new Oscilloscope choice

[kcbrown]

No 'scope is perfect but it takes extra steps to achieve something that other devices simply do then that's the opposite of "intuitive".

"Intuitive" is dependent at least in part on one's experience, and that's the point here.  If all you've used is analog scopes and you start to use a digital scope, you're already used to "what you see is all you get", because that is how analog scopes work.  The Siglent's approach will match your expectations in that case, and you won't be surprised when elongating the timebase when the scope is stopped doesn't show you more of the waveform.

But if you've used most digital scopes, then your expectations will be that you can zoom out, and a scope like the Siglent will be unintuitive to you as a consequence.

Honestly, each approach has advantages and disadvantages, and which dominates is quite obviously arguable (hence the continuous arguments ).  Nothing in principle prevents any scope manufacturer from implementing both strategies and allowing the user to select his preferred approach.  They don't do that because they don't believe they'll get enough additional sales from it to make it worth the effort, and they're almost certainly right about that.

In any case, it should be clear that the disadvantages to Siglent's approach, such as they are, aren't nearly enough to impact their sales enough to matter.  If they were, Siglent would almost certainly have changed their approach in response.

[kcbrown]

Just two weeks ago I had to debug a problem with a chip (64 pin QFN so not the easiest to probe) not wanting to work and I wanted to check whether it was getting SPI messages for it's configuration. I used a Tektronix TBS2000 scope set to full memory length. The SPI messages just come in one burst during startup so capturing an entire sequence isn't easy. I set the scope up but with Linux and a whole bunch of other software layers on top the timing of the SPI messages isn't very predictable. Instead of the expected burst I only got one SPI message so the time between the SPI messages was longer than expected. By twisting the time scale knob (zooming out) I got the rest on screen too. So I didn't need to count to the right pin, push the probe onto a tiny pad and restart the software again in order make a new measurement. I don't get why people still want to insist being able to zoom-out isn't useful! 

How well would it have worked out if you had set up the scope to trigger on SPI messages and used segmented memory to capture batches of them?  It sounds like you had the scope configured to look at something else, or to trigger on something else.   (But maybe the TBS2000 doesn't have segmented memory, and if that's the case then your example here is really an example of how one feature can compensate for the absence of another).

There's no question that being able to zoom out after the fact is a very handy thing, and adds flexibility, but it's pure luck as to whether or not what the scope manages to capture is enough.  It's better than WYSIAYG for that, of course, but it's still luck all the same.

[Zlotnik]

Guys.
Please.
Let it rest!

You're polluting another thread with endless repetitions of the same arguments. Signal to noise is dropping fast! Again.

I think you all got the point across to the thread starter that the Siglent memory management implementation is controversial, but some like it a lot - in particular considering the price point and feature set of the SDS2k+. And I say that as a not-fan of the Siglent implementation, and a big fan of subtle UI goodness over headline features: The SDS2k+ is great bang-per-buck and not everybody is bothered by its idiosyncrasies.

To the thread starter:
It really is horses for courses. Take a look at my post half a gazillion memory/zoom arguments earlier.

[tv84]

Nothing in principle prevents any scope manufacturer from implementing both strategies and allowing the user to select his preferred approach.  They don't do that because they don't believe they'll get enough additional sales from it to make it worth the effort, and they're almost certainly right about that.

You've nailed it with the analog evolution comparison. 

But, I don't totally agree with you on this quote. I recon that it's pretty easy for Keysight to limit it's capture to "what you see" (with a couple of instructions). I don't think it would be easy for Siglent to do the opposite. I guess it would be a huge modification (Siglent, you can always prove me wrong...  ).

Maybe as more people get started to scopes in the digital world, Siglent's way starts to look as more "limited"...

[tv84]

Guys.
Please.
Let it rest!

You're polluting another thread with endless repetitions of the same arguments. Signal to noise is dropping fast! Again.

