Siglent SDS800X HD 12 bit DSO's

[Furna]

I am starting to play with the save function in binary format.
I just attached CH1 probe to the Compensation Signal Terminal and used Auto Setup; as expected 500mV/div for vertical, 500µs/div for horizontal, 2.00GSa/s and 10.0Mpts

I then tried to save in binary format to the attached NAS share I use sucessfully for PNG on a daily basis.
I got the error "Not enough storage available!" while on the NAS I have more than 3 TB free 

I was able to save the binary file (20,004,096 bytes bytes) on local storage and then transfer it to the NAS; same issue with MatLab format. 

I am using firmaware V1.1.3.8

Did anyone had the same issue?

Edit: added values for GSa/s, Mpts and file size; also reported the exact error.

[eTobey]

A familiarization question really- What does the 12-bits cover? IOW, how much of that is beyond the screen upper and lower boundaries? What's the expected number of vertical steps over just the screen?

What about the ENOB of that scope?
Thats just what i am (kind of) looking at. Especially the offset, in relation to the vertical position.
At 5V/div, the offset ranges from -0.06V to +0.06V with a constant fluctuation that is in the range of just below 50mV.
With 40V range, 50mV steps would be just below 10bits...   

The graph shows offset values retrieved with an averaging count of 32! And a step of 0.1V.

[ebastler]

What about the ENOB of that scope?

8.4 bits (typical) according to the datasheet.

There's more to "ENOB" than you might think. And there's fewer effective bits (in any DSO and ADC) than you might expect intuitively. 

[eTobey]

What about the ENOB of that scope?

8.4 bits (typical) according to the datasheet.

There's more to "ENOB" than you might think. And there's fewer effective bits (in any DSO and ADC) than you might expect intuitively. 

Shouldnt the ENOB be increased by averaging???

[ebastler]

Shouldnt the ENOB be increased by averaging???

Yes indeed -- and you see that, right? Look at any open (or better 50-Ohm terminated or shorted) channel, let the scope's Measure function calculate the standard deviation. Then look at a Math function which takes e.g. a 4-fold or 16-fold average of that channel; again let the scope measure its standard deviation. Every 4x averaging should reduce the standard deviation by a factor of 2. Best observed by switching measurement statistics on in addition, so you get a more stable time-average over the standard devs.

When estimating SNR or effective resolution, you can't eyeball peak-to-peak noise. You need to use the standard deviation, which the scope can conveniently measure. 

[eTobey]

I got to explain a bit more:
The screenshot represents a single point of the graph. Every little step results in quite a huge offset jump. This might not be such a problem, but i believe this inaccuracy also applies to any signal, that "moves through those steps". Looks like some rather bad ADC nonlinearity.

The measured levels of the highlevel (F1) do not correspond with the values i set up, and measured with other devices (2.5V/div):
Siggen set to:                          8.00V
This scope measure:                7.87V
My cheap DSO:                        8.00V-8.08V
My multimeter (duty 99.99%): 7.97V

(Yes, its a screenshot of the duty of less than 99%)

Resoultion of 12 bits at 20V span means about 5mV. Deviation is about 130mV. 

I dont know what i should think of the zoom-mode feature then...

EDIT:
Tried now with 18V with MCP1407:
This scope:  17.81V
Multimeter:  17.97V
Cheap DSO: 18.0V

Deviation: almost 200mV (with math average).

Moral of the story: Dont expect too much, even with math averaging...

[ebastler]

I got to explain a bit more:
The screenshot represents a single point of the graph. Every little step results in quite a huge offset jump. This might not be such a problem, but i believe this inaccuracy also applies to any signal, that "moves through those steps". Looks like some rather bad ADC nonlinearity.

The measured levels of the highlevel (F1) do not correspond with the values i set up, and measured with other devices (2.5V/div)

I thought you were concerned about SNR and resolution. So is this about the offset error and the scale deviation instead? 

I don't think either of those point to an ADC non-linearity. The offset you set (with the small "vertical" knob on the front panel) is an actual offset applied in the analog domain in my understanding; it is not applied after ADC conversion. Which is a good thing, so the full ADC range is still available afterwards. There are apparently small offset errors in this analog subtraction.

And the vertical scale error (which you see when measuring the absolute value of a voltage) is, well, a vertical scale error. Even an ADC with perfect linearity will produce such a deviation when the reference voltage it is comparing to is not quite right. Doesn't matter; it is a scope, not a multimeter.

