Siglent SDS1104X-E In-Depth Review

[kahe40]

yeah, a really outstanding work, there is another performance about SDS2000-v2
https://www.eevblog.com/forum/testgear/siglent-sds2000-new-v2-firmware/50/
from the same @Performa01, there one can harvest a lot of knowledge too.

[saturation]

Teardown and earlier review:

https://www.eevblog.com/forum/blog/eevblog-1044-siglent's-$499-sds1104x-e-4ch-oscilloscope-teardown/50/

[lundmar]

For some this might be an important point not touched up in this review.

The remote (LXI/VXI11) programming speed of the SDS1000X is more than 10 times faster than e.g. the Rigol DS1000Z series.

This is useful if one needs to implement a poor mans data logger polling sample data from the scope at e.g. 100Hz.

I mean, we are talking about ~25 requests/second vs. 250+ requests/second here. A significant difference.

[Performa01]

Thank you all for the nice comments! Today I can finally provide some updates.

New revisions of all documents have been uploaded in the old posts (see EDIT notes there) and review parts 6 and 7 have been newly released in Reply #3.

The new parts cover Math, FFT, Mask test and Web Server. Unfortunately, FFT is such a huge topic so it didn’t fit into a single 1MB document and I had to split it between parts 6 and 7.

The review is still not complete and I’ll certainly not stop at this point.

[Performa01]

What is Siglent's reasoning for no AC trigger level indicator, do you know ?

Well, the AC trigger indicator has been there in an early version of the V2 firmware, but back then the AC trigger level was static, hence didn’t quite work as expected. I’ve been in touch with Siglent about this and it took several iterations until I was finally completely happy – but the trigger level indicator somehow vanished during this process and never returned, even though I’ve mentioned it a few times

It probably is a speed problem. Just like tracking cursors cannot follow a dynamic signal very well (only the cursor lines, the measurements do!) I suspect it would be the same with the trigger level indicator, so Siglent may have decided to better hide it

As I said in the review, I’m not happy with that, but it really is only a minor inconvenience in my book. On the old analog scopes we had no trigger level indicator at all, no matter the trigger coupling

[Performa01]

Some are probably waiting for the serial decoding review, which I am certainly planning to publish eventually. As I’ve already stated in my initial posting, the serial decoders still need refinement, as they have some issues and at least one missing feature. Siglent is aware of that and promised to implement the necessary bug fixes and improvements.

We’ll probably get a new firmware soon, where at least sequential mode is fully working again and some other bugs from my list have been addressed, but not sure if serial decoding has already been improved as well, as I’ve reported these issues fairly late in December. So we need just wait and see…

Until then, I thought I’d show a little teaser here:


SDS1104X-E_SPI_UART_Zoom_2

What you can see from this screenshot are two full duplex decoders working in parallel, one SPI and one UART. Of course this is an artificial setup, as this would not be possible with just 4 channels (it will be possible with the MSO option though). I just configured a custom baud rate of 2Mbit/s for the UART so it can find valid data in the 2.36MBit/s SPI data stream. The UART RX cannot work continuously, because it’s looking at the MISO stream, which in turn is just the inverted MOSI and has the wrong idle level for the UART decoder.

There is a record length of 700kpts for a time span of 7ms, zoomed in at 5µs/div so that we can read the decoding at the bottom of the screen. In the top area there is the decoder list for SPI, which is automatically adjusted for the center of the zoom window, showing three 8-bit data elements (which could be up to 32 bits and the list could be enlarged up to 7 entries), number 1740-1742 out of a total of 3492 entries for the complete record.

We can only display one decoding list at a time, but the one for the UART is quite similar, just with slightly less entries.

There was no protocol trigger used in this test, but a trigger on the falling edge of the ~CS signal. As a consequence, the trigger frequency counter displays the number of messages per second (27895).

Overall I think Siglent is on the right track with this.

EDIT: total time for one record corrected - it is of course 14 x 500µs = 7ms. Exact number of list entries added.

[rigol52]

Excellent (rare seen) review, thanks a lot Performa01.

Regarding "an external USB powered 25 MHz AWG module (Option)" from SDS1000X-E data sheet.

Is this "Option" related only to some sort of universal driver software, suit to other brand external AWG too,
or it is strictly and only Siglent module / external AWG required?

[tautech]

Excellent (rare seen) review, thanks a lot Performa01.

Regarding "an external USB powered 25 MHz AWG module (Option)" from SDS1000X-E data sheet.

Is this "Option" related only to some sort of universal driver software, suit to other brand external AWG too,
or it is strictly and only Siglent module / external AWG required?

There's two purchasable options:
The External SAG1021 USB AWG 25 MHz module.
The license to allow full functionality AWG control of it.

Bode plot usage requires no licenses, only the USB AWG HW or any other Siglent AWG. The scope takes charge of Siglent AWG's to do the Bode plot sweeps.

