set up that the signal came from. This is a very similar concept to using a lock in amplifier, where you know the frequency of the buried signal and use a reference of that frequency on a separate channel to help pull the signal out of the noise.

[@subhasarkar8823]

Nice demonstration.

[@HA7DN]

Why this reminds me to FT?

[@markschnittker4770]

Well, by taking the RMS value of a repetitive time averaged signal, we are basically finding the frequency component of of that signal. The more averages, the narrower the frequency range (bandwidth) that we are measuring. Like a narrow bandpass filter. By changing the reference frequency, we could find that same RMS value at that new frequency (spectral component). That becomes like a tunable bandbass filter. If we plot the measured RMS value of many reference frequencies we would eventually plot something that looked like an FT. Spectrum analyzers scan through frequencies as well, though I think they use different electronics. Conceivably one could write a program to step the reference frequency and measure the averaged RMS value and build a type of spectrum analyzer out of an Oscope and function generator.

[@ciprianpopa1503]

How come they are completely out of phase?

[@markschnittker4770]

The reference does not need to be in phase with the signal, it just needs to have a fixed phase relationship to the signal. Thus every time the Oscope triggers, it will overlap the signal in the same way and average out the noise leaving only the signal.