Siglent SDS800X HD Review & Demonstration Thread

[markone]

Here is my scope.
same settings, I even tuned the AWG for scope to show 50.00067 MHz so we can see if there are some specific beat frequencies..

Model of your scope ?

What do you mean? SDS824xHD of course. That is one we are discussing here...

Good, I know you have also something better ... anyway please align settings with ours, I mean :

[2N3055]

Here is my scope.
same settings, I even tuned the AWG for scope to show 50.00067 MHz so we can see if there are some specific beat frequencies..

Model of your scope ?

What do you mean? SDS824xHD of course. That is one we are discussing here...

Good, I know you have also something better ... anyway please align settings with ours, I mean :

FFT parameters are same:
Sa=125MSa/s
Curr=2MPts
delta f=59.60Hz
RBW=222.33Hz
Averages are 16 but makes no difference.

Is there anything else specific?

[markone]

Here is my scope.
same settings, I even tuned the AWG for scope to show 50.00067 MHz so we can see if there are some specific beat frequencies..

Model of your scope ?

What do you mean? SDS824xHD of course. That is one we are discussing here...

Good, I know you have also something better ... anyway please align settings with ours, I mean :

FFT parameters are same:
Sa=125MSa/s
Curr=2MPts
delta f=59.60Hz
RBW=222.33Hz
Averages are 16 but makes no difference.

Is there anything else specific?

Set SPAN to 1MHz, your spectrum image looks like SPAN = 500KHz.

[2N3055]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

[markone]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

Two quick questions :

1) Signal generator ?
2) Last but not the least  ... FW version IS  ... ? 

EDIT :

I would guess higher than mine :

[2N3055]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

Two quick questions :

1) Signal generator ?
2) Last but not the least  ... FW version IS  ... ? 

EDIT :

I would guess higher than mine :

AWG is SDG6052X.
FW version is latest one without public number.

[markone]

AWG is SDG6052X.

Ok, so for sure you signal is cleaner, do you know second and third hamonics level ?

Do you maybe have an SDG2000X ?

FW version is latest one without public number.

If the problem is not strictly harmonics related it could be that FW guys did some work on FFT math ... I want IT !

[gf]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

One difference I still see is 100mV/div versus 200mV/div.

[Martin72]

The RBW can only be achieved by manually setting the sample rate in the menu.
If you leave the settings on Auto, you get a different value - which RBW is the "right" one for a 50Mhz signal and 1Mhz span (50k, 10k)?
For example, if I leave the SDS3034X HD as it is, I have approx. 700Hz RBW, but also no "problems" with the signal display.

[ebastler]

the sad truth is still "selected RBW values not working" and "utterly wrong"

Is this one acceptable? SDS814X HD, upgraded to 824.

[markone]

the sad truth is still "selected RBW values not working" and "utterly wrong"

Is this one acceptable? SDS814X HD, upgraded to 824.

What have you changed from your previous spectrum image ?

[2N3055]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

One difference I still see is 100mV/div versus 200mV/div.

That is not problem. I use 200mV/div, because IMD for close in peaks is few dB better that way.
Even with SA you don't use full scale for distortion measurements.

With 100mv/div, slightly more distortion, nothing else.

[markone]

If you insist..
It makes no difference. I set it to 500KHz to better see details..

Attached 1, 2 and 10 MHZ span....

One difference I still see is 100mV/div versus 200mV/div.

I noted too, but in my case @ 200mV/div the spurious level up.

[markone]

-snip
With 100mv/div, slightly more distortion, nothing else.

I can confirm.

[ebastler]

What have you changed from your previous spectrum image ?

Adjusted the signal generator frequency such that the scope "sees" an exact 50 MHz. I wanted to get the point across that we are seeing harmonics beyond the Nyquist limit, folded back in the spectral range, if the input frequency is slightly off. Doesn't look like a bug to me.

My guess would be that 2N3055 sees less of them since he is using a better signal generator. 

[gf]

If the problem is not strictly harmonics related it could be that FW guys did some work on FFT math ... I want IT !

That's why I'd like to look at the raw samples, and why I asked for a .bin file with 100M points. And yes, I ask for 2 GSa/s first because I want to preserve as much information as possible from the original signal. I can downsample myself.

[markone]

What have you changed from your previous spectrum image ?

Adjusted the signal generator frequency such that the scope "sees" an exact 50 MHz. I wanted to get the point across that we are seeing harmonics beyond the Nyquist limit, folded back in the spectral range, if the input frequency is slightly off. Doesn't look like a bug to me.

Just replicated on my DSO, I had to shift down the generator frequency to 49.999350MHz to obtain a perfect 50MHz reading, normally I get 50.000650.

My guess would be that 2N3055 sees less of them since he is using a better signal generator.

