Author Topic: Siglent SDS1104X-E and SDS1204X-E Mixed Signal Oscilloscopes  (Read 156637 times)

0 Members and 2 Guests are viewing this topic.

Offline tautech

  • Super Contributor
  • ***
  • Posts: 14799
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #250 on: November 24, 2017, 07:45:42 am »
That is why I am interested in how the Siglent does when running at 200MHz & 500MSa/s - for all I know it could have similar behavior!
That's the same spec as the existing 2 ch SDS1202X-E.

Do remember that these 4 channel X-E's have the same inputs but doubled, 2 ADC's and 2 lots of 14 Mpts memory.
Avid Rabid Hobbyist
 

Offline MrW0lf

  • Frequent Contributor
  • **
  • Posts: 921
  • Country: ee
    • lab!fyi
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #251 on: November 24, 2017, 09:14:31 am »
What you see on the Rigol isn't a bug, it's exactly what the math says will happen if you only apply Sin(x)/x to a few samples on either side of the pixel of interest.

Nope. Implementation is not correct Sinc implementation. Direct proof: Impossible to determine analog bw of Rigol DS1052E with pulse gen method:
https://youtu.be/uBYMePUFinQ?t=20m43s
On normal scope this method works flawless and various ack modes and sampling rate have minimal effect on end result, if signal is repetitive and can be averaged.

 

Offline Hydron

  • Frequent Contributor
  • **
  • Posts: 382
  • Country: gb
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #252 on: November 24, 2017, 09:25:28 am »
That's the same spec as the existing 2 ch SDS1202X-E.

Do remember that these 4 channel X-E's have the same inputs but doubled, 2 ADC's and 2 lots of 14 Mpts memory.
Yeah that's what I understood, I just haven't seen how the SDS1202X-E does in this situation either (info may be in another thread, I had a quick look but didn't see it). Are the display/interpolation algorithms unchanged between 2 and 4 channel versions?
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #253 on: November 24, 2017, 09:32:01 am »
There we go. This is a Siglent SDS1202X-E with the coupling problem (not yet corrected).

Leo's pulse generator is connected to a T BNC connector, and the other end of the T has a BNC-SMA adapter and a 50 ohm terminator.

I have taken screenshots at 500 and 250 Msa/s. I have reduced the memory depth to the minimum and in order to go down to 250 Msa/s I have used a slower timebase. The screenshots have been taken with SINGLE trigger in all cases.

So here we have three sets of three files. 250 Msa/s, 500 Msa/s and 1 Gsa/s. For each one of them, a dots display, a vector display with sin(x)/x and a vector display with no sin(x)/x.

 
The following users thanked this post: tautech

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #254 on: November 24, 2017, 09:54:15 am »
Some Rigol ("upgraded" DS1074Z)

Interestingly, there is quite a lot of jitter at 500 and especially 200 Msa/s. I have made eye diagrams apart from the dots, sin(x)/x and no sin(x)/x.

(Corrected, at 500 Msa/s it's not jitter, but an artifact of reconstruction). However, I think there is a problem at 250 Msa/s, probably due to aliasing.

I would say that this scope has a very serious issue when sharing the A/D converter with more than one channel. The Siglent behaves much better.
Unless I have made any mistake, in which case I will be grateful for any corrections!


I would say that the Rigol has a serious aliasing problem at 250 Msa/s.
« Last Edit: November 24, 2017, 06:16:15 pm by borjam »
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #255 on: November 24, 2017, 10:00:44 am »
And eye diagrams on the Siglent, 500 Msa/s.

Again, enabling and disabling sin(x)/x

 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3014
  • Country: fi
  • Starting with DLL21
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #256 on: November 24, 2017, 10:10:15 am »


I have taken screenshots at 500 and 250 Msa/s. I have reduced the memory depth to the minimum and in order to go down to 250 Msa/s I have used a slower timebase. The screenshots have been taken with SINGLE trigger in all cases.


Also you can do single shot and in stop mode you can switch between vector or Sinc interpolation or turn it to without interpolation - just dots. There you can see how it solve Sinc. (also Siglent do not draw extra fake dots as some other scopes may do in "dots" mode).  Reducing sample speed and using window zoom you may find some results where it can visually see more clearly.  In these images we can see there is also real  overshoots and not only Sinc made overshoots what are other things. (front end BW filtering do not have enough steep stop before Nyquist for reject aliasing.)