[rigol52]

Bode plot usage requires no licenses, only the USB AWG HW or any other Siglent AWG.

Should this AWG be exclusivelly Siglent brand, or other brand of USB AWG can do too (to proper
communicate with scope over USB for Bode plot support)?

Curently I have Rigol AWG DG1011 and interested, if Siglent scope SDS1104X-E can support it
as external AWG for Bode plot?

[tautech]

Bode plot usage requires no licenses, only the USB AWG HW or any other Siglent AWG.

Should this AWG be exclusivelly Siglent brand, or other brand of USB AWG can do too (to proper
communicate with scope over USB for Bode plot support)?

Curently I have Rigol AWG DG1011 and interested, if Siglent scope SDS1104X-E can support it
as external AWG for Bode plot?

AFAIK it won't support your Rigol AWG, only Siglent models.

[rigol52]

Thanks.
Not really happy with such facts which lead to changing lab equipment (seems along with my nick name),
bu what to do - world is not perfect.
Still very impressed with capability of here reviewed scope taking into account its price/performance outcome.

[tautech]

Thanks.
Not really happy with such facts which lead to changing lab equipment (seems along with my nick name),
bu what to do - world is not perfect.
Still very impressed with capability of here reviewed scope taking into account its price/performance outcome.

Quite understand, yes not perfect at all but most brands do this proprietary stuff. 

Yes they are a feature packed little box so see if you can find one locally for a test drive.

[BillB]

Great in-depth review, Performa01 

Helped me decide to pick one up!

[maxwell3e10]

Thanks, Performa01, for a very detailed review. I am trying to figure out the input noise of the scope from your data, section 3 of the review.

When I look at figure on page 64 I would estimate that +/-2 sigma level is about 1 division, so the RMS noise is 125 uV. When I look at RMS noise on the plot on page 67, 16 LSB corresponds to 16*0.5mV*10/250 =  320 uV.

I am wondering if the RMS noise estimate on page 67 comes from the RMS measurement on the scope? In that case it can include the offset as well. It would be interesting to find the standard deviation of the data with shortened input.

[tautech]

Thanks, Performa01, for a very detailed review. I am trying to figure out the input noise of the scope from your data, section 3 of the review.

When I look at figure on page 64 I would estimate that +/-2 sigma level is about 1 division, so the RMS noise is 125 uV. When I look at RMS noise on the plot on page 67, 16 LSB corresponds to 16*0.5mV*10/250 =  320 uV.

I am wondering if the RMS noise estimate on page 67 comes from the RMS measurement on the scope? In that case it can include the offset as well. It would be interesting to find the standard deviation of the data with shortened input.

If you are going down this rabbit hole of DSO baseline noise as it compares to other DSO's, please consider a further two things:
    Is max input sensitivity at full BW and not BW limited.
    Is the max sensitivity an artificial magnification of real/actual maximum sensitivity ?

They both have an impact on apparent baseline noise.
Only then can you compare apples with apples.

[Performa01]

Thanks, Performa01, for a very detailed review. I am trying to figure out the input noise of the scope from your data, section 3 of the review.

When I look at figure on page 64 I would estimate that +/-2 sigma level is about 1 division, so the RMS noise is 125 uV. When I look at RMS noise on the plot on page 67, 16 LSB corresponds to 16*0.5mV*10/250 =  320 uV.

I am wondering if the RMS noise estimate on page 67 comes from the RMS measurement on the scope? In that case it can include the offset as well. It would be interesting to find the standard deviation of the data with shortened input.

You are perfectly right that RMS measurements always include the DC offset error, which has been around 10 LSB at that particular time, as can be seen in the graphs on page 66. Unfortunately, I tend to forget that and the data in the graph on page 67 comes from the automatic RMS measurement indeed.

Thanks a lot for the hint!

Here is an example for 1µs/div with the correct measurement included:


SDS1104X-E_Noise_1us_500uV

Regarding the scope input termination, it makes no difference whether it is shorted or left open. A high voltage tolerant amplifier with its clamp protected high impedance input network cannot be low noise anyway, so the noise of the source impedance is absolutely negligible. Yet I’ve fitted a 50? through termination just to make the input insensitive to electrical noise that might be generated within my lab.

I will update the noise graph for the next revision of my review document, but here is a preliminary version as an immediate “hot fix”:


SDS1104X-E Noise_vs_time BW_full


So we’re actually talking about a noise level of 39µVrms @ 1ns/div slowly increasing up to 68µVrms @ 1ms/div. I’ve also taken more time to determine the exact mean pk-pk noise level and as can be seen, it is a much smoother trace now.

