Siglent SDS1204X-E bode plot glitches!

[jhiesey]

I recently got a Siglent SDS1204X-E with the SAG1021 AWG addon. Almost everything works well, but there's a very annoying problem with the bode plot function that I'm pretty sure is a firmware bug. 

I'm using a very simple test setup, where I have the AWG output going through a piece of RG58 to a tee on channel 1, which also goes through a second piece of RG58 to a 50 ohm terminator on channel 2:



When I run a bode plot, the first sweep after I press the run button comes out fine. But if I let it loop around back to the start frequency, the second and subsequent sweeps have significant glitches in both the frequency and phase plots:



That's a 14db spike in the amplitude! I've also attached the csv data file from the scope that corresponds to this run.

EDIT: 14db is the biggest glitch I've seen so far. Usually the glitches are smaller, more like 1-6db, sometimes positive and sometimes negative.

At first I thought it might just be loose connections, but it's very reproducible and only happens on the second and subsequent frequency sweeps. Pressing the run button again to stop it and then restarting the sweep makes it work again for one more sweep.

I tried it with various sweep settings (e.g. 20hz to 20Mhz at 20 points/decade, and 100hz to 1khz at 100 points/decade) and it's pretty consistent. Now here's the interesting part: the glitches always happen right after the scope timebase (as indicated in the upper left of the screen) changes. The three problematic switches are where it goes from 100ms/div to 50ms, 50ms to 20ms, and 20ms to 10ms.

That sure sounds like a bug to me. My guess is there isn't enough delay after the scope switches timebases before it tries to take a measurement. Unless I'm the one doing something stupid? 

Versions:
SDS1204X-E firmware version 6.1.33, hardware version 01-03
SAG1021 firmware version 1.02, hardware version 0-3-1

[Performa01]

Strange. I've done extensive tests with Bode Plot II and even though it's still not perfect, I've never experienced something even remotely severe like this.

There are plenty of tests and examples by rf-loop in this thread:
https://www.eevblog.com/forum/testgear/siglent-sds1x04x-e-bodeplot-ii-(sfra)-features-and-testing-(coming)/

For all these tests, several passes have been performed without a significant difference between them.

My first thought was an issue with the cables/plugs/adaptors, but then you've already considered this.

I've used an SDG6000X for my tests exclusively  and rf-loop used an SDG1000X, so it might be a problem with the SAG1021 - even though I have not had any issues with it in the past (with the previous Bode Plot module). I can only check this next weekend - maybe someone else can try to replicate your problem earlier than this...

[rf-loop]

I have seen some cases some small, mostly small, glitches (same time when SDS "click" relays or change sampling rate)

But it is very important that self cal is well done and scope have reached its temperature specially in some sensitive cases.


Here just made quick shoot using SDG1000X and SDS1104X-E
1Vpp, 20Hz to 20MHz, 40 steps decade, Vin/Vout mode and Auto input gain.

Just simple  SDG CH1 to SDS Ch1 with T and then cable with 50ohm to Ch2
Even with this setup, it produce some amplitude glitch when it change sampling.
at this time it did not exist but I know it exist some times.
But your glitch is really huge. Never seen this. (but different SDG, different timing or what ever... )
First be sure your selfal is fresh and good.

[CiscERsang]

hi there,
slightly off-topic, but...
Which way do I convert provided above csv. Is there a special software which can be used? Or?

BR

[Performa01]

hi there,
slightly off-topic, but...
Which way do I convert provided above csv. Is there a special software which can be used? Or?

BR

Many programs can import CSV data. Microsoft Excel would be a good candidate. Just make sure to get the import settings correct in order to get a useful table.

[rf-loop]

hi there,
slightly off-topic, but...
Which way do I convert provided above csv. Is there a special software which can be used? Or?

BR

You mean BodePlotII  data?
And just if look my previous image,  bottom line there is "Data". Press Data and after then there come new bottom line menu, there do your selections and press Save..   Also you can recall it bact to scope and analyze look again just as fresh done bodeplot (.CSV include all for reconstruct it fully and also data have full resolution).


Overall main menu structure can see here.
https://www.eevblog.com/forum/testgear/siglent-sds1x04x-e-bodeplot-ii-(sfra)-features-and-testing-(coming)/msg2453661/#msg2453661

.CSV file you can recall back to scope and analyze again or open it using example OpenOffice or LibreOffice (free) and many others. .CSV is  just plain text, Comma Separated Values.

[CiscERsang]

Many programs can import CSV data. Microsoft Excel would be a good candidate. Just make sure to get the import settings correct in order to get a useful table.

That's the point, so I'm asking about template or third-party software.

BR

[graybeard]

That sure sounds like a bug to me. My guess is there isn't enough delay after the scope switches timebases before it tries to take a measurement. Unless I'm the one doing something stupid? 

