RE: Siglent SDS1104X-E Oscilloscope Performance

shabaz — Thu, 06 Feb 2020 22:55:14 GMT

Hi Andrew,

This is a great set of experiments! It's the best way to understand your instrument.

Regarding the sample rate, I think it used to be that it needed to be much higher than the analogue bandwidth, but although that might have been the case with early digital scopes (I'm guessing - I've not used extremely old digital scopes), all current scopes use the 'sinc' formula as you've mentioned for digitally filtering, meaning that even if they sample at only ~2.5 times the analogue bandwidth, they can near-perfectly reproduce the analogue signal (up to the analogue contents' max bandwidth component - no higher, provided the input filter is sharp enough to stop any aliasing)

For sure for a square wave the scope analogue bandwidth should be much higher than the fundamental frequency, to capture the components of it, but once that required analogue bandwidth is determined, the sample rate doesn't need to be higher than about 2.5 times that value.

This depends on the anti-aliasing filter though. Whether the filter is good enough is another story but it should be, because many scopes are rated at 2.5 times the analog bandwidth, for their sample rate (typically when using multiple channels simultaneously). In summary, you can use the 'scope with confidence right up to the rated analogue bandwidth, if the sample rate is >= 2.5 times that value, or if the sample rate is >=2.5 times the maximum component in the signal being captured.

The join-the-dots option is weird.. I don't know if my 'scope has it, it might, but I've never used it (nor sure why it's needed, because it will give a misleading representation.

Here's an example of join-the-dots, at a valid but low sample rate (i.e. at about 2.5 times higher than the signal input):

[View:/resized-image/__size/620x304/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-0b164ffa-ecf9-4517-8b7d-c00aedcb68a3/8203.contentimage_5F00_206045.png:620:304]

This is it with sinc:

[View:/resized-image/__size/620x307/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-0b164ffa-ecf9-4517-8b7d-c00aedcb68a3/7607.contentimage_5F00_206046.png:620:307]

It looks very close to the source signal, which was a sine wave. It mostly looks accurate, apart from the ends where the levels look slightly wrong.

This would apply to any signal, not just a sine wave, provided the sample rate was approx 2.5 times the max frequency component in the input signal.

Regarding waveform update rate, I don't know what others feel, but I think (this is just a theory) maybe most 'scopes are already at good update rates, and nice triggering will capture to make visible infrequent signals anyway, depending on what options are available with the triggering. The high waveform update rates can make for some nice screenshots though, and are useful if there's no expectation of what will be in the signal too I guess, but probably good triggering options could be more important.

I had a quite high-speed (500 MHz analogue bandwidth) HP 'scope with a very low waveform update rate but I could still capture great screenshots, provided I could trigger on it (or trigger on another event related to it). I've given it away since, but I kind of miss it.

RE: Siglent SDS1104X-E Oscilloscope Performance

fmilburn — Thu, 06 Feb 2020 21:41:53 GMT

Hi Andrew,

Great writeup. The Keysight position on memory depth is that they have segmented memory but I agree with your assessment and most of the time would rather have a long capture and the ability to zoom in. Does the Siglent have masks? If so, look into that for catching glitches and excursions.
Frank