Siglent SDS1000X and (MSO) SDS1000X+ series oscilloscopes

[rf-loop]

Edit:Add
SDS1000X is base model
(SDS1000X-S is base model with 25MHz Function/Arb generator hardware.)
New: SDS1000X+  is like SDS1000X-S  added with 16 channel LA hardware. (Activation need buy license key and also need logic probe SPL1016)



Of course there is EEVblog #797 - Siglent SDS1000X Review for watch with all goods, bads and using mistakes etc.


It is good to note that SDS1102X (100MHz)  and SDS1202X (200MHz)  only difference in practice  is there in analog front end frequency response. (same for X+ models) 

Also 100MHz model have 50ohm input selectable. Both have 2ns/div.

[rf-loop]

Here wfm/s speed table. Mesurements using older FW but I have not seen any change in FW....39.5

Removed obsolete table.

In this message table is ok.

[rf-loop]

I will add to this thread some my own tests.

For the record can you state the model you have?

In my own use  SDS1202X after september. (also edited first msg)

[rf-loop]

Images about SDS1000X series segmented memory acquisition cycle.
Maximum speed is aound 12 - 16 sequences/sec. (sequence may have 1 to 1024 segments and minimum segment interval is 2us.
If there is 1GSa/s, 50ns/div 700pts segment lenght, segment time is 700ns and dead time is 1300ns. (segment speed is 500ksegment/s)

Images from SDS2304.  Signal source SDS1202X Trig output.
Signal to SDS1202X 50MHz sine.
Mode: continuous run with Sequence acquisition on.
t/div 50ns and display dots, Sinc off.

First image SDS1202X set for 10 segments.
Next image set for 1000 segments.
Note that SDS1000X give also trig out for every segment and not only one pulse for every whole sequence as some other models.


10 segment sequence


1000 segment sequence

[rf-loop]

There is available new FW for SDS1000X series.
SDS1000X_UpdateCurrent Version: 1.1.2.1R1 | Published?2015-12-18
Downloadable from
http://siglentamerica.com/gjjrj-xq.aspx?id=2579&tid=15

[tautech]

There is available new FW for SDS1000X series.
SDS1000X_UpdateCurrent Version: 1.1.2.1R1 | Published?2015-12-18
Downloadable from
http://siglentamerica.com/gjjrj-xq.aspx?id=2579&tid=15

2 Mb

From the changelog:
1. Add the adjust the trigger level function in the Vedio trigger.;
2. Add CAN&LIN decode function;
3. Standardized measurement parameter name;
4. Fix Bug: When upgrade the new firmware to 100%,The old version firmware No prompt the message about “Update successful”.

[rf-loop]

Perhaps images are self explanating.






Btw, this unknown other scope is factory original 200MHz. I do not know its 500uV bandwidth but there is not forced BW filter and some suspect that it is around same as scope other voltage bands(?). Using higher voltages there is -3dB point around 240MHz.  In picture can see also things what leads to suspect  that this "ultra low noise" scope 500uV/div is not true analog channel...it is vertical digital zoom only.

Siglent 500uV/div -3dB point is around 270MHz.  Higher voltages(eg 50mV/div) -3dB is over 280MHz.

Of course testing is required to ensure that the environmental EMI has not affected the high impedance input.  These are so obvious basic fundamentals.

[rf-loop]

Here also some test images also about noise things.

Signal is square from SDG5082 function generator.

Noise p-p is easy to show. Pure sinewave RMS is easy to calculate. But, random noise is bit more difficult case.
If really want calculate it, there need take lot of samples to memore, more better. Then take every sample and start computing.
But, if I look image I can quess noise RMS is somewhere between 60 and 100uV. Between SDG and SDS there is 40dB attenuator.
Every V/div have two images. Unlimited BW (in this case - 3dB point 270 - 280MHz. 200MHz -1dB.  Then BW Limited to 20MHz using oscillosscope own BW limit. This I have measured and it have -3dB point @ 20MHz (measured).

[rf-loop]

Rigol DS1074  (100MHz)



Siglent SDS1202X  200MHz

About images: NOTE: Both have resolution 800x480. ON Computer display you see both same dimensions. (I have not resized images for show difference)
Rigol DS1000Z real TFT size is 7"  and Siglent 8".


Signal exactly same. Settings all as same as possible also both selfcal before test. 
Both signals with 50ohm coaxial from same very good quality RF generator, and sure there is not noise from generator what any this kind of oscilloscopes can detect. Rigol do not have 50ohm inputs so both scopes set for 1Mohm input.
Of course Siglent have more noise (if front end quality is equal) because its analog side measured bandwidth -3dB is ~280MHz and Rigol ~ half.  But no, Siglent is in its own class.

