Siglent SDS800 Zoom out question (problem?)

[nctnico]

"Zoom out" is term related to using zoom function. Nobody cares about "changing timebase after the stopped capture" except people that don't know how to set timebase right in the first place.

We could all learn to use left-handed scissors as well, but it's not optimal.

I fail to understand the comparison.

That is indeed your problem. Now just accept that you don't understand and all is well. There is lots of stuff I don't understand but I don't go around saying people are idiots because they can't make me understand.

[2N3055]

"Zoom out" is term related to using zoom function. Nobody cares about "changing timebase after the stopped capture" except people that don't know how to set timebase right in the first place.

We could all learn to use left-handed scissors as well, but it's not optimal.

I fail to understand the comparison.

That is indeed your problem. Now just accept that you don't understand and all is well. There is lots of stuff I don't understand but I don't go around saying people are idiots because they can't make me understand.

And you do understand what scissor comparison means?

Why are you even arguing here?
This so called "zooming out" in video has nothing to do with your method of capturing long with manual memory....
Why are you triggering to something that has nothing to do with you?
OP didn't ask about that.

And you gave him very good way of measuring his problem, that I very much agreed with.
That needs no "zoom out" or even manual control of memory.

So what is the problem. Nobody called you out.

Mentioned  video is based on wrong data and should not be taken seriously.
If you think that manual control of memory is important for OP, than explain it and state your case. You have a lot of knowledge and I would rather OP gets your advice from real experience, than that clickbait video.
Even if we disagree on many things, opinion wise.
 

[Someone]

SOMEONE asked why I need that zoom function so I decided to recheck my RF sweep generator that I made a few years ago. That had some challenging waveforms. Surprisingly, the frequency steps are not measuring correctly or consistently so I have to figure out what's wrong, it's probably me. I've  tried to describe the waveform details at the end of this note, but it's probably better not to get sidetracked in those details.

Anyway, the purpose was to determine if a zoom out function would be useful. And so far I'm not sure. IOW, the off-screen capture in this case would probably not be large enough to be useful. Further, overall I'm starting to think the few times a zoom-out would be nice is not a big enough benefit to justify buying a different mfr's scope and having to learn it's operation and perhaps hit other shortcomings. If the zoom-out was a life-saver, I might thing differently, but now I don't think that's the case. I haven't reached a decision yet, but I'm leaning toward the subject Siglent scope.

BTW, the few times I needed support the vendor TEquipment was very helpful and even got answers from Siglent for me. Although I never tried to get answers directly from Siglent myself. And I have to say, the SDS1104x-e I have has been very satisfactory.

-------------------------------------------------------------------------------------------------------------------------------
Waveform details: Basically I capture a 360khz wide, 27 step sweep with a 10.7mhz center frequency. The sweep takes about 1/2 second to complete the 27 steps. The 27 step marker pulses are on chan 3, with the RF on chan 1. The goal is to verify the frequency is correct at each of the 27 markers.
-------------------------------------------------------------------------------------------------------------------------------
Thanks, again, to all who contributed. Any further comments are welcome.
Frank

Do consider the other ways to capture that waveform as suggested above, and in relation to the diagram in the main thread on this topic (https://www.eevblog.com/forum/testgear/oscilloscope-zoom-out-quirk/msg4484857/#msg4484857)
for your specific example of 500ms, the SDS800 with 50Mpts of memory allows for a sample rate of 100MS/s which might be enough to capture 10.7MHz cleanly. Depending on noise that may resolve the frequency accurately (easiest to just test, it won't be far off your current scope). Or you might find it better/quicker to use segmented capture as mentioned above to store only a short segment of each stable frequency after it has settled. Depends what you are looking at.

Rarely is capturing around the window ("zoom out") actually improving anything at all. You're still limited by the available memory either way. All it does is save a little bit of screen space by not showing the full data all the time.

THAT IS ALL IT DOES, people need to stop pretending like there is some magical unsaid benefit.

