Siglent's new product- MSO/SDS2000 Series

[EEVblog]

What happened to 'Remove' button for messages?

It was disable today because people were deleting their first post in a thread which would delete the thread.

[rf-loop]

Because FW is preliminary version there is perhaps many things what need adjust. I hope they also adjust some parameters for better optimized  intensity grading and optimized also separately for different time bases.

This time I do not have time for tests , but I was just working with some freg modulated signal and fast take one random image with scope.

This:

(open image in separate window for full resolution (perhaps depending what browser in use)

[marmad]

Because FW is preliminary version there is perhaps many things what need adjust. I hope they also adjust some parameters for better optimized  intensity grading and optimized also separately for different time bases.

From what I've seen, the intensity-grading seems very nicely done - better and more fluid than the Rigol UltraVision or Agilent X-Series, I think.


That being said, given all of the information that has (or has not) been posted, I think the answer to the question I've been asking since November 13  (when Herman posted the first wfrm/s chart) might be:

Q: How does the Siglent achieve intensity-graded throughput speeds much faster than the Agilent 3000-X?
A: It doesn't.

It seems that the DSO might not use the intensity buffer (or at least, not in the same way) when sample sizes are >= 1.4M (in other words, at time bases >= 50us/div with those sample sizes). In any case, there are no images (from Siglent or anyone else) of the DSO using grading with deep sample lengths.

If this is true, then this is exactly at the same point where (shown in this previous chart I posted) the intensity-graded throughput speeds of the DSO would far surpass the Agilent 3000-X (and every known competitor).

This might also help explain why it's exactly beyond this same point (<= 560k samples @ <= 20us/div) when the Siglent starts experiencing a steep falloff in it's wfrm/s speeds between Dots and Vectors - since the processing time difference in displaying Dots or Vectors is more significant if intensity-grading is used:




Now, if this is all true (and of course, it's speculation based on the evidence presented) - is it a bad thing?

Not necessarily - there's an argument to be made that, when samples per pixel ratios gets greater than a certain amount, the capture rate should take precedence over grading. As a user, I would like that capability for certain applications - my only wish would be that it was a user-selectable option - as opposed to automatic behavior.

[tautech]

Because FW is preliminary version there is perhaps many things what need adjust. I hope they also adjust some parameters for better optimized  intensity grading and optimized also separately for different time bases.

From what I've seen, the intensity-grading seems very nicely done - better and more fluid than the Rigol UltraVision series, I think.


That being said, given all of the information that has (or has not) been posted, I think the answer to the question I've been asking since November 13  (when Herman posted the first wfrm/s chart) might be:

Q: How does the Siglent achieve intensity-graded throughput speeds much faster than the Agilent 3000-X?
A: It doesn't.

It seems that the DSO might not use the intensity buffer (or at least, not in the same way) when sample sizes are >= 1.4M (in other words, at time bases >= 50us/div with those sample sizes). In any case, there are no images (from Siglent or anyone else) of the DSO using grading with deep sample lengths.

If this is true, then this is exactly at the same point where (shown in this previous chart I posted) the intensity-graded throughput speeds of the DSO would far surpass the Agilent 3000-X (and every known competitor).

This might also help explain why it's exactly beyond this same point (<= 560k samples @ <= 20us/div) when the Siglent starts experiencing a steep falloff in it's wfrm/s speeds between Dots and Vectors - since the processing time difference in displaying Dots or Vectors is more significant if intensity-grading is used:




Now, if this is all true (and of course, it's speculation based on the evidence presented) - is it a bad thing?

Not necessarily - there's an argument to be made that, when samples per pixel ratios gets greater than a certain amount, the capture rate should take precedence over grading. As a user, I would like that capability for certain applications - my only wish would be that it was a user-selectable option - as opposed to automatic behavior.

Your are making these statements based on data supplied and if Firmware does not change all our work will remain valid. 

It is great that we have this venue to asses new products and use the collective wisdom/knowledge. 

But if there any subtle performance tweaks still to come....well one wonders.
Time will only tell.
Further more there is published functionality that is not yet enabled in SDS2000.
There are still to come a number of forecast firmware updates.

