Siglent SDS2000X Plus

[rf-loop]

Memdepth : It´s manual selectable...

That sounds very promising. Does it always stay that way?

AFAIK this is misunderstanding or careless expression.  (nice to see you also ask this question even when you know answer already)

I don't know the answer and that is why I ask the question.  To me it is utterly and totally unclear on whether this oscilloscope can be set to have a fixed memory depth or not (without needing to resort to workaround by using zoom mode c.q. even if that results in data being outside of the display). The original incarnation of this oscilloscope would revert back to automatic memory depth in many cases which was very annoying. In later incarnations the memory depth seems to be always automatic according to your information. Waiting for Martin to (finally) get a clear answer on this topic....

I think perhaps your memory is too short.
Here Performa01 answer your own question November 2019...in this same thread.

Re: Siglent SDS2000X Plus coming
« Reply #16 on: November 16, 2019, 01:51:21 am »

Or perhaps you think this can just change suddenly as some random small FW fix... No.

Also this whole question have handled many other times and in every corner you rise up this same question just for advertise this question.
Also one adverse effect IF user can force max memory for fast timebases is that he loose partially or whole wfm history buffer aka always background working slow segmented memory acquisition or dramatically change it, but of course this is not reason because when user can select something he is also responsible about consequences. Most of scopes keep only last  capture data and DPO brightness information in display memory is only what are left from previous acquisitions. Siglent works very different. With pros and cons.

What I hope is that Siglent change TFT for 4:3 or even turn it 90 degree and develop this hardware based zoom window even more better with more vertical room for display. Also need deveop perhaps bit more sophisticated zoom window positioning system and finally so that times are related tightly to trigger position and wash and clean brains from center of (main) display thinking. Trigger position need be always time zero. As we know in Siglent trigger can today select for fixed position or  (usual, default) fixed offset  (fixed offset is from center what is imho also partially wrong. It must be fixed offset from user selected trigger position what is always as time zero. But this is for  other discussion)

[nctnico]

Or perhaps you think this can just change suddenly as some random small FW fix... No.

You never know as Siglent keeps adding new features to firmware. As you can see from the replies from other forum members the way Siglent works is different compared to what is standard amongst oscilloscope manufacturers. And worse: it is causing loss of sales opportunities.

[thinkfat]

On the RTB you can say 10(20)M memory and even at 1ns timebase it'll capture that amount of memory at 2.5gs/s. So you can single capture then pull the timebase out and see more data.

It's the same even on the lowly Rigol DS1054Z. If I set the memory depth to 24Mpts, then even at 5ns/div the full 24Mpts are captured. After stopping the sweep, I can zoom out up to 2ms/div and have a look what happened long before the trigger event. I can also engage the second timebase and zoom back in and have the full details sampled at 1GSa/s available for inspection. This is one of the true merits of deep memory scopes and I cannot understand why Siglent would choose a different approach.

[Martin72]

For me it seems "normal" that memdepth will reduce the shorter the timebase is.

And now I know why...

Today at work, I grab some of our scopes ( all from lecroy), turning them on and having a look.
All lecroy models, 20yrs old waverunner, our new flagships waverunner 9054 ( 20GSa/s), hdo 6034 (12bit), a wavesurfer 432 ( from iwatsu 2005), a wavesurfer 3024 (from 2018)……
They all got one thing in common:
Same behaviour as the siglent scope….memdepth will decrease when the timebase becoming shorter.
Even by our newest scopes, the expensive waverunner and hdo.
So I couldn´t believe that this would be a failure or a bug in general over the years and models from lecroy.

Btw., the only scope in our house that "keeps" the full memdepth at any timebase is the rigol ds1054z...

Apart from that, how realistic are the chance, that this behaviour could be "fixed" via firmware ?

[maginnovision]

There is virtually no chance this couldn't be changed if they wanted to change it. How much work that'd be I have no idea. I believe the micsig scopes also retain memory depth when explicitly set.

[supperman]

My Tektronix MDO3k seems to capture the full memory depth that you set regardless of time-base. I guess that is the point to setting the memory in the first place.. since that setting would serve no purpose if it was time-base dependent.