You've stepped into this bizarre exaggerated theatre of a topic, it is being pushed by some people who want attention/views and once again trigged the same old crap that comes up every single time. Plenty of misleading noise which only makes it more confusing.

[electr_peter]

On Siglent SDS800X HD there is no setting of time/div, memory size, etc. that results in possibility to zoom out, ever. Only visible screen width is captured (that can be zoomed in with long up to 50 or 100M memory) and that's it. Screen edges will be always be blank if you try to zoom out. EDIT: not all Siglent models behave this way, on some you can set fixed memory depth to capture beyond display.

On Keysight InfiniVision scopes, there are specific conditions under which zoom out is possible - fast time base at max or max/2 sample rate, screen width that does not fully fill available memory and pressing STOP/SINGLE with new trigger event. Then there is data beyond screen edges which you can zoom out and analyze in any way you like. In all other conditions zoom out is not possible, only screen width is captured (including RUN mode at fast timebases). Of course, you can zoom in as well in any mode.

Is zoom out useful? Definitely, it allows quicker understanding of signal behavior at fast time base (main focus) and at slower time base without re-capturing again. With Siglent scopes (models having no fixed memory setting), you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still). It is kinda frustrating that Siglent scopes (models having no fixed memory setting) at fast time bases record only measly few hundred or k points on screen in STOP/SINGLE mode (with massive 50-100M in reserve) and you should be somehow happy about it -  that leaves you blind in some cases.
Is zoom out used much? In my case, sometimes. I do not really on it much, but it speedups analysis.
Is zoom out reverse of zoom in? No, it's a memory management thing mainly and usage/usefulness is a bit different than zoom in function.
Is zoom out critical to have? It is more of nice to have than must have. You can get around it with slower time base and zoom in instead, but I would prefer to have this feature in a scope.

IMO, confusion about zoom out stems from auto vs manual memory setup philosophy. Automatic memory management in Keysight scopes is very well implemented (with many "tricks" or "cheats") with no manual control given to the user and it covers >99% scope use cases (plus extra capabilities like zoom out). Some just do not like this in principle and instead prefer to have manual control to set acquisition parameters themselves (with a bit more effort and essentially the same results in the end, but with no extras like zoom out). As mentioned, zoom out capability may or may not be available in a specific timebase - this ambiguity is not liked by some.

EDIT: added clarification on Siglent models as they differ in memory management.

[Someone]

With Siglent, you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times does this need to be repeated?
You do not have to use that workflow, if you wish to navigate around a larger window then capture it in the first place using zoom (or whatever method you prefer). When stopped all the data is there and there is no need to recapture as you claim.

[frank2644]

Thanks, again for more good info.

I did manage to successfully confirm the marker frequencies of my sweep generator. I had impose the 20mhz bandwidth  limit and the frequency measurements became stable and correct. It was a slow process to measure each segment, but iI was able to do it. No doubt others here could have figured out a faster way.

More importantly, I don't think a zoom out capability would have helped. nctnico and 203055's recording method sounds good.

Thanks again,
Frank


 

[KungFuJosh]

20mhz bandwidth

To keep communication clear, please follow standard labeling. Hertz is Hz. Lowercase m is millihertz, uppercase M is Megahertz. That's a slightly large difference. 😉

[electr_peter]

With Siglent, you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times does this need to be repeated?

Full quote:

Is zoom out useful? Definitely, it allows quicker understanding of signal behavior at fast time base (main focus) and at slower time base without re-capturing again. With Siglent scopes (models having no fixed memory setting), you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times context of posts will be selectively ignored to discredit zoom out capability?

You do not have to use that workflow, if you wish to navigate around a larger window then capture it in the first place using zoom (or whatever method you prefer). When stopped all the data is there and there is no need to recapture as you claim.