I think you all got the point across to the thread starter that the Siglent memory management implementation is controversial, but some like it a lot - in particular considering the price point and feature set of the SDS2k+. And I say that as a not-fan of the Siglent implementation, and a big fan of subtle UI goodness over headline features: The SDS2k+ is great bang-per-buck and not everybody is bothered by its idiosyncrasies.

I agree BUT extruding the mem management part, this whole thread is also "polluting another thread with endless repetitions of the same arguments" for the same scope choices. All because of the OP's endlessly repeated question.

Totally agree on the advantages of the SDS2k+.

[nctnico]

Just two weeks ago I had to debug a problem with a chip (64 pin QFN so not the easiest to probe) not wanting to work and I wanted to check whether it was getting SPI messages for it's configuration. I used a Tektronix TBS2000 scope set to full memory length. The SPI messages just come in one burst during startup so capturing an entire sequence isn't easy. I set the scope up but with Linux and a whole bunch of other software layers on top the timing of the SPI messages isn't very predictable. Instead of the expected burst I only got one SPI message so the time between the SPI messages was longer than expected. By twisting the time scale knob (zooming out) I got the rest on screen too. So I didn't need to count to the right pin, push the probe onto a tiny pad and restart the software again in order make a new measurement. I don't get why people still want to insist being able to zoom-out isn't useful! 

How well would it have worked out if you had set up the scope to trigger on SPI messages and used segmented memory to capture batches of them? 

That would have been way overkill for a simple measurement like I needed. Also, triggering/recording specific parts of a signal might skip important clues on what might be going wrong. A simple, single trace showing the number of messages (10 or so in total), idle state and the spacing in between is good enough for a measurement like this. Anything else you throw at it is just a waste of effort.

[Fungus]

You've nailed it with the analog evolution comparison. 

It's a pretty picture but I don't believe for a minute that that's the reason Siglent did it. Other 'scopes have been doing it for a long time before the Siglent appeared.

Honestly, each approach has advantages and disadvantages

I fail to see any advantages at all in the Siglent way.

Guys.
Please.
Let it rest!

You're polluting another thread with endless repetitions of the same arguments. Signal to noise is dropping fast! Again.

The other thread is already finished.

[kcbrown]

Nothing in principle prevents any scope manufacturer from implementing both strategies and allowing the user to select his preferred approach.  They don't do that because they don't believe they'll get enough additional sales from it to make it worth the effort, and they're almost certainly right about that.

You've nailed it with the analog evolution comparison. 

But, I don't totally agree with you on this quote. I recon that it's pretty easy for Keysight to limit it's capture to "what you see" (with a couple of instructions). I don't think it would be easy for Siglent to do the opposite. I guess it would be a huge modification (Siglent, you can always prove me wrong...  ).

Maybe as more people get started to scopes in the digital world, Siglent's way starts to look as more "limited"...

Yeah, it might or might not be easy to do depending on how the firmware is architected, what assumptions were made when designing the FPGA internal setup, etc.  Even the memory interface design might have some assumptions built into it with respect to the capture implementation.

With respect to Siglent appearing to be "limited" in the digital world, I dunno.  The main advantage to Siglent's approach (and note that this advantage isn't dependent upon WYSIAYG) is that segmented memory is always active.  If you're doing digital work and are triggering on bus messages, the end result is that you'll see the beginning of each individual message in each segment (whether you capture the whole message or not depends on your timebase relative to the amount of time between message start and message end). 

The way I see it, scopes like the Instek and others that use the user-specified amount of memory per capture should *also* automatically divide the total memory into segments of that configured capture size.  This would get you the advantages of being able to zoom out while also getting you the primary advantage of Siglent's "always on" segment implementation.  It would also maximize the chance that every feature of the scope would work with segments, something that isn't necessarily the case for scopes that implement segmented memory as an afterthought (the Instek GDS-1054B I have, for instance, will not show the FFT on a per-segment basis when displaying the segments, while the Siglent obviously does).