Finally, I’d also like to show the absolute horror scenario for any general purpose scope input, i.e. 100ms/div, which should be equivalent to a lower noise bandwidth limit of <1Hz:


SDS1104X-E_Noise_100ms_500uV

Now we have 705µV pk-pk noise, but rms noise is even lower (compared to 1ms/div) at 64µV. That’s most likely because the sample rate has now dropped to 10MSa/s (whereas it still was 1GSa/s at 1ms/div) and quite obviously the aliasing caused by the noise above 5MHz does not have much impact for a number of reasons.

[maxwell3e10]

Thanks for quick measurements!

I wouldn't trust the rms noise for very short time scale (1 ns/div), because its beyond the bandwidth of the scope, so the data are largely correlated. The flat level of 60 uV rms is more representative of true noise.

As tautech points out, its also important to compare the BW. On many scopes the 1 mV/div scale is BW limited by default to 20 MHz, so for fair comparison it would be good to know the noise level on 20 MHz BW setting.

[Performa01]

Thanks.
Not really happy with such facts which lead to changing lab equipment (seems along with my nick name),
bu what to do - world is not perfect.
Still very impressed with capability of here reviewed scope taking into account its price/performance outcome.

Yes, I know it’s a pity and I also don’t like to change an instrument that does the job and that I’m used to. But you just cannot expect instruments from different manufacturers to remote control each other.

There is no universal standardized protocol to remote control signal generators (or any other instruments). It is sometimes not even the same for all instruments of the same class within one brand. So it’s no big surprise that T&M companies don’t make an effort in supporting instruments from competing brands, except when they have been established as a (quasi) industry standard.

You can only combine devices from any brands you like when you write your own application (on a PC) and remote control all instruments involved. By doing this, the many pitfalls associated with controlling instruments for a more complex application like network analysis will become clear quite quickly – apart from the fact, that it is way more efficient if that application runs directly on the instrument that has to do the majority of the work, i.e. the DSO in this and most other cases.

[Performa01]

Thanks for quick measurements!

I wouldn't trust the rms noise for very short time scale (1 ns/div), because its beyond the bandwidth of the scope, so the data are largely correlated. The flat level of 60 uV rms is more representative of true noise.

As tautech points out, its also important to compare the BW. On many scopes the 1 mV/div scale is BW limited by default to 20 MHz, so for fair comparison it would be good to know the noise level on 20 MHz BW setting.

Well, even though the SDS1104X-E is only a 100MHz scope, the figures get much better with the 20MHz bandwidth limit. Here’s the graph:


SDS1104X-E Noise_vs_time BW_20M


And here’s a screenshot for 1µs/div:


SDS1104X-E_Noise_1us_500uV_BW20M

[tautech]

Well, even though the SDS1104X-E is only a 100MHz scope, the figures get much better with the 20MHz bandwidth limit.

Dots or vectors ?

[Performa01]

Dots or vectors ?

Vectors!

This is what it looks like with dots display:


SDS1104X-E_Noise_1us_500uV_BW20M_Dots

[nctnico]

On many scopes the 1 mV/div scale is BW limited by default to 20 MHz, so for fair comparison it would be good to know the noise level on 20 MHz BW setting.

Which ones? The only scope I owned which switched the 20MHz BW limit on at 1mV/div was a Siglent SDS2000 series.
However comparing with the 20MHz bandwidth enabled does give a good comparison because more bandwidth usually means more noise.

[rf-loop]

Thanks, Performa01, for a very detailed review. I am trying to figure out the input noise of the scope from your data, section 3 of the review.

When I look at figure on page 64 I would estimate that +/-2 sigma level is about 1 division, so the RMS noise is 125 uV. When I look at RMS noise on the plot on page 67, 16 LSB corresponds to 16*0.5mV*10/250 =  320 uV.

I am wondering if the RMS noise estimate on page 67 comes from the RMS measurement on the scope? In that case it can include the offset as well. It would be interesting to find the standard deviation of the data with shortened input.

Here is added SDS1104X-E noise (bit explanation and 4 images)
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1290409/#msg1290409
Also explanation why RMS is not right for noise with more or less DC offset (non zero mean)

[Performa01]

Here comes an in-depth review of the optional SAG1021 AWG.

It’s a bit ridiculous, but I had to split the review for this humble little AWG into four parts (three would have worked too, but I didn’t want to tear the sine wave chapter apart). So here are the first two parts:

SAG1021 Review 1-8
- Operation
- Zero Adjust
- Frequency Accuracy


SAG1021 Review 8-25
- Waveforms
-- Sine
--- Amplitude Accuracy
--- Harmonic Distortion
--- Phase Noise

[Performa01]

Here are the second two parts of the SAG1021 AWG review:

SAG1021 Review 25-32
- Waveforms
-- Square
-- Ramp
-- Pulse
-- Noise
-- DC

SAG1021 Review 32-46
- Arbitrary Waveforms
-- Internal
-- External