Yeah, no need to recapture because you ignored first capture which was deemed too short/fast for some reason (how did you arrive to such conclusion without capturing, anyway?) and then recaptured on purposely slower time base. Oh, wait

Keysight can capture at fast timebase AND gets more memory (aka "zoom out") AND can do whatever analysis with this capture. OR it can capture at slower timebase and do whatever.
Siglent scopes (models having no fixed memory setting) can capture at fast timebase AND gets less memory thus FORCING you to recapture again in slower timebase if fast timebase capture memory if insufficient for analysis. Only after recapturing at slower timebase (which now gives adequate memory) you are free to do whatever.

Keysight memory management is more flexible at fast timebases (while having no downsides) compared to Siglent. On Siglent scopes (models having no fixed memory setting) you get less memory at fast timebases. On Siglent scopes (models having fixed memory setting) you get same or more memory at fast timebases, but at slower update rate.
Keysight scope gives more capture memory at fast timebases or is faster in RUN mode, obvious advantage. This fact does not change if some users do not appreciate/choose to ignore it.

Let's say you are in hypothetical T&M cafe drinking coffee. Waiter asks if you want anything else and you say: "Can I have a biscuit, please?"
If T&M cafe is run by Siglent (considering no fixed memory setting scope like SDS800HD), waiter responds: "Biscuit is not available to customers who have ordered coffee already. However, we will make biscuit if you order it with an additional coffee."
If T&M cafe is run by Keysight, waiter responds: "I will check if biscuit is available and get it for you. If not, we will make biscuit if you order it with an additional coffee."
Which cafe provides better service?
In this analogy exchange coffee to single capture and biscuit to memory beyond display limits and you will get the gist.

EDIT: added clarification on Siglent models as they differ in memory management.

[frank2644]

20mhz bandwidth

To keep communication clear, please follow standard labeling. Hertz is Hz. Lowercase m is millihertz, uppercase M is Megahertz. That's a slightly large difference. 😉

Thanks, I never knew that.

It's gonna be a challenge to remember that, I'm lucky to avoid using cps.
https://en.wikipedia.org/wiki/Cycle_per_second#References

Frank (very old guy)

[Fungus]

]How many times does this need to be repeated?
You do not have to use that workflow

How many times does it need to be repeated that it's the most natural, logical, sensible workflow.

There is NO POSSIBLE EXCUSE for just taking seven samples at maximum zoom and nothing on either side.

Hence the quip about scissors. Given the wide availability of right-handed scissors, why would anybody force themselves to learn and use left-handed ones?

[2N3055]

]How many times does this need to be repeated?
You do not have to use that workflow

How many times does it need to be repeated that it's the most natural, logical, sensible workflow.

There is NO POSSIBLE EXCUSE for just taking seven samples at maximum zoom and nothing on either side.

Hence the quip about scissors. Given the wide availability of right-handed scissors, why would anybody force themselves to learn and use left-handed ones?

You can repeat false statement milion times, it is not going to be less false...

Logical to you means nothing. Logical based on what premises?
That you choose to ignore how instrument should be used?

And no, it is not sensibile workflow. It your chosen workflow.
And it makes no sense to me, for instance. Deliberately slowing down the scope all the time because I might need some more data because I twiddle the knobs randomly without thinking so basically I need all the data I can so maybe I stumble on something by sheer luck...

You want to be able to look around event that lasts 100 µs? Set time base to 10µs/div and fit it on the screen. Then choose which part you want to look into detail. That is for stopped event.

You want to look at the detail and whole picture at the same time? Set time base to 10µs/div and fit it on the screen. Then enable zoom and look at the detail while seeing overview of the whole event all the time for easy navigation.

That is simplest way of doing it. It guarantees that you will 100% be able to see what you need, and guarantees optimum work of the scope at all times.
That applies to Keysight and Rigol too.

Look, you can use scope any way you want . Even stupid way. Other people will do it their way.
If you can achieve result using your way and you are happy with process good for you.

[tautech]

With Siglent, you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times does this need to be repeated?

Full quote:

Is zoom out useful? Definitely, it allows quicker understanding of signal behavior at fast time base (main focus) and at slower time base without re-capturing again. With Siglent, you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times context of posts will be selectively ignored to discredit zoom out capability?

