Siglent SDS800X HD 12 bit DSO's

[Performa01]

Any suggestions what to test thats missing from a MSOX3104, what I used at my different jobs for the past 14 years or so?

I have great sympathy for people who are just content with what they have – and only having what they really need. Requirements for T&M gear in general (and DSOs in particular) can vary widely, depending on the tasks on hand. And of course, why should an instrument that met your requirements 14 years ago not still be the tool of choice today, as long as your requirements and expectations haven’t changed during all these years?

Yet it sometimes cannot hurt to reconsider the requirements, otherwise one might miss some opportunities to make life easier. My Review & Demo thread contains lots of examples of what even a lowest end analytical scope can do nowadays:

https://www.eevblog.com/forum/testgear/sds800x-hd-review-demonstration-thread/

What about 12-bit resolution and a nice vertical zoom implementation as demonstrated in the chapters “Pulse Response”, “Vertical Zoom Demo”, “Probe Bandwidth” and “Zoom Expectations”?

And wouldn’t it be nice sometimes to have true 500 µV/div full resolution sensitivity, instead of just 4 mV/div and everything below that just fake, as demonstrated in “True Vertical Sensitivity”?

Could it help sometimes to have a low noise instrument with <2.4 nV/√Hz voltage noise when 50 ohm terminated as demonstrated in chapters “Noise & Spurs”, “Noise Density” and “Noise Density 2” as well as “ Granular Noise”?

Is History something useful, where you can find up to 80000 past trigger events (records), play them back and analyze them to your heart’s content using all the tools the scope has to offer? Look at “History & Sequence Mode”.

What about deep measurements, which have deep memory as a prerequisite? Sections “Counting Pulses” and “Deep Measurements” give some examples of tasks that cannot be completed with short memory and measurements on heavily decimated data.

Aren’t the small Histicons in the measurement statistics convenient, as demonstrated in chapter “Measurement Histograms” and many other sections where automatic measurements are used?

You might value measurement Trend plots at times – wouldn’t Track plots be very useful as well, as they allow you to clearly visualize frequency and phase modulated signals as demonstrated in “Measurements 2 (Trend/Track)”?

Wouldn’t it sometimes be nice to be able to make accurate measurements? For very low frequencies and DC we can pull out a (good) DMM, but is it really necessary if you can have a DSO with 0.5% DC-accuracy, as confirmed in “DC Check”?

When I still worked with analog CROs, I made sure to have one with trigger signal output, where I could connect an 8-digit frequency counter, so that I’ve always been able to know the exact signal frequency, as it was important for many of my tasks. That’s become a habit and since trigger signal outputs were never very common and completely went out of fashion at one point, I wouldn’t accept any scope without a permanently visible precision frequency counter, like the 7 digits in any contemporary Siglent. Also mentioned in the “Counter”-section and numerous examples can be found throughout the Review thread.

4 highly sophisticated math channels with formula editor allow various experiments as well as extensive data conditioning (not only) for the FFT.

A deep (2 Mpts) FFT in turn allows good frequency resolution, and it’s all the more useful if we can have four FFTs at the same time as demonstrated in some of the noise measurements. A high dynamic range of up to 100 dB as demonstrated in “FFT Dynamic range” should also be welcome.

What about a powerful Frequency Response Analysis (Bode Plot) with up to 3 channels up to 120 MHz and >100 dB dynamic range, as demonstrated in “Bode Plot at a glance” and “Bode Plot Example”? That’s for free, by the way, not part of an expensive Power Analysis package.

Maybe the two custom probe definitions would be more convenient than a bunch or predefined probe factors, see “Custom Probe Factors”?

Of course it would not be fair to compare a $700,- instrument like the SDS824X HD to any of the Keysight MSOX3000 series, because it naturally lacks sample rate, bandwidth, 50 ohm inputs and active probe interfaces. But other than that...

[EvgenyG]

Thanks for the photos!
Looks like there's a bit of overshoot when in 824X mode compared to 804X. I wonder if the real 824X would show same behaviour.
I'd rather have correct signal display at 70Mhz bandwidth, than observing overshooting that is not there.

[Martin72]

I wonder if the real 824X would show same behaviour.