[Fungus]

The main advantage to Siglent's approach (and note that this advantage isn't dependent upon WYSIAYG) is that segmented memory is always active.

What's the maximimum number of segments?

[kcbrown]

It's a pretty picture but I don't believe for a minute that that's the reason Siglent did it. Other 'scopes have been doing it for a long time before the Siglent appeared.

Even Siglent's earlier scopes, such as the SDS1000CML series, used the traditional approach.

I can't say what reason Siglent had for changing their approach, but I have a suspicion that it wasn't their idea, but rather LeCroy's.

Honestly, each approach has advantages and disadvantages

I fail to see any advantages at all in the Siglent way.

The primary advantage to WYSIAYG is obviousness.  You know just by looking what you're going to get.  There's no guessing or computation needed to understand how much time a given capture is going to represent.  It's right there in front of you.

Standard implementations require that you know the sample rate, know the amount of memory per capture, and divide the former into the latter in order to derive the capture time.  If there are scopes that will show you the total capture time, I don't know of them (but my experience is limited to Siglent, Rigol, and Instek, though I've seen plenty of videos in which Tektronix and Keysight scopes were used, and never noticed such a thing on any of them).

Now, maybe it's important to know how much time is being captured and maybe it's not.  But obviously that's going to depend on the specific use case.

You asked for an advantage to the Siglent approach, and there it is.  Such as it is.

[kcbrown]

The main advantage to Siglent's approach (and note that this advantage isn't dependent upon WYSIAYG) is that segmented memory is always active.

What's the maximimum number of segments?

It's dependent on the scope.  For the 2104X+, it appears to be 90K.  That's with a 1K segment size, which requires that you set your timebase to 50 ns/div if your channel configuration is getting you 2GS/s, and a single channel active.  It seems that Siglent has two separate banks of acquisition memory, each of which gets two channels.

Setting the timebase smaller than that doesn't get you any more segments.  Weirdly, it seems doing that can decrease them, which I don't understand at all.

I can understand why the maximum number of segments on the Siglent would not be just the amount of memory divided by the capture size, since each segment is going to have some storage overhead associated with it. 

[nctnico]

It's a pretty picture but I don't believe for a minute that that's the reason Siglent did it. Other 'scopes have been doing it for a long time before the Siglent appeared.

Even Siglent's earlier scopes, such as the SDS1000CML series, used the traditional approach.

I can't say what reason Siglent had for changing their approach, but I have a suspicion that it wasn't their idea, but rather LeCroy's.

I think so too. And owning a Lecroy Wavepro 7000 series myself I also have an idea why: Lecroy scopes are geared towards signal analysis. However, the more memory you have to go through, the longer it takes to process and the lower the update rate. Even with a beefy Pentium 4 processor the Wavepro scope slows down to a crawl with some stacked math functions enabled. So by limiting the acquisition length to what is minimally needed to fill the screen Lecroy makes sure to give the user the best performance possible from the scope when using it for signal analysis. Siglent seemed to have blindly copied it (mainting Chinese tradition) without considering all the use cases.

And having history mode always enabled isn't an advantage of the Siglent approach. Yokogawa and R&S also have history mode while allowing to set a fixed memory length. AFAIK Lecroy doesn't have history mode on the Wavepro 7000 series. You really need to seperate various aspects of oscilloscope memory management.

[kcbrown]

I can't say what reason Siglent had for changing their approach, but I have a suspicion that it wasn't their idea, but rather LeCroy's.

I think so too. And owning a Lecroy Wavepro 7000 series myself I also have an idea why: Lecroy scopes are geared towards signal analysis. However, the more memory you have to go through, the longer it takes to process and the lower the update rate. Even with a beefy Pentium 4 processor the Wavepro scope slows down to a crawl with some stacked math functions enabled. So by limiting the acquisition length to what is minimally needed to fill the screen Lecroy makes sure to give the user the best performance possible from the scope when using it for signal analysis. Siglent seemed to have blindly copied it (mainting Chinese tradition) without considering all the use cases.