Versions:
SDS1204X-E firmware version 6.1.33, hardware version 01-03
SAG1021 firmware version 1.02, hardware version 0-3-1

I see the same thing when I use mine with my SDS1104X-E and SDG2042X.  It sometime happens on the first sweep, but clears the second time  through.   However the first point (lowest frequency) point often glitches on subsequent passes.

[jhiesey]

I've done a bit more investigating and have some updates.

First, at the suggestion of some here, I let it warm up for over 30 mins and reran the self-cal function. The first time I tried that it hung at 98% for at least 10 minutes and wouldn't respond to any key presses, so I had to force power off the scope. I tried again today and it finished much faster. Maybe the hang was because I had a USB flash drive inserted the first time? Not confidence inspiring though. At any rate the self-cal didn't affect the bode plot issue.

I also tried different stimulus amplitudes (10mV up to 3V) with no difference.

However, I did notice that the issue happens much more consistently in the "Channel Gain: Hold" setting than "Channel Gain: Auto". It does happen in both modes, but sometimes it's ok in auto mode. It is still always fine for me on the first sweep after I hit run. Whenever I set the gain to "hold" I exited the bode plot menu and manually adjusted the vertical gain to ensure the signal amplitude was the largest possible non-fine setting, e.g. 200mV per division for a 1Vpp signal. It seems like an oversight that the gain can't be set manually without exiting the bode plot function. Or am I missing something here?

I also tried different sweep settings (linear and logarithmic) and numbers of points, and the only consistent behavior is the glitches always occur right after the three specific timebase changes I mentioned in my original post, not at exact specific frequencies.

[tinhead]

as i have SAG1021 as well, i did exact the same measurments and got same, random errors.





However, this is somehow related to SAG calibration. Once i did Zero Adjust calibration of the SAG, the BodePlot II result was ok.




As the SAG1021 have some issues with USB power/DC offset, i use USB Isolator. But this does not have any positive/extra influence on the random spikes





The run the AWG Zero Adjust, close BodePlott II, go to AWG, Settings, Zero Ajust and click once on Automatic. The SAG1021 needs to be connected directly to CH1.

[Performa01]

I also did some tests with the SAG1021 now.

I was able to get a glitch immediately after powering up the scope:


SDS1104X-E_SAG1021_Initial

This is most likely because of the automatic Quick-Cal feature of the SDS1000X-E.

You can turn Quick-Cal off on page 1 of the Utility menu, but you should still wait at least 15 minutes until the instrument has warmed up a bit before using the Bode Plot.

Even with Quick-Cal enabled, it delivers solid results after 30 minutes of warm-up time:


SDS1104X-E_SAG1021_30min

EDIT: The test setup wasn't quite the same for these two screenshots, this is why there is some amplitude and phase error in the first one.

[Performa01]

First, at the suggestion of some here, I let it warm up for over 30 mins and reran the self-cal function. The first time I tried that it hung at 98% for at least 10 minutes and wouldn't respond to any key presses, so I had to force power off the scope. I tried again today and it finished much faster. Maybe the hang was because I had a USB flash drive inserted the first time? Not confidence inspiring though. At any rate the self-cal didn't affect the bode plot issue.

That sounds very strange. I just ran a Self-Cal with an USB stick inserted (as I've often done in the past) and it still works for me with the 6.1.33 firmware. However the behavior might depend on the specific USB stick used. Please try again with your stick inserted and see if you can reproduce the hang.

However, I did notice that the issue happens much more consistently in the "Channel Gain: Hold" setting than "Channel Gain: Auto". It does happen in both modes, but sometimes it's ok in auto mode. It is still always fine for me on the first sweep after I hit run. Whenever I set the gain to "hold" I exited the bode plot menu and manually adjusted the vertical gain to ensure the signal amplitude was the largest possible non-fine setting, e.g. 200mV per division for a 1Vpp signal. It seems like an oversight that the gain can't be set manually without exiting the bode plot function. Or am I missing something here?

You are right: setting/checking the correct vertical gain (required for Hold mode) is cumbersome.

On the other hand, any manual gain change would have to stop and resume the Bode Plot, so there needs to be an implicite resume function. A menu item for the user to get a peek on the y-t traces together with access to a channel selection and the vertical gain control could then be implemented.

Another solution could be an automatic setup mode, which acts pretty much like Auto mode, but only lowers and never increases the gain. So it would start at the highest sensitivity, then automatically reduce it as required by the signal until the final steady gain setting is reached.

With regard to the glitches, the Hold mode should be less prone to it in theory. The original Bode Plot sometimes produced some minor glitches on vertical gain changes - whenever relay actions were involved - but this has been pretty much elimitated in the Bode Plot II. And after an extensive test I can confirm that Hold mode is rock stable and the gain curve even smoother than with Auto Gain.


SDS1104X-E_SAG1021_Hold
 

I also tried different sweep settings (linear and logarithmic) and numbers of points, and the only consistent behavior is the glitches always occur right after the three specific timebase changes I mentioned in my original post, not at exact specific frequencies.