[Martin72]

Before the asian low cost scope came up, I´ve done measurings with lecroy models only, so I don´t care about the memdepth "problem" of the siglent.
But....
Having 200mpts memory is somekind of worthless, when you can use it very limited.
You only got it in ms-ranges….
The selectable maximum memdepth is only vaild for slower timebases - above it´s functionless….

[supperman]

Before the asian low cost scope came up, I´ve done measurings with lecroy models only, so I don´t care about the memdepth "problem" of the siglent.
But....
Having 200mpts memory is somekind of worthless, when you can use it very limited.
You only got it in ms-ranges….
The selectable maximum memdepth is only vaild for slower timebases - above it´s functionless….

Why? I'm not suggesting it is a good thing.. but if you want to record all that data why not zoom out, capture and zoom back in. As long as you are capturing at 2 Gs there is no loss of fidelity in the Y domain. It may be a UI decision.

Siglent has segmented memory too.. Are we sure that people are using the memory options correctly??? I would think there is a lot that can be configured.. perhaps a "frame" of segmented memory is limited to the zoom level.. and the # of frames depends on your overall memory depth setting.. etc.. This thing can "save" hundreds of triggered frames.. for later review, right?

[Martin72]

This is the point where I get a little bit confused today.
Obviously lecroy, a "A-Brand", does the memory thing handle like siglent ( and for what they´re "blamed") for decades.
I did a capture of a 1Mhz sine in ms-range and zoomed in with no problems because 200mpts memory was active.
But in reverse this wouldn´t go on the siglent because of the memory decreasing.
Is this a really problem....

[nctnico]

Before the asian low cost scope came up, I´ve done measurings with lecroy models only, so I don´t care about the memdepth "problem" of the siglent.
But....
Having 200mpts memory is somekind of worthless, when you can use it very limited.
You only got it in ms-ranges….
The selectable maximum memdepth is only vaild for slower timebases - above it´s functionless….

Why? I'm not suggesting it is a good thing.. but if you want to record all that data why not zoom out, capture and zoom back in.

That is way too tedious. Just one example: often I'm triggering on a signal detail in order to tweak it until it is right. Every now and then I want to zoom out to check whether the rest of the timing is still OK. Your method would mean to zoom in/out every time. That means twisting the time/div knob hundreds of times. Way too slow and way too much things to deal with. With >4 signals, decoding,  measurements, etc zoom mode eats away way too much of the screen. And it is not like the Siglents are really cheap compared to the competition which does memory management in a normal way.

[Martin72]

It was stated (and proofed) that rigol uses the full memory at any time, R&S probably too, Tek probably also.
Keysight/Agilent/HP I don´t know.
But Lecroy do it the same way like siglent….Lecroy, a brand named in a row with Tek and Keysight and of course R&S.
This makes me confusing…

[thinkfat]

This is the point where I get a little bit confused today.
Obviously lecroy, a "A-Brand", does the memory thing handle like siglent ( and for what they´re "blamed") for decades.
I did a capture of a 1Mhz sine in ms-range and zoomed in with no problems because 200mpts memory was active.
But in reverse this wouldn´t go on the siglent because of the memory decreasing.
Is this a really problem....

That depends largely on your habits (and what you typically do). The Siglent way means, if you want to make use of the deep memory, you need to remember starting from a slow timebase, setting up a potentially complex trigger for an event that happens once every blue moon and not seeing what you captured until you zoom in. I personally find that a little impractical.

[maginnovision]

I'm in agreement with thinkfat and nctnico. I trigger on one thing then I want to make sure others are in line after that. I don't typically zoom out because especially when you have lots of stuff on screen that hurts visibility(even with the R&S resizing ability), but I do scroll forward and back. I'm typically using 2-4 analog and 12-16 digital, so the higher the resolution and the bigger the screen the better. It's a big part of why I still have the RTB not the RTM or RTA. For some things I'd love the low noise RTA, and I'd like to have power analysis from either... But considering the BW is fine they just aren't offering me enough to switch. The SDS2000X plus has more downsides and, honestly, having the "bigger" memory doesn't even help because I use the segmented(history) memory anyway which gives me much more overall memory.

[Martin72]

Thankyou, your post makes it clearer to me.
The "issue" that the siglent got is not a general one, only specific wise.