Seeing how Siglent is an OEM for LeCroy, even if they considered the use cases, they might easily have decided that it was more economical in the long run to stick with the LeCroy approach and to refine that.  There's probably no way to really know without having an insider view of things. 

And having history mode always enabled isn't an advantage of the Siglent approach. Yokogawa and R&S also have history mode while allowing to set a fixed memory length. AFAIK Lecroy doesn't have history mode on the Wavepro 7000 series. You really need to seperate various aspects of oscilloscope memory management.

Most certainly, the "always on" history segment capability isn't dependent upon WYSIAYG, and I did note that.  But it is a part of Siglent's approach to things, and I think it's a good choice because, firstly, it ensures that memory is always used to its fullest even if the user configures a smaller capture size, and secondly, it makes segments a first-class mechanism in the system, something that will be accounted for in every other operation that's implemented.

I can't remember the specifics now, but Instek's segment implementation (as an example) has some arbitrary limitations, and conflicts with some of the other features the scope makes available.  And that almost certainly wouldn't be the case if segments were a first-class object in the implementation, instead of an afterthought.

[tv84]

The primary advantage to WYSIAYG is obviousness.  You know just by looking what you're going to get.  There's no guessing or computation needed to understand how much time a given capture is going to represent.  It's right there in front of you.

Agree. Conceptually, I find it much easier to implement.

[rf-loop]

There is idea about optionally curtains for peoples who do not like to see whole capture length as need also look with WICIWYS principle.


Here first displayed standard model out from factory.
After stop user can zoom in and out. And also look interesting details what he in this case know they exist. Because he see whole acq. all time.
Of course, if there is a sequential performance model of the brain, it can be difficult to detect both windows. No need worry and feel bad with scope. There is solution. That is why curtains are available as a DIY option.



Here with optional retrofit curtains. All available from many kind of needlework DIY shops.
I will recommend this Option for peoples who really do not want see whole length in runtime.
After stop user can zoom in and out. And also look IF there is some interesting details.


Legal notice: Anyone is free to use the revolutionary invention presented here for their own private or commercial purposes without any restriction or royalties.

 

[2N3055]

 

[kcbrown]

There is idea about optionally curtains for peoples who do not like to see whole capture length as need also look with WICIWYS principle.

You should patent this.   You'll make a fortune!

 

[nctnico]

And having history mode always enabled isn't an advantage of the Siglent approach. Yokogawa and R&S also have history mode while allowing to set a fixed memory length. AFAIK Lecroy doesn't have history mode on the Wavepro 7000 series. You really need to seperate various aspects of oscilloscope memory management.

Most certainly, the "always on" history segment capability isn't dependent upon WYSIAYG, and I did note that.  But it is a part of Siglent's approach to things, and I think it's a good choice because, firstly, it ensures that memory is always used to its fullest even if the user configures a smaller capture size,

How is that any different than configuring the DSO to use full memory size and using it as a single acquisition? At one end you reason you want to see what you get but on the other end you find a variable (!) number off-screen acquisitions in history mode good to have. Why is having multiple individual acquisitions off-screen in memory good while having off-screen data from a single acquisition is not good?

I can't remember the specifics now, but Instek's segment implementation (as an example) has some arbitrary limitations, and conflicts with some of the other features the scope makes available.  And that almost certainly wouldn't be the case if segments were a first-class object in the implementation, instead of an afterthought.

You need to seperate two things here: user interface and acquisition engine.

For the acquisition engine segmented recording is nothing more than cycling the acquisition position counter between a lower and higher boundry (start & end point). In fact most (if not all DSOs) have segmented recording at the base of the operation of the acquisition engine in order to have double buffering (*). Once the acquisition is done the acquisition position counter and trigger point are stored and the acquisition is handed over to the rendering engine. This is not rocket science to implement (been there, done that).