My guess too, reason why I asked him levels of 2nd and 3rd harmonics, anyway I would be glad to try the new incoming FW 

In this case the folded back harmonic/s would Zero beat with central carrier producing a clean near spectrum image.

[markone]

If the problem is not strictly harmonics related it could be that FW guys did some work on FFT math ... I want IT !

That's why I'd like to look at the raw samples, and why I asked for a .bin file with 100M points. And yes, I ask for 2 GSa/s first because I want to preserve as much information as possible from the original signal. I can downsample myself.

I would be glad to provide it but I have no time now to write an application program, do you guys know if exist something usable for the purpose ?

EDIT : foldback should be feasable with decimated vector that supplies FFT Math, not with original data sampled @ 2GSa/s, right ?

[gf]

If the problem is not strictly harmonics related it could be that FW guys did some work on FFT math ... I want IT !

That's why I'd like to look at the raw samples, and why I asked for a .bin file with 100M points. And yes, I ask for 2 GSa/s first because I want to preserve as much information as possible from the original signal. I can downsample myself.

I would be glad to provide it but I have no time now to write an application program, do you guys know if exist something usable for the purpose ?

User manual:

29 Save/Recall

The SDS800X HD supports saving setups, reference waveforms, screenshots, and waveform data
files to internal storage, external USB storage devices (e.g. U disk), or SMB File Share. Saved setups
and reference waveforms can be recalled as needed.
For details of SMB File Share, refer to the section “SMB File Share”
...

Save the trace as Binary Data (.bin file).

EDIT:  foldback should be feasable for decimated data vector, not for original data sampled @ 2GSa/s, right ?

Already sampling the analog signal folds the spectrum.
And down-sampling folds it again to an even smaller frequency range.

Folding back is not possible in general, since the folding is a N:1 mapping in the frequency domain.

[markone]

-snip
Already sampling the analog signal folds the spectrum.
And down-sampling folds it again to an even smaller frequency range.

Folding back is not possible in general, since the folding is a N:1 mapping in the frequency domain.

I mean in respect to "real" sampleable harmonics that should be 2nd, 3rd and 4th, all below Nyquest @ 2GSa/s.

Anyway :

1) link to compressed bin file 100MSa @ 2GSa/s (size is about 95MB) : https://file.io/su83ZcbhPr4d
2) link to compressed bin file 40MSa @ 2GSa/s (size is about 40MB) : https://file.io/hxDe9PqKBLGY

[Performa01]

I don't expect the original sample data to reveal the true cause, which is already registered in Siglents internal bug list.

In short: a problem can arise whenever the FFT sample rate is not equal to the acquisition sample rate. There seems to be an exception for the 1:2 ratio.

In this particular test case, you can try using constant sample rate of 250 MSa/s and then, with 5 Mpts record length, the FFT will use 125 MSa/s, resulting in the smallest possible RBW at 50 MHz, without excessive spurious signals:


SDS824X HD_FFT_C50M_S1M_Sa250_125

[gf]

I don't expect the original sample data to reveal the true cause, which is already registered in Siglents internal bug list.
In short: a problem can arise whenever the FFT sample rate is not equal to the acquisition sample rate. There seems to be an exception for the 1:2 ratio.

If it is already a well-know and accepted bug, then there is not need to investigate further.

[markone]

I don't expect the original sample data to reveal the true cause, which is already registered in Siglents internal bug list.

Good to know, this is exactly what I was hoping to with my lame post activity 

So now we can officially say that there is some margin of improvement in FFT performance, could you "disclose" if we will see something in regard with the incoming FW release?

[Martin72]

The RBW can only be achieved by manually setting the sample rate in the menu.
If you leave the settings on Auto, you get a different value - which RBW is the "right" one for a 50Mhz signal and 1Mhz span (50k, 10k)?
For example, if I leave the SDS3034X HD as it is, I have approx. 700Hz RBW, but also no "problems" with the signal display.

Was this question too stupid ? 
Edit:Pics(1M,50k,10K)..

[joeqsmith]

...
The commonly agreed industry standard for the maximum acceptable input VSWR is 1.5:1, which is equivalent to 14 dB return loss.
...

I've never heard of such a standard.  Could you provide a reference?  If you are making measurements, 1.5 seems really high. 

Looking at the manual for my LeCroy 7200, they spec the VSWR at 1.2, BW is 4GHz with a 125pSec transition for 50mV and up.  Isolation is 60dB min at 1GHz.

I recently looked at SJL's sampling scope and verified the port match.

Wasn't sure how good your port match is. 

Not the best, to be honest: -17 dB @ 3 GHz (although we don't characterize this on a per-unit basis).
We have a design with better matching (-30 dB) at the cost of 2x noise floor. We opted for the lower-noise design, since you can get essentially the same results with a 6 dB attenuator (but not the other way around).

Once you get up to 5GHz, the return loss is around -7 and you really need to add an attenuator to improve it.