Siglent's new product- MSO/SDS2000 Series

[Siglent]

New firmware for SDS2000 series

https://www.eevblog.com/forum/testgear/siglent-technical-support-join-in-eevblog/msg573377/#msg573377

[rf-loop]

SDS2304 peak detect (for correct previous wrong information)

Peak detect works also with 100ms/div and lower horizontal speeds.
Pulses 20ns wide. (first image is only for show what kind of pulses these are)
Peak detect done with 20ns width 1Hz and 0.4Hz pulse freq. (CH1 1Hz and CH2 0.4Hz)

[Hydrawerk]

Will Dave Jones get one scope for review?

[nctnico]

I think Siglent will need to bring the firmware in a more finished state first!

[Mark_O]

I think Siglent will need to bring the firmware in a more finished state first!

I totally agree with Nico on this.  There's literally no point in Siglent providing a unit for Dave to test.  All he will do is rip them a new one.

The hardware on the SDS2000 is solid, and has some unique capabilities.  But the firmware is barely finished, has loads of rough edges, and isn't very usable.  Just look to comments from Nico and Marmad for all the evidence of that which anyone should need. 

Once the firmware addresses those issues, THEN getting Dave a unit to review would be effective advertising for Siglent.  Sooner than that would just damage their rep, even further than selling a unit in this condition already has.

[Hydrawerk]

http://www.siglent.com/ENs/qyxwxx.aspx?id=919&sid=25
According to their website, the SDS2000 was released in November 2013 (in China maybe). The firmware should be OK nowadays more than one year after release.

[tautech]

Will Dave Jones get one scope for review?

I understand he will but I'm not at liberty to say when. 
AFAIK this was always going to happen.

I think Siglent will need to bring the firmware in a more finished state first!

I'm tempted to send him my SDS2304 just to hurry the FW along as Dave will find and publicize any remaining bugs.
However mine is a pre-release unit without a MSO port, so Dave could do only half a job. 

I totally agree with Nico on this.  There's literally no point in Siglent providing a unit for Dave to test.  All he will do is rip them a new one.

The hardware on the SDS2000 is solid, and has some unique capabilities.  But the firmware is barely finished, has loads of rough edges, and isn't very usable.  Just look to comments from Nico and Marmad for all the evidence of that which anyone should need. 

After the 1054Z mess all manufacturers will be a little shy one would imagine.
And this is the way it should be, if your gear is not up to scratch....prepare to get a new one ripped. 

Many of these reported issues have been dealt with, however not all.
Shame Mark does not have his loaner anymore, just to confirm the fixed bugs, also the more pressure from EEVblog members on Siglent the better.

We have asked for FW change logs, in the past they have been held in-house without translation for all to see.
I understand this is to change  and not before time.

[tautech]

http://www.siglent.com/ENs/qyxwxx.aspx?id=919&sid=25
According to their website, the SDS2000 was released in November 2013 (in China maybe). The firmware should be OK nowadays more than one year after release.

It was released as a pre-release unit (HW 3.3) then to dealers and beta testers in the Western markets.
I got my pre-release unit in Dec 13.

[nctnico]

I think Siglent will need to bring the firmware in a more finished state first!

I'm tempted to send him my SDS2304 just to hurry the FW along as Dave will find and publicize any remaining bugs.
However mine is a pre-release unit without a MSO port, so Dave could do only half a job. 

Dave could borrow mine for a while...  Put it to good use and yes I will want it back.

[rf-loop]

Both scopes get exactly same signal. Same level, same cable, same environment all as much same as possible. Only big difference is that with Rigol there is external 50ohm termination and with Siglent its own internal.
Both have 2mV/div, 20MHz BW filter on. Both have 500ns/div and 1GSa/s. Normal sampling mode. Of course with Rigol can not turn all other channels on because in this case it drops samplerate to 250MSa/s.

Result.
Note: in live Rigol screen is smaller than Siglent. Because both have 800x480 RGB pixels it looks same size on the Computer.




Both have 20MHz BW so noise levels are also comparable  enough accurately related to bandwidth.
Note also that CH2,3 and 4 do not have any meaning here in Siglent image. They are just on for some other test left on (just as also forget AC coupling on), do not look them. Only look CH1 signal.  (and CH2 need be on for drop samplerate to same what is Rigol max)

Really there is not any hidden trick for show Siglent better than Rigol. This difference is coming exactly from hardware quality alone. Stop.  (in Rigol Ad: "Low noise floor...")


(this is not preliminary   hw "prototype" version from Nov-Dec 2013 = very different)

[nctnico]

To clarify: which Rigol have you used? If it is a 1000Z then the comparison isn't very realistic because the SDS2000 is in a different class.

