Siglent SDS2000X Plus

[jemangedeslolos]

Hello Performa01,

Thank you for the explanations.
It will be my main scope and I have a MSO5000 as a second scope when I have to work at home who does not have shared control.
I need to see by myself if it is not so disturbing. I will wait a review to see a side by side comparison between the SDS2354X Plus and the SDS5034X.
The price difference between these two will be decisive 

[Martin72]

Hi,

Reading this thread, I don´t see any advantages against

I have a MSO5000

except the 50Ohm inputs on the hardware side.

Apart from this, I don´t like the "A-Brands" discussions every time a new model from rigol or siglent comes up.
Siglent and rigol makes decent scopes for hobbyists affordable, not less or more.
We all should keep this in mind, when moaning/judging about some supposed quirks.
If we could spend for example 2000 bucks, what do we get from siglent, rigol, r&s, tektronix, lecroy, keysight, this is the right question.
And this is the discussion framework within which we should move.
At company, I work with a waverunner 9054 from lecroy everyday since we got it - At home there is my MSO5 from rigol waiting.
It´s no comparison at all, but this is "natural" .
Nevertheless I could also work with it.
At home.
Doing hobbyist things.
If I need/want to do professional things, I buy a professional scope.
That´s all.

[Frex]

Hello tautech,

I look to replace in some time my MSO2072A and this new SDS2000X Plus would fit well.
Anyway, what about noise floor of this new scope ?
What aquisition modes are available too ?
Does 10 bits mode improve FFT dynamic range ?
I would really appreciate if you are able to take some measurements about this. Regards.

Frex

[tautech]

Hello tautech,

I look to replace in some time my MSO2072A and this new SDS2000X Plus would fit well.
Anyway, what about noise floor of this new scope ?
What aquisition modes are available too ?
Does 10 bits mode improve FFT dynamic range ?
I would really appreciate if you are able to take some measurements about this. Regards.

Frex

Frex, we'll have to wait some 6 weeks before public release unless Performa01 can share with us these details.

[rf-loop]

@Performa01: Does Siglent have any plans to get rid of the automatic memory depth setting? For me the automatic memory depth adjustment is a total productivity killer. It annoyed the hell out of me during the short time I had the SDS2000.

The short answer is no, sorry.


The zoom window supports math, automatic and cursor measurements as well as mask test.

The maximal record length can be limited in order to reduce the sample rate and/or get more history segments.

Any acquisition strategy that would hide a single sample from instant view would annoy the hell out of me. Thankfully, Siglent will continue to follow PicoTech and LeCroy in this regard.

We get a lot of feedback from professional Siglent users and I think it's most indicative that not one has ever questioned this acquisition strategy.

And how about full memory decoding? It is about time Siglent addresses this too.

This is a (mis)leading question.

 

Strong Yes!

[nctnico]

And how about full memory decoding? It is about time Siglent addresses this too.

This is a (mis)leading question. While there are certain (protocol dependent) limits for the total number of table entries (bytes, words, double words, packets) which in turn obviously limits the maximum record length that can be fully decoded (at least if oversampling is low), not many apart from you would consider the following screenshot with a total of 45 decoded I2C messages as not full memory decoding:

But it still needs the zoom crutch. Everyone can walk faster without a crutch. Also decoding only what is on screen gives you the problem that it is nearly impossible to correlate messages which may be in different parts of the acquisition memory. Get a Keysight or R&S scope on your bench and see how easy it is to be able to go through the entire record with timestamps relative to each packet. Same for the automatic memory length. For the millionth time: limiting the memory size to the visible screen only is a serious productivity limitation. There is no arguing around that other than using yet another crutch (going back & forth between time/div settings). Everyone can walk faster without a crutch.

Remember I used to own an SDS2000 and the automatic memory selection annoyed the hell out of me. And no, automatic memory length is not a typical Lecroy feature. The Wavepro 7200A I have allows to select between automatic and fixed memory length just like any other oscilloscope I have. Except for the SDS2000 I don't recall having owned any other DSO which did not allow to select a fixed memory length.