You do not have to use that workflow, if you wish to navigate around a larger window then capture it in the first place using zoom (or whatever method you prefer). When stopped all the data is there and there is no need to recapture as you claim.

Yeah, no need to recapture because you ignored first capture which was deemed too short/fast for some reason (how did you arrive to such conclusion without capturing, anyway?) and then recaptured on purposely slower time base. Oh, wait

Keysight can capture at fast timebase AND gets more memory (aka "zoom out") AND can do whatever analysis with this capture. OR it can capture at slower timebase and do whatever.
Siglent can capture at fast timebase AND gets less memory thus FORCING you to recapture again in slower timebase if fast timebase capture memory if insufficient for analysis. Only after recapturing at slower timebase (which now gives adequate memory) you are free to do whatever.

Keysight memory management is more flexible at fast timebases (while having no downsides) compared to Siglent. Scope gives more capture memory at fast timebases, obvious advantage. This fact does not change if some users do not appreciate/choose to ignore it.

Let's say you are in hypothetical T&M cafe drinking coffee. Waiter asks if you want anything else and you say: "Can I have a biscuit, please?"
If T&M cafe is run by Siglent, waiter responds: "Biscuit is not available to customers who have ordered coffee already. However, we will make biscuit if you order it with an additional coffee."
If T&M cafe is run by Keysight, waiter responds: "I will check if biscuit is available and get it for you. If not, we will make biscuit if you order it with an additional coffee."
Which cafe provides better service?
In this analogy exchange coffee to single capture and biscuit to memory beyond display limits and you will get the gist.

Peter, you name Siglent several times in your chapter about scope zoom use.

Respectfully I would like to inform you not all Siglent DSO's use the same zoom mechanisms and those with deep memory have a magnitude more zoom out than most on the market when set to do so.

However like 2N3055 mentions, there is a performance price to pay for this capability which is the reason why Siglent, LeCroy and others prefer not to capture off screen or employ tricks to do so.

[electr_peter]

Peter, you name Siglent several times in your chapter about scope zoom use.

Respectfully I would like to inform you not all Siglent DSO's use the same zoom mechanisms and those with deep memory have a magnitude more zoom out than most on the market when set to do so.

Fair point, I wrote with Siglent SDS800X HD model in mind (as in initial question of this topic) just not to repeat full name all the time. Some Siglent scopes can be set up to have "zoom out" behavior manually with fixed memory length setting (but slowing update rate). I will adjust post to mention this. EDIT: posts were updated

However like 2N3055 mentions, there is a performance price to pay for this capability which is the reason why Siglent, LeCroy and others prefer not to capture off screen or employ tricks to do so.

Sampling full memory all the time will slow down things, no doubt. I am familiar with Keysight memory tricks which allow both fast update in RUN mode and full memory in STOP/SINGLE mode with no downsides, which proves that auto memory management is better than manual in most cases. These points are mentioned in my posts above.

[Someone]

With Siglent, you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times does this need to be repeated?

Full quote:

Is zoom out useful? Definitely, it allows quicker understanding of signal behavior at fast time base (main focus) and at slower time base without re-capturing again. With Siglent scopes (models having no fixed memory setting), you would have to set slower timebase, capture again and zoom in - more steps and effort (not much, but still).

How many times context of posts will be selectively ignored to discredit zoom out capability?

You do not have to use that workflow, if you wish to navigate around a larger window then capture it in the first place using zoom (or whatever method you prefer). When stopped all the data is there and there is no need to recapture as you claim.

Yeah, no need to recapture because you ignored first capture which was deemed too short/fast for some reason (how did you arrive to such conclusion without capturing, anyway?) and then recaptured on purposely slower time base. Oh, wait

You are the one saying there is a certain number of ways to use a scope and listing them. Completely ignoring and avoiding the case of using zoom from the start:

User wants to capture a trace longer than what they are viewing on the screen (for whatever reason)
User sets their scope to capture that longer window, for "around screen" method they set a longer memory depth and adjust timebase for desired view, for zoom method they set the horizontal time base and then zoom in to the desired view
Now when the magic single capture arrives where they want to "zoom out"
"around screen" method adjusts horizontal timebase, zoom method adjusts zoom control

User did not need to recapture another trace/event/trigger. That is an imaginary problem that you invent and pretend like there is no possible way around it.

