Siglent SDS1202X-E odd behavior and questions

[borjam]

Hi,

I received a Siglent SDS1202X-E yesterday. I updated the firmware to the last version and I've been playing a bit in order to get familiar.
However, I have found something really puzzling:

I set it up just to play with signal capture using the built in calibration signal. DC coupled, 1 µs/div, 1V/div (10X), edge triggering on the raising edge, 1.5 V. Nothing fancy.

What I've found is, regardless of the memory depth I specify on Acquire->Mem Depth, it always captures just what I can see on screen. Let's say I want to fill the 14 Msample buffer. For instance, this is the behavior I see and which I wouldn't expect. I set the time base to 50 µs/div. I set it up in single mode and let it trigger. As I can see on screen it has filled just 700 pkts.

On a Rigol DS1000Z, however, if I set a large memory depth, it will fill it no matter what is visible on screen. So, the same settings on the DS1000Z result on 24 Mpts captured. With the trigger frozen I can use the horizontal position to peruse the captured signal.

Is this normal? Maybe it's because I am used to it, but the Rigol way in this case is really convenient. Fire the trigger and fill up the memory with pre-trigger signal up to the event that triggered it, and keep capturing until you will up the whole memory you have configured.

The Siglent, on the other side, just captures enough data to fill the screen. With those settings, for example (50 µs/div) it just stored 700 µs in memory. The Rigol stored 12 ms of data. What am I missing?

A question about the Siglent as well, as the documentation is really incomplete. May someone from Siglent explain what is ACQ->Acq Mode, which can be "slow" or "fast"'

[oleh]

I'm in almost same puzzled state. I even went further and bought another scope for the same price minus $19 discount from Owon. I'm researching now which one is going back to the store. I the thread which I created elder people said that there is an "history" mode in Siglent which supposed to store data in the entire buffer. It would take an extra cost, I suspect.

[borjam]

I'm in almost same puzzled state. I even went further and bought another scope for the same price minus $19 discount from Owon. I'm researching now which one is going back to the store. I the thread which I created elder people said that there is an "history" mode in Siglent which supposed to store data in the entire buffer. It would take an extra cost, I suspect.

The SDS1202X-E certainly has a "History" mode which works in a similar way to segmented capture. But it's not an extra. The SDS1202X-E is sold at a "closed" price, no additional licenses.

I can understand that the History mode can be convenient for some applications, but Siglent's implementation makes single shot trigger almost useless.

An example of what I did trouble free with a Rigol is this. I was debugging a Modbus (RS485) problem. So I connected the probe, set the maximum record depth, configured the start bit triggering, set the scope to "single" trigger mode and just fired the program. The scope captured the whole Modbus dialog and I could indeed see what had happened.

Of course I see how to do it with the Siglent (move the time base adjustment until the scope says it will fill the amount of memory you want) but the lazy method that you can use with the Rigol is not bad.

[oleh]

I believe there is the answer:

Select Sampling Mode

The oscilloscope only supports real-time sample. In this mode, the oscilloscope samples
and displays waveform within a trigger event. The maximum real-time sample rate is
1GSa/s

http://www.siglentamerica.com/USA_website_2014/Documents/UserManual/SDS1000X&Xplus_UserManual_UM0101X-E02A.pdf , p.39

[MrW0lf]

I'm quite sure many users also hang in "more memory than fits screen" mode not actually realizing what it does and wfm/s rate suffers + scope does useless work. Siglent and some other scopes assume use of zoom if you want see fine detail from large detailed capture. I do not know if Siglent supports maximizing zoomed view somehow or you always "lose" half of screen?

[borjam]

I'm quite sure many users also hang in "more memory than fits screen" mode not actually realizing what it does and wfm/s rate suffers + scope does useless work. Siglent and some other scopes assume use of zoom if you want see fine detail from large detailed capture. I do not know if Siglent supports maximizing zoomed view somehow or you always "lose" half of screen?

I understand the reasons for that, although you could control the amount of useless work done by reducing capture memory. For example you can do it on the Rigol.

And anyway when you set up the trigger in "single mode" there's no captures per second because you just want a single event, that's it.  "Single" is very useful to capture events you can control. Like, for example, actions produced by a program when you run it.