Reasoning that none of the customers has asked for it is more indicative of the kind of customers Siglent has than automatic memory selection should be standard. Power users like me that need to have an oscilloscope which works efficiently will ask for it (or silently just don't buy the oscilloscopes from Siglent). Siglent has come a long way in the past few years and now they really have to start addressing the productivity issues (at least on the higher end models) in order to really compete with the A-brands.

[Performa01]

Anyway, what about noise floor of this new scope ?

It's in line with other Siglent X-series DSOs.

Here are some measurements with 1Mpts, 2GSa/s (resulting measurement bandwidth is 2kHz ~ 1GHz) at 1mV/div:

Full bandwidth (570MHz): 80µV RMS
Bandwidth limit 200MHz: 54µV RMS
Bandwidth limit 20MHz: 30µV RMS

As expected, 10 Bit Mode does not change much at high input sensitivities, because the noise from the analog frontend is dominating:
Full bandwidth (100MHz): 36µV RMS
Bandwidth limit 20MHz: 27µV RMS

See attached screenshots.

What aquisition modes are available too ?

Only Normal (aka Sample) and Peak Detect.
Average and ERES are now implemented as math functions. I'll post some demonstrations of ERES later.

Does 10 bits mode improve FFT dynamic range ?

No, It cannot. 10bit mode is accomplished by oversampling, taking advantage on the highly linear ADC.
FFT implicitely does something similar and vastly increases the dynamic range by itself, so 10bit mode is no benefit here.

Here's an example of a 30MHz carrier, 5% amplitude modulated at 100kHz. Visible (and spurious free!) dynamic range is 80dB and the peak table confirms that it is fully operational. That's better than many old swept SA boat anchors out there...


SDS2354X+_30MHz_3dBm_AM100kHz_5%

[Performa01]

ERES

ERES means enhanced resolution and enhancement is available from 0.5 bits to 3 bits in 0.5 bit steps. You don't specify the vertical resolution with this, but the (theoretical) enhancment of the ENOB (Effective Number Of Bits). The vertical resolution is twice as high, so 3.0 bits enhancements get you 6 bits additional vertical resolution, hence a total of 14 bits.

Look at the screenshot below:


SDS2354X+_Math_ERES2_8bit

This shows a zoomed view of the input signal (1kHz sinewave) which shows exactly one sample per division at a 16mV/div scale. So the LSB of the original sample data is equivalent to 16mV.

The ERES math trace with 2.0 bits enhancement in the zoom window shows two samples per division at a 2mV/div scale, hence one LSB is now only 1mV. This is a 16 times improvement, equivalent to 4 bits.

Effectively the math trace shows the signal with 12bit resolution.

[Performa01]

ERES 2

The noise reduction effect of ERES.

The following experiment uses the base noise at 100mV/div, where the contribution from the analog frontend is almost negligible.

At 100mV/div, one LSB is equivalent to 3.33mV. The zoom window at 5mV/div shows the granular +/-1 LSB noise of the input signal as +/-0.66 div (green trace).

The orange ERES math trace is scaled at 5mV/div, so it is directly comparable to the input signal.  The granular noise after 1.0 bits of resolution enhancement is correspondingly lower:


SDS2354X+_Noise_50_100mV_ERES_1bit

The following screenshots show the same view, but with 2.0 and 3.0 bits of resolution enhancement:


SDS2354X+_Noise_50_100mV_ERES_2bit


SDS2354X+_Noise_50_100mV_ERES_3bit

[Performa01]

Average

Average is another function that has been changed from an acquisition mode to a math function. One of the advantages is that we can now combine ERES and Average for the same signal. We can also use the new 10 bit acquisition mode which will theoretical increase the ENOB by 1.0 bits.