[0xdeadbeef]

All lecroy models [...] got one thing in common:
Same behaviour as the siglent scope….memdepth will decrease when the timebase becoming shorter.

The better Lecroys (e.g. Waverunner 6zi)  offer a selection between "Maximum Memory" and "Fixed Sample Rate" (in real-time sampling mode).
In the "maximum memory" setting, you select the "maximum sample points" and the scope adjusts the sample rate when changing the time scale (or pre/post trigger delays) to make use of the given sample points.
In the "fixed sample rate" setting, you select the sample rate which is kept whatever time base you select. "Lesser" Lecroys like the WS3000 don't offer this kind of flexibility but support only the "maximum memory" mode.

Anyway, note that when using a fixed memory length, even if this means that your record length is ten times longer that what you seen on the screen, this also decreases the update rate and increases the dead time. So while it can be handy to force a scope into doing so, this is usually not what you want as only or even default setting in normal trigger mode. In single trigger mode though, there is no real reason to not fill the whole memory. So when comparing the behavior of different scopes, the trigger mode could make a difference.

Of course you could argue whether acquiring offscreen data after the trigger event is always more interesting than offscreen data before the trigger event. So if capturing offscreen data was supposed to be an important feature, you'd also need a setup that allows to somehow define a maximum pre/post trigger offset.  I would think that no scope on the market offers something like that.

[rf-loop]

My Tektronix MDO3k seems to capture the full memory depth that you set regardless of time-base. I guess that is the point to setting the memory in the first place.. since that setting would serve no purpose if it was time-base dependent.

If it have 10M memory length and 5GSa/s and IF it always use full memory regardless of timebase how it reach nearly 300kwfm/s capturing speed. Yes of course it can not. With 5GSa/s alone 10Msamle takes 2ms. (I believe there is other mode for max wfm speed)
If user have 1ns/div setting and IF screen is (just for easy numbers) 1 div wide... user can see 10ns and 1999980 ns is out from display and rest are invisible included to true visual blind time.

Difference with Siglent is that you need look both on the screen at same time, long memory and then detailed slice as need,  if you want long memory and short timebase details. Often these cases do not need maximal wfm speed so this is not problem at all (and as previously told and is obvious no one can do fast long memory), still Siglent hw based second timebase is quite usable. You set 1st timebase and memory limit setting for wanted memory length  and then 2nd  timebase as fast as need for wanted details and then position 2nd timebase where ever you need inside long trace. Also force trigger position horizontally to fixed position so that you have ok amount of pre and post trigger lenth. Long memory trace (slow timebase)and then this narrow window, including that you can also in runtime look what ever place in whole lengtht and also use different time bases if need.  Why need hide this long trace and look only this tiny slice.
Disadvantage in Siglent is that zoomed window have less vertical room. Well enough for 8 bit resolution itself but if there is 16LA and 4 analog  then it is really too small window. What Siglent can do if they want is: When zoomed (and this is important mode in siglent) this long trace aka main window vertical size can reduce more and give much more room for this hardware based zoom window. Why they keep is as half screen. Also is it difficult to use some 3:4 TFT or otherways more vertical room, specially when we talk MSO scopes. This is optimal for situations where "normal mode" is zoomed window open.



Due to heavy lack of internet speed just now I can not search who told something about segmented acq. vs long mem or something like it.

1. Of course Siglent have fast segmented acquistion mode, up to 90000 segments depending segment length and some other things.

2. Siglent have also always background running slow segmented acquisition(current wfm speed). (wfm history buffer fifo what user can inspect after stop and also there every segment have time stamp and max amount of segments is depending about used memory length and some other settings but unlike fast segmented mode here user can not set segment limit number just segments fifo with max segments.)

[rf-loop]

Of course you could argue whether acquiring offscreen data after the trigger event is always more interesting than offscreen data before the trigger event. So if capturing offscreen data was supposed to be an important feature, you'd also need a setup that allows to somehow define a maximum pre/post trigger offset.  I would think that no scope on the market offers something like that.

Of course Siglent do....