Pretty much the only difference between these is, "around screen" does not show the zoom window so the view is a little larger, zoom method has full control of where the wider capture is placed. They are barely different but we have people like you consistently adding confusion and noise and claiming there is no possible way to achieve anything like "zoom out" when that has been debunked and explained so many times it is getting silly. Differences are incredibly minor, thats all that needs to be said. Instead we seem to end up with this misdirection and shouting about "CANT ZOOM OUT" and people new to these products end up confused and mislead rather than informed and confident.

Keysight scope gives more capture memory at fast timebases or is faster in RUN mode, obvious advantage. This fact does not change if some users do not appreciate/choose to ignore it.

Expect it does NOT do that in run mode, only when you press stop and the trigger is periodic and another the trigger occurs in a brief moment after the user presses stop.

Which was explained several times in the linked mega thread on this topic. Why keep posting the same things which have been debunked and shown to be incorrect so many times before?

[Someone]

There is NO POSSIBLE EXCUSE for just taking seven samples at maximum zoom and nothing on either side.

Capturing data around the view is an important parameter which the user should be able to choose. Adding memory around the screen reduces the waveform update rate. The nonsense claims of zero downside are plainly false.

Why keep coming back with already debunked claims? Why would anyone trust your comments when these ones are so plainly lies?

edit* Here is the specific example of this specific point being explained in depth with examples and context, 4 years ago in the mega thread:
https://www.eevblog.com/forum/testgear/oscilloscope-zoom-out-quirk/msg3499730/#msg3499730

[Someone]

Look, you can use scope any way you want . Even stupid way. Other people will do it their way.
If you can achieve result using your way and you are happy with process good for you.

Yes, this is a big part of the fake argument. Continually pretending that other people wouldn't use features and capabilities available to them and not explaining why. Then turning it around and telling people there are only a limited number of ways to use a scope if you don't use my preferred method the other methods I describe have HUGE PROBLEMS.....    while failing to mention the comparable methods.

Same thing happens with the selective comparisons some users make between brands, picking some clearly poorly matched/suited example and ignoring/rejecting the more appropriate comparisons.

[thm_w]

Zoom is great but saying its optimal to use when you have a smaller 7" screen, sometimes its not. This is one trace, which is about the limit I'd want to go:

Capturing data around the view is an important parameter which the user should be able to choose. Adding memory around the screen reduces the waveform update rate. The nonsense claims of zero downside are plainly false.

So maybe an ideal would have two memory modes:
- Auto - Performance
- Auto - Capture area

[2N3055]

Zoom is great but saying its optimal to use when you have a smaller 7" screen, sometimes its not. This is one trace, which is about the limit I'd want to go:

Capturing data around the view is an important parameter which the user should be able to choose. Adding memory around the screen reduces the waveform update rate. The nonsense claims of zero downside are plainly false.

So maybe an ideal would have two memory modes:
- Auto - Performance
- Auto - Capture area

SDS800xHD has 2 memory modes: Auto and fixed sample rate.
Manual memory mode is reserved for higher models.

It is seriously inexpensive scope. With tons of capability. Can't have everything.

[tautech]

Zoom is great but saying its optimal to use when you have a smaller 7" screen, sometimes its not. This is one trace, which is about the limit I'd want to go:

Zoom this is but both Horizontal and Vertical Modes and very cool for an entry level DSO. 

[electr_peter]

@Someone, @tautech, I respectfully would like to ask to read my posts carefully. I do not claim at all that there is some best process of how to zoom in/out to view both long and short timebase events, it is up to the user entirely. All I say is that scopes with zoom out (or more precisely, memory beyond display) are more flexible in some circumstances than scopes without this feature. Especially Keysight implementation.