[Mark_O]

To clarify: which Rigol have you used? If it is a 1000Z then the comparison isn't very realistic because the SDS2000 is in a different class.

Nico, he's using his DS1054Z DS1074Z.  And comparing his $2600 SDS2304 against the $400 $600 Rigol.  Though I doubt the cheaper "comparable" SDS2074 at only ~$1200 performs much any differently in this test.

What he's really trying to show though is that the residual noise on the Siglent is lower than the Rigol.  And looking at the "fuzz" on the Rigol waveform, one can see that the 3x 2x costlier Siglent does appear to be better in this regard.  That is always good to know, that for $800 $600 more one can get a cleaner trace.

The screen notation about "Extremely difficult to get any stabile [sic] trig" though is an unwarranted, and unfair criticism.  It's pretty obvious that the trigger level on the Rigol is set higher (0.7mV), and closer to the peak than on the Siglent (0mV).  Also, he doesn't mention it, but bumping the vertical to 1mV/div on the cheap Rigol would make triggering even easier.  But he can't do that, because the 3x twice as expensive Siglent is already maxed out at 2mV/div, and can't do 1mV/div at any noise level. 


[BTW, I'm not trying to set rf-loop off here, and generate a barrage of defensive posting.  I think his intentions were honest, and he simply wanted to share his observation that his 1054z 1074z has a higher residual noise level than his Siglent scope, that can potentially obscure signal detail, or imply there is noise in a source signal, that's really inside the Rigol.  And that's always a good thing to know.  (It's also unsurprising, in an instrument in this price class.)

But anyone who has a vested interest (and rf-loop does, in the SDS) has to be really careful in posting any comparisons, because they will get nit-picked to death.  So I definitely give him credit for trying, and sharing his observations.  I wouldn't want to discourage that.]

[NB: updated with strikeouts to correct errors.  Sorry.]

[nctnico]

The screen notation about "Extremely difficult to get any stabile [sic] trig" though is an unwarranted, and unfair criticism.  It's pretty obvious that the trigger level on the Rigol is set higher (0.7mV), and closer to the peak than on the Siglent (0mV). 

That is true but I've noticed that on some oscilloscopes you need to play with the trigger level a bit to get a stable trigger. Especially on signals with a low amplitude or exceeding the bandwidth of the oscilloscope. So I'm not surprised the trigger level on the Rigol is set to 0.7mV. I assume Rf-loop adjusted the Rigol's trigger level so he got the most stable triggering. In my review I have tested my 200MHz SDS2204 with a 450MHz signal at -30dBm input level and it still could trigger perfectly even though the amplitude of the signal was extremely small.

[rf-loop]

To clarify: which Rigol have you used? If it is a 1000Z then the comparison isn't very realistic because the SDS2000 is in a different class.

What ever class but Rigol write this: in Rigol Ad: "Low noise floor..." 
This just one good example  what is different between good expensive HW and elcheapo HW.

Of course other way it is not "fair" due to fact that with one Siglent price can buy 5 Rigol.
But when I go to buy equipment what I need to go from other place to other place  I also compare in my mind bicycle, snowmobile, skis, .... Fiat, Nissan, Geely, BMW and MB...  even if and when  they are in different price class. Why I buy expensive if I can do all wanted with cheap. For this I need know what kind of they are related to  my needs. So it is fair comparison but also many times very important...  Last time I compare expensive SLR camera and quite cheap phone camera. I find that phone camera is enough and better for me in some cases... no one tell me that comparison is not realistic.  Very last time when I need compare products for me was washing machine... other was 10 times more expensive than other.  After compare I find that this elcheapo was ok for my needs.  No one tell me that this comparison was not realistic. Many times I need think what is ok... also when I buy DS1074Z it was perfectly ok for this use what it was for.  After use it can throw out to recycle without any big sad.

[Mark_O]

Also, on the triggering issue, with low-level signals near the threshold of the instrument's sensitivity...

it's not unusual for any DSO, even those much more expensive than the 1000z series, to have trouble with triggering as the signal level approaches or drops below 1 vertical division in amplitude.  That's par for the course.  And the solution is to bump the v-sensitivity (as far as you can), to compensate.

But what rf-loop may have been trying to point out (and I'm sure he will come back and clarify) may have been that the SDS2000 series scopes have an enhanced trigger circuit, which functions well in these marginal situations, where other scopes do not.  [We already know that Siglent made other improvements in their Trigger design, for reduced jitter, that they are proud of.]

If that IS the case, then it's a very good thing.  Because a signal you can't trigger on is a signal you can't see.

~~

OTOH, I just realized that both of his traces were live, and therefore the 'fuzz' on the Rigol trace could be a result of intensity grading turning a small horizontal triggering jitter into a wider band of what appears to be vertical "noise".  Which may not actually be noise at all, but small h-jitter in the Rigol triggering system (which we already knew to be a plus on the SDS2000 series).