(note. In some previous images where is undefined scope ffor compare. It is NOT Rigol DS1000Z series oscilloscope. It is well known bit more higher class equipment.)

[tautech]

Rigol DS1074  (100MHz)



Siglent SDS1202X  200MHz

About images: NOTE: Both have resolution 800x480. ON Computer display you see both same dimensions. (I have not resized images for show difference)
Rigol DS1000Z real TFT size is 7"  and Siglent 8".


Signal exactly same. Settings all as same as possible also both selfcal before test. 
Both signals with 50ohm coaxial from same very good quality RF generator, and sure there is not noise from generator what any this kind of oscilloscopes can detect. Rigol do not have 50ohm inputs so both scopes set for 1Mohm input.
Of course Siglent have more noise (if front end quality is equal) because its analog side measured bandwidth -3dB is ~280MHz and Rigol ~ half.  But no, Siglent is in its own class.

(note. In some previous images where is undefined scope ffor compare. It is NOT Rigol DS1000Z series oscilloscope. It is well known bit more higher class equipment.)

WOW, That's a vast difference in displayed noise. 

I'd guess you are referring to an Owon for the previous images. ?

[rf-loop]

WOW, That's a vast difference in displayed noise. 

I'd guess you are referring to an Owon for the previous images. ?

No, they are not Owon.  (Owon  do not have 500uV/div in any model)
Here is Owon XDS3102A (ADC resolution selection 8 bit because Siglent is 8bit)

In all images 5mV/div because this is lowest V/div in Owon what have full BW. (Owon have 2 and 1mV/div but 20M BW rejected. Siglent have full BW down to 500uV/div)
Both scopes get parallel exactly same signal (splitted).
Both scopes run 1ms/div with full 1GSa/s, acquisition normal
Then Stopped and zoomed to 50ns/div. (due to lack of memory Siglent can not do this with more low t/div)
(with this method scopes are "face to face" so that very different wfm/s update rate with this setting and lack of intensity grading in Owon etc do not affect result on the screen)

Next images about same but used window zoom.

Here need note that Siglent is 200MHz and Owon here is 100MHz model. (part of random noise, thermal noise,  is related to BW and thumb rule is that this part of total noise power is 3dB higher when BW doubles)

[rf-loop]

SDS1202X Trigger Out delay and jitter. 
trigger set for rising edge in CH1

Trig Out in CH2

Dealy is near 310ns. (add around 5ns to display due to stupid mistake with cable delay compensation.)

Jitter peak to peak around 8ns. 

Not so bad, not so good... most important is that user know it so that do not design to use this signal for set-up  where need near zero delay and jitter like in old analog oscilloscopes.

With slow signals you can use it for some other equipment synch (if this kind of delay and amount of jitter is negligible)

[rf-loop]

After FW update to version 1.1.2.1.1  (1.1.2.1R1 | Published?2015-12-18)   The maximum number of segments is changed from 1024 to 80000.

Here new table. 



Note: This table show "up to" maximum speed.
More important in Sequence mode is maximum speed where all waveforms what meet trigger  are also captured.
We can name it as something like "guaranteed maximum". For testing this, test must use maximum amount of segments in evvery t/div speeds.
This table here is not guaranteed maximum. In this test have not counted possible events what havve dropped out. Here have calculated from time stamps amouunt of used time / amount of captured segments.

When test guaranteed speed there must be (example) burst what include exactly 80000 pulse. Then look what is maximum pulse speed where no one pulse is missing from sequence acquisition. There is later other table about this test. In this test highest speed was 485ksegment/s.  Using various kind of signals and trigger settings etc, including some possible variations between individual scopes, Siglent specifications tell 400kwfm/s (400ksegm/s) what is, imho, just ok to tell. Not "up to", but "least faster than"..

[rf-loop]

Due to new FW 1.1.2.1.1 there was also chance in Sequence mode so that maximum number of segments inside sequence is chanced (improved from max 1000 segment to max 80000 segment) I have tested if it change some other things in Sequence aacquisition.  I have not seen other notable changes in Sequence mode. Here some random images collected during tests.

In this case there was around 50MHz sinewave input to SDS1202X CH1
Horizontal speed 50ns/div (1GSa/s). Segment lenght is 700ns (one waveform)
Display mode dots.
Acquisition mode Sequence and with this setting, maximum number of segments in sequence is 45526.
Sequence mode running continuous mode (repeating until user stop).





