Oscilloscope Zoom Out Quirk

[Fungus]

But I do think it is MAJOR change in the core of the system

No it isn't. The capture routine has to work with any memory size (you can zoom out until you get all the memory, right?)

All they need to do is always use all the memory instead of computing a smaller memory size and using only part of it.

In reality it should simplify the code.

[Martin72]

When it´s so simple to change, when it seems it got only advantages when always using the full memory amount- Why do a major brand like teledyne lecroy not use it?
When it´s so important for many one in the world, why won´t they change it in the last decades, instead they still got satisfied customers around the world..Are these all blind dumbasses ?
There must be a advantage when lecory handle it in this way - and a disavantage on the other side.

  (...) Which (being perfect on entry level products) for some reason is not requirement for big brands like Keysight and R&S and LeCroy.  They are allowed to make products that are deliberately limited for marketing purposes. With them it is business decisions and "small niggles". With Siglent and Rigol it's "crap I wouldn't buy"...

True, true...
And it´s a very annoying thing.
A real superb scope model like the "new" 2K siglent doesn´t have a feature like this zoom out thing - Reading this thread, people who are interesting to buy it may get the impression that this absence is a very bad flaw that would make it unusable.

[nctnico]

When it´s so simple to change, when it seems it got only advantages when always using the full memory amount- Why do a major brand like teledyne lecroy not use it?

Because traditionally Lecroy serves a different market; the market where waveforms/s don't matter so much but signal analysis does. More geared towards scientific use. AFAIK Lecroy offers the widest number of signal analysis tools on their scopes (after you pay for all the options). But there are a few things Lecroy doesn't support (on their own designs): peak detect and AFAIK averaging and high resolution are available as (slow) math traces only instead of being produced by the acquisition engine. All in all Lecroy has a very purist approach towards signal acquisition. This is fantastic for signal analysis (I really like my Wavepro 7300A for that) but as a general purpose bench scope it is less than ideal. And calling Lecroy a major brand is a bit of an overstatement; to me they seem to be aiming towards high profit niches. Much of their other gear consists of rebadges.

[Martin72]

And calling Lecroy a major brand is a bit of an overstatement

...........
 

[tautech]

And calling Lecroy a major brand is a bit of an overstatement..........

Name the 2 manufacturers making 100+ GHz scopes.

[Martin72]

GW Instek and Micsig  

[tautech]

GW Instek and Micsig  

Only in their dreams.

[nctnico]

And calling Lecroy a major brand is a bit of an overstatement

...........
 

Just look at Lecroy's sales revenue. A quick Google search learns me that it is dwarfed by that of Keysight, Tektronix, Rhode & Schwarz and Yokogawa by a factor of 5 to 20 (not combined but compared to each!). Lecroy really isn't a big player in the test equipment market. And you may laugh at GW Instek but they do half of Lecroy's revenue and equal to that of Rigol and Siglent combined.

[Fungus]

When it´s so simple to change, when it seems it got only advantages when always using the full memory amount- Why do a major brand like teledyne lecroy not use it?

Who knows?

But that's "argument from authority". It's not a reason for Siglent to follow them or an argument that Siglent is "doing the right thing".

All it means is that TDL is doing it wrong, too.

[tautech]

No SDS2104X Plus left ...customers have them all so out comes zoom on a SDS5104X.......
What's not to like......still an 8" display of which half is still usable.
Cluttered display ? 

All available memory used.
2 displays each with its own timebase and each accessible by clicking or touch.
~160 packets consisting of ~920 bytes and zoom in or out to your heart's content.

[2N3055]

When it´s so simple to change, when it seems it got only advantages when always using the full memory amount- Why do a major brand like teledyne lecroy not use it?

Who knows?

But that's "argument from authority". It's not a reason for Siglent to follow them or an argument that Siglent is "doing the right thing".

All it means is that TDL is doing it wrong, too.

What is an "argument from authority" ? Fact that Nico and you say it's easy so that's it?