So is this really vertical noise, as rf-loop implied, or horizontal jitter?

[Mark_O]

That is true but I've noticed that on some oscilloscopes you need to play with the trigger level a bit to get a stable trigger. Especially on signals with a low amplitude or exceeding the bandwidth of the oscilloscope. So I'm not surprised the trigger level on the Rigol is set to 0.7mV. I assume Rf-loop adjusted the Rigol's trigger level so he got the most stable triggering.

Right.  I completely agree, and tried to clarify in my followup.  I wasn't trying to fault rf-loop, or imply he hadn't done everything he could to optimize the triggering on the Rigol.  My apologies if it appeared otherwise.

In my review I have tested my 200MHz SDS2204 with a 450MHz signal at -30dBm input level and it still could trigger perfectly even though the amplitude of the signal was extremely small.

Yes!  I've said more than once that I felt the SDS2000 series had some unique hardware design advantages.  And their triggering is definitely one of those.  (I will omit this time the obligatory knock on the condition they've left the firmware in. )  I think the hardware is quite impressive.

[rf-loop]

The screen notation about "Extremely difficult to get any stabile [sic] trig" though is an unwarranted, and unfair criticism.  It's pretty obvious that the trigger level on the Rigol is set higher (0.7mV), and closer to the peak than on the Siglent (0mV).  Also, he doesn't mention it, but bumping the vertical to 1mV/div on the cheap Rigol would make triggering even easier.  But he can't do that, because the 3x as expensive Siglent is already maxed out at 2mV/div, and can't do 1mV/div at any noise level. 

No.

Trigger level was set for only position whaty it can even some amount trig.  Before this I did this first and find this situation and after bthen do selfcal and it did not help any.

I have tried also with exatly same signal using Rigol 1mV/div joke. It did not help anything. (and it is natural because it do nothing real)

You can easy check it, 1mV and 2mV noise levels are equal. Noise figure is equal. Only what change is "bit trick". (and rendering image for hide it ... it can easy see if take anough snaps shots and look signal figure. Of course it is easy hide due to "injected" high amount of random noise what nearly hide this trick.)  (my suspect is that they zoom it before rendering),  and tell that now we have 0.5mV/div or even something more.. But this I can not proof without opening this DS1074Z. Yes, it is not DS1054Z, it is DS1074Z (if it matters anything but I pay more for support Rigol because they do not get enough subvention money from goverment)

But. Only point was noise... and Rigol Ad.
Perhaps get idea from GW who also start claim ultrasuperlownoise without any real and then they show some image comparison what was last days best joke.

 

[Mark_O]

But when I go to buy equipment what I need to go from other place to other place  I also compare in my mind bicycle, snowmobile, skis, .... Fiat, Nissan, Geely, BMW and MB...  even if and when  they are in different price class. Why I buy expensive if I can do all wanted with cheap. For this I need know what kind of they are related to  my needs. So it is fair comparison but also many times very important...  Last time I compare expensive SLR camera and quite cheap phone camera. I find that phone camera is enough and better for me in some cases... no one tell me that comparison is not realistic.  Very last time when I need compare products for me was washing machine... other was 10 times more expensive than other.  After compare I find that this elcheapo was ok for my needs.  No one tell me that this comparison was not realistic.

I knew this was coming.     I also happen to agree with you that it's helpful to know what the differences are between instruments in different price classes.  As long as it's made clear that they ARE in different classes.

[tautech]

But. Only point was noise... and Rigol Ad.
Perhaps get idea from GW who also start claim ultrasuperlownoise without any real and then they show some image comparison what was last days best joke.

Mentioned joke link:
https://www.eevblog.com/forum/testgear/gw-instek-gds-2000e-released/msg612529/#msg612529

[Mark_O]

I have tried also with exatly same signal using Rigol 1mV/div joke. It did not help anything. (and it is natural because it do nothing real)

Thanks!  If the increased sensitivity was real, it should have increased the triggering stability.  I didn't know 1mV/div was a joke, but if so, Rigol is not the first to tell it.  HP/Agilent/Keysight have often used this ploy (with an open admission) on some quite expensive scopes.  So all Rigol has done is take a lesson from the masters.

Like you, I'd much prefer they kept to reality.  A "digital zoom" does, in no way, increase the vertical sensitivity.

(my suspect is that they zoom it before rendering),  and tell that now we have 0.5mV/div or even something more.

That may be the case on their DS2000 series(?), but the 1000z doesn't claim 0.5mV.

But. Only point was noise... and Rigol Ad.

Are we sure though it is really noise?  I.e.,  it could as easily be due to inferior triggering stability, compared to the SDS series.