Yet my clear statements are ignored to push for "zoom in is better process anyway, so why care about zoom out benefits" agenda. This shifting of focus from case of "user looks at fast timebase as a MAIN focus and decides to look at slower timebase as a secondary check" to "user needs to capture slower timebase all the time and look for full details" is tiring. In former case, zoom out capability allows not to recapture again - this fact is conveniently ignored, I wonder why? Of course, user then can go to latter case (with MAIN focus on slower timebase) where zoom in can be used straight away.

That is an imaginary problem that you invent and pretend like there is no possible way around it.

That is your mental gymnastics to ignore the circumstances where "zoom out" allows more options. I do not say there is no way around it, please don't put words in my mouth.

Keysight scope gives more capture memory at fast timebases or is faster in RUN mode, obvious advantage. This fact does not change if some users do not appreciate/choose to ignore it.

Expect it does NOT do that in run mode, only when you press stop and the trigger is periodic and another the trigger occurs in a brief moment after the user presses stop.

Expect that it exactly what I am saying in quoted post (plus all explanation before which you selectively ignore). To clarify for you, "Keysight scope [gives more capture memory] at fast timebases OR [is faster in RUN mode], obvious advantage." Due to automatic memory management, in RUN mode Keysight is fast with short memory, but in STOP/SINGLE mode gives full memory. You cannot do both with Siglent without changing settings all the time.

Which was explained several times in the linked mega thread on this topic. Why keep posting the same things which have been debunked and shown to be incorrect so many times before?

I only stated facts which in no way were showed to be false or debunked. All I get is mental gymnastics to counter the facts and selective misquoting.
I do not claim there is some superior process of how to use scope. Some try to diminish Keysight scopes where they have slight advantage in auto memory management to improve look of Siglent scopes. Yes, advantage is there, it is small, but it is there which allows some extra flexibility. You don't have to change your process of using scope because of that.

[tautech]

@tautech, I respectfully would like to ask to read my posts carefully. I do not claim at all that there is some best process of how to zoom in/out to view both long and short timebase events, it is up to the user entirely. All I say is that scopes with zoom out (or more precisely, memory beyond display) are more flexible in some circumstances than scopes without this feature. Especially Keysight implementation.

Yet you keep ignoring my clear statements to push for your "zoom in is better process anyway, so why care about zoom out benefits" agenda. You simply shift goal post from case of "user looks at fast timebase as a MAIN focus and decides to look at slower timebase as a secondary check" to "user needs to capture slower timebase all the time and look for full details". In former case, zoom out capability allows not to recapture again - this fact is conveniently ignored, I wonder why? Of course, user then can go to latter case (with MAIN focus on slower timebase) where zoom in can be used straight away.

Zoom out is just one, repeat just one analysis process !

I much prefer thoughtful use of a DSO with emphasis on storage where at slow timebase settings provide the best advantage.
However, let's also not forget intelligent use of the trigger suite available, let alone the Search capability that can also be engaged in order to make the term look redundant !

Why engage the brain when the processing power in the modern DSO is far more powerful ? 

[electr_peter]

Why engage the brain when the processing power in the modern DSO is far more powerful ? 

Why put blinders on and mentally limit available options from scope suite?
Anyway, user chooses process that is most suitable for application at hand, I not arguing what one "should use".

[tautech]

Why engage the brain when the processing power in the modern DSO is far more powerful ? 

Why put blinders on and mentally limit available options from scope suite?
Anyway, user chooses process that is most suitable for application at hand, I not arguing what one "should use".

Nor am I.
A strong feature set is there to use not to ignore.

Okay, so you like the Keyshite way, I get that, so show us a $440 model with that capability ?

[2N3055]

Why engage the brain when the processing power in the modern DSO is far more powerful ? 

Why put blinders on and mentally limit available options from scope suite?
Anyway, user chooses process that is most suitable for application at hand, I not arguing what one "should use".

