Siglent SDS800X HD 12 bit DSO's

[Martin72]

What do we see ?
Edit now we see

[Mechatrommer]

Fact check from rigolna.com: The entire DHO900 series comes equipped with digital channels as standard, allowing digital signal analysis when paired with the PLA2216 logic analysis probe.

Which is almost as expensive as the SPL-1016 + the software license.    It seems that for certain customers the ability to separate them would be advantageous.  For us we just get the license for a few minutes work saving quite a bit.

please note the increased price in dho900 series is not just for LA, its also for higher BW (100-250MHz). if comparing below pictures... granted SDS800 is $100 cheaper at first, but when including LA license, the SDS price become same or more expensive... same with DHO900 with S series, price is up more $100 but already with AFG HW inside, for siglent, we need buy more license and HW separately, more expensive hidden trap. btw the similarity is i suspect siglent wants to beat rigol in price (we know and remember price history), so the coincidence in price similarity is like them try to copycat rigol's price. to be fair, siglent is 2GSps, thats imho is the selling point, and more stable bode plot and LA SW... ymmv.. btw i'm not the one starting “xxxx is better than yyyyy” discussion and some people may not like facts. though i'm trying to be as factual as i am cheers.

(ps: we (or i) are waiting someone accept hack challenge call to reverse eng, or post tear down photos of siglent AWG and LA HW for even more cheaper 4th party alternative because we like cheap so kudos to siglent for taking some steps)

Do you understand that  RT-ZL04, RTA4004, MXO4 and MXO5 are not Siglent SDS800X HD 12 bit DSO's or accessories? Why is it so impossible to stay even remotely on topic? Maybe I should go over to the R&S threads and post about my Calphalon cookware.
Yah, I know this is also off topic. But for once I just had to rant. Sorry.

go to rigol threads and count how many siglenism chime in for comparison. if you cant accept comparing, you can live in your own world, shut off the internet. the current price proved siglent also comparing with competitor

[tszaboo]

And now back to the topic. This morning my new black toy arrived. Yes it is small related to my SDS2000X or SVA1032, but not too small.
It found its place on my bench. Network connection and time server were easy to manage as all Siglent gear.

Checked it out with my Leo Bodnar Pulser (I searched the whole weekend for it...).
Connected with a 50 ohms feedthrough adaptor.

You see the progress from a 70 MHz -> 100 MHz -> 200 MHz scope   

PS: Sorry I run into problems with the images, try to fix it

Does your pulser have those overshoots or is that some sort of interpolation error? Or reflections? Why does it appear with the increased bandwidth?

[Bad_Driver]

Does your pulser have those overshoots or is that some sort of interpolation error? Or reflections? Why does it appear with the increased bandwidth?

Seems to be the pulser, I got the same overshoot on my SD2000X+ (500 MHz; 50 ohms input) but with around 700ps rise time.

[Mechatrommer]

we've been discussing to death that the overshoot is the combined effect of the pulser and dso front end, not the pulser alone. ymmv.

[pdenisowski]

Perhaps we could outsource the topic of options, as there seems to be increased interest in it across all brands.

Agree - this thread should really be for the Siglent scope.  I've started a thread on oscilloscope options (all manufacturers)

https://www.eevblog.com/forum/testgear/oscilloscope-option-discussion

[Bad_Driver]

First impressions:
User Interface as the SDS2000X+, no problems at all, but USB mouse support helps.
The fan is ....not silent, needs improvement.

The new FFT is impressive, I like the logarithmic freq-scale, I wish to have this for my SDS2000!

I tried the new filter function but it needs some reading and training to understand the limits (>0.01x fs).
The new (for me) sample rate setting is helpful. Maybe there is an easier way...
 
Has someone tried to use the screen zoom function remote via web?
 

[2N3055]

Has someone tried to use the screen zoom function remote via web?

Zoom is in Acquire menu on top...

Or down, add Channel, type Zoom.

[Bad_Driver]

Has someone tried to use the screen zoom function remote via web?

Zoom is in Acquire menu on top...

Or down, add Channel, type Zoom.

Thanks, I got it!

[Bad_Driver]

From my playground:

The noise on my bench, probe without signal on channel 1 but ground connected.
Low pass filter 20 kHz -> FFT

[lofe]

From my playground:

The noise on my bench, probe without signal on channel 1 but ground connected.
Low pass filter 20 kHz -> FFT
(Attachment Link)

Is this good or bad? What do you think?

[2N3055]

From my playground:

The noise on my bench, probe without signal on channel 1 but ground connected.
Low pass filter 20 kHz -> FFT
(Attachment Link)

Is this good or bad? What do you think?