Why should it look any different with a "real" 824...
The hardware is the same, at least in that area, and the connections from the pulser to the input would also be the same.

I'd rather have correct signal display at 70Mhz bandwidth, than observing overshooting that is not there.

What is true and correct in this context....

[EvgenyG]

Why should it look any different with a "real" 824...

I am pretty sure hardware is same, but factory calibration of a 824X could be different to 804X, no?

[Martin72]

I find that hard to imagine, because then people buying a bandwidth upgrade for their 2000X+ and above model would have a problem.
But that reminds me to do a self-calibration(after the "upgrade") and then run the last sweep again.

[Martin72]

That's a good idea, the feature has never been shown before, as far as I can remember.
I'll think of something.

I already have a certain idea.
We (at work) once designed a switching power supply, which was ages ago, and it has a two-part design.
One part is the controller, the other plate contains the switching FETs and rectifier diodes, both plates are connected with cables - it couldn't be more ideal for testing the PA option.
That was one of the reasons why I hadn't paid much attention to the PA before, because many of the scenarios described in the manual are not possible with conventionally constructed/layout-designed, ready-made power supplies.
Now the memory of this antique power supply...
But it may take some time to have the opportunity to do so, so please be patient.

[Performa01]

Looks like there's a bit of overshoot when in 824X mode compared to 804X. I wonder if the real 824X would show same behaviour.
I'd rather have correct signal display at 70Mhz bandwidth, than observing overshooting that is not there.

It is just not possible to meet everyone’s taste. There is one fraction of people, who thinks Siglent “does it right this time” by fitting a “proper” anti-aliasing filter in the frontend, where “proper” usually means “effective”, like in reply #417 here:

https://www.eevblog.com/forum/testgear/siglent-sds3000x-hd-and-upgraded-sds1000x-hd/msg5376662/#msg5376662

Then there is another fraction, complaining whenever the slightest pulse distortions (in form of overshoot) become visible.

Anyone who has ever tried to design a “proper” AA-Filter (and thoroughly analyzed the outcome) will have learned that this is just not possible, without severely compromising the pulse fidelity, for a number of reasons.

The only near perfect solution would be to use massive oversampling, like in the ancient boat anchors, that didn’t have sufficient processing power to apply a true sin(x)/x reconstruction, hence had to make do with simple linear interpolation, which in turn only works as long as the sample rate is at least ten times the signal bandwidth.

In the latter scenario, we could use a 7th order Bessel filter to get 60 dB attenuation at the Nyquist frequency – or a 3rd order Bessel to get about -34 dB together with a digital FIR filter to provide additional attenuation in the 1st Nyquist zone, i.e. from the rated bandwidth up to Nyquist, which would then be about five times that bandwidth.

But with that “ideal” approach, another (huge) fraction of people would pop up and loudly complain about a scope with only max. 50 MHz bandwidth (and 4 x 500 MSa/s), or alternatively, a 200 MHz scope that’s sooo expensive (because it has to have 4 x 2 GSa/s).

That’s not a Siglent-specific issue; all the top models of a product line of any serious manufacturer have to make the same tradeoff, either strong aliasing or bad pulse distortions. Yet this problem seems to hit Siglent in particular, because their products are still affordable and, even more importantly, easily hackable (at least the low-end devices) and everyone who takes pride in themselves feels inclined to hack it for the highest possible bandwidth.


The SDS 800X HD has a moderate(?) AA-filter in the frontend, presumably it’s a 3rd order Butterworth, which of course causes significant overshoot. It’s even specified in the data sheet: 10% with a 150 ps rise time pulse from a 50 ohm source. Not many manufacturers are brave enough to even specify that in their data sheets!

When more than 2 channels are activated and the sample rate drops to just 500 MSa/s, an additional digital filter is activated to limit the bandwidth to true 200 MHz (~245 MHz otherwise) and aid aliasing suppression in the first Nyquist zone (200-250 MHz).

The lower bandwidth models just activate the most appropriate bandwidth limit in the integrated PGA (Programmable Gain Amplifier). The bandwidth limit has to be much higher than the specified bandwidth, because there is a limited number of choices and the corner frequency is subject to high tolerances of ±20% or even more (because we cannot have accurate resistors on a silicon chip) and we don’t want to risk any number of individual scope samples to not meet their specifications.