[MrW0lf]

And anyway when you set up the trigger in "single mode" there's no captures per second because you just want a single event, that's it.  "Single" is very useful to capture events you can control. Like, for example, actions produced by a program when you run it.

Well, "zoom" style scopes have some other important aspect that makes them far superior.
Ok, suppose you captured 24M of some protocol on Rigol. Zoomed in on fine detail. All you get is decoding (or auto-measurements) on microscopic window from whole capture + lots of knob work.
Now "zoom" style scopes sort of "suggest" that you never capture more than actually need, but when you do it will decode (or auto-measure) full record. With protocol it means decoding of whole captured block and you get full event table or auto-measurements across whole record in great detail. It's like having superpowers. Seeing all the needles in forest at once from mile away...
I'm sure master rf-loop will provide some luxorious example of that. If not I can provide one from my "zoom" scope (128MB fully decoded into Excel style table). So you actually you may not even need to zoom because if comms run smooth you are not interested in analog part at all, only actual messaging decoded into nice table with errors highlighted (if any).

[oleh]

And anyway when you set up the trigger in "single mode" there's no captures per second because you just want a single event, that's it.  "Single" is very useful to capture events you can control. Like, for example, actions produced by a program when you run it.

Well, "zoom" style scopes have some other important aspect that makes them far superior.
Ok, suppose you captured 24M of some protocol on Rigol. Zoomed in on fine detail. All you get is decoding (or auto-measurements) on microscopic window from whole capture + lots of knob work.
Now "zoom" style scopes sort of "suggest" that you never capture more than actually need, but when you do it will decode (or auto-measure) full record. With protocol it means decoding of whole captured block and you get full event table or auto-measurements across whole record in great detail. It's like having superpowers. Seeing all the needles in forest at once from mile away...
I'm sure master rf-loop will provide some luxorious example of that. If not I can provide one from my "zoom" scope (128MB fully decoded into Excel style table).

Scope can easily do "zoom" when it stores data in a buffer longer than the screen size. Opposite is not possible,

[rf-loop]

Siglent works differently.

If you run 1ns/div or 1ms/dif whole captured length is visible, same width as screen.

Of course oscilloscope can make for working so that it use long memory and t/div setting is nothíng more but "zoom" inside this length. What can win with this method. Imho nothing. But these equipments need diffeerent practice from user to operate. Also RPN calculator need different operation than arithmetic. Front and back drive cars need diffent hadling in some cases.

There is zoom for look details. Userr can select speed and memory so that he get length what he want and if need look insde acq length with more fst timebase just zoom. Even with 1ms/div you can zoom down to 1ns/div. And what important even in this case trigger is accurate and use fine interpolation between sampled points.

It is only if I want turn this knob or other. But if user is adapted to use different then he need learn new prctice so that it can do without thinking just as we walk without thinking how to set legs.

There is one important advantage in this "WICIWYS" principle working scope. In all cases whole captured length is visible. There is zero visual blind time.  Many times this same leads to wondering that Siglent example decode only what is screen byt then these peoples forget that whole capture lenght is on the screen.

Also Siglent do not decimate  data. All is mapped to display memory what give some advantages for better intensity gradation. Not only how many sequential acq. is overlayed but also how much one acq, data is is overlayed for image.

So or so, one like orange and one like banana. But also after learn and adapted to this system, some may start like it. Some not even then.



About history. Always when scope run and capture it push also captured wfm to  wfm history buffer, without slowing scope, it works backround just with wfm/s speed what it capture. How many wfm it keep, depends wfm length and some other things.
Also stored wfms use just raw ADC samples. Sinc on or off and vectors dots can set afterwards as want. Also because it have full rw data there can use full resolution measurements and FFT for every individual wfms in history buffer.

Normal wfm history is not  same as Sequence acquisition. In sequence acquistion it do not handle data for display at all during sequence. Only after whole sequence is captured, amount of segments user have set, then it process whole sequence for display, it stack (overlayed) all segments  even 80000 to one image after sequence ready and processed for display. After then it stop or continue to new sequence. Depending settings. When user want he can look individual segments using hotory view function and THIS viewer is same for wfm history buffer and for sequence acq.
Also both have time stamps.
User can not select amount of history buffer. Sequence mode he can selectamount of segments what is also some times important.