What do you mean good or bad?

If you connect a probe and make a loop to tip with ground wire it is same as you connected antenna to scope..
It shows EMC environment at his desk...

Interesting thing is that you can get same analysis in single Math channel.  In Math menu you go to Formula Editor and type FFT(Filter(C1))..
You can nest expressions... In single channel, no need to pipe Math CH to MathCH.

[Bad_Driver]

From my playground:

The noise on my bench, probe without signal on channel 1 but ground connected.
Low pass filter 20 kHz -> FFT
(Attachment Link)

Is this good or bad? What do you think?

The noise on my bench is bad but the scope is impressive!
With this device you get an incredible amount of performance for very little money.
I think it will take me months to even try out all the features once.

PS: I know that this setting with open tip was an EMC antenna, but I want to catch some noise. For serious measurements I have some famous Jim Williams-like tin cookie cans 😉

[pdenisowski]

d) Mask test doesn’t even seem to be available at all – and if it was, it most likely would be another expensive option.

Mask test for the MXO is coming out later this month.  It's taken a while because it's being done in hardware (extremely fast)

And it's free

[gf]

Does your pulser have those overshoots or is that some sort of interpolation error? Or reflections? Why does it appear with the increased bandwidth?

There is no such thing as a lowpass filter (or frontend) whose frequency response is "maximally flat" and whose step response shows no overshoot. If the frequency response has a pronounced flat top before it rolls-off (steeply), then overshoot in the step response is inevitable. The 250 MHz frequency response obviously falls into this category.

OTOH, the 70MHz response seems to roll-off much smoother (more like Gaussian or similar), thus almost avoiding overshoot.

And the 100 MHz response seems to be somewhere in between.

[Performa01]

d) Mask test doesn’t even seem to be available at all – and if it was, it most likely would be another expensive option.

Mask test for the MXO is coming out later this month.  It's taken a while because it's being done in hardware (extremely fast)

And it's free

Early adopters of the MXO will be happy about that 

Just to make sure Siglent customers don't get a sense of inferiority, I need to stress the fact that the mask test in the SDS800X HD is free, HW-accelerated (it works at the same rate as the acquisition engine) and was already available right from the start:

https://www.eevblog.com/forum/testgear/sds800x-hd-review-demonstration-thread/msg5338655/#msg5338655

[tszaboo]

Does your pulser have those overshoots or is that some sort of interpolation error? Or reflections? Why does it appear with the increased bandwidth?

There is no such thing as a lowpass filter (or frontend) whose frequency response is "maximally flat" and whose step response shows no overshoot. If the frequency response has a pronounced flat top before it rolls-off (steeply), then overshoot in the step response is inevitable. The 250 MHz frequency response obviously falls into this category.

OTOH, the 70MHz response seems to roll-off much smoother (more like Gaussian or similar), thus almost avoiding overshoot.

And the 100 MHz response seems to be somewhere in between.

Thanks I'm familiar with filter theory. I was just expecting the scope to be (close to) critically dampened for the max frequency it can do, and overdampened at the lower frequencvy options. But seems like Siglent had other design goals in mind. It's a bit strange, that they decided that way, for me having ~280MHz bandwidth in single channel mode, for a 200 MHz scope is not necessary. I would rather have the bandwidth of the nameplate, and less overshoots. Or maybe it's just more visible because of the 12 bit.   
But then, this is a direct connection to into an unterminated scope with a sharp edge pulse generator. With a scope probe, I don't think we would ever see such pronounced overshoots. Not an usual use case.

[BillyO]

But seems like Siglent had other design goals in mind. It's a bit strange, that they decided that way, for me having ~280MHz bandwidth in single channel mode, for a 200 MHz scope is not necessary. I would rather have the bandwidth of the nameplate, and less overshoots. Or maybe it's just more visible because of the 12 bit.   

Pretty well everybody does this and I believe the practice may have been started by Tek.  I know with my "improved" SDS2104 you get gaussian response at 350MHz and at 200MHz (of course at 20MHz too).  Your probes will act a lot like a gaussian filter too. 

[BillyO]

The pulse generator is a well known and well characterized device that can produce sub 40ps rise times.  It's not just some random device.

The step response seen is the step response of the system.  Pulser + load + scope + connectors, etc..

This has been analyzed and discussed for some time.  Nothing new or not understood here.

Below is the same pulser (Bodnar) directly into the input of a scope with more BW but with an internal 50 ohm load, so a little less hardware.

[BillyO]

The pulse generator is a well known and well characterized device that can produce sub 40ps rise times.  It's not just some random device.

