Siglent's new product- MSO/SDS2000 Series

[marmad]

I saw mjlorton's video on the SDS2072 where he mentioned the trigger out problem.  I have a SDS2074 and captured some small bits on video.  Here they are.

Profuse apologies for production values.  I don't have a decent video camera yet and I didn't intend on publishing any of this.  Note: I never got the unit to trigger any faster than 78K wfrms/sec. 

As mentioned elsewhere in this thread (and others), the Siglent seems to reach around 100k wfrm/s only with BOTH channels enabled, in DOTS mode, @ 50ns per division.

[FrankenPC]

I saw mjlorton's video on the SDS2072 where he mentioned the trigger out problem.  I have a SDS2074 and captured some small bits on video.  Here they are.

Profuse apologies for production values.  I don't have a decent video camera yet and I didn't intend on publishing any of this.  Note: I never got the unit to trigger any faster than 78K wfrms/sec. 

As mentioned elsewhere in this thread (and others), the Siglent seems to reach around 100k wfrm/s only with BOTH channels enabled, in DOTS mode, @ 50ns per division.

Of course.  Cut the sampling down from 2Gs/s to 1 and  it speeds up. 

[marmad]

Of course.  Cut the sampling down from 2Gs/s to 1 and  it speeds up.

Well, it's still 'sampling' at the same speed - it's just interleaved between the 2 channels. In fact, you would normally expect the wfrm/s rate to be slower with 2 channels because of more processing time needed for both waveforms (and it is on the Rigol and other DSOs), but for some unknown (as of yet) reason, the Siglent seems to always be faster with both channels on (maybe display FPGA for each channel?)

[kebogen]

Hey guys, I get information that the Siglent has solved the SDS2000 series in Europe, anybody got a SDS2304? Is it can reach the 110,000wfms/s? And how about the MSO function.

[Hydrawerk]

Mjlorton is probably going to do a proper review in future, he already owns this Siglent scope. This is only first impressions video. Just wait...

[rf-loop]

Hey guys, I get information that the Siglent has solved the SDS2000 series in Europe, anybody got a SDS2304? Is it can reach the 110,000wfms/s?

Yes it can up to 110kwfms/s
(test made using old preliminary FW)
 
Also it can mask test with full wfms/s speed.
Automatic measurements on,  not affect wfms/s speed.
Color on, no affect.
Persist on, no affect.

Beginning of december 2013 I get 2304 (in EU of course but this time there was preliminary version FW. Now it have of course  normal  updatwed FW but I have not repeated this test with new end user version FW due to extremely busy with extremely high priority level work project)

Attached old test result.
NOTE: Table is OBSOLETE! This was tested with old "very limited functionality" preliminary FW! But this is fairly ok. Some number values are perhaps wrong.
Highest 109kwfms/s (average value) is ok and most of other values. 
Average means that mostly it capture around 114kwfms/s but because around every 37ms there is small dropout due to system some busy this average is just under 110kwfm/s.
 
Also I have noted that this is some amount difficult to measure specially with some speed and channel settings. This is because time intervals between trig outs is variable and this is some amount problem for time interval counter. For better aaccuracy and reliability it is better to monitor Trig out with external oscilloscope and use conventional time gated pulse counting frequency meter. But, even this is in troubles if equipment is not good enough. Pulse width may have big variations.
(do not even try with some total crap "multimeter counter".  Even Agilent/HP 53131A is in troubles if use frequency counter function. But it have also time gated pulse counter (Totalizer) and it can do job (with proper settings).
(I start doupt this my old test after I find one time that with same settings I get some times around 56kwfms/s and sometimes around 44kwfms/s. (in this time HP counter was in time interval mode and also with auto trigger. After deep inspection with external oscilloscope I find that HP counter calculate wrong due to pulses width variation and pulses interval fast (but small) changes)

New a fuller, more accurate and more comprehensive table is under work.
In the old table is not anything quite radical errors due to its limited outreach.
In addition, a new FW may also affect the results.

[Terry]

Does anyone know the sequence mode? SIGLENT claimed that at sequence mode the maximal capture rate is 
more than 300,000 wfs/s. why?http://www.siglent.com/en/server/download.aspx?nodecode=119004002001

[rf-loop]

SIGLENT claimed that at sequence mode the maximal capture rate is 
more than 300,000 wfs/s. why?

Explanation is in image (text) you attached.

Normal waweform capture rate (wfms/s) means captured and (real time) displayed waveforms / s

Sequence mode do not diplay in real time these captured seqments.

[Terry]

Explanation is in image (text) you attached.