This means, we will usually measure a bandwidth significantly higher than specified, but it will vary from sample to sample and we cannot rely on just a few individual reports.

In any case, this integrated bandwidth limiter is just first order, hence will have Gaussian response and not cause any pulse distortion.

What has been stated above is also true for the switchable 20 MHz bandwidth limiter.

[2N3055]

Why should it look any different with a "real" 824...

I am pretty sure hardware is same, but factory calibration of a 824X could be different to 804X, no?

No.

[tautech]

[...] initial tests with TL-WR802N mini router in bridge mode proves NTP automatically sync'ing at boot is more reliable.
Tested with TL-WR802N powered before boot and powered from the scope at boot.
When powered by the scope NTP sync is not immediate but syncs shortly after.

For those without LAN connection at the bench and with multiple units needing LAN, the TL-WR802N mini router in bridge mode and a cheap switch is a better connectivity solution than just the USB WiFi dongle.

I am trying the same WiFi/Ethernet bridge, TP-Link TL-WR802N, with my SDS814X HD. It works fine when I power the TL-WR802N separately, and give it a headstart of a minute or so before I power on the scope.

But when I power the adapter from the scope's USB port -- which I would prefer for a tidier wiring and not using an extra power outlet --  the scope does not establish a network connection at all. Neither NTP sync nor accessing the scope from another computer on the network will work. I have tried both DHCP and static IP address entry; it doe not seem to make a difference.

Any ideas what I can try? Can the scope be configured to delay its attempt at the initial network connection? Can the TL-WR802N be configured to only activate its Ethernet port after it has established the WiFi connection maybe? (I can't see anything promising in its configuration interface.)

Thank you for any hints!

TL-WR802N must be set to Bridge mode and becomes invisible to your LAN while providing a WiFi connection to your access point.

Mine connects fine from boot with any Siglent device however you need correctly set whatever device you're using IP correctly. Double check the scope is retaining the correct IP for your LAN.

I normally just leave it to auto negotiate an IP and enter that IP into any browser for webrowser access.
Once you have WiFi access and a working scope IP your router should retain the IP for next time connection.

NTP takes a minute or so to Sync each boot.

If you like capture a screenshot and let's see if there's anything that looks outta place.

[tautech]

Have you tried a wired LAN connection ?
The scope should retain its IP address for instant connection, is yours not ?

Pop up a screenshot of your I/O LAN settings.

[tautech]

TBH you could try a different approach, that is to have TL-WR802N permanently ON and have it feed the WiFi connection to a cheap local switch at the bench to supply all your devices with a wired LAN connection.

[Martin72]

But that reminds me to do a self-calibration(after the "upgrade") and then run the last sweep again.

The changes after that are only marginal, i.e. to be regarded as null and void.
Operation via web server:
Slightly delayed, depending on whether the device is operated via WLAN or LAN.
You can forget about fine-tuning cursors/markers with this, only going to the scope itself will help.
Overall, I don't think the feature is very important as far as the instrument control function is concerned.

[tautech]

TBH you could try a different approach, that is to have TL-WR802N permanently ON and have it feed the WiFi connection to a cheap local switch at the bench to supply all your devices with a wired LAN connection.

Yes, that's the fallback. It does not even to be on all the time, as long as I turn on the master power switch for the bench a minute before I power up the scope.

But I should track down the WiFi-to-WiFi connection issue first. If that's the root cause, it could still occur sporadically even when I power the TP-Link separately: Move something around on the bench, and suddenly the connection to the WiFi repeater becomes the favored one, gets established but does not actually work...

Anyway, it seems most likely that the connection issues have nothing to do with the scope at all, but are compatibility problems between different brands of WiFi equipment. I will check tomorrow and post an update!

Yep, I suspect that is the issue....only 1 access point here so don't see these issues.....200m² timber framed residence so 1 WiFi AP penetrates it well.

[tautech]

Any ideas what I can try? Can the scope be configured to delay its attempt at the initial network connection? Can the TL-WR802N be configured to only activate its Ethernet port after it has established the WiFi connection maybe? (I can't see anything promising in its configuration interface.)