What is miss is: True sample dots highlight On/Off function  when Sinc or Vectors in use.
If practice and theory is not equal it tells that used application of theory  is wrong or the theory itself is wrong.
It is much easier to think an apple fall to the ground than to think that the earth and the apple will begin to move toward each other and collide.
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #257 on: November 24, 2017, 10:20:59 am »
Also you can do single shot and in stop mode you can switch between vector or Sinc interpolation or turn it to without interpolation - just dots. There you can see how it solve Sinc. (also Siglent do not draw extra fake dots as some other scopes may do in "dots" mode).  Reducing sample speed and using window zoom you may find some results where it can visually see more clearly.  In these images we can see there is also real  overshoots and not only Sinc made overshoots what are other things. (front end BW filtering do not have enough steep stop before Nyquist for reject aliasing.)

What is miss is: True sample dots highlight On/Off function  when Sinc or Vectors in use.
True.

Now I see that the Rigol draws fake dots when I enable or disable sin(x)/x, which is silly. Dots are dots, unless I am terribly wrong it should represent just A/D sampled values. Interpolation only makes sense in vector mode.

 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 9580
  • Country: 00
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #258 on: November 24, 2017, 11:35:45 am »
I overlaid the Rigol waveform on the Siglent waveform in a single image. As you can see they're identical when you view the wave with the same settings.



I would say that this scope has a very serious issue when sharing the A/D converter with more than one channel.

There's more than one channel enabled in the image above...  :popcorn:

The Siglent behaves much better.

There's a major difference: The Rigol drops down to 250Ms/sec when you turn on the third channel, the Siglent doesn't (because it's only 2 channels).



At 5ns per division there's 1.25 samples per division on screen. I'm not sure how they're aligning the trigger point so exactly but it looks like they're not offsetting the rest of the wave to match. The "jitter" looks to be about 1/1.25 of a square.

« Last Edit: November 24, 2017, 12:03:07 pm by Fungus »
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3014
  • Country: fi
  • Starting with DLL21
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #259 on: November 24, 2017, 05:39:12 pm »
I overlaid the Rigol waveform on the Siglent waveform in a single image. As you can see they're identical when you view the wave with the same settings.



I would say that this scope has a very serious issue when sharing the A/D converter with more than one channel.



There's more than one channel enabled in the image above...  :popcorn:

The Siglent behaves much better.

There's a major difference: The Rigol drops down to 250Ms/sec when you turn on the third channel, the Siglent doesn't (because it's only 2 channels).



At 5ns per division there's 1.25 samples per division on screen. I'm not sure how they're aligning the trigger point so exactly but it looks like they're not offsetting the rest of the wave to match. The "jitter" looks to be about 1/1.25 of a square.

Identical?  This is total false.  If you use identical input then they can compare.  Do these image have  exactly same signal?

Then

Quote
There's a major difference: The Rigol drops down to 250Ms/sec when you turn on the third channel, the Siglent doesn't (because it's only 2 channels).

Reason is not because Siglent have only two channel. Root reason is construction.
When max 1GSa/s Siglent have 4 channel on it do not drop to 250MSa/s.

Rigol construction is cheapest possible:
1ADC1MEM/4

Siglent 4 channel construction is semi cheap:
1ADC1MEM/2 + 1ADC1MEM/2

If compare Siglent 100MHz 4 channel model and Rigol 100MHz 4 channel model:

Rigol  1Channel on,  1GSa/s max mem 24M (officially without option 12M)
Rigol  2Channel on,  500MSa/s 12M both channels. (without option 6M)
Rigol  3Channels on, 250MSa/s 6M all channels (without option 3M)
Rigol  4Channel on, 250MSa/s  6M  all channels. (without option 3M)

Siglent 1Channel on, 1GSa/s max mem 14M
Siglent 2Channel on, 1GSa/s max mem 14M both channels
Siglent 4Channel on, 500MSa/s max mem 7M for all channels.

Some other oscilloscope may have other construction
example 1ADC1MEM + 1ADC1MEM + 1ADC1MEM + 1ADC1MEM
this kind of oscilloscope have always same sampling speed and memory independent of how many channels is simultaneously in use.  This kind of construction is of course most expensive (And best).

