EEVblog #1220 - Siglent SDS5000X 1GHz Oscilloscope Review

[EEVblog]

Siglent have entered the 1GHz oscilloscope market with the SDS5000X, is it a worthy competitor?

[maginnovision]

I was cracking up when you wanted to do sequence mode with zone trigger and couldn't. The ONE THING you wanted to do you couldn't. Pretty funny.

Edit: Teardown

[EEVblog]

I was cracking up when you wanted to do sequence mode with zone trigger and couldn't. The ONE THING you wanted to do you couldn't. Pretty funny.

It didn't seam unreasonable!
I was shocked when it didn't do it, had no inkling that it wouldn't, just seemed like an obvious practical test and cool combination of features.
I can't really think of any architecture thing that would prevent it, yet the software knows and the has the message a the ready in case you try.

[TheSteve]

It seems to be a jack of all trades, but a master of none.

[JPortici]

(Glad you were using 8.2R1 which is indeed the latest firmware available)

I think zone trigger is implemented in software. Trigger -> processor get data, compare with zones -> actual trigger
This would explain why sequence mode is not available with zone

Dave, can you do a teardown of the active probe? I'm really interested in it.. or at least in the communication protocol. The interface on the hardware level is veeeeeery simillar to probus (i measured +12,-12,+5,0 and two 3.3V which may very well be SDA and SCL)
Can't find the thread.. there is one on reverse engineering probus

Re: bugs
cool, didn't know of that one. But i NEVER use gestures. I tried once just to see if they were implemente but after that... knobs or manual entry on the bottom left channel menu.. same for horizontal/vertical adjust.
I'm very fine with no acceleration on the encoders, as i can use the touch screen or manual input for coarse adjustment

[thmjpr]

$2,900 the 2ch version is in a bit of a no mans land. Rigol MSO5000: 4 channels, 2 channel sig gen built in (with modulation), better update rate, better intensity/color grading, higher sample rate. Of course, Siglent has a lot of features: histogram charts and zones is unique, 50 ohm, active probe, better responsiveness, front end. And the updates/feature requests via forum is huge too.

For the 4CH model, one of those things has to jump out at you to pay the $3500 vs $1000 price. If 350MHz is hackable to 500MHz, then you've surpassed the rigol spec, and really in another market.
The 350 and 1G have different front ends according to the siglent thread, so you may not be able to get the full 1G via unlock.

BTW if you want to clear something thats gone awry, have to hit 'Default' :p. With the Auto button I would just expect it to set horizontal/vertical settings, trigger level, maybe trigger type and leave everything else as-is. Considering the ridiculous number of settings scopes have these days, I wouldn't be surprised if they start implementing profiles to store custom setups.

I think zone trigger is implemented in software. Trigger -> processor get data, compare with zones -> actual trigger
This would explain why sequence mode is not available with zone

Thats not the greatest solution, but may have been a compromise that had to be made given the acquisition structure.
And they call the FFT "Hardware accelerated", but I wonder to what degree, its not wildly fast.

[JPortici]

mm.. I have the 4 channel 350 MHz and that's 2.9k here

Considering the ridiculous number of settings scopes have these days, I wouldn't be surprised if they start implementing profiles to store custom setups.

Ah, it's already been there for quite some time. I have used a bunch of scopes with this capability

[2N3055]

Thats not the greatest solution, but may have been a compromise that had to be made given the acquisition structure.
And they call the FFT "Hardware accelerated", but I wonder to what degree, its not wildly fast.

2Mpoints FFT is slow by virtue of gate time. Somebody should try speed on Siglent with 64Kpoints FFT like Keysight has. And usually Keysight doesn't even have full 64K you have to set it so.
So until then , we don't know what the speed is .

Yeah, it's rough around the edges.. We'll see..

[tautech]

(Glad you were using 8.2R1 which is indeed the latest firmware available)

Version 8.4R4 has been out with the beta testers for a few days already.

[TurboTom]

Dave, thanks for the review 

But I've got to place some small criticism here -- your input noise comparison is somewhat dodgy (34:00 into the video): Since the R&S had some visible DC offset of the input signal (round about -300µV), this will clearly contribute to the RMS measurement. Try checking noise with the inputs in AC coupled mode. Depending on the internal processing, the average offset will get subtraced in this mode (not sure if it works this way on the R&S, at least the "mature" Rigol products do it like this).

Then I'm sure you'll find that the R&S is at least on par with the new Siglent scope.

