Strange scope results, ch. 56

[MikeK]

Rigol DS1052E, 10X probe, scoping the output of a 10MHz Colpitts crystal oscillator running at ~6V.  The only difference in the scope screens is that I changed the timebase.  Nothing else.  The first screen doesn't make sense to me.  How is oscillation showing up at that timebase?  Very confusing.

[ataradov]

It is called aliasing - https://en.wikipedia.org/wiki/Aliasing

Scope samples at a certain sample rate, it does not know what your signal is doing between the samples, so it only shows the sampled points.

[MikeK]

But at 1Gsa/s it should have plenty to know that the signal is not 2.7kHz.  Scope has also been hacked to 100Mhz.

[ataradov]

Scopes adjust sampling rate based on the time base. Some higher end models have ability to select sample rate independent of the time base, not sure about this one.

[bdunham7]

The issue probably isn't the maximum possible sample rate, it is the total memory available and the resulting sample rate at that timebase.  Not familiar with your model, so what is the record memory in this example?

[CaptDon]

Slow the timebase way down and the scope may stabilize as if it were a 10Khz sinewave. I hate aliasing!!! Trying to look at NTSC video with a digital sampling scope is almost useless!! One of the many reasons I keep a 100mhz pure analog scope in my stable.

[ataradov]

Yes, this scope has 1Mpts of memory depth, so at 1 Ghz it will be able to store 1 ms worth of data. So, this would not fill the full screen on your first picture. And processing 1 M points each time is also resource intensive, so update rates would be abysmal.

This is why deeper memory and independent setting of a sample rate is important. I think even DS1054Z can do that. I remember capturing frames worth of data of a video signal with no issues.

[MikeK]

Wow, I changed the memory from "Normal" (16kpts) to "Long Mem" (1Mpts) and the problem disappeared.  Thanks!  This wouldn't happen if I had an analog scope. :/

[MikeK]

So...If I was scoping this without knowing the crystal was 10MHz...let's say there was no marking on the crystal...How would I know that I should scan deeper?

[TimFox]

Another example of aliasing:  we were tracking down some unexpected 60 Hz hum in a rack full of equipment with an '80s digital scope.
Weird:  the frequency was not exactly 60 Hz (damned close, though) and it did not synchronize with line trigger.
It looked like a good sine wave, however.
It turned out the 10 MHz reference oscillator in the scope and the 10 MHz reference oscillator in the rack were mutually off by about 6 ppm (within spec).
Of course, when we changed the time-base setting on the scope and the apparent frequency changed, we knew it was an alias.

[ataradov]

So...If I was scoping this without knowing the crystal was 10MHz...let's say there was no marking on the crystal...How would I know that I should scan deeper?

General experience and understanding of what you should roughly expect from the observed circuit. Aliased signals often make no logical sense, so you would naturally look deeper. If you are absolutely not sure, start with the fastest time base and work your way back.  This is usually not a problem in real life.

A lot of things would not happen if you had an analog scope, including many useful observations. If you look at sine waves all day, then sure analog scopes rule.

[MikeK]

And I found out while scoping some digital signals that Long Mem needs to be off; it is a speed PIG.  Maybe I'll bump up to a better/faster scope for Xmas.

[ataradov]

Using the whole memory all the time is slow on most scopes. It might work on stuff prices at $10K, I don't know, never used that. But most consumer scopes would be similar.

For most tasks automatic memory selection is sufficient.