  (some models with new FW) you can force trigger to wanted position in memory.
This can partially solve this need.
There is two modes in Siglent for trigger horizontal position.
1. Time offset mode from center of screen.
2. Fixed horizontal (memory/display) position mode and because display width from  left border to right border in Siglent = acquisition length aka memory length it can say that trigger can set to fixed position in currently used memory length. Of course if then user do other settings and memory length change its proportional position in memory length stay same. With this feature it can set even for " analog scope" style where trigger stay start of sweep (exept amount of fixed (if not adjustable in some specials...) trig delay what normal analog scopes have)

[rf-loop]

About new SDS2000X Plus SFRA aka Bode Plot.

SDS2kX+ have 50MHz internal signal generator.

I have not SDS2000X Plus and here in China all who knows are now spring holidays.

Who can check if this model can use internal signal generator for BP use only without FG license installed. Even with scopes with external SAG USB generator do not need SAG  license if use this only for BodePlot.

If use SDS2kX+ internal signal gen only for BP (BP controls it automatically) do it work now with latest FW without FG licence.

[edigi]

This is the point where I get a little bit confused today.
Obviously lecroy, a "A-Brand", does the memory thing handle like siglent ( and for what they´re "blamed") for decades.
I did a capture of a 1Mhz sine in ms-range and zoomed in with no problems because 200mpts memory was active.
But in reverse this wouldn´t go on the siglent because of the memory decreasing.
Is this a really problem....

In couple of cases I had the situation where the point where I could trigger and what I really wanted to check was far apart (concrete practical case: SPI messages but could be anything else as well). To be able to adjust the trigger I had to set a small timebase (zoom in) in then set a bigger timebase (zoom out) for capture then set a smaller timebase for the captured signal (zoom in) again and then scroll. A bit tedious (compared to just scrolling), but nothing I'd really call a problem (considering the price point).

Maybe it can be solved similarly how I did it with FFT, where FFT is also done using the samples that you see on the screen: Set a timebase that all needed samples (for example 1M) fit onto the screen and use zoomed signal window to see the actual signal shape (done via pushing the timebase button). In case of FFT it means 3 different windows (please don't ask why I want to see both the signal shape and FFT output; simply SA would have been complex to use) or alternatively continuously switch between FFT output and 3 orders of magnitude timbase difference for checking signal shape.

Naturally this DSO has an excellent display with much better resolution (compared to what I've used) that probably helps but I still could imagine some improvements in this regard.

I couldn't care less how the top DSO vendors do it as based on their price point (even used) I'm not their target customer and for a big pile of money they have probably a solution for just about anything.

[edigi]

1. Time offset mode from center of screen.
2. Fixed horizontal (memory/display) position mode and because display width from  left border to right border in Siglent = acquisition length aka memory length it can say that trigger can set to fixed position in currently used memory length. Of course if then user do other settings and memory length change its proportional position in memory length stay same. With this feature it can set even for " analog scope" style where trigger stay start of sweep (exept amount of fixed (if not adjustable in some specials...) trig delay what normal analog scopes have)

Do you mean when Delay displayed in the top left corner (see attached picture)? Hmm, this sounds great then.
It would be good to know which model/firmwares support this.

[tautech]

1. Time offset mode from center of screen.
2. Fixed horizontal (memory/display) position mode and because display width from  left border to right border in Siglent = acquisition length aka memory length it can say that trigger can set to fixed position in currently used memory length. Of course if then user do other settings and memory length change its proportional position in memory length stay same. With this feature it can set even for " analog scope" style where trigger stay start of sweep (exept amount of fixed (if not adjustable in some specials...) trig delay what normal analog scopes have)

Do you mean when Delay displayed in the top left corner (see attached picture)? Hmm, this sounds great then.
It would be good to know which model/firmwares support this.

All X-E, 2kX+ and 5kX models.
X-E: Utility>Reference Pos>Horizontal>Fixed Position.
Then no matter what timebase settings are used the H Pos stays in the same position on the display.

[nctnico]

All lecroy models [...] got one thing in common:
Same behaviour as the siglent scope….memdepth will decrease when the timebase becoming shorter.

The better Lecroys (e.g. Waverunner 6zi)  offer a selection between "Maximum Memory" and "Fixed Sample Rate" (in real-time sampling mode).
In the "maximum memory" setting, you select the "maximum sample points" and the scope adjusts the sample rate when changing the time scale (or pre/post trigger delays) to make use of the given sample points.
In the "fixed sample rate" setting, you select the sample rate which is kept whatever time base you select.