As a demonstration I’ve set up a 883µV RMS sine signal buried under a 5mV RMS noise carpet and 10 bit acquisition mode has been used. Triggering on the weak signal is accomplished over the sync output of the signal generator.

In the screenshot below we can see that the working intensity grading in this scope already reveals that there is a signal hidden under the noise. The math trace shows the input signal with 1024 times averaging, which makes it almost noise free, especially when considering the low signal level and the 1mV/div scale.

Automatic measurements show the properties of the original signal with a maximum amplitude of ~25mVpp and only ~2.33mV RMS (mean statistics value) – this is because the noise signal was set for 500MHz bandwidth, but the actual input bandwidth is only 100MHz in 10 bit mode, which results in a theoretical level of 2.24mV RMS.

The math trace measures the recovered sinewave as 2.78mVpp which is a bit too high because of the remaining noise (should actually be 2.5mVpp). Yet the RMS measurement is very accurate at 888µV (should be 883µV).


SDS2354X+_Noise_5mV_sine_883uV_10bit_IG

Since the intensity grading already revealed that there is a sinewave hidden under the noise, color grading shows this even better:


SDS2354X+_Noise_5mV_sine_883uV_10bit_CG

[Emo]

Hi Performa01,

If I understand this correctly the new 10 bit acquisition mode is accomplished in software. Do you expect this to become available in other(high-end) models as well?

Eric

[Performa01]

Hi Eric,

the SDS2000X Plus is the first Siglent DSO that implements 16 bit data processing and display interface, which is the prerequisite for features like vertical zoom and opens up further opportunities, like the new HR (10 bit) acquisition mode.

The 16 bit data has been on the todo list for the SDS5000X right from the start and will be ported from the SDS2000X Plus ptototype implementation as soon as it is mature (which has already been the case for quite a while now from my point of view).

Contrary to my initial reservations, the 10 bit mode has proven to be quite useful (because of the excellent DNL of the ADC), so I expect it to be added to the SDS5000X and future DSOs in and above the 2000X class.

[Performa01]

HR (10bit) + ERES

The good performance of the ADC in this DSO led me to trying something crazy: why not combine the 10bit HR acquisition mode with ERES 3.0, for a total of 16 bits of resolution?


SDS2354X+_Math_ERES3_10bit_St

You can see the original trace in the upper window at a vertical gain of 960mV/div. Because in 10bit HR mode we get exactly 120LSB per division and one LSB is already reduced to 8mV, which is just a quarter of what it would be in 8bit mode.

In the zoom window at a vertical scale of 8mV/div we can see exactly one green dot per division, just as expected.

But now look at the orange math trace. Here the sensitivity is set to 500µV/div and we see exactly 4 dots per division, which is equivalent to 125µV for one LSB.

Look how evenly spaced the orange dots are. The total resolution is 16 bits and linearity looks still excellent.
Of course, the total bandwidth has dropped to just 14MHz now.

[mysiak]

I am tempted to get the Rigol MSO5074, hope Siglent will release this guy soon 

Is the bandwidth software unlockable ?
How is the screen's resolution? Similar to R&S quality? This is the only thing which bother's me a bit about Rigol.

Thanks!

[tautech]

I am tempted to get the Rigol MSO5074, hope Siglent will release this guy soon 

In just a few weeks are the hints I have.

Is the bandwidth software unlockable ?

There will be BW upgrade options from what we can see on the Chinese datasheet.

And welcome to the forum.

[mysiak]

Thank you so much for an extremely quick reply.

I edited my very first post to ask about the screen's quality before I noticed you already replied. This forum community is just awesome!

[tautech]

How is the screen's resolution?

1024x600 10.1" TFT-LCD

Similar to R&S quality?

Siglent's displays have always used a matt toughened glass so don't have the issues with mirroring the surrounding environment or lighting like some R&S models do.

Apparently some R&S owners have resorted to using a screen overlay to address the display reflections but that's plastic and with use might need replacing and for a touch display that might get lots of use this becomes an additional cost of ownership.

