New Siglent SDS1202X-E oscilloscope based on Xilinx Zynq-7000 SoC architecture

[karkoon]

Hi,

I just released my 2 worth of video regarding my new SDS-1202X-E. Hope you enjoy!

The issue you identify towards the end with being unable to trigger to channel 2 after auto setting on channel 1 is normal, you need to press the setup button and change the trigger source to channel 2, as it will still be set to ch1 even if you turn it off, and so it will not be trying to lock onto the signal on ch2, this is why it works if you press auto again, as that will be changing the trigger to ch2 for you.

Thanks for the mentions too [emoji6]

Thank you so much. Your video convinced me that I can go for the scope despite of all the noise.

Your videos are awesome.


Sent from my iPhone using Tapatalk

[karkoon]

Released an errata video about my auto-setup feature understanding.

https://youtu.be/rd_sfPpiZJk

Thanks.

[rf-loop]

About FFT. Tiny example.




(there is 455kHz (Radio receiver IF) signal first with 100Hz and after then 120Hz modulation. (Carrier level 1Vrms (0dBVrms) and in both cases modualtion 20%. (with 20% AM mod both sidebands are -20dBc)

Of course 100Hz from carrier can see, and as can see resolution is well enough for separate 20Hz (100Hz and 120Hz peaks) with this frequency. Of course if we rise f Nyquist also resolution change - this is FFT.

Pity there is not 50ohm inputs so it can not show dBm (power) scale.


Please Siglent:

Add function that if user use external 50ohm termination then user can tell it to scope (select External 50 ohm) and after then dBm is available as example in SDS1000X(X+ series.
With rf works dBm is normal and very common way to tell levels what everyone knows and use.  Of course every user can calculate but this is waste of time when scope can do it simply for me.

[Loboscope]

Another observation I could make: When I looked at some higher frequencies, for example 25 MHz (this is the maximum of my SDG1025) and I hit the "Run/Stop"-button to stop and freeze the image, I did notice, that the curve-trace will be deformed a little bit. Please look at picture SDS1202X-E-11.
This will not occur, when I will print the image without stopping the aquisition. (Pic. SDS1202X-E12).
Also at lower frequencies below about 15 MHz I could not notice this phenomenon. Unfortunately I can not do some tests above 25 MHz because I do not have a signal-generator which will deliver more than 25 MHz, but I guess this phenomenon will become more worse the higher the frequency will be.
The Rigol 1104Z will perform here without problems (Pic. DS1Z_QuickPrint1).

I worry that this behavior will be more than a little flaw, but probably a serious bug. Because you will always be sure that the waveform the scope shows in Stop-Mode will be exactly the real shape of the signal and not a more or less randomly deformed pattern down to the whim of the scope. A scope must be reliable (within its specs) or it is worth nothing ...

[Glowtape]

From how it looks the Siglent freezes and specifically displays a single waveform at the trigger point, where as Rigol stops on an average it got while unfreezed. Note the blurred waveform. I'd wager that if you can make the Rigol display a single waveform, it'll look the same.

[IAmBack]

About FFT. Tiny example.

(there is 455kHz (Radio receiver IF) signal first with 100Hz and after then 120Hz modulation. (Carrier level 1Vrms (0dBVrms) and in both cases modualtion 20%. (with 20% AM mod both sidebands are -20dBc)

Of course 100Hz from carrier can see, and as can see resolution is well enough for separate 20Hz (100Hz and 120Hz peaks) with this frequency. Of course if we rise f Nyquist also resolution change - this is FFT.

Can You show us, how looks FFT for a low frequency signals, like 1kHz square wave?

[mikeselectricstuff]

As we discussed though Rob, it could easily be added back if it has a use once again, if the button has to stay, then at least it needs to have a screen pop up saying something like "All options are enabled" or "No options available" etc. a button has to have a function, otherwise it causes confusion, as seen.

Or turn it into a quick-access button for a user-defined function, like the "Quick action" button on the Keysight models.

[kcbrown]

From how it looks the Siglent freezes and specifically displays a single waveform at the trigger point, where as Rigol stops on an average it got while unfreezed. Note the blurred waveform. I'd wager that if you can make the Rigol display a single waveform, it'll look the same.