Perhaps it need bit learning so that can find how to use these so that can get out all what these features can give for use in different test cases.

If someone think history and segmented works same then there is misunderstooded totally some things (and it happen easy due to Siglent quite poor user manuals. ).

In sequence mode trigger re-arm time  is down to around 2us. Around ten times faster than example Keysight bottom models can do.
Rigol can not do anything like this.

Pity and shame is that user need himself study and learn using lot of exercises for find how it work and how to use some features right so that can find how to operate with these so that get all advantages what they may give out. 

One good example was when Dave did his one video where he use, if I remember right, SDS1000X and he did some mistakes when he talk about segmented acq. It was some kind of mess.



May I ask what you mean with this:

I can understand that the History mode can be convenient for some applications, but Siglent's implementation makes single shot trigger almost useless.

How it affect so that single shot is useless? Totally diffent things. How you connect these in your thinking.

[rf-loop]

I'm quite sure many users also hang in "more memory than fits screen" mode not actually realizing what it does and wfm/s rate suffers + scope does useless work. Siglent and some other scopes assume use of zoom if you want see fine detail from large detailed capture. I do not know if Siglent supports maximizing zoomed view somehow or you always "lose" half of screen?

This is what hope Siglent develop better.

There need be more than just one kind of zoom display divided vertically to two same height window. This is NOT optimal in many cases. This I hope they do development work, and hardly. With some other things what need develop better evev when we are in bottom price class oscilloscope group.

They CAN add feature where is works so that useer can force more capture lenght what is display width --  it is one window zoom.

They can give selection where user can select if zoomed window is master window what have more vertical height. Now there is only 1:1 in use. It is nice if user can select if he normally want 1:1 or 2:1 or 1:2 height. Take example Owon XDS what have always zoomed window more height than upper unzoomed trace.

Other things what need develop is more sophisticated history and segments viewer where is some functions for better search an find anomalies or do some other things and analyzing.

[MrW0lf]

Scope can easily do "zoom" when it stores data in a buffer longer than the screen size. Opposite is not possible,

In Rigol world, maybe, not in Siglent and some others...
Arduino => MCP4725 DAC I2C comms. Did choose 24MS/3 => 8MS buffer size to mimic DS1000Z with 3 active ch.
50ms/div, 15.63MSa/s: 2706 packets decoded & analyzed single-shot. Can navigate clicking on packet-of-interest in table.
"Zoomed" window is not mandatory. "Non-zoomed" window represents "screen size".



So guess nothing wrong with topology as such...

[borjam]

There is zoom for look details. Userr can select speed and memory so that he get length what he want and if need look insde acq length with more fst timebase just zoom. Even with 1ms/div you can zoom down to 1ns/div. And what important even in this case trigger is accurate and use fine interpolation between sampled points.

First, please forgive me if my "useless" comment below was too blunt. I was a bit puzzled, that's it.

Understood. I see how it works now. When intending to do something like this, the right way to do it with Siglent is to select
a time/div setting while looking at the memory depth and sample rate instead of looking at the time/div itself, and after
capturing using the time base knob for zoom. Keeping the proper sample rate and memory depth will ensure that the data
is correctly captured.

I admit that maybe I was a bit lazy with the Rigol (which is not bad anyway, just a different approach) with the "just press SINGLE and zoom in to explore the captured data".

Also Siglent do not decimate  data. All is mapped to display memory what give some advantages for better intensity gradation. Not only how many sequential acq. is overlayed but also how much one acq, data is is overlayed for image.

Sounds good

So or so, one like orange and one like banana. But also after learn and adapted to this system, some may start like it. Some not even then.

I do own both, so I'll have to develop some double personality disorder

About history. Always when scope run and capture it push also captured wfm to  wfm history buffer, without slowing scope, it works backround just with wfm/s speed what it capture. How many wfm it keep, depends wfm length and some other things.
Also stored wfms use just raw ADC samples. Sinc on or off and vectors dots can set afterwards as want. Also because it have full rw data there can use full resolution measurements and FFT for every individual wfms in history buffer.