It is not "who knows why". LeCroy has dozens of whitepapers that explains why, but that requires reading and accepting the fact that one might not be smartest guy in the room and may be wrong.

So we have two punters who claim they know better than engineers that make 100 GHz scopes....

There is no argument here. We have Nico who have digital view of the world : his way and wrong way, and you who join any controversial discussion with no intention of contributing anything useful, just riling up emotions and provoking people.....Like Bugs Bunny, but without charm....

LeCroy, Siglent, Picoscope, Rigol (by default AUTO strategy), R&S (by default AUTO strategy), Keysight Infiniivision (Run mode), Micsig (by default AUTO strategy) and all others that have AUTO memory mode use that same method of only grabbing only amount of data that is equivalent to length of screen in current timebase. They all do it to minimize amount of data taken in every trigger to get faster reacting scope and keep sampling rate high to avoid aliasing.   It is like analog scope, you see what you see. There is NO data before or after screen, and there is no need to. If you want to see more, you put more time ON the screen. Benefit of digital scope is that you can stop it and then zoom in and still see details. 

There is no simpler or more logical way of doing it, and as I said before, most of the scopes come by default set for that use pattern. Tektronix added AUTO mode on latest models, because people were asking for this "LeCroy" mode, because twiddling memory length manually all the time is pain in the ass.  Why would stupid scope always collect 0,1 second of data, update 10 times a second (like a multimeter), and then throwaway everything except 1 usec of data ( throwing away 99,999% of data) and then show that on the screen. How is that efficient, good engineering.

What Nico does is similar to Rube Goldberg machine, where he achieves same result by setting several setting, doing mental math and paying attention that they all sit together in the end by side effect to give him his data.
Instead of using the doorknob to open the door, he devised a system of levers and pulleys that opens the door from the lever on the wall next to the door , because he doesn't like how doorknob makes the door look. And then one day, he realized, if he moves that lever low enough, he can open the door with a foot. Which is kind of neat when your hands are full. And then he saw the light and now it is the only righteous way..
Rest of use just use doorknobs, and open the door with an elbow if our hands are busy. He likes his handles and pulleys, his complicated ways of doing simple things ? Great, glad for him. Me, I'll keep using doorknobs, like millions of others.

[Fungus]

What is an "argument from authority" ?

I did link to the Wikipedia page, but the short version is "not an argument based on logic or evidence".

LeCroy, Siglent, Picoscope, Rigol (by default AUTO strategy), R&S (by default AUTO strategy), Keysight Infiniivision (Run mode), Micsig (by default AUTO strategy) and all others that have AUTO memory mode use that same method of only grabbing only amount of data that is equivalent to length of screen in current timebase.

Nope. All of them grab the amount of data which is the current memory depth. That can easily be more data than fits on screen.

They all do it to minimize amount of data taken in every trigger to get faster reacting scope and keep sampling rate high to avoid aliasing.

Are you saying Keysights are slowed down and/or suffer aliasing because of this? My Rigol certainly wasn't.

It is like analog scope, you see what you see. There is NO data before or after screen, and there is no need to.

 

If you want to see more, you put more time ON the screen. Benefit of digital scope is that you can stop it and then zoom in and still see details. 

What if the event is a rare one and you can't simply "put more time on screen" then re-record it?

There is no simpler or more logical way of doing it

Um, yes there is. It's a much more natural workflow to simply zoom out from what you're looking at.

[2N3055]

What is an "argument from authority" ?

I did link to the Wikipedia page, but the short version is "not an argument based on logic or evidence".

LeCroy, Siglent, Picoscope, Rigol (by default AUTO strategy), R&S (by default AUTO strategy), Keysight Infiniivision (Run mode), Micsig (by default AUTO strategy) and all others that have AUTO memory mode use that same method of only grabbing only amount of data that is equivalent to length of screen in current timebase.

Nope. All of them grab the amount of data which is the current memory depth. That can easily be more data than fits on screen.

They all do it to minimize amount of data taken in every trigger to get faster reacting scope and keep sampling rate high to avoid aliasing.