TBH I had issues getting a connection today.
It turned out Auto DHCP was giving an incorrect IP so it got turned OFF and IP configured manually. 

After getting to the workstation I opened a command box and double checked IP with 'ipconfig'.
The subnet 255.255.255.0 was correct but didn't provide a reliable reconnection so I dropped a 255 to have it 255.255.0.0 and problem was gone.

At one stage I had the scope IP set to 192.168.0.100 and it popped up a message of a possible IP conflict    so I changed it to .101
.100 turned out to be my workstation so thank heavens the scope identified that. 

Boot of TL-WR802N powered by the scopes USB results in them both finishing boot quite close together however if everything is set correctly NTP time will sync a minute or so later.

[tszaboo]

Any suggestions what to test thats missing from a MSOX3104, what I used at my different jobs for the past 14 years or so?

I have great sympathy for people who are just content with what they have – and only having what they really need. Requirements for T&M gear in general (and DSOs in particular) can vary widely, depending on the tasks on hand. And of course, why should an instrument that met your requirements 14 years ago not still be the tool of choice today, as long as your requirements and expectations haven’t changed during all these years?

The MSOX3000 series has definitely served me in my job.
Let's see if the SDS800HD will do the same at home. It should arrive on Wednesday.
Thanks for the lengthy writeup. I usually treat scopes as tools, it just have to do the job. No time to investigate new technologies. I'm not even sure what people talk about when they are talking about analytical scopes. Or how that translates into time savings in the design or verification of a part, or more reliability of the verification.

[RAPo]

They've just mailed: a Siglent SDS814X HD oscilloscope and Siglent SLA1016 logic analyser, which will be soon on my doorstep.

This indeed is no fun. I've emailed Eleshop: deliver scope+probe now

[2N3055]

Any suggestions what to test thats missing from a MSOX3104, what I used at my different jobs for the past 14 years or so?

I have great sympathy for people who are just content with what they have – and only having what they really need. Requirements for T&M gear in general (and DSOs in particular) can vary widely, depending on the tasks on hand. And of course, why should an instrument that met your requirements 14 years ago not still be the tool of choice today, as long as your requirements and expectations haven’t changed during all these years?

The MSOX3000 series has definitely served me in my job.
Let's see if the SDS800HD will do the same at home. It should arrive on Wednesday.
Thanks for the lengthy writeup. I usually treat scopes as tools, it just have to do the job. No time to investigate new technologies. I'm not even sure what people talk about when they are talking about analytical scopes. Or how that translates into time savings in the design or verification of a part, or more reliability of the verification.

For instance I really like histicons. They are saving time because not only you measure and have statistics, but also distribution of measurements.

[Swift Tester]

Perhaps a little strange question, but nevertheless: can anybody provide the dimensions of the packaging box?

[tszaboo]

Please ignore the bottom box, nothing to see there.
First impression: That is a very-very sharp sine wave.
Playing a little bit with it. It can do FFT on ERES math channel?

[Martin72]

It can do FFT on ERES math channel?

Yepp....

[2N3055]

Please ignore the bottom box, nothing to see there.
First impression: That is a very-very sharp sine wave.
Playing a little bit with it. It can do FFT on ERES math channel?

You can even combine them on single math channel:
FFT(ERES(C1))

[tautech]

From another thread, questions of the Power Analysis feature.

PA is an optional feature also found in the more expensive Siglent models and as in those you have 30 trial uses of PA without the need for licensing.
PA is a dedicated special mode of operation, much like Bode plot and takes a second or 2 to launch and return to normal scope use.

Probes needed to make full use of the PA feature are:
Std voltage probes
Differential probes
Current probes.

In each of the 12 PA measurement types there is a Connection guide to guide correct usage.
Screenshots dumped via webserver below display an indication of menus and how to use PA.

[Martin72]

@ebastler:
I was able to recreate this, but I don't know why this is the case.

[tautech]

I was able to recreate this, but I don't know why this is the case.

NTP server not yet set ? 

Looked one up for you:
TU Berlin   times.tubit.tu-berlin.de   130.149.7.7

[egonotto]

Hello,

Martin lives in a country with a mains frequency of 50 Hz and ebastler lives in a country with 60 Hz.

Best regards
egonotto

PS: My assumption proved to be wrong