Normal waweform capture rate (wfms/s) means captured and (real time) displayed waveforms / s

Sequence mode do not diplay in real time these captured seqments.

Do you mean it  can not be measured by the trigger out signal because it is not real time captured waveform?
In this case I think  the waveform capture rate in sequence mode can be measured by 1/?T in history mode. I tested it  on my SDS2102.
For example, at the timebase of 50ns, the waveform capture rate is 1/3us=333,333waveforms /s.  Did I make a mistake?

[rf-loop]

Do you mean it  can not be measured by the trigger out signal because it is not real time captured waveform?

I'm not sure about this. I have not tested this. I do not want quess.
(but, of course it is real time captured. But it is not real time displayed)
It is totally different case if we talk about displayed waveforms/s speed of scope.

Agilent: "Waveform update rate = the number of waveforms an oscilloscope can capture and display in one second"
Good paper for read
http://cp.literature.agilent.com/litweb/pdf/5989-7885EN.pdf

In this case I think  the waveform capture rate in sequence mode can be measured by 1/?T in history mode. I tested it  on my SDS2102.
For example, at the timebase of 50ns, the waveform capture rate is 1/3us=333,333waveforms /s. 

In segment mode we do not see captured waveforms on the screen so it is totally different case.
Using waveforms/s speed for this in Siglent manual is some amount misleading.

As can read in Siglent text, it is over 300000.
If I'm Siglent, I write: ...over 300000 segments/s.
(1s / 3us   in your example)

Quite fast, however. (This class of oscilloscope)

[Robert_Mark]

Hi Guys!

I'am total newbie here so maybe this reply is in the wrong place.

First time when I read about this scope, I just wonder about the specs and features and price.
Small stupid but useful question.
When Dave will do a detailed review? 

I see the siglent guys are here and part of this topic. Can You guys send one 4 ch 100 - 200 MHz version to Dave? This will do really deep impact on the selling!
I already waiting for someone do some deep review because I want to buy some lowmid level 100 Mhz MSO scope. Like Agilent X2000 or Rigol DS4000 or this puppy.
My opinion is that as a hobbyist:
- Agilent great, excellent precision, layout, screen. Constant update speed (asic..) Real equipment! but so expensive when You  need some features (almost a triple price of the scope), no hacks,
- Rigol hackable, more background social knowledge's everywhere but small screen, slower update rate.
- Siglent seat somewhere in between two, but more features, higher specs, higher update speed, cheap feature upgrading (don't need to hack), MSO capable, and cheaper then anything else on the market in this category.

Better if Dave do some  compare video
If Dave do this, I will send some interesting 80's puppy for him!

[Hydrawerk]

I am not a Rigol fan, but the DS2000 and DS4000 have decent displays. DS2000 is 8 inch, DS4000  is 9 inch and Agilent DSOX2000 is 8,5 inch. Not a great difference.

[rf-loop]

- Rigol .... small screen, ....

?


Rigol      DS2000 TFT panel is 8"  800x480 
Siglent  SDS2000 TFT panel is 8" 800x480   500:1 contrast, back light 300nit, color 24bit

[FrankenPC]

- Rigol .... small screen, ....

?


Rigol      DS2000 TFT panel is 8"  800x480 
Siglent  SDS2000 TFT panel is 8" 800x480   500:1 contrast, back light 300nit, color 24bit

The SDS2000 has a really nice screen.  Very easy on the eyes.

[Hydrawerk]

8 inch LCDs are cheap nowadays. It is not like back in 1998 when TDS3000 was bloody expensive...

[edavid]

- Rigol .... small screen, ....

?
Rigol      DS2000 TFT panel is 8"  800x480 
Siglent  SDS2000 TFT panel is 8" 800x480   500:1 contrast, back light 300nit, color 24bit

He is looking for a 4 channel scope, and of course the DS1074Z has a 7" LCD.

And Robert_Mark, another difference is that the Agilent and Siglent scopes have 4 sets of vertical controls, where the Rigol only has 1, so you have to select the channel first to change anything.