IMO Siglent is brave to enter this class of an established market and as it seems, deliver a competitive product. 

[marmad]

Your are making these statements based on data supplied and if Firmware does not change all our work will remain valid. 

Of course - I can only make assumptions based on the available data SDS users have posted. And if no data had been published until the FW was finished and the DSO was officially released - then I wouldn't be making assumptions at all.    And the only reason I have been wondering about this for over 6 weeks is because I noticed that the published numbers for deep memory @ slower time bases were very high compared to other DSOs.

But if there any subtle performance tweaks still to come....well one wonders.
Time will only tell.
Further more there is published functionality that is not yet enabled in SDS2000.
There are still to come a number of forecast firmware updates.

Sure, I would expect Siglent to improve performance and offer new features with each new FW release. Rigol just released new FW for the DS2000 3 weeks ago that increased the wfrm/s rates very nicely at time bases < 50ns - and this is about ~1.5 years after initial release.

But if you're supplying feedback to Siglent about future development for the SDS, I think the following feature would be popular for many users:

BLIND TIME percentage displayed on screen. It doesn't have to be updated - it could just be an approximate average - but it can be useful for gauging the way all settings (time base, vectors, etc) affect what you're seeing - and what you're looking for. I wish all manufacturers would have this.

[tautech]

Sure, I would expect Siglent to improve performance and offer new features with each new FW release.    Rigol just released new FW for the DS2000 3 weeks ago that increased the wfrm/s rates at < 50ns - and this is about ~1.5 years after initial release.

FYI the mentioned forecast updates are for pre-market release models such as mine & rfloop's. That does not include any updates that may be released after market release. 

Edit:

But if you're supplying feedback to Siglent about future development for the SDS, I think the following feature would be popular for many users:

FYI Siglent spent some time yesterday following this thread and will be watching for anything useful we contribute.

BLIND TIME percentage displayed on screen. It doesn't have to be updated - it could just be an approximate average - but it can be useful for gauging the way all settings (time base, vectors, etc) affect what you're seeing - and what you're looking for.

Edit:
At this time the UI is quite fast, but one must take care of what extra features are included as not to slow the UI to speeds that could frustrate the user. 

rfloop made this statement:

Real visible time is nearly 95% of total time.

If this is so across all timebase settings, and published as such, there would be little need for that on-screen statistic, one would be using valuable real-estate that may be used for a more valuable statistic. Maybe % ahead of Rigol xxxx?. 

[marmad]

rfloop made this statement:

Real visible time is nearly 95% of total time.

If this is so across all timebase settings, and published as such, there would be little need for that on-screen statistic, one would be using valuable real-estate that may be used for a more valuable statistic. Maybe % ahead of Rigol xxxx?.

      I think you have a lot to learn about DSOs, yes?

rf-loop was talking about a couple of the sloooow time base settings. The Siglent, like every DSO, is blind much of the time:



* = based on published single-channel tables by rf-loop and tautech


EDIT: BTW, if you want me to post comparisons (correct ones - not like those posted by Herman) against Rigol DS2000, I can. But I figured that might make more sense when the Siglent is actually on the market.

[tautech]

I think you have a lot to learn about DSOs, yes?

Thats why we rely on prominent users like yourself to keep us in check. 

But I figured that might make more sense when the Siglent is actually on the market.

Yep.  We look forward to a full review/teardown from Dave or others.(yourself included of course) 
Maybe we start a poll as to who will do it? 

[Carrington]

Yep.  We look forward to a full review/teardown from Dave or others.(yourself included of course) 
Maybe we start a poll as to who will do it? 

Mark (marmad) is certainly an expert. He has demonstrated an extensive knowledge about DSO.
Siglent should send one SDS2304 for free.

[marmad]

I think you have a lot to learn about DSOs, yes?

Thats why we rely on prominent users like yourself to keep us in check. 

It's no problem - we all start from the same place.     I used analog oscilloscopes for 30 years - but only started using DSOs 2.5 years ago.