This first image is from SDS2304 what is connected to SDS1202X Trig Output.
In upper window there can see whole sequence period including small part of next period starting near right side of window.
Whole period is bit over 260ms.  Segments acquisition phase take around 91ms in this case. Segment  aacquisition repeat period is 2 us. (45526x2us) This 2us  segmeent period can see in bottom window where is separate individual segment trig out signals visible. After this sequence acquisition time scope is busy and can not start new sequence.  At this time it is processing what it have acquired  and also update to image some segment so that user can see something). Also it do things what have been waiting because scope is very busy so that it can minimize trigger recovery time and take asap new segment. In this case there is 700ns segments and between segments 1300ns dead, but speed itself is 500kwfm/s (ksegm(s).



Next 3 images are from SDS1202X screen.



This is what is visible in SDS1202X TFT during this sequence mode running. (sequence number counting and waveform image updated slowly (looks like once per whole sequence) At this time nearly 4 times/second something like "single shots"...




Next two images show something about stored segments watching.

After get single shot or continuous mode wanted segment acquisitions and want look what there. Scope stops if push History button (if not stopped due to single shot sequence) After pushing history, you can look stored segments just as you can do if you stop scope from normal mode and want look waveform history. Only difference is how they have acquired. Wiew and operate with these "segments" or "waveforms" or "frames" (how ever we want call these)   is same.



In this image, Sequence mode captured segments (waveforms)*)  watched using FFT and here in this image segment number 29371. (yes FFT settings are not optimal here, just for show it can do for segments and of course same same can do for normal  history waveforms. )
segment = waveform = string of ADC samples  what lenght is current used memory lenght.



One segment (top window). In this case there is 45526 this kind of segments what is displayed in upper window (in this case segment number 10300).
Of course there can do measurements and pan-zoom details. In this case segment is 700 data points captured using 50ns/div setting.  These can watch whole display or windowed display zoomed.  Also measurements can use. Of course these are full undecimated raw ADC sample strings, not display images, even if segment/waveform  lenght is 14M..

[rf-loop]

Of course everyone has heard of thousands of rumors about how bad digital oscilloscopes are when they should trig to the AM modulated signal .

SDS1000X triggering directly to  AM modulated RF signal (not used modulating signal separately for trigger)
Upper window  50MHz signal AM modulated using around 3kHz sine and lower window zoomed to carrier. In live running it also looks rock solid, just nearly like still image.

[alsetalokin4017]

How well does it perform in X-Y mode?

[tautech]

How well does it perform in X-Y mode?

Haha, kind of flattering you want to follow this thread so I'll ask you to post a screenshot of your Rigol in X-Y mode and we'll duplicate it. 

[alsetalokin4017]

How well does it perform in X-Y mode?

Haha, kind of flattering you want to follow this thread so I'll ask you to post a screenshot of your Rigol in X-Y mode and we'll duplicate it. 

You and I both know that the Rigol 1054z can't hold a candle to a basic analog scope in X-Y mode. It fails miserably to display the ScopeClock and barely makes a legible image with the Oscillofun file.
 
But the Siglent should be better, no? So let's see how it does with Oscillofun or something like the ScopeClock.

ETA: Actually I was able to get a fair-to-middling display of the ScopeClock on the Rigol after all, after playing around with the Mem Depth and etc.

[znroot]

No news about scope hacks??? O0

[smarteebit]

How well does it perform in X-Y mode?

Haha, kind of flattering you want to follow this thread so I'll ask you to post a screenshot of your Rigol in X-Y mode and we'll duplicate it. 

You and I both know that the Rigol 1054z can't hold a candle to a basic analog scope in X-Y mode. It fails miserably to display the ScopeClock and barely makes a legible image with the Oscillofun file.
 
But the Siglent should be better, no? So let's see how it does with Oscillofun or something like the ScopeClock.

ETA: Actually I was able to get a fair-to-middling display of the ScopeClock on the Rigol after all, after playing around with the Mem Depth and etc.

Hello alsetalokin4017,

The ScopeClock looks interesting. Is it a static display, or a real clock which can count the time? I'm curious how you generate the 2 signals if it is a real clock.

Thanks.

[tautech]

New FW for the SDS1000X series:
http://www.siglentamerica.com/USA_website_2014/Firmware&Software/firmware/SDS1000X%20P06.rar

From the website changelog:

1.Add ASCII display function;
2.Fix Some bugs:
   a) CSV file wrong timestamp problem
   b) Inverted channel setting’s issue
   c) Timeout error when using Labview Driver

[rf-loop]

Just tiny wfm/s speed chec with new P06 FW.

