Rigol DS1054z mystery

[evava]

First: I am sorry if this belongs in the beginners section
Secondly: I am sorry for the somewhat clickbait title

Namely, I recently came across an interesting thing that I would like some more experienced people to explain/clarify to me:

Mem depth is set manually to 120k. Then, I apply 1Mhz to the input of the oscilloscope, successively a sinusoidal waveform, a rectangular waveform and a triangular one.
After pressing Auto on the oscilloscope, the oscilloscope correctly displays the waveform with a time base of 200ns and the frequency meter shows 1Mhz. However, when I manually extend the time base to 50ms, the oscilloscope shows the same waveform, with the frequency meter reporting a frequency of about 7Hz!
All this, of course, only with a manually set mem depth of 120Kbit (or less, on this particular oscilloscope).
The waveform is preserved (sine/rectangle/triangle), including voltage values. ​​

I tried googling this phenomenon, but I guess I'm asking the wrong question.
I believe that this mystery applies and appears apparently on all digital oscilloscopes where we deliberately set a small mem depth.
Please explain it to me - thank you.

[RoGeorge]

It is called spectrum folding or aliasing.  Usually it happens when the sampling rate is too small, when it's less than twice the max frequency of the signal.  Turn on the 'Anti-aliasing' on the oscilloscope (last button on the right column in your pics), this should help, though there might be some tricky frequency in sync with the sweep frequency that might still appear wrong.


Image from:  https://en.wikipedia.org/wiki/Aliasing

For example, if the signal is similar with the blue sinusoid, but the sampling rate of the oscilloscope is much rare (like the black dots in the image), apparently you'll see on the oscilloscope a much lower frequency (you'll see on the oscilloscope only the orange line determined by the sampling black dots).

[evava]

Thank you very much for your answer!

I thought it was something with aliasing, but I can't imagine how it is possible to keep exactly the same waveform?
Every waveform I tried, I got exactly the same waveform, during this aliasing..
Even pulse, etc.

[RoGeorge]

The waveform is preserved.  To convince yourself, you can try drawing by hand on a paper, or write a few lines of code to simulate it.

The signal is repeating itself, so it doesn't matter if you sample now, or if you sample now + 100 periods in between, you'll see the same amplitude.

[evava]

Yes, it sounds plausible. You are right.

Now I tried on my new ZEEWEII DSO2512G, and the same phenomenon. Only, it cannot be suppressed by increasing mem depth 

How to get around this with digital oscilloscopes with really little memory?  Each time to check if there is not the same waveform with a much higher frequency? Or what else?
How function AUTO works that it always finds the right frequency waveform?

[KrzysztofB]

Memory depth is not that critical. What is important is Sampling rate.
You want first to look at your signal with high sampling rate, and then if you see signal is slow you can decrease sampling and extend you time window.
Pressing Auto is most of the time good starting point also.

If you start with little Sampling rate you might end up being tricked like here.
Google / Wiki : Nyquist theorem

[evava]

Thank you.
By high sampling rate you mean probably short time base?
Because on simple scopes, e.g. on DSO2512G, you cannot do anything about sampling rate.

[2N3055]

Keep memory size on Auto.. this will keep sample rate high..

[Mechatrommer]

Because on simple scopes, e.g. on DSO2512G, you cannot do anything about sampling rate.

nor you can on rigol ds1054z, its automatic based on time/div and mem depth set by you. always take note of the red circled figure in attached picture and also keep in mind the nyquist minimum sampling rate criteria (sampling rate must be more than 2X of signal frequency) so your 200kSa/s will only can show up to 100kHz (a little bit less) signal correctly, if you give your DSO equal or more than this frequency, you'll get ghost (aliased) signal, and ghost will always be related to mystery, and mystery can get you nuts if you overthinking it.

[evava]

mystery can get you nuts if you overthinking it.

Thank you!

Trying to remember this..

(sampling rate must be more than 2X of signal frequency) so your 200kSa/s will only can show up to 100kHz (a little bit less) signal correctly, if you give your DSO equal or more than this frequency, you'll get ghost (aliased) signal

I was confused that just by spinning the timebase wheel I made a ghost 
From now on Memory size Auto is my best friend.

[Anthocyanina]

it is not just all that was mentioned above, but also that the 1054z unfortunately doesn't have good antialiasing. Even with it turned on, it will still alias. Other scopes do have much better antialising, like the AD2 for example, even with its very low memory and sample rate it doesn't show an aliased waveform, like shown in this video with a 1MHz sine. This particular scope will show a darker color to show there is information there not properly captured by the acquisition system. If you do stop the acquisition then zoom in, you will see an aliased waveform, but if the acquisition is running constantly, instead of aliasing, you'll see the darker color showing the right amplitude and it will let you know you should use a faster timebase to properly see the signal. So it's also important to know that memory and sample rate are not everything that matters when it comes to avoid displaying an aliased signal

[RoGeorge]

Antialising works well enough in ds1054z if the acquisition is left on auto.
When one sets custom memory depth, this will affect sampling rate, too.
Aside that, in the above pics Anti-alising was set to OFF (default is on).

[evava]

To sum it up, in the end it turned out, as it always does, that when all else fails,

RTFM !

I apologize for the unnecessary thread..

[2N3055]

To sum it up, in the end it turned out, as it always does, that when all else fails,

RTFM !

I apologize for the unnecessary thread..

Don't worry. It was honest learning step and everybody was polite and nice to everybody. This is GOOD thread.