If practice and theory is not equal it tells that used application of theory  is wrong or the theory itself is wrong.
It is much easier to think an apple fall to the ground than to think that the earth and the apple will begin to move toward each other and collide.
 

Offline HalFET

  • Frequent Contributor
  • **
  • Posts: 477
  • Country: 00
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #260 on: November 24, 2017, 05:51:40 pm »
Even Keysight and Tek share AD converters usually. Keep in mind you also need the hardware to support that converter and to process the data from it. These things run hot, are power hungry, and dump up to 8 or 16 Gbps of data into your FPGA and memory, not to mention the timing constraints. So it's perfectly reasonable to share ADCs between channels if you're short on money.
 

Offline simone.pignatti

  • Frequent Contributor
  • **
  • Posts: 313
  • Country: it
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #261 on: November 24, 2017, 05:54:58 pm »
Early next week I will be able to do a test video of the SDS1204X-E and SDS1104X-E. I will use the SDG6052X (video already online for it) and run frequency test on multiple channels as 1+2 and 1+3. Let see what will happen  :popcorn:
At max 200MHz standard cable quality shouldn't be a problem.
Technical Support
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #262 on: November 24, 2017, 05:57:36 pm »
Identical?  This is total false.  If you use identical input then they can compare.  Do these image have  exactly same signal?
The signal is the same, yes. Leo Brodnar's pulse generator connected to a T BNC adapter with a 50 ohm terminator.

I tried to compare the behaviors of both at the same sampling rate. The eye diagram is a different story, though, I can't do it at 250 Msa/s with the Siglent.
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #263 on: November 24, 2017, 06:02:04 pm »
Even Keysight and Tek share AD converters usually. Keep in mind you also need the hardware to support that converter and to process the data from it. These things run hot, are power hungry, and dump up to 8 or 16 Gbps of data into your FPGA and memory, not to mention the timing constraints. So it's perfectly reasonable to share ADCs between channels if you're short on money.
I understand that it's perfectly reasonable, but I think the Rigol has an aliasing problem at 250 Msa/s. This pulse generator is a nasty source for this kind of test because it has lots of harmonics. Remember that the rise and fall times are less than 60 ps.
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #264 on: November 24, 2017, 06:13:56 pm »
I overlaid the Rigol waveform on the Siglent waveform in a single image. As you can see they're identical when you view the wave with the same settings.
My apologies, it's true at 500 Msa/s and it's an effect of interpolation, not jitter. I was about to go to bed when I wrote that.

The Rigol has a jitter problem at 250 Msa/s, though. I think an anti aliasing filter is missing.
 

Online JPortici

  • Super Contributor
  • ***
  • Posts: 2506
  • Country: it
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #265 on: November 24, 2017, 06:18:59 pm »
Anyway, I was mainly trying to point out why the premium for the SDS1104X-E over the DS1054Z might be worth it even at 100MHz.

good luck with some dudes.
a premium that gives you access to better equipment in the long run, such as decodes you don't think you'll need, an AWG that doesn't look terrible on paper (is it worth it? we'll see) and an MSO option.
but the pricetag on the other is lower.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 1773
  • Country: hr
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #266 on: November 24, 2017, 07:01:25 pm »
I don't have Leo's gen.. (It's on a list :-)
This is my old Hameg 8030-6, at 11 something MHz.. Horrible square..

But let's put that on the side...

Running 4 ch, no Jitter...  Dot mode and vector mode...

When running 1GS an 500 GS/sec it is bit cleaner but looks the same.

You should run self cal to see if it helps. I realized that it is necessary to do self cal often.

Regards,

Sinisa
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3014
  • Country: fi
  • Starting with DLL21
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #267 on: November 24, 2017, 07:28:55 pm »
I don't have Leo's gen.. (It's on a list :-)
This is my old Hameg 8030-6, at 11 something MHz.. Horrible square..

But let's put that on the side...

Running 4 ch, no Jitter...  Dot mode and vector mode...

When running 1GS an 500 GS/sec it is bit cleaner but looks the same.

You should run self cal to see if it helps. I realized that it is necessary to do self cal often.