Horizontal timebase changes have never been a problem so far.

I start wondering if there's something wrong with your particular instrument. Self-Cal sometimes not working properly together with the huge Bode Plot glitches even with the DSO warmed up - all this doesn't sound quite right.

[Performa01]

Here's the result of some further investigations:

After some hours of continuously running the Bode Plot with the SAG1021, keeping an eye on the plot and watching out for glitches, I suddenly got this:


SDS1104X-E_SAG1021_1h

The reason should be obvious: there was no output signal from the SAG1021 anymore!

An investigation of the AWG settings showed the output as still on, but it was actually off. Pushing the Output On/Off button once changed the display to "Off", but no relay click could be heard. Pushing once again turned the SAG1021 back on and the relay clicked as it should.

This means, the SAG1021 shuts off its output for no reason after a random timespan. Since I've not experienced this in the past, it might have something to do with the unterminated operation into High-Z inputs. For tests like this I normally always use a proper power splitter instead of just a BNC-T and 50 ohm inline terminators at the scope inputs.

After switching the SAG1021 output back off again and restarting the Bode Plot, the SAG1021 output got automatically turned on and the result was good again:


SDS1104X-E_SAG1021_1h_Rst

However, 15 minutes later the output shut down once again. The temperature of the SAG1021 "brick" was certainly not high (just 32.6°C measured), so I would be surprised if this was a thermal shutdown.

I will run a long term test with the correct standard setup now for a while and see if the unmotivated output shutdown happens again:


Bode Test Standard Setup 03

[Performa01]

Update:

With the proper 50 ohm setup, the bode plot produced more glitches in my perception compared to the high-Z setup used before.  I even have turned Auto-Cal off to be sure that this is not the culprit. The SAG1021 shut down its output again after several hours.

So my initial assumptions have been wrong and the following is true:

-   Even a well warmed up instrument with Auto-Cal turned off produces glitches every now and then.
-   The SAG1021 output shuts down randomly for no reason regardless of the termination.

I still firmly believe that all these issues are specific for the SAG1021, but now I've started a long term test using the SDG6052X and will report back should I find any issues.

Siglent R&D is aware of this thread and they are also investigating right now.

[maginnovision]

I also have a problem, I can't differentiate those two colors.

[tautech]

I also have a problem, I can't differentiate those two colors.

Then pick any one of the 3 other channels to Bode plot with and use Ch2 for the hidden Ref input.

[Performa01]

I also have a problem, I can't differentiate those two colors.

I've already requested - among other things - the visual appearance of the plot to be reworked. This also includes the option to control the visibility of the individual graphs. This way amplitude-only and phase-only plots could be viewed and printed.

We usually want the phase plot to be less prominent, e.g. less width, less brightness, but there are more ideas:

 - Customizable appearance of each individual trace (width, brigthness)
 - A split screen option with separate grids for amplitude and phase

[maginnovision]

I also have a problem, I can't differentiate those two colors.

Then pick any one of the 3 other channels to Bode plot with and use Ch2 for the hidden Ref input.

That would be a good idea, except those aren't my pictures and screenshots. Thanks anyway.

[graybeard]

I also have a problem, I can't differentiate those two colors.

I've already requested - among other things - the visual appearance of the plot to be reworked. This also includes the option to control the visibility of the individual graphs. This way amplitude-only and phase-only plots could be viewed and printed.

We usually want the phase plot to be less prominent, e.g. less width, less brightness, but there are more ideas:

 - Customizable appearance of each individual trace (width, brigthness)
 - A split screen option with separate grids for amplitude and phase

I like the simultaneous amplitude and phase plot.  The main thing I would like to see changed in the format is eliminating the excess trailing zeros in the frequency axis markings.  I like using the full screen for both since the resolution is so limited.

There are two types of glitches I observe:

There are tiny ones that are really just small <0.3dB errors in the fine (gain/attenuation) levels.  These are acceptable since they are really only visible when you look at very fine amplitude scale and are probably not correctable.

Then there are large random glitches.  I don't know for sure but I theorize these are due to not giving the circuitry time to settle before making a measurement after a gain/attenuation range adjustment.  On my scope these seem to occur the most when the sweep starts the second pass. 
It also seems to be worse on some measurements and not present on others.  Perhaps this is because some of gain/attenuation level take longer to settle than others.

Chris

[Performa01]

Siglent engineers have now identified the reason for the glitches. It will be fixed in the next firmware.

[jhiesey]

I can confirm this is fixed in the new firmware 6.1.35R2. Thanks Siglent! 

And the comment in the release notes on the fix (tried to quote it here but it doesn't copy properly from the PDF) is almost word-for-word what I guessed the problem was haha.

Anyway I'm impressed! This was the only reason I wasn't totally happy with my scope, and now it's fixed!