Cheers,
Thomas

[maginnovision]

No matter what the RTB is disadvantaged since the Siglent drops to 200MHz bandwidth at small v/div. The only like for like comparison could be done is with both having the 20MHz filter on. However, I would expect the Siglent to have better noise floor anyway.

[Performa01]

Maximum waveform update rate of 110kWfm/s can be reached at 20ns/div in dots mode only. The sin(x)/x reconstruction filter takes its toll in terms of processing time on all the Siglent scopes.

There certainly is a History button on the front panel and it’s even lit when in history review mode.

Zone Trigger is just an addition to any of the existing triggers (so it can be used to refine a complex trigger even further), but not an independent trigger function itself. This is why it doesn’t belong into the trigger selection menu.

BTW, I don’t think Zone Trigger is a software function, as this would be far too slow. It is in the FPGA, like pretty much anything that has to do with triggering.

I cannot know for sure, but I’m positive that History for zone triggered acquisitions could be supported. I guess it could be just like Mask Test in the Zoom window – I’ve long requested that and we’ve got it now on the SDS5000X. I’ll have to check with Siglent R&D.

To get rid of an accidentally placed Histogram region (which automatically enables the Histogram), we need just disable the Histogram function in the Analysis menu.

Intensity Adjust as the default function on the Intensity Adjust knob (as we have it on the older SDS2000X) is on Siglent’s todo list, just like the search for serial triggers.

Finally, I think we can clearly see from the video that the R&S is slightly lower noise despite the higher bandwidth at 1mV/div – but it just so happened to have a far worse DC offset than the Siglent, which didn’t show any noticeable offset at all in this test. So if the proper measurement (standard deviation = RMS without DC component) would have been used, then Siglent would have lost that duel – only by a small margin, but still...

[nctnico]

Regarding reviewing more expensive equipment: I think this is useful. Ofcourse you get a unit on loan but nevertheless it is good to know whether it is worth considering before spending time on getting a piece of equipment in and testing it thouroughly.

Finally, I think we can clearly see from the video that the R&S is slightly lower noise despite the higher bandwidth at 1mV/div – but it just so happened to have a far worse DC offset than the Siglent, which didn’t show any noticeable offset at all in this test.

DC offset is just a matter of running the self-calibration (and/or let the instrument warm up before use).

[BillB]

Good review.  I imagine it must be difficult to cover so much functionality while trying to prioritize what you cover, all while keeping it from being a four hour video. 

Regarding reviewing more expensive equipment: I think this is useful. Ofcourse you get a unit on loan but nevertheless it is good to know whether it is worth considering before spending time on getting a piece of equipment in and testing it thouroughly.

Definitely agree.  Though, I wouldn't want to watch exclusively high-end reviews either.  It would probably bankrupt me. 

[TK]

Siglent has a lot of features: histogram charts and zones is unique, 50 ohm, active probe, better responsiveness, front end. And the updates/feature requests via forum is huge too.

Rigol MSO5000 has histogram charts and zone trigger.  It lacks 50 ohm input option.

[JPortici]

(Glad you were using 8.2R1 which is indeed the latest firmware available)

Version 8.4R4 has been out with the beta testers for a few days already.

Glad he didn't use that one

[David Hess]

So it has a dedicated roll mode button but not dedicated trigger slope button.  Which one is more likely to be used?

I am not sold on drop down menus combined with a touch screen.  What advantage does that have over using physical menu buttons other than cost when the touch screen is going to be included anyway?

The index graded display looks terrible.

At least they used a non-glare display.  That puts them ahead of many of their competitors and Apple.

Update:

The loss of signal integrity of channel one at about 29:20 looks to me like aliasing of the intermodulation products between the frequency content of the fast edge and sampling clock when the sample rate is reduced to 2.5 Gsamples/second.  Try taking single shot captures in that configuration to see what is going on.  The sample points are correct but the sin(x)/x reconstruction fails when aliasing is present.

The 500 MHz response of the Keysight should not display it as much simply because of lower ratio of bandwidth to sample rate and the same goes for the Rigol.

The shown transient response is not very good.


I would not trust the RMS noise measurement until I confirmed that the DSO measures RMS noise correctly.  Rigol for instance does not.

Better results may be available by measuring the DC standard deviation of a single shot capture and this should agree with the RMS measurement.


I chuckled at the smudged screen of the Keysight toward the end.  Is that a feature?

[JPortici]

I am not sold on drop down menus combined with a touch screen.  What advantage does that have over using physical menu buttons other than cost when the touch screen is going to be included anyway?