Are you saying Keysights are slowed down and/or suffer aliasing because of this? My Rigol certainly wasn't.

It is like analog scope, you see what you see. There is NO data before or after screen, and there is no need to.

 

If you want to see more, you put more time ON the screen. Benefit of digital scope is that you can stop it and then zoom in and still see details. 

What if the event is a rare one and you can't simply "put more time on screen" then re-record it?

There is no simpler or more logical way of doing it

Um, yes there is. It's a much more natural workflow to simply zoom out from what you're looking at.

It's yellow.

[nctnico]

What Nico does is similar to Rube Goldberg machine, where he achieves same result by setting several setting, doing mental math and paying attention that they all sit together in the end by side effect to give him his data.
Instead of using the doorknob to open the door, he devised a system of levers and pulleys that opens the door from the lever on the wall next to the door , because he doesn't like how doorknob makes the door look. And then one day, he realized, if he moves that lever low enough, he can open the door with a foot. Which is kind of neat when your hands are full. And then he saw the light and now it is the only righteous way..

And now back to ridiculing other people's workflow  I was hoping my previous message would appeal to your maturity but appearantly it hasn't. How are we supposed to take you serious when you keep pulling nonsense out of your ass? Just accept the fact that you don't understand everything. Understanding why 'zoom out' is more efficient is one of them.

A simple example: looking at the data in the middle of an I2C message (only part of the message on screen). A scope with deep memory length will capture the start (and thus is able to decode the message) without explicitely needing to setup a zoom window and think what kind of time/div that zoom window needs to have. I2C usually runs at tens of kHz at least. Even a relative short span like 10Mpts of memory @1Gs/s gives you a total time of 10ms. That is enough to capture an I2C message a few bytes long (clocked at 10kHz) several times over. Why make setting up an oscilloscope more troublesome than it has to?

[rsjsouza]

No SDS2104X Plus left ...customers have them all so out comes zoom on a SDS5104X.......
What's not to like......still an 8" display of which half is still usable.
Cluttered display ? 

In my opinion, yes. I have less screen area to what I really want to see from the waveform itself, not the entire capture. Sure, a Pico will never have this issue, but an 8'' screen (just like my DS4014) this is not good.

LeCroy, Siglent, Picoscope, Rigol (by default AUTO strategy), R&S (by default AUTO strategy), Keysight Infiniivision (Run mode), Micsig (by default AUTO strategy) and all others that have AUTO memory mode use that same method of only grabbing only amount of data that is equivalent to length of screen in current timebase.

Sinisa, that is where you lose me. No question about the benefits or characteristics of the Auto setting, but another is the fact that in a Rigol I can set a memory depth and the oscilloscope will obey my command. That is not what is seen in Dave's video: Siglent will size it according to what it thinks it is better.

And no, I don't need to do "mental math and paying attention that they all sit together" either.

Otherwise, I have been in lenghty interface and ease-of-use discussions to know these things are hard to convince if there is not a very clear benefit one way or another.

[tautech]

No SDS2104X Plus left ...customers have them all so out comes zoom on a SDS5104X.......
What's not to like......still an 8" display of which half is still usable.
Cluttered display ? 

In my opinion, yes. I have less screen area to what I really want to see from the waveform itself, not the entire capture. Sure, a Pico will never have this issue, but an 8'' screen (just like my DS4014) this is not good.

You may have misunderstood. 
10" native display but in Zoom mode 8" of corner to corner usable display still remains where in this image only half is used by 3 traces including the decode display.
This is a pile of screen real estate in which to add a further trace, apply Maths, Histograms, Statistics, Measurements etc any of which can be positioned on the remaining half of the display to ones liking.



A stopped/captured waveform record is where the power of a DSO can then be applied to magnify amplitude, Pan or Search the whole record with a range of parameters to ensure the waveform integrity meets requirements of which for SDS5000X we have the minimum 125 Mpts at this primary timebase setting. SDS2000X Plus will show a minimum 100 Mpts record with the same scope settings.