That's what I would have thought, too, but the presented waveform doesn't look quite like what I would expect to see from sin(x)/x interpolation.

I suppose the only way to really know would be to see the dot mode capture under the same circumstances.

@Loboscope: can you repeat your first test, and show the same stopped capture in both sin(x)/x and dots modes?   It has to be the same stopped capture because a different capture is likely to have different points and thus a slightly different shape.  I'd think it should be possible to switch sin(x)/x on and off even with the scope stopped.

[tautech]

As we discussed though Rob, it could easily be added back if it has a use once again, if the button has to stay, then at least it needs to have a screen pop up saying something like "All options are enabled" or "No options available" etc. a button has to have a function, otherwise it causes confusion, as seen.

Or turn it into a quick-access button for a user-defined function, like the "Quick action" button on the Keysight models.

There are spare buttons in the UI that could be assigned to any number of things.
For the time being the Option button need only be changed to indicate there are NO additional options at this time.

[rf-loop]

There is nothing wrong. In stop mode you get one single shot, one waveform. One of these waveforms what are stacked in one frame. Nothing is changed in capture data. Curve is NOT deformed, only displayed one single curve instead of display many stacked curves overlayed.

In run mode there is perhaps  lot of waveforms stacked in one display frame.

Example, if scope capture in some situation 10000 wfm/s, one TFT frame  may have 400 last captured waveforms stacked. (refresh cycle 40ms)

Of course it can work so that if stop scope it keep these all stacked wfms - but this keep last one.  If you want more after stop...   go to history viewer  and look how many previous waveforms there are in history buffer. When you playback these you can also get image where last wfms are stacked on the screen.  And if you make playback very slow you can see these individual wfms separately (use example 100 or 200ms interval in playback)

Also you can look if you use slow or fast acquisition mode (selection can find in acquisition menu)

[rf-loop]

From how it looks the Siglent freezes and specifically displays a single waveform at the trigger point, where as Rigol stops on an average it got while unfreezed. Note the blurred waveform. I'd wager that if you can make the Rigol display a single waveform, it'll look the same.

That's what I would have thought, too, but the presented waveform doesn't look quite like what I would expect to see from sin(x)/x interpolation.

Think again. There is 1GSa/s and 10ns/div.  Sinc function is not "smooth" function over several sample points.

[Loboscope]

From how it looks the Siglent freezes and specifically displays a single waveform at the trigger point, where as Rigol stops on an average it got while unfreezed. Note the blurred waveform. I'd wager that if you can make the Rigol display a single waveform, it'll look the same.

That's what I would have thought, too, but the presented waveform doesn't look quite like what I would expect to see from sin(x)/x interpolation.

I suppose the only way to really know would be to see the dot mode capture under the same circumstances.

@Loboscope: can you repeat your first test, and show the same stopped capture in both sin(x)/x and dots modes?   It has to be the same stopped capture because a different capture is likely to have different points and thus a slightly different shape.  I'd think it should be possible to switch sin(x)/x on and off even with the scope stopped.

Indeed, this "exercise" was very insructive!
I did set aquire to dot-mode and sin(x)/x at both scopes (on the Rigol here you cannot change sin(x)/x to x like at the Siglent). Then I stopped the scopes and took the first screenshots in dot-mode (SDS1202X-E18 and DS1Z_QuickPrint3). Then I changed the displays to vectors and voilà: now the Rigol also shows the slightly deformed pattern like the Siglent (SDS1202X-E19 and DS1Z_QuickPrint4).

Insofar the Siglent would have shown me the true pattern of the SDG1025 (obviously working at its limit, but I never noticed this before, I mostly use lower frequencies working with audio-stuff) in Stop-Mode and it was the Rigol who did show me a "lie".
Thanks for revealing this phenomenon to me!
And thanks @rf-loop for your explanation!

By the way: Saving the screenshots to USB works really fast on the 1202X-E.
 

[StillTrying]

As a compromise between the single pixel drawn lines and the blur-o-vision I like to use dot mode with 1 or 2 sec. of persistence when possible.

[kcbrown]

Think again. There is 1GSa/s and 10ns/div.  Sinc function is not "smooth" function over several sample points.