[Robert_Mark]

Hi guys!
thx for the  answers. My question is not about what is better, what I need to buy and so on. I will decide based on Dave's review ))

Back to the answers. About the screen.
Looks to me, somehow the Siglent SDS2000 has better more visible screen ratio and arrangement then the Rigol's (Agilent copy?)
I have an old color CRT Yokogawa witch is smaller but thanks to the smart (time to time too smart ) ) arrangement is much more usable then nowadays cheap digi scopes.
So until Dave do a video (and I can decide about the buying) I can survive with my junkyard analog and half analog scopes (banch of old TEK, HP, Fluke, Yoko, Croy)

Most of nowadays under 2K USD scopes annoying me. The lag of the controlling, screen refresh, junk 8bit AD-s etc.
Up to now only the on screen measurement and math functions a real benefit for me by this cheap digital scopes.
Easier then hook up a lot of multimeter, counter, uA uV meter etc and then copy all the values into the computer and do the math.
And this is why the screen is come important for Me. Look and read lot of datas, 3-4 channel waveform together in lab light can be horror-movie if the arrangement, ratio, fontsize, luminance and color bad designed.
But the Siglent SDS2000 serie looks to me interesting. Price, speed, quality, screen and features. Cons is the wobbling knobs by MjLorton:)))

For Me as a hobbiest, the toy investion has to be long and reasonable.   
so please please please Dave,  I (or We?) definitely need to do a video about this scopes
The Agilent ignite a big war on this class and the Siglent who is answered fast and cheap. Lets Dave  judge this war

Siglent guys! When when when Dave will get this scope???

[tautech]

Siglent guys! When when when Dave will get this scope???

Would you send one to Dave before you were sure it was perfect as it could be? 

[tautech]

My question is not about what is better, what I need to buy and so on. I will decide based on Dave's review ))

 
Why not let the specs speak for themselves. 

Cons is the wobbling knobs by MjLorton:)))

This can be found in several new DSO's.
But how much trouble do they give?
How many threads are there here seeking advice on how to repair failed pots?
Yes there are a very few, but mainly stemming from accidental damage. 

[idpromnut]

Siglent guys! When when when Dave will get this scope???

Would you send one to Dave before you were sure it was perfect as it could be? 

What, and miss out on free feedback that can help me make the product better or avoid some really stupid mistakes before i go into mass production? I would be silly not to!

[tautech]

Siglent guys! When when when Dave will get this scope???

Would you send one to Dave before you were sure it was perfect as it could be? 

What, and miss out on free feedback that can help me make the product better or avoid some really stupid mistakes before i go into mass production? I would be silly not to!

I'm quite sure Siglent don't need Dave to hold their hand when they are working closely with Lecroy who make scopes to 100 GHz bandwidth.   
However they will take note of comments he may make, just as they have for the appraisers around the world that have been involved with development.

[rf-loop]

Explanation is in image (text) you attached.

Normal waweform capture rate (wfms/s) means captured and (real time) displayed waveforms / s

Sequence mode do not diplay in real time these captured seqments.

Do you mean it  can not be measured by the trigger out signal because it is not real time captured waveform?
In this case I think  the waveform capture rate in sequence mode can be measured by 1/?T in history mode. I tested it  on my SDS2102.
For example, at the timebase of 50ns, the waveform capture rate is 1/3us=333,333waveforms /s.  Did I make a mistake?

Tested now after I get newest FW and short visit and time in my workshop.
1000 segment. 50ns/div (700pts/segment), Ch1 and  Ch2 on. input 10MHz sine.
Speed around 400000 segment/s (calculated from total time used for 1000 segment using history window timestamps. It need calculate from group  of segments because segment timestamp resolution is only 1us.)

[rf-loop]

What I know after some tests about wfms/s speed, about mask test speed and segmented memory capture speed.

-- Waveforms update rate up to ~110kwfms/s  (average in 1s)
(it have around 115kwfms/s but during scope working cycle there are short interrupt around every ~27ms and it reduce average to ~110kwfms/s)

Measurements do not reduce speed.

-- Segmented memory speed up to >400000 segments/s  (50ns /div) and over 50k segments/s from 1us/div to 1ns/div. (look table)
(in this mode it do not show all separate (segments) waveforms in real time display.)
Segments lenght are same as normally displayed waveforms lenght (1 screen width)

-- Mask test (pass/fail) it is fast as normal wfms/s rate BUT with new FW have some bug in this function. I can not continue this function performance tests.



More about sequence acquire speed.  (some may also name it as Segmented memory acquire)

1ns/div - 50ns/div thre can be max 80000 segments (1 segment lenght is depending time/div setting and is same as display horizontally (one displayed waveform) using same amount of sample points (example 50ns/div 1GSa/s in use (2 channel) 700 points)

1-50ns/div max 80000 segments each
100ns  max 63157 segments
200ns 36585
500ns 16042
1us 8108
2us 4067
5us 1630
10us 815 
and so on.

Speed measured using 100 segments time  using segment time stamps.

