The Siglent SDS1102X oscilloscope noise issue

[radiolistener]

I tried to update my SDS1102X with the latest firmware 1.1.1.2.15R10 and found that it has much worse noise floor.
Previously I was used firmware 1.1.1.2.13R5 and it was ok.

The good thing is that downgrade is possible, so I downgraded firmware to 1.1.1.2.13R5 and noise issue disappears.
I reported it to Siglent support, and hope it will be fixed in the next firmware release.

To test the noise floor I was used open inputs and the following settings: 1GS/s, probe 10x, 20V/div, 1 ms/div.
I record 14 Mpoint binary file and then analyze it with 4 Mpoint FFT (for example Sigview can do it).

Usually I'm using such sequence as a spectrum analyzer. And this is the reason why I catch it after upgrade.

Note: SDS1102X uses headers for the binary format. It consists of some data include oscilloscope settings.
I researched it and found how to extract key information from this header.
I used it to write a small tool which extracts a raw data and shows oscilloscope settings used to take BIN file.
But the header format was changed with a new firmware, so there is need different tool for each firmware.

See screenshots in attachment.

[tautech]

Did you run the Self Cal after the update ?
30 mins warm up time before you run it. 

[radiolistener]

Did you run the Self Cal after the update ?
30 mins warm up time before you run it. 

Yes, I performed Self Cal.
I also noticed that the latest firmware take much more time for Self Cal (for about 2-3 times longer or something like that).
But I didn't find any pros for such long Self Cal. On the contrary, it has a higher noise.

When I flashed the old firmware 1.1.1.2.13R5, the noise come back to the normal level.

[tautech]

At what V/div do you see the noise as an issue ?
Can the BW limit manage it ?

Just sold my last 1102X  with the latest FW installed so bugger I can't play along and check it for you.

[radiolistener]

I tried different V/div, but the noise present for all of them.
Firstly I thought that I'm doing something wrong.
But later I opened the signal in the time domain and noticed that there is a higher noise (see screenshot for the time domain).

[tautech]

I tried different V/div, but the noise present for all of them.
Firstly I thought that I'm doing something wrong.
But later I opened the signal in the time domain and noticed that there is a higher noise (see screenshot for the time domain).

Well, displayed noise in a DSO can be seen as sensitivity so you should be able to manage the displayed trace with a touch of Averaging.

Post a screenshot please.....oh yes bugger, you'll have to change it from BMP to something smaller to fit the forum 1MB file limit.

[radiolistener]

Here is the tool that I wrote to convert BIN files into RAW with source code in C#.
It removes headers and writes ADC raw data only into a RAW file.
Also it shows the oscilloscope settings used to take a BIN file.
This is very useful in order to see the original sample rate.

The BIN file has firmware specific headers, so there is need for a different tool for different firmware version.
This zip file contains the tools for 1.1.1.2.13R5 and 1.1.1.2.15R10.
The version for 1.1.1.2.15R10 may show incorrect values, because I just fixed the code to extract RAW ADC data and didn't tested it more deep.

[radiolistener]

Well, displayed noise in a DSO can be seen as sensitivity so you should be able to manage the displayed trace with a touch of Averaging.

Post a screenshot please.....oh yes bugger, you'll have to change it from BMP to something smaller to fit the forum 1MB file limit.

I cannot take screenshot for 1.1.1.2.15R10 right now, because I roll back to 1.1.1.2.13R5.

Here is the picture of the magnified waveform from BIN file. Lower noise is taken on the old firmware 1.1.1.2.13R5.
The only difference between these two waveform is just firmware. All other conditions is the same.

As you can see:
Firmware 1.1.1.2.13R5 = 1 noise bit. And the noise IS WHITE. This is good.
Firmware 1.1.1.2.15R10 = 2 noise bits. But the noise IS NOT WHITE. This is bad.

It looks that FW 1.1.1.2.15R10 has some kind of nonlinear distortions. Probably it comes from signal processing.
I cannot use the latest firmware for automatic measurement due to this bug... 

[tautech]

Well, displayed noise in a DSO can be seen as sensitivity so you should be able to manage the displayed trace with a touch of Averaging.

Post a screenshot please.....oh yes bugger, you'll have to change it from BMP to something smaller to fit the forum 1MB file limit.

I cannot take screenshot for 1.1.1.2.15R10 right now, because I roll back to 1.1.1.2.13R5.

Here is the picture of the magnified waveform from BIN file. Lower noise is taken on the old firmware 1.1.1.2.13R5.
The only difference between these two waveform is just firmware. All other conditions is the same.

As you can see:
Firmware 1.1.1.2.13R5 = 1 noise bit. And the noise IS WHITE. This is good.
Firmware 1.1.1.2.15R10 = 2 noise bits. But the noise IS NOT WHITE. This is bad.

It looks that FW 1.1.1.2.15R10 has some kind of nonlinear distortions. Probably it comes from signal processing.
I cannot use the latest firmware for automatic measurement due to this bug... 

Well if I still had a SDS1kX I'd roll back the FW to take screenshots and then do more with the new FW.
Less than half an hour can accomplish this IF you wish to put in the time.

I really think 'you' can manage this noise with Averaging or ERES but failing that we'll have to wait for replies from Siglent to you to get a better understanding of what's 'really' happening.

[radiolistener]

I really think 'you' can manage this noise with Averaging or ERES but failing that we'll have to wait for replies from Siglent to you to get a better understanding of what's 'really' happening.

Averaging and ERES is not relevant here, because it's oscilloscope display functions.
I'm capture raw ADC data with oscilloscope and processing it on the PC.
I'm working in the frequency domain, I can average it or do whatever I want, but the problem is that the new firmware adds noise distortion to the original ADC raw data. And there is no way to avoid it...

All what I need is just to get raw ADC data with no modification/distortion/processing. In the form as it comes from the ADC.
The same as it works in firmware 1.1.1.2.13R5.

[tautech]

Thanks I see things a little clearer now.

Have you tried using other than Sigview SW or must you use it ?
Post acquisition processing is 'not' my strong point and I'm only trying to explore other possible solutions for you.

Maybe someone else that has good experience with data capture can chime in with some advice.

[radiolistener]

Have you tried using other than Sigview SW or must you use it ?

You can process it with any software, for example you can use LabVIEW or any tool for FFT.
The FFT is just highlight what is wrong with the data. It shows that the noise is not white.
It has significant spikes at some frequencies and looks like intermodulation distortion.

But you can see that the noise is much higher even by opening file in any text viewer.
You will found that the new firmware produce noise with values 126, 127, 128, 129, 130.
The old firmware produce noise with values 128, 129, 130.
The difference is pretty clean and obvious.

I have some assumptions about the root of cause for this issue:
1) The new firmware performs ADC data processing which add distortions.
2) The new firmware trying to perform some actions in parallel with ADC capture. It leads to RF interference.
3) The new firmware write corrupted BIN file.

But in any case this is a bug and should be fixed.