These are relatively new and powerful scopes that like any new device/tool/component takes time to comprehend and properly utilize in a manner we may have not used or fully understood when compared to previous experience.
 

[nctnico]

@Tautech: please stop making a fool out of yourself by pretending everyone but you is stupid. DSOs have had zoom mode and math since the late 80's. Oscilloscope users are well aware that it exists; there is nothing new or special about zoom mode and math these days.

[2N3055]

Sinisa, that is where you lose me. No question about the benefits or characteristics of the Auto setting, but another is the fact that in a Rigol I can set a memory depth and the oscilloscope will obey my command. That is not what is seen in Dave's video: Siglent will size it according to what it thinks it is better.

And no, I don't need to do "mental math and paying attention that they all sit together" either.

Otherwise, I have been in lenghty interface and ease-of-use discussions to know these things are hard to convince if there is not a very clear benefit one way or another.

Rafael, it is probably me not being able to explain well. Rigol comes by default in Auto, and I met more than few people that keep using it only in Auto, because for interactive work it is better, than having to think about capture memory, and still have scope work as fast as possible. If you want to understand how Siglent memory works, put your Rigol in Auto memory and observe how it behaves.
That is what I wanted to explain. I also repeated many times that Rigol is better in this regard, because you can choose fixed memory too. Which you use sometimes also, which is OK, because sometimes it can be useful. That is fine.

About mental math, that is maybe debatable. Why? Because, when you set 14 Mpts memory depth, how long SPI (or any other event) sequence can you get with it ? And by long I mean how much time you get in total? 10 ms, 20 ms, 330 ms ?

Please see attached excel table and verify if I calculated wrong...

In short , if you're looking at short timebase and you don't need fast refresh, you will have a lot data captured off the screen.
But that is because your scope have huge 140(70)MPts  memory. Nicos scope has 80(40)MPts, so you loose some advantage.
If you have 10 Mpoint scope, it looks less good. But still usable.

Problem is that, after hundreds of messages, I managed to make Nico say what is his main use for it. What he said it was capturing long data on serial buses, while looking at very short part of that capture that has some event or packet of interest. He keeps scope in RUN mode (not Single), and if he sees the event (error, or whatever is of interest) he manually stops the scope (events are sparse, so there is time to react) and then can move inspect whole capture for details. Captures are quite long, on order of 0,1 -2 seconds, done on low sample rates.

To achieve that, you cannot run too fast timebase, because at 80Mpts you need 50Msps/sec to achieve 1,6 sec, for instance. How do you setup your scope to do that?

Regards,

[Someone]

LeCroy, Siglent, Picoscope, Rigol (by default AUTO strategy), R&S (by default AUTO strategy), Keysight Infiniivision (Run mode), Micsig (by default AUTO strategy) and all others that have AUTO memory mode use that same method of only grabbing only amount of data that is equivalent to length of screen in current timebase.

Nope. All of them grab the amount of data which is the current memory depth. That can easily be more data than fits on screen.

2N3055 is correct, and provided the specific modes which cause this to occur. The "current" memory depth on many scopes varies and can't always be set explicitly. This entire thread is that extremely narrow point.

They all do it to minimize amount of data taken in every trigger to get faster reacting scope and keep sampling rate high to avoid aliasing.

Are you saying Keysights are slowed down and/or suffer aliasing because of this? My Rigol certainly wasn't.

Why auto? Lets go way back in time:
https://www.eevblog.com/forum/testgear/new-tektronix-tbs2000-oscilloscopes/msg1283038/#msg1283038
Thats a Rigol (the specific brand you asked for) causing aliasing or slow update rates if the user selects an inappropriate memory depth. On older Tek scopes there wasn't an auto memory setting and it was quite annoying to go into the menus to adjust the memory depth every time you wanted to change capture lengths (some debugging does well to look at small details and then larger ones, back and forth). Keysight megazoom scopes? They don't let the user set any of that and always maintain the highest update rate (in run mode, in single trigger they fill the full depth), aliasing isn't an issue with them as they have very effective filtering to prevent it.