- All measurements on (continuously updating table left top on display)
- Gated measurements on (Max, Min, RMS can select more measurements and no affect speed)  measuring continuously between A and B (Gate)  (on or off, same speed)
- Statistics on. (on or off, same speed)
- Tracking Cursors on, continuously tracking. (on or off, same speed)
- Persistence on. (on or off, same speed)
- Sin(x)/x on (on or off, same speed)
- Display mode dots. (do you see dots in image)
(if lines, instead of dots, is on then wfm/s pseed drops down lot of. But with these speed mostly do not need lines. With more low horizontal speeds lines may be more iportant and there lines/dots have same wfm/s speed, change happend somewhere around 500ns/div)

SDS1000X reach maximum wfm/s speed with 50ns/div.

Continuous (average including processing time period between display refresh) 60.9kwfm/s (60.7 - 70)
Peak speed 70.8kwfm/s  (continuous speed inside one display-processing period without this time cap. Some scope manufacturers tell this max speed)



Always there in bacround is working history recording to history buffer, with this speed or what ever is current speed, oscolloscope save every single wfm to history buffer.
In this case (50ns/div) backround buffer lenght is ~45000 waveforms (max 80000 wfm). If you stop scope you can scroll back all these individual waveforms. Measurements, zooming etc can also use with these. History buffer can not turn off or on. It is there, always working and available when normal mode running is stopped. If use segmented memory acquisition (Sequence mode). It use this same memory so in this mode there is not separate "history".  Segments are "history" in this mode. (in sequence mode, it do not update display or processing during segments acquisition until selected or maximum amount of segments are acquired. Only after then it update display and do other time consuming things for reach maximum 500kwfm/s speed. Do not mix normal mode History  and Sequence mode "History".  After sequence (amount of segments) are captured then user select "History" player for look these segments. (this "History" player is same for looking normal mode history buffer and for looking sequence mode acquired segments.)

[rf-loop]

Other test.

Here window zoom

- All measurements on (continuously updating table left top on display)
- Gated measurements on (Max, Min. User can select more measurements and no affect speed)  measuring continuously between A and B (Gate)  (on or off, same speed)
- Statistics on. (on or off, same speed)
- Tracking Cursors on, continuously tracking. (on or off, same speed)
- Persistence on. (on or off, same speed)
- Sin(x)/x on (on or off, same speed)
- Display mode dots. (dots can see in zoomed window but still due to enough fast wfm/s speed there is so many sample points in one trace that individual sample dots draw nearly continuous trace)
(look previous message)


Continuous (average including processing time period between display refresh) in this case using windowed zoom. wfm/s speed is dropped to  1.2-1.25 kwfm/s.
Peak speed is 1.8 kwfm/s  (continuous speed inside one display-processing period without this time cap.)



50ns/div without zoom scope reach maximum wfm speed ~60kwfm/s. When change to zoom with this 50ns/div it drops lot of but still keep >1kwfm/s.  With lower t/div speeds drop ratio in speed is typically much less.

EDIT: Corrected wfm/s  (wfm/s --> kwfm/s)

[pxl]

if lines, instead of dots, is on then wfm/s pseed drops down lot of. But with these speed mostly do not need lines.

But this screenshot has been made in lines mode, right? It is in 700 pts, which means 70 dot/dev and in the screenshot we can see continuous lines, not dots.

UPDATE: I read your next comment, I got the point.

[rf-loop]

if lines, instead of dots, is on then wfm/s pseed drops down lot of. But with these speed mostly do not need lines.

But this screenshot has been made in lines mode, right? It is in 700 pts, which means 70 dot/dev and in the screenshot we can see continuous lines, not dots.

UPDATE: I read your next comment, I got the point.

good.


Some more explanation for someones who may wonder what is this..
It IS dot mode.

If this image where is 50ns/div and not zoom.
There display mode is dots.
You are right there is 700 dots on the screen (horizontally)
(Btw, Siglent is mapping always what ever memory lenght to display width. Also in SDS1000X)

It means that 50 dots for one div. (14 div)
Where sinewave is rising or falling there must see dots and it is true IF we take one shot one waveform.
But it draw 60000 waveform in one second. But peak vaue is around 70000. If we take one TFT image (I do not know exactly TFT refres rate but lets think it is 25 updates/s. 70000/25 = 2800. In one TFT image there is 2800 waveforms overlayed (in simplified theory). Samples are not exactly same position (due to - many reasons) and it draw like continuous line. There is visible dots if we stop scope and after then take one waveform, example if we after then look waveforms history buffer, there is last over 40000 single acquisitions. In history viwewer we can run it slowly and now we can see these walking dots clearly and if we playback it more and more fast, they disappear.