Rigol 1054Z user-error or oscilloscope problem?

[rpetitho]

So I was doing some testing on a project and had a strange issue. Here are three photos of the response curve I was testing. All are the exact same curve with very different oscilloscope measurements. First photo was a single shot at 500mv resolution, second photo is the same shot zoomed out to 1v resolution, third is another shot of the same response at 1v resolution. Did I screw something up? Why do the last 2 photos look so different when they should look very similar? Let me know if you need more info.

[ Specified attachment is not available ]

[Fungus]

Did I screw something up? Why do the last 2 photos look so different when they should look very similar? Let me know if you need more info.

I think what you're seeing is op-amp overload/recovery.

Your signal goes way off the top of the screen somewhere and is overloading the op-amp responsible for vertical scale. Anything that happens before it has chance to recover from that is undefined. Switching to a different vertical scale will change the op-amp gain setting (or maybe even use a completely different op-amp) so you see different results.

nb. This happens to all op-amps and all oscilloscopes.

https://www.google.com/search?q=op+amp+overload+recovery+time

[JohnPen]

I was just about to say the same thing but Fungus beat me to it.  The return edge on the original shot is the poor op-amp struggling to recover!

[rpetitho]

Always something new to learn. I think my project may be performing better than the oscilloscope would have me believe. Thanks!

-Bob

[hgjdwx]

I don't think that's normal
http://v.youku.com/v_show/id_XNDQyNDg2NzgzNg====.html

[Miti]

I think even in the second shot (third picture) you saturate the input. You should always keep the trace within the screen limits to trust what you see.

[Fungus]

I think even in the second shot (third picture) you saturate the input. You should always keep the trace within the screen limits to trust what you see.

And if a detail is so small that you need to push the rest of the signal far enough off the screen to cause problems? That detail isn't important...