True.  I was mainly surprised to see the artifacts given that the signal generator should be emitting a reasonably pure sine wave (not completely spectrally pure, of course) even though 25MHz is its highest frequency.  I would have expected any spectral impurities in a single capture to not actually be so visible to the eye.  But I suppose, in hindsight, the expectation one should have would be dictated by implementation details of the generator that aren't likely to be published.

The artifacts are obviously random, though, enough so that the persistence-based display shows a sine wave that is missing the artifacts (or, at least, the artifacts are diminished enough to make them difficult for this neophyte's eyes to detect).

Anyway, the supplied screenshots make it clear what's going on, and it's interesting to see how the Rigol and the Siglent differ in their displayed results here.

[rf-loop]

About FFT. Tiny example.

(there is 455kHz (Radio receiver IF) signal first with 100Hz and after then 120Hz modulation. (Carrier level 1Vrms (0dBVrms) and in both cases modualtion 20%. (with 20% AM mod both sidebands are -20dBc)

Of course 100Hz from carrier can see, and as can see resolution is well enough for separate 20Hz (100Hz and 120Hz peaks) with this frequency. Of course if we rise f Nyquist also resolution change - this is FFT.

Can You show us, how looks FFT for a low frequency signals, like 1kHz square wave?

Here some examples with 1kHz and 100Hz square, risetime and fall time 6ns.

[IAmBack]

About FFT. Tiny example.

(there is 455kHz (Radio receiver IF) signal first with 100Hz and after then 120Hz modulation. (Carrier level 1Vrms (0dBVrms) and in both cases modualtion 20%. (with 20% AM mod both sidebands are -20dBc)

Of course 100Hz from carrier can see, and as can see resolution is well enough for separate 20Hz (100Hz and 120Hz peaks) with this frequency. Of course if we rise f Nyquist also resolution change - this is FFT.

Can You show us, how looks FFT for a low frequency signals, like 1kHz square wave?

Here some examples with 1kHz and 100Hz square, risetime and fall time 6ns.

Looks pretty nice. Thanks.

[tautech]

Some comments from the Xilinx forum:
https://forums.xilinx.com/t5/Xcell-Daily-Blog/Dave-Jones-tears-down-the-new-lt-400-Zynq-powered-Siglent/ba-p/758656

[nctnico]

That is not a forum but a Xilinx employee spouting marketing BS trying to get as many buzzwords in there for Google to pickup!

[boggis the cat]

This is what made the GW Instek models better than Rigol and Siglent.  By using the same more powerful processor, Siglent have made a 'scope that responds quickly.  Cutting the component count down has also brought the cost down.

Rigol will certainly follow suite, and use either this product line or something similar in their new models -- which would be a good thing.

There are no advantages to a slow, erratic, UI.  It also makes me wonder if the 'scope with such behaviour is able to keep up with the data processing, or if issues may be present from missed samples.  (Not stuff that is checked during calibration, by the way.  Basic functions may meet spec and be fine, but the product still be poor and ill-suited for particular tasks.)

[pascal_sweden]

Any forum users out here, who want to do a detailed video review of the serial decoding?

[karkoon]

Any forum users out here, who want to do a detailed video review of the serial decoding?

I can volunteer.


Sent from my iPhone using Tapatalk

[pascal_sweden]

I look forward to your detailed review on decoding!

It would be nice to check if the decoding is implemented in hardware.

Maybe it is possible to telnet or open an ssh connection to the scope, and run the Linux "top" command to check the overall processor load and the load of individual processes.

This way you could have the "top" command running in a terminal window, to monitor the processor load for various decoding scenarios, and confirm or at least have a good indication whether the decoding is implemented in software or in hardware.

You could do this for different time bases and memory settings, and for single and dual channel configurations.

You could also do this for the decoding of various supported protocols, to check if the processor load is the same for all the protocols, or if there is a significant difference between different protocols.

[karkoon]

Sure. Thank you very much for instructing me on how to do this. I am an oscilloscope noob.


Sent from my iPhone using Tapatalk

[MatCat]

Just got my SDS1202X-E 2 weeks ago.   

[pascal_sweden]

Just got my SDS1202X-E 2 weeks ago.   

MatCat becomes HappyCat