Perhaps get idea from GW who also start claim ultrasuperlownoise without any real and then they show some image comparison what was last days best joke.

No, I 'm sure they got it from HPagiSight. 

[nctnico]

Something to consider when comparing noise levels is the bandwidth of an oscilloscope. More bandwidth usually means more noise. To make things worse: noise isn't bad perse because it is required to make oversampling a.k.a. the high resolution mode work. I used to have a tektronix TDS744A; with the bandwidth limit on the high resolution mode would produce odd results because there wasn't enough noise for proper oversampling.

[rf-loop]

Something to consider when comparing noise levels is the bandwidth of an oscilloscope. More bandwidth usually means more noise. To make things worse: noise isn't bad perse because it is required to make oversampling a.k.a. the high resolution mode work. I used to have a tektronix TDS744A; with the bandwidth limit on the high resolution mode would produce odd results because there wasn't enough noise for proper oversampling.

It is true that for hi-res there need be small amount of "noise". Best is   if injected noise (if need injected extra noise) is not white or pink noise but hf noise where low freq are filtered out before injection. In this case we naturally filter out this injected noise from final result doing averaging. (and it is not at all "over sampling". We only take for averaging these true sampled points what we in normal mode  flush out when use lower than true ADC working sampling speed. ADC work example 1GSa/s (typically ADC works with its nominal fixed speed and we do decimating for lower sample rates) and scope horizontal speed is low, say example 10MSa/s. It means that we drop out 99 points and use 1 point. In simplest case we can take all these 100 points and calculate average and use this result. Of course this can do (or better to do) also using more intelligent way.   But if we do not have enough amount random noise, we can not get more resolution. Specially if noise (including ADC itself) is less than ADC resolution we do not get anything)  Also some other equipments use noise for better result, example some HP counters.

--------
If this noise in picture is because horizontal jitter. Of course I have checked this. Im not people who just have take out from box my first oscilloscope and start ranting. I have made several checkings that I have ground under my legs. Also compared this with single shot and zooming in and out, also compered this with open input single shot and and zooming in and out for look something about this noise figure. Looking also sample points in dot mode. (but this I really do not know what Rigol is doing after ADC. Are these data manipulated or true raw ADC data.)

About 1mV/div
I do not have real proofment what they really have done. But it looks like bit trick.
I have used several HP scopes where they have used only "bit stretch" for most low vertical V settings. I do not know if it was HP or Tek who start this stupidity (in very rare cases it may still be useful for visibility) and of course then someones ape.

But, I did not NOT WANT Rigol  poor-mouth  here! 

This was totally about Ad superlatives,  and only show that more expensive may some times give something better.



 

[nctnico]

Actually your test inspired me to compare the trigger sensitivity between my SDS2204 and an Agilent 7000 series oscilloscope. I used my SDG1010 generator to output a 1MHz sine wave and a 20dB (factor 10) attenuator to have a better adjustment range on the generator. Both oscilloscopes set to 50 Ohm input and bandwidth limit on.
With the Agilent set to 2mV/div (the most sensitive setting) the smallest signal it will trigger on is 2.6mVpp.
With the SDS2204 set to 2mV/div (again the most sensitive setting) the smallest signal it will trigger on is 0.9mVpp (with a stable reading on the frequency counter). When disregarding the frequency counter I can get a stable waveform down to 0.5mVpp.

[tautech]

Actually your test inspired me to compare the trigger sensitivity between my SDS2204 and an Agilent 7000 series oscilloscope. I used my SDG1010 generator to output a 1MHz sine wave and a 20dB (factor 10) attenuator to have a better adjustment range on the generator. Both oscilloscopes set to 50 Ohm input and bandwidth limit on.
With the Agilent set to 2mV/div (the most sensitive setting) the smallest signal it will trigger on is 2.6mVpp.
With the SDS2204 set to 2mV/div (again the most sensitive setting) the smallest signal it will trigger on is 0.9mVpp (with a stable reading on the frequency counter). When disregarding the frequency counter I can get a stable waveform down to 0.5mVpp.

Sounds like the start of a new thread....have a think about pre-conditions, we don't want all sorts of random results. 

[ElektronikLabor]

I found something strange while searching for the sds2000.
On Amazon the picture of the scope looks strange:
http://www.amazon.com/Siglent-Technologies-SDS2072-Oscilloscope-Channels/dp/B00LGD7ZWQ

Compare to traditional view:
http://www.siglent.eu/siglent-sds2304-4kanaals-scope-p-611.html

Compared to traditionaly design the knobs on amazon are different.
Also I can find other differences:

• No curved arrow under the "Intensity" knob
• The rubber buttons are white instead of grey and aren't curved on top and bottom
• SPO label is orange

Have Siglent updated the design?