Single Channel oscilloscope

[sako243]

Normally these kinds of devices use SPS to represent digital bandwidth and Hz to represent analogue bandwidth.

If you're looking at purely digital signals or perfect sine waves then you can get away with being close to Nyquist (half the bandwidth) to fully represent the signal. If it's a more complex signal then ideally you want a greater oversampling ratio.

So any signal will be limited by the front end, there's no way you can get around that without changing the front end. Having a higher sample rate than the front end bandwidth allows for extra signal processing techniques to be applied that can be used to improve the fidelity of the displayed waveform.

Make sense?

 

[Robbie260]

Normally these kinds of devices use SPS to represent digital bandwidth and Hz to represent analogue bandwidth.

If you're looking at purely digital signals or perfect sine waves then you can get away with being close to Nyquist (half the bandwidth) to fully represent the signal. If it's a more complex signal then ideally you want a greater oversampling ratio.

So any signal will be limited by the front end, there's no way you can get around that without changing the front end. Having a higher sample rate than the front end bandwidth allows for extra signal processing techniques to be applied that can be used to improve the fidelity of the displayed waveform.

Make sense?

Ummmm sort of, I must confess it was in high-school that I last used an oscilloscope.

So the front end is referring too the 1Mhz band width in this case and that is my limiting factor.

Mega samples per second for want of a better analogy is like the resolution of a camera, the higher it gose the more accurate the picture is too real life/more detailed it will be.

So looking at digital on of square wave signals for example or pure sine waves that are 2MHz on a 1MHz scope will work and give quite an accurate picture of what's going on still as all you have too do is double all the values given.

 

[sako243]

Thread diversion but perhaps useful to some people

So the front end is referring too the 1Mhz band width in this case and that is my limiting factor.

Mega samples per second for want of a better analogy is like the resolution of a camera, the higher it gose the more accurate the picture is too real life/more detailed it will be.

In a general sense yes.

With regards to digital vs. analogue representation. If you've got an "analogue" input then it's purpose is to accurately represent the waveform on the probe. If you want to display a digital signal you only need to make a decision is it above a certain threshold or not.

Why digital can be faster
All modern scopes are digital storage oscilloscopes (DSOs) which means that there is a lot of processing that goes on in the background. Representing an analogue waveform on a digital platform to put it bluntly is more difficult and requires more resources.

E.g. lets say that the oscilloscope has a display that's 1024x768 pixels. Well on a digital signal we only need 1024 bits (128 bytes) to represent the waveform across the full screen (assuming one sample per pixel). If you want to represent 1V increments on a 1kV signal then each sample needs to be able to represent one of 1000 values. So you need 10 bits per sample - so you need 10x the memory requirements simply to display the signal let alone process it etc. So as a "free" feature some scopes allow faster sampling of digital signals (since they tend to be faster).

Why is the sample rate much higher than required by Nyquist?
Simple version - oversampling. Oversampling is a technique that can be applied to improve the signal-to-noise ratio (among other things).

The analogue front-end's (usually stated as bandwidth in MHz) purpose is to reject out-of-band signals. Simply put because the signal is being sampled the higher frequency components "fold" back in and corrupt the actual signal you want (aliasing).

However some noise will still be introduced by downstream components, generally speaking noise is assumed to be white-noise and thus if you repeatedly add the same "signal" samples to random noise samples the noise should average out - effectively you are reducing the effects of the noise and improving the signal to noise ratio.

So the example of 1MHz input bandwidth and 20MSPS ADC means that for every sample at 1MHz you actually get 20 that are being summed together to produce an "average" but in the process you're improving the quality of the signal.

I can explain in even more detail if you want but I think you understood the end result.

 

[Robbie260]

Thread diversion but perhaps useful to some people


In a general sense yes.

With regards to digital vs. analogue representation. If you've got an "analogue" input then it's purpose is to accurately represent the waveform on the probe. If you want to display a digital signal you only need to make a decision is it above a certain threshold or not.

Why digital can be faster
All modern scopes are digital storage oscilloscopes (DSOs) which means that there is a lot of processing that goes on in the background. Representing an analogue waveform on a digital platform to put it bluntly is more difficult and requires more resources.

E.g. lets say that the oscilloscope has a display that's 1024x768 pixels. Well on a digital signal we only need 1024 bits (128 bytes) to represent the waveform across the full screen (assuming one sample per pixel). If you want to represent 1V increments on a 1kV signal then each sample needs to be able to represent one of 1000 values. So you need 10 bits per sample - so you need 10x the memory requirements simply to display the signal let alone process it etc. So as a "free" feature some scopes allow faster sampling of digital signals (since they tend to be faster).

Why is the sample rate much higher than required by Nyquist?
Simple version - oversampling. Oversampling is a technique that can be applied to improve the signal-to-noise ratio (among other things).

The analogue front-end's (usually stated as bandwidth in MHz) purpose is to reject out-of-band signals. Simply put because the signal is being sampled the higher frequency components "fold" back in and corrupt the actual signal you want (aliasing).

However some noise will still be introduced by downstream components, generally speaking noise is assumed to be white-noise and thus if you repeatedly add the same "signal" samples to random noise samples the noise should average out - effectively you are reducing the effects of the noise and improving the signal to noise ratio.

So the example of 1MHz input bandwidth and 20MSPS ADC means that for every sample at 1MHz you actually get 20 that are being summed together to produce an "average" but in the process you're improving the quality of the signal.

I can explain in even more detail if you want but I think you understood the end result.

Thanks very much for the explanation sako that makes it all make a lot more sense.