Yep.  We look forward to a full review/teardown from Dave or others.(yourself included of course) 

Yes, I would love to try one some time - it looks like very good value for the money.

One of the biggest problems with Chinese test equipment (and this includes all the Chinese manufacturers - including Rigol, Instek, etc.) is the slow trickle of information about the product (reviews, opinions, etc) to the public - and back to the company. I guess it's more complicated because of language and culture differences.

[marmad]

Mark (marmad) is certainly an expert. He has demonstrated an extensive knowledge about DSO.
Siglent should send one SDS2304 for free.

Well, I don't expect anything for free, but as I've mentioned, I'd love to do a comparison review versus the Rigol DS2000.  I can promise to be fair - I have nothing to gain either way - but I'm truly hoping that the Siglent SDS series is a much better DSO then the Rigol DS2000, Agilent DSOX2000, and Instek GDS2000A DSOs - because it will force everyone to make better and less expensive DSOs (just as the Agilent X-Series did first and the Rigol UltraVision series did after).

[Hydrawerk]

Agilent DSOX2000 is still the most modern scope... It is ASIC based.  But you must buy other scope, if you want:
a) memory longer than 1MB/channel
b) more than 4 automatic measurements
c) better sensitivity than 4mV/div
d) two channel wavegen
e) more bang per buck
d) 65000 segments of memory, or so
It is hard to say in general which scope is the best. It depends on what you need. Someone might need 8 automatic measurements and cursors at the same time, but he does not need much DPO capability. Then he should buy GW Instek GDS-2000A scope...

[Hydrawerk]

For some reason the SDS2000 scopes are still not in stock. http://www.siglent.eu/oscilloscopes/sds-2000-series.html
The manufacturer is till working on the firmware, probably.

[sync]

Agilent DSOX2000 is still the most modern scope... It is ASIC based.

It's the most modern one because it's using ASIC??? What a strange reason.
IMHO it's inflexible. An FPGA based designs can be updated more easily. New models with more features and hardware (memory, etc.)

[Carrington]

About Agilent DSOX2000, watch this video (minute ~10):


Agilent engineers have reviewed this video and noted the autotrigger timer on the 2000 X-Series oscilloscopes is the reason it is missing triggers. They are now actively looking at and working on ways to enhance the autotrigger timer.

Is a ASIC's error? I hope not.

[sync]

Is a ASIC's error? I hope not.

I wouldn't call that an error at all. It's how scopes traditionally works. The auto trigger sweeps after a time without a trigger signal. During a sweep it does not trigger. So the pulse is displayed at random position. The GW-Instek can somehow trigger during the sweep. This surprised me when a saw this. It's advanced.

Normally you would use the normal trigger mode. Then you get a stable display.

[Carrington]

Is a ASIC's error? I hope not.

I wouldn't call that an error at all. It's how scopes traditionally works. The auto trigger sweeps after a time without a trigger signal. During a sweep it does not trigger. So the pulse is displayed at random position. The GW-Instek can somehow trigger during the sweep. This surprised me when a saw this. It's advanced.

Normally you would use the normal trigger mode. Then you get a stable display.

What I want mean is: And if it had been an ASCI's error? 

IMHO it's inflexible. An FPGA based designs can be updated more easily. New models with more features and hardware (memory, etc.)

In some of those things I don't agree, but as you say: "an FPGA based designs can be updated".

[sync]

What I want mean is: And if it had been an ASCI's error?

Then try a software workaround or put the error in the errata sheet as unfixable.

[Carrington]

What I want mean is: And if it had been an ASCI's error?

Then try a software workaround or put the error in the errata sheet as unfixable.

Sure, but not everything can be fixed with software.

[tautech]

I'm not sure why, but it doesn't seem as if there's any intensity-grading in that image.

I did not have it on, but fresh attempts with it on make little difference.
However with it on the same AM waveform @ 2 uS shows 3 Graded traces.
I have used Intensity Grading on other than AM waveforms and it works fine.

As I have stated before we will have to wait for all Firmware tweaks to be installed before we can see this DSO's full potential. 

