Siglent SSA3000X Plus math trace unexpected res

[g00se]

I’m trying to understand how the Math trace function is supposed to work in spectrum analyzer mode on the SSA3000X Plus.

To test, I enable a math trace of Power Diff, X-Y+0, where the X and Y traces are exactly the same (Clear Sweep, Max Peak).  I expect to see the math output trace to just be a flat trace at 0dB, but instead I see a flat third trace that looks to be about -95dB offset.

If I try to “capture” an ambient trace and switch it to View, and subtract this from a second Clear Sweep trace of the same general source, I expect to see minimal peaks from 0dB representing the differences, but instead I see a messy trace that jumps up and down all the way from -95dB to the Clear Sweep trace value…

If I try a simple offset value for my math function, like 10dB offset of one trace, it displays as expected.

The manual is not much help, and I can’t find any videos of the Siglent math trace in action.  There are some videos of the Rigol DSA doing a math trace, and it looks to behave how I would expect.

Has anyone here used the Siglent math trace function to subtract two traces successfully?

[Performa01]

I don't think this is something SSA-specific, it's just fundamentals.

0 dB = 1 = 100%

whereas

0 = -infinity dB.

If the result of a math operation is zero, then you should get the maximum negative value for a dB result, which seems to be about -95 dB in this case.

[rf-loop]

I’m trying to understand how the Math trace function is supposed to work in spectrum analyzer mode on the SSA3000X Plus.

To test, I enable a math trace of Power Diff, X-Y+0, where the X and Y traces are exactly the same (Clear Sweep, Max Peak).  I expect to see the math output trace to just be a flat trace at 0dB, but instead I see a flat third trace that looks to be about -95dB offset.

If I try to “capture” an ambient trace and switch it to View, and subtract this from a second Clear Sweep trace of the same general source, I expect to see minimal peaks from 0dB representing the differences, but instead I see a messy trace that jumps up and down all the way from -95dB to the Clear Sweep trace value…

If I try a simple offset value for my math function, like 10dB offset of one trace, it displays as expected.

The manual is not much help, and I can’t find any videos of the Siglent math trace in action.  There are some videos of the Rigol DSA doing a math trace, and it looks to behave how I would expect.

Has anyone here used the Siglent math trace function to subtract two traces successfully?

Are you measuring dB values with Spectrum analyzer? Least I do not know how much is just one dB. Sorry but I have not seen any SA what measures dB.  Or are you perhaps measuring example dBm levels as we usually measure with rf-spectrum analyzers.

How much is 10dB or 0dB or what ever dB. My answer is "nothing".
How much is 10dBm or 0dBm.  0dBm is 1mW  and 10dBm is 10mW  and then how much is 100 dBm. No it is not 100 mW. It is 1e10 mW aka 10000000000mW (10 MW  yes Mega Watt)

Now I ask you, how much is  a dBm - a dBm (you told you have two equal traces).   Is it 0dBm 
Or how much is 0dBm - 0dBm. Is it 0dBm. No it is not. It is 0mW.  How many dBm it is?

You tell there read -95dB  perhaps you mean -95dBm (95dB below 0dBm)
0dBm is 1mW power
-95dBm is 0.000000000316 mW !  0.316 pW.

[g00se]

Yes, sorry the units are dBuV.  I was hoping to “subtract” the ambient signals from my scan of device under test for EMC pre-compliance.

I was hoping to do something similar to what is shown here using a Rigol analyzer:

[g00se]

I captured some screens to try and illustrate:

Here is just an ambient sweep:


Here is ambient  + device sweep:


and this is what the math trace looks like, trying to subtract the first trace from the second:

[g00se]

I am going to try and contact Siglent today and see how the math trace is actually supposed to work.

I tried switching to different units and scales, but the behaviour is the same in all cases.

[rf-loop]

I am going to try and contact Siglent today and see how the math trace is actually supposed to work.

I tried switching to different units and scales, but the behaviour is the same in all cases.

I will recommend some lessons about math. In my previous message I have tried give indirect hint.... but with dBm.


Think this attached image what all things you see there. Think carefully! Think again. I can tell, it is right. If you think it is wrong please (kindly  ) repair your thinking. 

Btw, 7 points hint:  how much is -93dBuV  ... you may know... it is much less than nearly nothing... in practice

Think other: How much is 0dBuV - 0dBuV

It is just math and my opinion is that it is just right. But we have problem in practice because SA can not draw infinite and can not work as ideal machine doing infinitely ideal math. It have limits, as all have.  We can not draw   -infinite level... but  -93dBuV is so small we can keep it as "nothing" ... our bottom limit -200dBm  ~2.24x10^-11 Vrms




Note. Both traces have used trace averaging, before set to hold (aka View), for reduce noise for make it more clear how this Math works.
First set hold one and then small level change and frequency change and after this trace average set this trace to View (hold) mode.

[g00se]

Thanks for that.  I am reading about from this post, which helps: https://www.reddit.com/r/rfelectronics/comments/nt616h/answers_for_0db0db_and_0dbm0dbm_are_different_why/

[g00se]

Maybe you can also help me understand what is different about the way the Rigol is performing the operation shown in that video?

[rf-loop]

Maybe you can also help me understand what is different about the way the Rigol is performing the operation shown in that video?

You have selected math where is level1 - level2 = level (in your case level is in dBuV but also same if you use dBm level only numbers are different but principle same)
And your SA do it and it do it right.

But have you ever thinked what you see in your example video where SA is Rigol.

Now; look it carefully and think what you see. Is it possible that -20dBm - -20dBm is 0dBm.  NO!  But as you can estimate this happen in this Rigol video.

Now I ask why you select Math Power diff where result is power. (level difference where result is level)  You know that 1kg - 0.5kg and result unit is kg, in this case 0.5kg. 1kg - 1kg result is 0kg.

But if you calculate  example 0dBuV - 0dBuV it is  0uV  and now I ask how many dBuV it is. Or example 0dBm - 0dBm result is 0mW but really not 0dBm but -infinite dBm
Think again what you see in Rigol video.

If there is -20dBm - -20dBm result is same as -30dBm - -30dBm  in both cases difference is 0dB! and so on. 
The answer is the  levels difference  deciBels and not the difference in level. dB is a ratio but dBuV and dBm is the signal level (relative to a reference of 1uV and 1mW).

Look carefully next image (yes my SSA is different but principle same)
Look these traces and relations... how it is calculating. Now you see difference in dB (look what math was now in use)
And this is same principle what you see in Rigol diplay....

[g00se]

Thank you!  That was genuinely helpful.   

It was a good learning experience for me.