I agree. Let's not put blinders on.

So why are you ignoring that Keysight does not "capture outside the screen".

What you see on the screen is separate, later capture from separate trigger than the one that made you press Stop.
If you have repetitive signal, like clock or even some AWG waveform that may be complex but 100% repetitive and with stable well defined trigger, you won't know the difference.

Example 1. 
- Signal is single sine period burst repeating every 1 µs
- Normal trigger mode.
- Capture at 20ns/div.  (Image 1)
- Stop
- Change timebase (not zooming out, not using Zoom feature of 3000T) to 50µs/div.
- (Image2) We see scope actually captured cca. 40µs before and cca. 360µs after trigger.  It is 400µs total, balance before/after trigger is connected with where trigger point is.

Great. This i what you say you use and it is very useful to you.

Example 2.
- Signal is single sine period burst repeating every 500 ms
- Normal trigger mode.
- Capture at 20ns/div. (Image 1 as previously, nothing changed here)
- Stop
- Change timebase (not zooming out, not using Zoom feature of 3000T) to 100ns/div.
- (see Image 3) Wait, there are no samples outside the area defined by the screen at 20ns/div! There is only 200ns worth of data. Not 400µs.

How come?
Because scope DOES NOT capture outside the screen. Ever.

When you press stop it waits for next trigger event and captures some other, random part of waveform, uncorrelated to what was on the screen when you press Stop.
It performs following action, automatically and without warning to the user (pretending that it captures more data than it does):

- Once you press Stop, scope waits for another trigger. It can wait for as long as 200-250ms for it so that can be distance between two events.
- If there is a trigger within that timeout period, it will take one 400µs long (short?) single capture and then go to STOP state.
- If there is NO trigger within timeout period, it will simply show last captured trigger, but just screen bound 200ns

So you get something only on timebases faster than 20µs/div, you never get anything more than 400µs worth, and it some random capture up to 250ms away from where original screen was when you pressed stop.
That is not 1µs or 100µs, but a whooping quarter of a second...

So basically, to observe the neighborhood of some event of the screen, it as useful as simply changing timebase to whatever you want to look at while scope is still in the RUN mode and then STOP.
That is exactly what Keysight does, but automatically. I guess it is good if you cannot be bothered to change timebase then stop instead of just single lazy STOP.

But result is the same, you get some random uncorrelated time in signal when trigger activates. If that is OK for you because signal is monotonic and auto repetitive to such degree that it does not matter where you look, then what is the point of "looking around". It is all going to be the same. Perfect replica. Once you put whole period on screen, all others are the same.
If signal is highly agile, than Keysight is useless. Dangerous I may say. It will show you false data somewhere else than what you were looking at...
It is trick, and the scope lies to you. literally.

Also, by changing timebase, I can get ANY time period, not just puny 400µs.
Capturing long, using long memory to keep sample rate at maximum, is equivalent of 0 blind time for period captured. In the captured period nothing can hide.

Note, I have MSOX3104T for years now. I like it and think at the time it was really good scope. I know in detail how it works.
And for doing deep signal inspection of the type you are talking about I don't use it.

I use Siglent  or Picoscope. That are deep memory scope. You get long capture and STOP. Nothing can hide in there.

[nctnico]

Why engage the brain when the processing power in the modern DSO is far more powerful ? 

Why put blinders on and mentally limit available options from scope suite?
Anyway, user chooses process that is most suitable for application at hand, I not arguing what one "should use".

Indeed.   

The whole discussion reminds me of when I worked at the local chocolate factory for one day when I was a teenager. I had to pull bars of chocolate onto trays from a conveyer belt using an piece of aluminium. I had to reach out quite far so I used one hand to support myself and the other to rake in the chocolate so my back wouldn't hurt at the end of the day. Boss lady comes around and tells me to use two hands because god gave me two hands. FFS!    I guess she either never had to do this task herself or just wanted to check my ass when I bent over.