[2N3055]

I am starting to play with the save function in binary format.
I just attached CH1 probe to the Compensation Signal Terminal and used Auto Setup; as expected 500mV/div for vertical, 500µs/div for horizontal, 2.00GSa/s and 10.0Mpts

I then tried to save in binary format to the attached NAS share I use sucessfully for PNG on a daily basis.
I got the error "Not enough storage available!" while on the NAS I have more than 3 TB free 

I was able to save the binary file (20,004,096 bytes bytes) on local storage and then transfer it to the NAS; same issue with MatLab format. 

I am using firmaware V1.1.3.8

Did anyone had the same issue?

Edit: added values for GSa/s, Mpts and file size; also reported the exact error.

Did not have problems saving BIN files to windows PC share, even a 100MBytes ones.
But that was some time ago, so I need to try again.
Will let you know.

[gf]

The measured levels of the highlevel (F1) do not correspond with the values i set up, and measured with other devices (2.5V/div)

Keep in mind that the resolution does not tell anything about accuracy / measurement uncertainty.
Note that the datasheet does not promise an accuracy of 1LSB (as you obviously expect), but only

DC gain accuracy 0.5 mV/div ~ 4.95 mV/div: ±1.5 %; 5 mV/div ~ 10 V/div: ±0.5 %;
Offset accuracy ± (0.5% of the offset setting + 0.5% of full scale + 1 mV)

And that's only promised for DC. At higher frequencies (and your signal is non-DC) there can be additional uncertainty due to the dynamic behavior.
I'm also unsure if this is for the scope alone, or including the probes. I guess the former - then the probes will add an additional uncertainty.

If you want to assess the vertical scale error, you must measure both, the upper and the lower level and calculate the difference.
From the screenshot, I cannot see that the lower level is really measuring zero, but there could be an offset error.

And btw, did you run a self-calibration?

Averaging can only reduce the uncertainty due to random noise, but it cannot mitigate systematic errors.

[tautech]

The measured levels of the highlevel (F1) do not correspond with the values i set up, and measured with other devices (2.5V/div)

Keep in mind that the resolution does not tell anything about accuracy / measurement uncertainty.
Note that the datasheet does not promise an accuracy of 1LSB (as you obviously expect), but only

DC gain accuracy 0.5 mV/div ~ 4.95 mV/div: ±1.5 %; 5 mV/div ~ 10 V/div: ±0.5 %;
Offset accuracy ± (0.5% of the offset setting + 0.5% of full scale + 1 mV)

And that's only promised for DC. At higher frequencies (and your signal is non-DC) there can be additional uncertainty due to the dynamic behavior.
I'm also unsure if this is for the scope alone, or including the probes. I guess the former - then the probes will add an additional uncertainty.

If you want to assess the vertical scale error, you must measure both, the upper and the lower level and calculate the difference.
From the screenshot, I cannot see that the lower level is really measuring zero, but there could be an offset error.

And btw, did you run a self-calibration?

Averaging can only reduce the uncertainty due to random noise, but it cannot mitigate systematic errors.

Quite true and can be corrected some by using the Probe Check feature to apply a modified attenuation to correct error.
Not unusual to see 10.1:1 for a 10x probe or 103:1 for a 100x probe.

But even imperfect probe compensation will impact P-P measurements.

[eTobey]

...

0.19V of 20V is 0.95%, not quite 0.5%, and at 99.99% duty cycle, i would consider this DC. (It was at the last portion of the high level)
Subtracting an offset error from my graph (-10V vertical position), 0.17V of 20V is still 0.85%

Yes, i did a calibration. It was much worse before.

I could also imagine, that my scope did age already? I estimate, that i had let it run for like 1000-1300 hours already.

[eTobey]

Quite true and can be corrected some by using the Probe Check feature to apply a modified attenuation to correct error.
Not unusual to see 10.1:1 for a 10x probe or 103:1 for a 100x probe.

That is actually a fix for this issue. But there is a catch: The vertical scale has now some ugly numbers. 

Now i just have to remember not to touch the vertical position knob between measurments, for the offset to be somewhat constant.

I knew, that it was influenced the the vertical position, but not with such erratic fluctuations.

The probcheck gave me a wrong value: It gave me 10.2, when it was actually 10.08

[Furna]

I am starting to play with the save function in binary format.
I just attached CH1 probe to the Compensation Signal Terminal and used Auto Setup; as expected 500mV/div for vertical, 500µs/div for horizontal, 2.00GSa/s and 10.0Mpts

I then tried to save in binary format to the attached NAS share I use sucessfully for PNG on a daily basis.
I got the error "Not enough storage available!" while on the NAS I have more than 3 TB free 

I was able to save the binary file (20,004,096 bytes bytes) on local storage and then transfer it to the NAS; same issue with MatLab format. 