[nctnico]

Apparently some R&S owners have resorted to using a screen overlay to address the display reflections but that's plastic and with use might need replacing and for a touch display that might get lots of use this becomes an additional cost of ownership.

Now you are being overly dramatic. From my own experience with several 'glossy' touch screen scopes the reflections are not an issue at all. Your eyes will focus on the display and by doing so get rid of the reflections. It is basic optics really.

[jjoonathan]

Matte screens can fuzz out the reflection of a light source situated right behind you, but by the same token, they gather light from sources that aren't right behind you and fuzz it into your display. Pick your poison.

Note that cell phone displays tend to be glossy to let you deal better with the reflection of the sun (!). If the sun is right behind you, its' reflection in your phone screen is a huge problem, if the sun isn't right behind you, its' reflection is a small problem. If you made the screen matte, the sun's reflection would always be a medium problem. Since it's usually possible to tilt your phone screen or change your orientation to pick the small problem over the huge problem, {small,huge} wins over {medium,medium}, and that's why smartphone screens tend to be glossy, because under those circumstances glossy provides better reflection control.

Of course, if the boss man had me chained to a desk across from a window, I'd want a matte screen

[tautech]

Noteworthy and come release only 100, 200 and 350 MHz models have been made available to western markets. 

Edits with models and pricing made to the OP.

[Martin72]

Noteworthy and come release only 100, 200 and 300 MHz models have been made available to western markets. 

To keep it below the bandwiths of the SDS5000 series, I guess..

[rf-loop]

Noteworthy and come release only 100, 200 and 300 MHz models have been made available to western markets. 

Edits with models and pricing made to the OP.

100, 200 and 350MHz  and 350MHz model have 500MHz option.

Option SDS2000XP-4PW50  is availablein western markets and also english version datasheet tell it.
It is for SDS2354XPlus for upgrade it to 500MHz.*)

*)
as all know 1GSa/s fnyquist is 500MHz, so it is "useless" for 1GSa/s operation using rt sampling and Sinc, but afaik it can run 2GSa/s with 2 channels simultaneously and in this case 500MHz is ok with Nyquist and Sinc)

[Martin72]

https://www.batronix.com/versand/oszilloskope/Siglent-SDS2000X-plus.html

Rigol´s mso5000 is getting into trouble...

[Martin72]

So who is going to research if the hardware is the same and 100Mhz = 500Mhz. That would be a Rigol killer... for sure..

https://www.eevblog.com/forum/testgear/new-siglent-sds2000xplus-launched/msg2874388/#msg2874388

Comparing to the mso5000....

50Ohm Inputs, probe-sensing, external trigger, bode plot integrated, bigger 10.1" screen....price for the 100Mhz version 1200€ (excl. VAT)....MSO5104 : 1100...
Makes me nervous…

What hacking to 500Mhz concerns:
Only the 350Mhz versions seems to be upgradable to 500 Mhz, so I think there will be a difference in hardware.

[tv84]

So who is going to research if the hardware is the same and 100Mhz = 500Mhz. That would be a Rigol killer... for sure..

https://www.eevblog.com/forum/testgear/new-siglent-sds2000xplus-launched/msg2874388/#msg2874388

Comparing to the mso5000....

50Ohm Inputs, probe-sensing, external trigger, bode plot integrated, bigger 10.1" screen....price for the 100Mhz version 1200€ (excl. VAT)....MSO5104 : 1100...
Makes me nervous…

What hacking to 500Mhz concerns:
Only the 350Mhz versions seems to be upgradable to 500 Mhz, so I think there will be a difference in hardware.

I think it will be an interesting battle!  Of course you should compare with 5074 since it's the lowest we can get.

Regarding the HW difference: maybe it's only on the factory calib.  How do they distinguish a 350 (upgraded from 100) and a 350 bought in the shop, in HW?

The ones that didn't reach the aprox 570 MHz BW, are rated 100 MHz models?