I like that, reminds me of my old LeCroy 9400 and the way it considers captured data "sacred", performing manipulations on a "wave processing" virtual channel.

Perhaps it need bit learning so that can find how to use these so that can get out all what these features can give for use in different test cases.

Understood, thank you.

If someone think history and segmented works same then there is misunderstooded totally some things (and it happen easy due to Siglent quite poor user manuals. ).

Indeed. I am not in the Chinese oscilloscope bashing fetish (as I said, I own two and I consider them useful tools) but the documentation is all the Chinese products I have seen is just awful. They must learn the value of transparent communication with
customers and proper documentation.

In sequence mode trigger re-arm time  is down to around 2us. Around ten times faster than example Keysight bottom models can do.
Rigol can not do anything like this.

And no, I wasn't in "Rigol vs Siglent" mode, but just pointing out something that looked a bit confusing.

Pity and shame is that user need himself study and learn using lot of exercises for find how it work and how to use some features right so that can find how to operate with these so that get all advantages what they may give out. 

That's the next step for the newcomers to the instrumentation market. The Big Names have professional writers in staff and engineers with teaching experience who produce excellent documentation.

One good example was when Dave did his one video where he use, if I remember right, SDS1000X and he did some mistakes when he talk about segmented acq. It was some kind of mess.

Anyway remember that the best tools in the world can be useless and even dangerous without proper documentation.

May I ask what you mean with this:

I can understand that the History mode can be convenient for some applications, but Siglent's implementation makes single shot trigger almost useless.

I was clearly wrong. Now I see how I can apply the Siglent to do the same stuff I've been doing with the Rigol and I will obtain as a bare minimum the same functionality. My apologies.

[tautech]

I admit that maybe I was a bit lazy with the Rigol (which is not bad anyway, just a different approach) with the "just press SINGLE and zoom in to explore the captured data".

Which seems a strange way to capture the data. SINGLE use is for preset trigger conditions, while that it should not make any difference IMO it's more intuitive to use the STOP/RUN control.

[borjam]

I admit that maybe I was a bit lazy with the Rigol (which is not bad anyway, just a different approach) with the "just press SINGLE and zoom in to explore the captured data".

Which seems a strange way to capture the data. SINGLE use is for preset trigger conditions, while that it should not make any difference IMO it's more intuitive to use the STOP/RUN control.

In my case it wasn't strange. I wanted to capture a serial (Modbus) dialogue. So I set the trigger in serial mode to detect the first start bit, left the oscilloscope waiting in SINGLE mode and ran the program that sent the Modbus response and waited for the response. Once the first start bit was detected the oscilloscope triggered and I had the whole request-response dialog captured, ready to be examined.

In this case I wasn't trying to capture a random event, but looking at the behavior of a program that I executed myself.

[tautech]

I admit that maybe I was a bit lazy with the Rigol (which is not bad anyway, just a different approach) with the "just press SINGLE and zoom in to explore the captured data".

Which seems a strange way to capture the data. SINGLE use is for preset trigger conditions, while that it should not make any difference IMO it's more intuitive to use the STOP/RUN control.

In my case it wasn't strange. I wanted to capture a serial (Modbus) dialogue. So I set the trigger in serial mode to detect the first start bit, left the oscilloscope waiting in SINGLE mode and ran the program that sent the Modbus response and waited for the response. Once the first start bit was detected the oscilloscope triggered and I had the whole request-response dialog captured, ready to be examined.

In this case I wasn't trying to capture a random event, but looking at the behavior of a program that I executed myself.

OK, gotcha. 

As rf-loop has stated each and every brand has their own way of obtaining the result you need.....multiple paths to the same destination.
Familiarity with the UI makes this fast or slow and if once I look at the display and see there is more detail needed, slow the timebase, capture (or ZOOM) then adjust the timebase to see the detail and pan along it as needed.
For the X-E there is sufficient processing capability to leave the Mem depth at max for all use.

It is a shame this animated gif no longer works as it gives some nice insight to how Sample points, memory depth and timebase relationship. Maybe rf-loop can fix it.



From: https://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg933997/#msg933997