I am using firmaware V1.1.3.8

Did anyone had the same issue?

Edit: added values for GSa/s, Mpts and file size; also reported the exact error.

Did not have problems saving BIN files to windows PC share, even a 100MBytes ones.
But that was some time ago, so I need to try again.
Will let you know.

I captured the network traffic on the NAS and compared the save operations for a PNG and for a BIN ... the behaviour of the oscilloscope is different.

- For PNG files the oscilloscope makes three different requests to create the file; the first and the second contain also other requests (the second doesn't even contain the filename) and fail while the third one is successful.
- For BIN file the oscilloscope makes four different requests to create the file; the first and the fourth contain also other requests (the second and the third don't even contain the filename) and all four are unsuccessful.

If you try to replicate, please test BIN and MatLab.

Thank you.

Edit
I can share the pcap files in private if needed.

[2N3055]

I am starting to play with the save function in binary format.
I just attached CH1 probe to the Compensation Signal Terminal and used Auto Setup; as expected 500mV/div for vertical, 500µs/div for horizontal, 2.00GSa/s and 10.0Mpts

I then tried to save in binary format to the attached NAS share I use sucessfully for PNG on a daily basis.
I got the error "Not enough storage available!" while on the NAS I have more than 3 TB free 

I was able to save the binary file (20,004,096 bytes bytes) on local storage and then transfer it to the NAS; same issue with MatLab format. 

I am using firmaware V1.1.3.8

Did anyone had the same issue?

Edit: added values for GSa/s, Mpts and file size; also reported the exact error.

Did not have problems saving BIN files to windows PC share, even a 100MBytes ones.
But that was some time ago, so I need to try again.
Will let you know.

I captured the network traffic on the NAS and compared the save operations for a PNG and for a BIN ... the behaviour of the oscilloscope is different.

- For PNG files the oscilloscope makes three different requests to create the file; the first and the second contain also other requests (the second doesn't even contain the filename) and fail while the third one is successful.
- For BIN file the oscilloscope makes four different requests to create the file; the first and the fourth contain also other requests (the second and the third don't even contain the filename) and all four are unsuccessful.

If you try to replicate, please test BIN and MatLab.

Thank you.

Edit
I can share the pcap files in private if needed.

I just tried quickly, 100MPts to BIN and MAT.
To a Win10 share. 1.8TB size.
Worked just fine.

What comes to my mind is either some size limit, or version of SMB..
What NAS is it?

[Furna]

I am starting to play with the save function in binary format.
I just attached CH1 probe to the Compensation Signal Terminal and used Auto Setup; as expected 500mV/div for vertical, 500µs/div for horizontal, 2.00GSa/s and 10.0Mpts

I then tried to save in binary format to the attached NAS share I use sucessfully for PNG on a daily basis.
I got the error "Not enough storage available!" while on the NAS I have more than 3 TB free 

I was able to save the binary file (20,004,096 bytes bytes) on local storage and then transfer it to the NAS; same issue with MatLab format. 

I am using firmaware V1.1.3.8

Did anyone had the same issue?

Edit: added values for GSa/s, Mpts and file size; also reported the exact error.

Did not have problems saving BIN files to windows PC share, even a 100MBytes ones.
But that was some time ago, so I need to try again.
Will let you know.

I captured the network traffic on the NAS and compared the save operations for a PNG and for a BIN ... the behaviour of the oscilloscope is different.

- For PNG files the oscilloscope makes three different requests to create the file; the first and the second contain also other requests (the second doesn't even contain the filename) and fail while the third one is successful.
- For BIN file the oscilloscope makes four different requests to create the file; the first and the fourth contain also other requests (the second and the third don't even contain the filename) and all four are unsuccessful.

If you try to replicate, please test BIN and MatLab.

Thank you.

Edit
I can share the pcap files in private if needed.

I just tried quickly, 100MPts to BIN and MAT.
To a Win10 share. 1.8TB size.
Worked just fine.

What comes to my mind is either some size limit, or version of SMB..
What NAS is it?

TrueNAS Scale ElectricEel-24.10.1 -> latest version
What I don't understand is the different oscilloscope's behaviour for PNG and BIN ...
Also keep in mind I am able to move the file to the NAS after it is saved locally using the oscilloscope File Manager; so no size limit or SMB version.

Are you able/willing to capture the network traffic on your Win10 (i.e. Wireshark)?

Thank you.

[2N3055]

TrueNAS Scale ElectricEel-24.10.1 -> latest version
What I don't understand is the different oscilloscope's behaviour for PNG and BIN ...
Also keep in mind I am able to move the file to the NAS after it is saved locally using the oscilloscope File Manager; so no size limit or SMB version.

