Kortuk instructed me with my new handheld scope DSO201:

You can use very high end chips and high working ranges, just at the end you will need a filter to bring it down to what your o-scope can handle, which is roughly 1/2.1 the bandwidth of your scope. (Chat discussion)

I did not understand this. Does this mean that it is possible to measure high-frequency samples with a handheld scope having "1MHz analogue bandwidth -- but -- 200kHz (1 Msps)"? What does it mean to "bring it down"? By which chip I can bring the frequency down to make it visible with a small scope?


What he is saying is what you need to do in order to avoid aliasing.

When you measure a high frequency signal with a low sample rate scope, you get an aliased signal, what is this alias thing? I'll try to explain with an example:

Lets say you have a 10MHz signal and you sample at exactly the same frequency (very unprobable, but illustrative for the example). What will happen in this case? You will see that all the samples have the same value, right? so, when you represent those samples on the scope screen they will appear as if it was a DC component, that is aliasing. The effect happens whenever you try to sample a signal at less than twice that signal's frequency.That same 10MHz signal sampled at 15MHz will produce aliasing and appear to be a 5MHz signal. If you want to get a bit deeper on this effect see this wikipedia entry.

So, why do you need a filter? because a filter eliminates those high frequency components, avoiding the aliasing effect. In fact the scope's analog bandwidth is an analog filter in that regard, but it may not be enough.

Anyways, what kortuk meant in that chat was that they are measuring the output from a circuit that does all the signal conditioning, giving a nice output at a sufficiently high voltage level and low frequency, in that sense, you could measure that signal with almost any oscilloscope, but you need the circuit, that's where the magic is!



This is related to sampling conversing the signal to digital, aliasing and Nyquist theorem. I am trying to filter the key points below.

Earlier answer referenced to an article with some condition. Suppose you want to generate your original signal from the sampled signal. Now an important condition is that the frequency is in the Nyqyist domain \$ f_{Nyquist} > |f| \$. The thing is to remove the extreme frequencies, I haven't yet fully understood this -- but let's simplify: we cannot have all frequencies in the frequency domain due to limitations in the scope so our fourier transform from time domain to frequency domain is not accurate and vice-versa.


if your higher frequencies alias down as lower frequencies you have no way to tell what is being caused by an actual low frequency and what is caused by a high frequency. So you corrupt your data, this is why people speak so negatively of aliasing. Kortuk

"Aliasing is the term for what happens there. -- What happens, is that as a frequency becomes higher then your sampling frequency it actually looks like a lower frequency.

A real example: if I change the state of a light ever second, I turn it on, then off... 1,0,1,0,1,0,1,0 [and so on]. But you look at me every 2 seconds, you sample at half the frequency of the signal then you will either see the light always being on, or you will see it always off."

*Sorry, that was not an accurate statement, you have to sample at a non harmonic, then you start getting very odd things you think I am doing(

@Kortuk Is that Niquist theorem?

@angelatlarge Yes, that is what I am explaining.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.