Signal 1 - 50ns/div 100MHz
Then 10MHz and 5MHz and 0.5MHz
Trigger Rising edge from CH1
CH1 and CH2 on.
Display Dots. *)
Sin(x)/x off *)
(* after captured, in stop mode   when looking/payback/measure  segments these can turn on)

Speed result:

time/div___segments per second calculated using 100 segment time.
1ns____50900 segm/s
2ns____91900 segm/s
5ns___176000 segm/s
10ns__246300 segm/s
20ns__239800 segm/s
50ns__404000  segm/s ! ( as we know live realtime displayed wfms/s speed is around 110000 )

100ns_325700 segm/s
200ns_222700 segm/s
500ns_111100 segm/s
1us____62500 segm/s
2us____32200 segm/s
5us____13150 segm/s
10us____6650 segm/s 

1 segment is same as 1 waveform.

wfms/s =  captured and displayed waveforms/s

segm/s is captured and stored waveforms/s 
in segment mode only some of them are displayed using slow waveforms update rate on the display)

note that example with 10us/div one seqment lenght itself 140us so if there is not any kind of blind time theoretical maximum speed is 7142 segment in second.

Later I will do more performance tests with this fuction. Speed and amount of segments related to selected memory max value and channels in use and horizontal speed is relatively complicated. Later I try do some table.



Then tested only with 50ns/div (signal 10MHz)

only CH1 on (2Gsa and now of course segment have douple amount of samplepoints)
result 312500 segm/s

All channels on and 10MHz
result 324600 segments

One public note to Siglent!

Please do something for universal knob useability. Try select 79999 frames from zero.
Do it tens of times in one working day with different numbers.
Step increment need self adjust much more clever!

Other thing is knob position on the front panel.
Ergonomic / useability things need really design and it is whole time serious work and who do this need have lot of practice/experience in real world working with equipments.

If I need turn knob just many times for adjust these parameters it may take my working time one hour or more and can not do anything else... I loose my this hour salary. My profession is not knob turner, I need analyze things.

With old ancient HP oscilloscope this kind of parameter setting take exactly 2.1 second. If I need do this same in several working days with this SDS, I need special doctor after it for treatment pain in my hand/fingers. Working ergonomy is important thing!

Oh yes, with this can survive but this is still some amount sad detail in nice equipment.



one screenshot just for example about this test

[Mark_O]

More about sequence acquire speed.  (some may also name it as Segmented memory acquire)

Thanks for performing these tests, rf-loop.  Most illuminating.

1ns/div - 50ns/div thre can be max 80000 segments (1 segment lenght is depending time/div setting and is same as display horizontally (one displayed waveform) using same amount of sample points (example 50ns/div 1GSa/s in use (2 channel) 700 points)

So segmented acquisitions capture into a larger Memory buffer, of >100 MB.  (80,000 x 700 x 2 ~107M.)

note that example with 10us/div one seqment lenght itself 140us so if there is not any kind of blind time theoretical maximum speed is 7142 segment in second.

So in this case, the 'visible' period is 93%, with only 7% Blind.

One public note to Siglent!

Please do something for universal knob useability. Try select 79999 frames from zero.
Do it tens of times in one working day with different numbers.
Step increment need self adjust much more clever!

Other thing is knob position on the front panel.
Ergonomic / useability things need really design and it is whole time serious work and who do this need have lot of practice/experience in real world working with equipments.

If I need turn knob just many times for adjust these parameters it may take my working time one hour or more and can not do anything else... I loose my this hour salary. My profession is not knob turner, I need analyze things.

With old ancient HP oscilloscope this kind of parameter setting take exactly 2.1 second. If I need do this same in several working days with this SDS, I need special doctor after it for treatment pain in my hand/fingers. Working ergonomy is important thing!

I agree that this is a significant oversight, and only possible because it hasn't been tested in real-world use before.  I second your appeal.  I'm sure that will be an immediate complaint from many, as real users get their hands on them.  Luckily, it's a problem that can be easily corrected in software.  (The knob placement, not so much.)

one screenshot just for example about this test

In this case, we're seeing a 700ns acquisition, every 2.5us.  Thus the re-arm time here is 1.8us, and the blind-time is 72%.

[Keeping in mind in all of the above that 'blind-time' is normally referenced in conjunction with live, interactive acquisitions.  So that an operator has an opportunity to observe anomalies in real-time, while exploring dynamic signals.  Still, even used as a post-acquisition analysis tool, this can be very useful.  But only IF there is a way to view those thousands of segments in a usable way.  (overlays, mask compares, Search & flag, etc.)]

[Hydrawerk]

Will there ever be a good Siglent promo like this?? http://info.tek.com/www_mdo3000oscilloscope-oneisthenewsix.html
That SDS2000 might be a good scope but I as a customer have no chance to get wind of that.