[2N3055]

2N3055 is correct, and provided the specific modes which cause this to occur. The "current" memory depth on many scopes varies and can't always be set explicitly. This entire thread is that extremely narrow point.

Thank you for summarizing that better than I ever could.

Thats a Rigol (the specific brand you asked for) causing aliasing or slow update rates if the user selects an inappropriate memory depth. On older Tek scopes there wasn't an auto memory setting and it was quite annoying to go into the menus to adjust the memory depth every time you wanted to change capture lengths (some debugging does well to look at small details and then larger ones, back and forth). Keysight megazoom scopes? They don't let the user set any of that and always maintain the highest update rate (in run mode, in single trigger they fill the full depth), aliasing isn't an issue with them as they have very effective filtering to prevent it.

EDIT. deleted.

[nctnico]

Sinisa, that is where you lose me. No question about the benefits or characteristics of the Auto setting, but another is the fact that in a Rigol I can set a memory depth and the oscilloscope will obey my command. That is not what is seen in Dave's video: Siglent will size it according to what it thinks it is better.

And no, I don't need to do "mental math and paying attention that they all sit together" either.

Otherwise, I have been in lenghty interface and ease-of-use discussions to know these things are hard to convince if there is not a very clear benefit one way or another.

About mental math, that is maybe debatable. Why? Because, when you set 14 Mpts memory depth, how long SPI (or any other event) sequence can you get with it ? And by long I mean how much time you get in total? 10 ms, 20 ms, 330 ms ?

It doesn't matter as long as it is enough. It is like the gas gauge in a car you have driven for a decent amount of time. If it points at 'half full' you know how far you can get (and need to refuel for the trip ahead or not) because you are familiar with the car. Not everything needs to be quantified to 10 digits after the decimal point to in order to be accurate enough, usefull and/or true. How much gas is there in the car? 'Enough' is sufficient to know.

Problem is that, after hundreds of messages, I managed to make Nico say what is his main use for it. What he said it was capturing long data on serial buses, while looking at very short part of that capture that has some event or packet of interest. He keeps scope in RUN mode (not Single), and if he sees the event (error, or whatever is of interest) he manually stops the scope (events are sparse, so there is time to react) and then can move inspect whole capture for details. Captures are quite long, on order of 0,1 -2 seconds, done on low sample rates.

Now you are mixing up several scenarios I outlined.

[SilverSolder]

The zoom out thing is obviously useful in some scenarios, where you have both "trees" and "forests" going on, and you are switching between the two perspectives. 

How did this end up being so hotly debated?  - got to love the Internet! 

[Someone]

How did this end up being so hotly debated?

Its not hotly debated, like has happened on other topics a minority (or single person) keeps reposting misleading information for some unknown reason. They can't fully justify or support their claims but keep coming back for more ridicule.

It'd be great if scopes offered the choice to set all these parameters, but absolutely none do. Its that all the different scopes have different ways to achieve the same thing, and their tradeoffs are different. What you like I might hate, which is fine. But I don't go out of my way to insist everyone else should like what I do, and all scopes should change their way of working to match my ideas. I also don't bomb every thread even slightly related to shout about how particular brands work in ways I don't prefer, blowing up tiny issues that no-one asked about.

The fun thing is there has been a manufacturer who radically changed their operation mode: Tek. Auto vs non auto, which is most popular? which causes more support requests?

[SilverSolder]

[...]
The fun thing is there has been a manufacturer who radically changed their operation mode: Tek. Auto vs non auto, which is most popular? which causes more support requests?

Is there anything quite as disheartening as staring at an oscilloscope screen that refuses to display any trace? 

[Martin72]

An elder Note from Lecroy about the advantage of long memory:

http://cdn.teledynelecroy.com/files/appnotes/lecroy_long_memory_in_digital_scopes_appnote010.pdf

The uploaded pic shows a part of the rigol ds1000z user manual...