Are you able/willing to capture the network traffic on your Win10 (i.e. Wireshark)?

Thank you.

Thanks.

Sure I can capture the sniffer trace, when I get a bit of time...
Will let you know. Don't worry, we'll figure it out.

Can you send me CAP files you captured?
There is an email in my profile you can send it to.
Best,

[Furna]

Thanks.

Sure I can capture the sniffer trace, when I get a bit of time...
Will let you know. Don't worry, we'll figure it out.

Can you send me CAP files you captured?
There is an email in my profile you can send it to.
Best,

I sent you a PM

[MathWizard]

I can still get by with my 8-bit Siglent DSO's, but I could get a SDS804X now. I don't notice the screen size is still the same size as the sds1000X-E's, they aren't too bad but now I've been spoiled by the 10.1" screens.

But then there's the SDS1104X HD, for ~3x the price tho, I guess the 1000X HD's are more the direct 12-bit upgrade of the sds1004X-E, but with the 10.1" also.

Well I have the datasheets for both, my wiser self is saying to save up and get the 1104X HD, then it would be my main DSO. It's a shame to have any of them sitting doing nothing most of the time tho. I should sell my old 1980's scope tho either way, I'm not using it.

[ebastler]

I can still get by with my 8-bit Siglent DSO's, but I could get a SDS804X now. I don't notice the screen size is still the same size as the sds1000X-E's, they aren't too bad but now I've been spoiled by the 10.1" screens.

But then there's the SDS1104X HD, for ~3x the price tho, I guess the 1000X HD's are more the direct 12-bit upgrade of the sds1004X-E, but with the 10.1" also.

Well I have the datasheets for both, my wiser self is saying to save up and get the 1104X HD, then it would be my main DSO. It's a shame to have any of them sitting doing nothing most of the time tho. I should sell my old 1980's scope tho either way, I'm not using it.

While the physical screen size of the 800X HD is still the same as in the 1000X-E, the resolution is increased from 800*480 to 1024*600. Actually it's the same resolution as in its larger sibling, the 1000X HD. So you won't get more information on the larger screen, but it will be easier to see.

And of course you will get some extra features in the larger scope, probably most notably the 50 Ohm inputs and probe detection. Performance-wise I think they are exactly the same (vertical and horizontal ranges, sampling rate, time base accuracy etc.). If you are not opposed to "hacking" the scope, it's a bit of a pity that you can only start with the 100 MHz version for the 1000X HD, which further widens the price gap.

[Conrad Hoffman]

When I need 50 ohms (rarely) I just use a feed-thru 50 ohm BNC terminator. Prices are all over but you can find double ended ones for about $10 if you shop around. FWIW, I dislike probe detection because the pins on the probes scratch and wreck scopes that don't support it. Visually, the vertical LCD resolution is still going to show you some stepping for gently sloped waveforms. We need 12-bit LCDs! Not holding my breath. It depends on what you need a scope to do. I sold off all my old boat anchors and plug-ins and downsized to the SDS824X HD. So far I'm completely happy with it for my audio and scientific/technical work. Also got the SDG2082X arb waveform generator and that's been a joy to use as well. I had my doubts about not having knobs for everything, but the design makes it a non-issue. Yes, I got rid of most of my other generators if the Siglent overlapped their function.

[ebastler]

Visually, the vertical LCD resolution is still going to show you some stepping for gently sloped waveforms. We need 12-bit LCDs!

I have written this too often here, but still: The limitation is not the actual LCD resolution but the fact that the SDS800X HD and 1000X HD always pair two pixels vertically when drawing traces, effectively cutting the vertical resolution in half. Rarely noticeable on the small screen of the 800 models, more visible with the larger pixels on the 1000 screen.

[Conrad Hoffman]

I was reading the in-scope help and stumbled across the VNA section, though it's not available in the SDS800X HD. Just curious, is it "just" software, or is different hardware required?

[tautech]

I was reading the in-scope help and stumbled across the VNA section, though it's not available in the SDS800X HD. Just curious, is it "just" software, or is different hardware required?

Not in the beta FW I'm using so must be a mistake that's been righted in the FW to come.

[eTobey]

We need 12-bit LCDs!

In certain situations (see my volt step shootout), you will then see signals, which are actually not real. At least für the SDS800X HD...

[tautech]

We need 12-bit LCDs!

In certain situations (see my volt step shootout), you will then see signals, which are actually not real. At least für the SDS800X HD...

Gibbs Ears ?