  @ marmad
From Page 15  re discussion on Intensity Grading
FYI
SDS2000 does not enable intensity grading, it is on by default.
The amount of grading is controlled by Trigger level alone.
The Colour grading is switchable however

If you examine all Intensity Grading images posted in this thread you will see Trig level set close to Ch baseline.

[marmad]

SDS2000 does not enable intensity grading, it is on by default.

Yes, this is normal for DSOs with intensity grading.

The amount of grading is controlled by Trigger level alone.

If you examine all Intensity Grading images posted in this thread you will see Trig level set close to Ch baseline.

I'm not sure what you mean. Do you mean the Trigger frequency (i.e. waveform captures)? Because intensity, just like on an analog oscilloscope, is a reflection of how often a signal crosses a certain point in a given period of time. For example, if during one second a waveform crosses pixel 1 50% of the time and pixel 2 100% of the time, then pixel 1 should have half the intensity of pixel 2.

[tautech]

SDS2000 does not enable intensity grading, it is on by default.

Yes, this is normal for DSOs with intensity grading.

The amount of grading is controlled by Trigger level alone.

If you examine all Intensity Grading images posted in this thread you will see Trig level set close to Ch baseline.

I'm not sure what you mean. Do you mean the Trigger frequency (i.e. waveform captures)? Because intensity, just like on an analog oscilloscope, is a reflection of how often a signal crosses a certain point in a given period of time. For example, if during one second a waveform crosses pixel 1 50% of the time and pixel 2 100% of the time, then pixel 1 should have half the intensity of pixel 2.

No not frequency just the vertical Level as displayed on screen.
A little hard to see as the Trig level indicator is the same color as the trace for any channel.
BTW on the Siglent it is a sideways pointing triangle on right hand side of display.

As I adjust the Trigger Level down the waveform the amount of Intensity grading increases.
This can be counter-productive with a singe period waveform as it will shift partly off screen of course.
However the amount of displayed Intensity Grading is controlled with Trigger Level knob.

And with Color turned on the trigger crossing points are further highlighted.

[rf-loop]

The amount of grading is controlled by Trigger level alone.
The Colour grading is switchable however

If you examine all Intensity Grading images posted in this thread you will see Trig level set close to Ch baseline.

Amount of grading is NOT controlled by Trigger level.
It is better to understand firszt how oscilloscopes works before...
Yes trigger level change may lead to situation where you see more or less grading dependent your signal.  (it is good to think causality...)

[rf-loop]

It seems that the DSO might not use the intensity buffer (or at least, not in the same way) when sample sizes are >= 1.4M (in other words, at time bases >= 50us/div with those sample sizes). In any case, there are no images (from Siglent or anyone else) of the DSO using grading with deep sample lengths.

If this is true, then this is exactly at the same point where.......

Why make so much opinions and thinking about things what are not yet tested or displayed. Is it better first tesat and investigate and after there is more data and tests then make opinions etc about things what are not yet ready showed.
Least I do not have around of clock time for playing just with one scope. I need first priority do my works  and after then this kind of hobby or semihobby if there is still free time after all more important things like family etc...  it may take time...

But here again some random samples... with some continuously variable signal.
And they show that intensity grading (and color grading) works also with slower speeds with more memory.


As  I told before, intensity grading need some small adjustments in FW for better work with different time and also some adjustments (levels scale - gradient)

User may also adjust it some amount using intensity adjustment - specially this effect is visible if color is on. (not separately demonstrated in these pictures)

Dependent your browser... better open pictures in new window for full resolution.

[tautech]

The amount of grading is controlled by Trigger level alone.
The Colour grading is switchable however

If you examine all Intensity Grading images posted in this thread you will see Trig level set close to Ch baseline.

Amount of grading is NOT controlled by Trigger level.
It is better to understand firszt how oscilloscopes works before...
Yes trigger level change may lead to situation where you see more or less grading dependent your signal.  (it is good to think causality...)

Thank you rfloop.
Have I got this right?
The displayed Intensity Grading is dependent on waveform complexity first and Trig level second.
Your images are stunning especially those with 14 and 28 M