But that still doesn't give you acquisition beyond the edge of the screen even if there is enough memory to do so. And Keysight proves that it is perfectly possible to have high update rates AND deep memory at the same time.

Anyway, note that when using a fixed memory length, even if this means that your record length is ten times longer that what you seen on the screen, this also decreases the update rate and increases the dead time. So while it can be handy to force a scope into doing so, this is usually not what you want as only or even default setting in normal trigger mode. In single trigger mode though, there is no real reason to not fill the whole memory. So when comparing the behavior of different scopes, the trigger mode could make a difference.

Of course you could argue whether acquiring offscreen data after the trigger event is always more interesting than offscreen data before the trigger event. So if capturing offscreen data was supposed to be an important feature, you'd also need a setup that allows to somehow define a maximum pre/post trigger offset.  I would think that no scope on the market offers something like that.

In many cases waveform update rate doesn't really matter when debugging digital / mixed signal circuits. Regulary I use normal mode and take single shots of specific events. It would be logical to have the centre of the screen as the centre point of the memory (half left, half right). At the point where this starts to matter you'll need to choose the time/div and trigger offset settings so you capture the part of the signal you want. But with deep memory this isn't difficult. The biggest benefit of deep memory is that you don't have to care so much about the trigger.

[rf-loop]

And Keysight proves that it is perfectly possible to have high update rates AND deep memory at the same time.

Keysight own Salesmens University have perhaps own basic math book. 1M memory need one M samples to fill. If sampling speed is 1GSa/s it takes 1ms and even with zero blind time (impossible) 1kwfm/s. But when deep memory is selected and scope is running for high update rates it can not capture deep memory, but salesmens forget to tell it....  Yes there can do example last acquisition trick.... and fool users (I do not know what is Keysight salesmens trick for this).  After I have seen tens of years how they fulfill sales brochures and example compare to others in appnotes I do not wonder anything anymore.

But how about deep full mem measurements during this high speed update rate with deep memory, how salesmens jump over this lazy fox. Oh... they never promised...

[nctnico]

And Keysight proves that it is perfectly possible to have high update rates AND deep memory at the same time.

Keysight own Salesmens University have perhaps own basic math book. 1M memory need one M samples to fill. If sampling speed is 1GSa/s it takes 1ms and even with zero blind time (impossible) 1kwfm/s. But when deep memory is selected and scope is running for high update rates it can not capture deep memory, but salesmens forget to tell it....  Yes there can do example last acquisition trick.... and fool users (I do not know what is Keysight salesmens trick for this).

There is no fooling users. Now you sound like the sales brochure you loath so much.

Just run two acquisition systems in parallel. One for storage and one for display. I have implemented something similar a couple of years ago. Works like a charm.

But how about deep full mem measurements during this high speed update rate with deep memory, how salesmens jump over this lazy fox. Oh... they never promised...

No oscilloscope can measurements / math realtime. And it seems Siglent does ERES and Sin x/x interpolation in software which is even worse. But then again I don't really care about waveform update rates. I rather have an oscilloscope which offers a high rate of productivity (=least amount of twiddling knobs). The latter is where Siglent needs to improve.

[rf-loop]

1. Time offset mode from center of screen.
2. Fixed horizontal (memory/display) position mode and because display width from  left border to right border in Siglent = acquisition length aka memory length it can say that trigger can set to fixed position in currently used memory length. Of course if then user do other settings and memory length change its proportional position in memory length stay same. With this feature it can set even for " analog scope" style where trigger stay start of sweep (exept amount of fixed (if not adjustable in some specials...) trig delay what normal analog scopes have)

Do you mean when Delay displayed in the top left corner (see attached picture)? Hmm, this sounds great then.
It would be good to know which model/firmwares support this.

If we talk about pos 2.
Answer with your image no.
In fixed position mode it is not possible to set trigger position outside main display window.
Your trigger position is far outside of screen. Fixed Position mode trigger reference position is always inside display in user defined fixed horizontal position (what can translate also as fixed position in memory length).