It depends on the user interface on how the segments are displayed & handled. There is a wild variety of ways and depending on what you need one DSO may be better suited than the other but this doesn't mean segmented recording as a base has been added as an afterthought. Displaying segments is not GW Instek's strong suit but they have a rather unique feature which allows to do statistic analysis on the recorded segments.

* If you dig deeper into the user manual of an oscilloscope you can spot the tell-tale signs of double buffering. On Keysight scopes you get double the memory in single mode and on the recent GW Instek scopes segmented recording mode can use twice the size of the specified memory length.

[Fungus]

The main advantage to Siglent's approach (and note that this advantage isn't dependent upon WYSIAYG) is that segmented memory is always active.

Segmented mode can be useful but I'd like to be able to turn it off and just use all the memory. I'm sure zooming out is the more useful of the two modes.

The real problem with the Siglent is that you don't have the option to do it. Even in Segmented mode you could still capture a bit more on either side (the trigger point can be in the center of the screen so that's not where it starts capturing). Wouldn't it be nice to set an "overcapture" value to be able to capture three or four more screens worth...?

Dave's video shows an example where the Siglent captures only 10 samples. That's ridiculous.

The primary advantage to WYSIAYG is obviousness.  You know just by looking what you're going to get.  There's no guessing or computation needed to understand how much time a given capture is going to represent.  It's right there in front of you.

I've never felt I needed to know exactly how much more there is to left/right of the screen. It's just nice to have something (anything!).

FWIW my Micsig has a little indicator at the top to give you an idea how much extra there is at any given time. The area inside the brackets [...] is what's currently visible on screen. In this image the memory depth is set to "Auto" and there's a couple more screenfuls on either side.

[2N3055]

This is exactly how all the hunting accidents happen... Shooting in the dark in random direction, hoping you hit something. Even your fellow huntsman is fine, as long you shot something, anything... 

Simply, deliberately scope wider and then inspect... Or you would like scope to be smarter than you..?

And every scope that has same sample rate captures 10 samples at that timebase. It is what YOU ordered scope to do. If you need more, order more.

Or, when you order coffee, a waiter brings coffee and water, whiskey, a steak and desert, because, you know, you might want those too...
If I want steak, I'll order a steak and get exactly that, instead of ordering coffee and hope waiter will telepathically recognize I'm hungry too...

With coffee, you'll get glass of water, and maybe one of those little cookies... That is reasonable. Everything else.. neeeh..

[Fungus]

This is exactly how all the hunting accidents happen... Shooting in the dark in random direction, hoping you hit something. Even your fellow huntsman is fine, as long you shot something, anything... 

Simply, deliberately scope wider and then inspect... Or you would like scope to be smarter than you..?

If only all the real-world signals were predictable, too.

What was that? You captured an infrequent glitch and wanted to zoom out to see what came before/after? Oh, dear.

[tv84]

With coffee, you'll get glass of water, and maybe one of those little cookies... That is reasonable. Everything else.. neeeh..

Having the possibility to refill is always a nice-to-have even if we end up not using it. Refill would be the zooming out... 

[rf-loop]

The real problem with the Siglent is that you don't have the option to do it. Even in Segmented mode you could still capture a bit more on either side (the trigger point can be in the center of the screen so that's not where it starts capturing). Wouldn't it be nice to set an "overcapture" value to be able to capture three or four more screens worth...?

False.

Look my last message bottom image with curtains option. It can use just as you like. There is most part of acquisition length now not visible (49ms in this image, left side and right side.). After you stop you can zoom out  etc.

[ROMUZ]

Hello!
Sorry for entering to this topic, but could you advise please - is the Tonghui TDO3102B enough for ripple measurements of low voltage DC power supplies?
It is quite old model, but I've got it almost for free as "NOS".
Thanks!

[2N3055]

With coffee, you'll get glass of water, and maybe one of those little cookies... That is reasonable. Everything else.. neeeh..

Having the possibility to refill is always a nice-to-have even if we end up not using it. Refill would be the zooming out... 

With espresso you don't get refills.. Just good coffee...