The step response seen is the step response of the system.  Pulser + load + scope + connectors, etc..

This has been analyzed and discussed for some time.  Nothing new or not understood here.

I'm not saying that new research and discovery is needed. I was questioning the conclusion in the prior posts -- I understood them right -- that the overshoot is only or mainly due to Siglent's choice of lowpass filter characteristic.

Well no as I clearly said: "The step response seen is the step response of the system.  Pulser + load + scope + connectors, etc.."

You can clearly see in the image I posted similar behavior but because the scope is faster, the frequency of the overshoot oscillation is faster.  The design of the scope front end has a LOT to do with it.

[roads]

Was super excited about this new release and decided to hop on the train. All these posts have been super helpful to make my decision easy between a DHO800, SDS800XHD, and stretching to a SDS2000XHD. The 800XHD fits the bill perfect for my first scope (and budget)!

Unfortunately there seems to be a long lead time with US distributors, I placed my order on TEquipment and the estimated ship date is 5/10/24 . It's too bad they didn't list that during ordering... I'd rather have my money in the bank and pay once its available to ship

[2N3055]

To make sure full data is available, my Bodnar pulser in spec sheet specifies 4.7% overshoot and 8.2% undershoot.
And that in almost perfectly terminated in SCA803 with SD-30 sampling head.

1MΩ/17pf + 50Ω passthrough  terminator and signal path not optimized for SWR1.0 at 50Ω will make all kinds of ringing on top of rising/falling edge...

But..

If I connect Bodnar->50Ω passthrough -> 10 dB terminator I get this:



EDIT:  ROV is Rising edge OVershoot and FOV is Falling edge OVershoot measurement. So no need to eyeball it...

[tautech]

........
Just to make sure Siglent customers don't get a sense of inferiority, I need to stress the fact that the mask test in the SDS800X HD is free, HW-accelerated (it works at the same rate as the acquisition engine) and was already available right from the start.....

Just as the Mask test feature has been in every Siglent DSO for at least a decade.......free !

[gf]

To make sure full data is available, my Bodnar pulser in spec sheet specifies 4.7% overshoot and 8.2% undershoot.
And that in almost perfectly terminated in SCA803 with SD-30 sampling head.

But the period of the ringing is in the order of 50ps, and the overshoot has well settled within the risetime of the scope. For a frontend with only 200 MHz bandwidth, the inherent Bodnar risetime and overshoot are IMO almost negligible (i.e. it can be considered an almost ideal square wave). For a 1GHz scope, it likely starts to matter, though.

If I connect Bodnar->50Ω passthrough -> 10 dB terminator I get this:

Shouldn't the order be be rather Bodnar -> attenuator -> 50R passthru -> scope, in order that the Bodnar is 50R terminated, the attenuator is 50R terminated as well at each side, and we get a 25R|15pF 1st order lowpass (~400MHz) at the scope input?

From which model and bandwidth is your screenshot? Is it a SDS824X? If not, is any artificial (digital?) BW limit activated?

[tszaboo]

You guys can read a lot from (or into?) that step response on the screen. How do we know whether the overshoots are originating in the scope's input lowpass vs. a small impedance mismatch vs. the pulse generator itself?

That's why I was asking the questions.

The pulse generator is a well known and well characterized device that can produce sub 40ps rise times.  It's not just some random device.

The step response seen is the step response of the system.  Pulser + load + scope + connectors, etc..

This has been analyzed and discussed for some time.  Nothing new or not understood here.

I'm not saying that new research and discovery is needed. I was questioning the conclusion in the prior posts -- I understood them right -- that the overshoot is only or mainly due to Siglent's choice of lowpass filter characteristic.

Well no as I clearly said: "The step response seen is the step response of the system.  Pulser + load + scope + connectors, etc.."

You can clearly see in the image I posted similar behavior but because the scope is faster, the frequency of the overshoot oscillation is faster.  The design of the scope front end has a LOT to do with it.

Well technically the scope will do the same sampling rate. I don't think anything will change in the analog path as well. So we kind of can conclude that it's the digital path that brings it up. Or it gets rid of it when the bandwidth is lower.

But as I said, I don't think it matters if you use it with an oscilloscope probe, it will not do this.

Shouldn't the order be be rather Bodnar -> attenuator -> 50R passthru -> scope, in order that the Bodnar is 50R terminated, the attenuator is 50R terminated as well at each side, and we get a 25R|15pF 1st order lowpass (~400MHz) at the scope input?

From which model and bandwidth is your screenshot? Is it a SDS824X? If not, is any artificial (digital?) BW limit activated?

Yes it should be that way.