There is a looooot of stuff that's not reachable with the buttons because it was added later and/or there wasn't a good place to put them in the buttons, and there's going to be more in the future.
This means that the scope can't effectively be used without the touch screen

I'm fine with it, i use the touch controls a lot more than the knobs these days

Intensity grading, i don't give a shit. I prefer "temperature" color grading a million times over DPO style intensity grading

Incidentally, none of the scope i have and use at work (picoscope, tek TPS, SDS-5035X, lecroy 7200A -the one from the 80s-) have intensity grading, at most they have color grading and in the rare times i need such a thing i prefer it over starndard intensity grading, but that's me

[genghisnico13]

Just for reference I removed the offset with Math. 1M memory.

[smarteebit]

Just for reference I removed the offset with Math. 1M memory.

Why not directly use the Stdev measurment?

[snoopy]

It seems to be a jack of all trades, but a master of none.

Plus lots of annoying bugs you don't usually find with the major brands. Looks like a rush job and another re-run of the Rigol MSO7000

[genghisnico13]

Just for reference I removed the offset with Math. 1M memory.

Why not directly use the Stdev measurment?

I wanted to replicate the measurements as close as possible, but you are right it would have probably been easier, either way the results are the same.

[2N3055]

It seems to be a jack of all trades, but a master of none.

Plus lots of annoying bugs you don't usually find with the major brands. Looks like a rush job and another re-run of the Rigol MSO7000

Keysight Infiniivision 3000T series release notes are 27 pages... R&S RTM3000 is 12 pages, in a year and half after release. Download them and read them.
All of them have bugs. It is about fixing them and how fast.

So it's not right to hold Rigol or Siglent to different standards. 
Siglent seems to be improving in fixing problems, Rigol could do better but you can't say they are not doing anything...
But SDS5000X is just recently released, and R&S RTM3000 was still fixing serious problems in first few releases.

[0xdeadbeef]

Yeah, e.g. there is this Youtube channel "Afug-Info" which seems to focus on finding even the smallest issue in the RTM 3004. While some problems might be fixed already, it's clear that also the RTM 3004 has/had lots of (minor) issues here and there. I fear that's how it is right now. Then again, even the Agilent 7000 series scopes had major (!) bugs in its initial firmware revisions >= ten years ago.
Still, honestly, some aspects of this review have been undermining my enthusiasm for the DS5000X quite a bit. It's also kinda shocking to hear in the 21st century that enabling sin(x)/x interpolation causes a major drop in acquisition (or waveform update) rate and that the specified values are only for dot mode. What about linear interpolation?
Anyway, in direct comparison to the RTM3K directly next to it, I like the (Lecroy-ish) control panel and matte screen of the SDS5000X much better but there's no way denying that everything on the RTM3K looks just sleeker and more polished. E.g. the FFT implementation on the Siglent just looks underwhelming compared to the one on the RTM3K (despite using much less points).

[Performa01]

Still, honestly, some aspects of this review have been undermining my enthusiasm for the DS5000X quite a bit. It's also kinda shocking to hear in the 21st century that enabling sin(x)/x interpolation causes a major drop in acquisition (or waveform update) rate and that the specified values are only for dot mode. What about linear interpolation?

There is no difference. it quite obviously is the additional display points that slow the acquisition rate down, no matter what's the math behind it.
Measurements with an old pre-release firmware (won't be much different today) show that at 20ns/div the trigger rate is 110k with single channel and 100k with both channels in dot mode. It is 23k4 for single channel and 2k8 for dual channel in vector mode.
That sounds bad, but then again we can use dots mode in most situations, because it looks nearly the same on the screen and avoids all the shortcomings of interpolation:
- no graphics processing overhead
- no unstable waveform display in the presence of strong aliasing products
- better intensity grading

Since the interpolation is just a post processing, we can switch it on anytime after the acquisition, e.g. in the history.

Anyway, in direct comparison to the RTM3K directly next to it, I like the (Lecroy-ish) control panel and matte screen of the SDS5000X much better but there's no way denying that everything on the RTM3K looks just sleeker and more polished. E.g. the FFT implementation on the Siglent just looks underwhelming compared to the one on the RTM3K (despite using much less points).

Yes, it is traditional Math FFT right now. Next FW release includes configurable peak and harmonic search with list view as well as configurable FFT length:


SDS5104X_FFT_Peaks_02_Avg10

A spectrum analyzer package is on the todo list, which should provide a simple and easy setup, similar to a real SA.