Regards,

Sinisa

As can see it can handle signal what do not have so high frequency components. This signal do not have. Signal rise and fall times are very slow. 250MSa/s even with false Sin(x)/x can handle this kind of signal thumbs down. Sample interval is 4ns.  Serious problems start with much faster signal rise/fall times.
If practice and theory is not equal it tells that used application of theory  is wrong or the theory itself is wrong.
It is much easier to think an apple fall to the ground than to think that the earth and the apple will begin to move toward each other and collide.
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 14799
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #268 on: November 24, 2017, 07:51:23 pm »
I shouldn't, NDA and all that......

SDS1104X-E
Source: SDG1032X 10 MHz terminated into 50 \$\Omega\$ feed-throughs for each channel



NO visible difference between SinX and X interpolation
Avid Rabid Hobbyist
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #269 on: November 24, 2017, 07:54:14 pm »
Yes, that's 1 Gsa/s. It gets interesting when you go to 500 Msa/s.

Note that I'm not criticizing the SDS1202X-E. sin(x)/x interpolation invents data of course. But it's a similar effect to that of an analog oscilloscope "filling the gaps"
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 14799
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #270 on: November 24, 2017, 07:58:03 pm »
Yes, that's 1 Gsa/s. It gets interesting when you go to 500 Msa/s.

Note that I'm not criticizing the SDS1202X-E. sin(x)/x interpolation invents data of course. But it's a similar effect to that of an analog oscilloscope "filling the gaps"
Is there a reason why your screenshots are done with AC trigger coupling ?  :-//
Avid Rabid Hobbyist
 

Offline Muttley Snickers

  • Supporter
  • ****
  • Posts: 1916
  • Country: au
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #271 on: November 24, 2017, 08:02:12 pm »
I am yet to read the manual but will the Siglent SDS 1000X-E series incorporate a real time clock and in turn allocate time and date stamps to recorded or saved events in stand alone operation, this is important as I regularly monitor systems for glitches and require time stamped events for correlation, leaving a laptop connected and running on a remote site for this particular purpose is not always practical.
 

Online borjam

  • Supporter
  • ****
  • Posts: 661
  • Country: es
  • EA2EKH
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #272 on: November 24, 2017, 08:11:55 pm »
Is there a reason why your screenshots are done with AC trigger coupling ?  :-//
Not at all, distraction. Should it matter much? I can try with DC trigger coupling. Anyway the pulse generator signal is really symmetric, so there shouldn't be a difference.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 1773
  • Country: hr
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #273 on: November 24, 2017, 08:26:44 pm »
As can see it can handle signal what do not have so high frequency components. This signal do not have. Signal rise and fall times are very slow. 250MSa/s even with false Sin(x)/x can handle this kind of signal thumbs down. Sample interval is 4ns.  Serious problems start with much faster signal rise/fall times.

I wasn't talking about ringing on rising and falling edges. Every Sin(x)/x interpolation will have those artifacts if fed signal with frequency content above the Nyquist limit...  There is no replacement for 10x times oversampling and tight filtering.. If you look into his captures, Siglent has worse ringing than Rigol at 500 MS/sec...

I was referring to the 5ns jitter on the falling edges of the pulse.. That is something not proportional to signal speed, quite the opposite, with signals that have slow edges jitter will be worse..  Again, there is no jitter on the right side of the screen..

Screen from a new one looks very good, but one with 500 MS/sec would be interesting one, to see how it looks worst case. Best case with 1 GS/sec looks perfect on Rigol too..
Anyways it does look promising and definitely is a step forward. It has potential to be a good scope.
It will be interesting to see how will it compare with new Rigols that will come out...

It is always a good thing when competitors come out with good product. In the end user are getting good products, from one company or another..

Regards,
Sinisa

 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 14799
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Siglent SDS1204X-E released for domestic markets in China
« Reply #274 on: November 24, 2017, 08:28:31 pm »
I am yet to read the manual but will the Siglent SDS 1000X-E series incorporate a real time clock and in turn allocate time and date stamps to recorded or saved events in stand alone operation, this is important as I regularly monitor systems for glitches and require time stamped events for correlation, leaving a laptop connected and running on a remote site for this particular purpose is not always practical.
Sorry Muttley the 1kX-E series don't have a RTC.  :(
Avid Rabid Hobbyist
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf