Siglent SDS800X HD 12 bit DSO's

[2N3055]

Btw, is it actually possible to use FFT(CHx) as operand of other math functions, so that FFT(CH3)-FFT(CH1) [in decibels] can be displayed directly with a trace?

Unfortunately no, at this moment you cannot use FFT result as input to other operations...
That would be my wish too, because it opens whole another set of things you could do....

[mawyatt]

Btw, is it actually possible to use FFT(CHx) as operand of other math functions, so that FFT(CH3)-FFT(CH1) [in decibels] can be displayed directly with a trace?

Unfortunately no, at this moment you cannot use FFT result as input to other operations...
That would be my wish too, because it opens whole another set of things you could do....

And same for Bode Function, such as including Calibration References (like LCR or VNA) as well as extending operation to limit of AWG (500MHz with SDG6000X).

Best

[2N3055]

Btw, is it actually possible to use FFT(CHx) as operand of other math functions, so that FFT(CH3)-FFT(CH1) [in decibels] can be displayed directly with a trace?

Unfortunately no, at this moment you cannot use FFT result as input to other operations...
That would be my wish too, because it opens whole another set of things you could do....

And same for Bode Function, such as including Calibration References (like LCR or VNA) as well as extending operation to limit of AWG (500MHz with SDG6000X).

Best

Yes we spoke about that. That would need deembedding and whole lot of additional math...

[iMo]

Frankly, I somehow doubt the mcus in those scopes are performant enough for some complex math stuff.. What would help is a fast data transfer off the scope into an external computer for the further processing (the same for rigols).

[BillyO]

therefore FFT(CH3)-FFT(CH1) [in dB] is the log magnitude of the frequency response of the probe

The problem with that is in 2N3055's screen shot FFT(CH3)-FFT(CH1) is positive starting at about 210MHz and increases to about +13dB at 420MHz and is still +5dB at nearly 500MHz.  When I ran it I ran it on a faster scope and ran it out to 650MHz and FFT(CH4)-FFT(CH1) was still +3dB.

What is more believable, that a $30 70MHz probe has BW over 650MHz (albeit not very flat), or is there something strange going on with this physical arrangement? 

[iMo]

Does the scope subtract FFT(CH3) - FFT(CH1) as complex values?
PS: I would download both channels FFTs data (or direct CH1 and CH2 waveforms) into python (or similar) and do the math there..

[BillyO]

The scope does not do the subtraction.  That's done manually and only real peak values are used to create the FFT curve.  Only peak voltage is being taken into consideration so there are no complex numbers to deal with.

[iMo]

The "peak values" as you wrote are the amplitudes of the spectral lines, but they have got a "phase" as well, so math with two ffts have to be done with complex representations of each spectral bin, imho. So the measurement you want to do is basically a Bode, that math has to be done with complex numbers..

[2N3055]

therefore FFT(CH3)-FFT(CH1) [in dB] is the log magnitude of the frequency response of the probe

The problem with that is in 2N3055's screen shot FFT(CH3)-FFT(CH1) is positive starting at about 210MHz and increases to about +13dB at 420MHz and is still +5dB at nearly 500MHz.  When I ran it I ran it on a faster scope and ran it out to 650MHz and FFT(CH4)-FFT(CH1) was still +3dB.

What is more believable, that a $30 70MHz probe has BW over 650MHz (albeit not very flat), or is there something strange going on with this physical arrangement? 

That is not 70 MHz probe.
That is Testec HF212, a 300MHz BW probe.
Sorry if I was not clear.
I don't have any probes that usually ship with 800xHD.
From Siglent passive probes I only have 500MHz ones...
I could try with 60MHz probe from Picoscope that came with 4262.
But it is not important.

I was not showing you probe response but layout of measurement.
And the fact that this measurement is more than capable to measure that 70MHz probe.

And also there is a point that passive probes are complex schematic, that will on one side load your circuit, but also might have peaking where it can extend BW of scope itself.

And yes, if source impedance is low enough, 500MHz passive probes can sometimes go to almost a 1 GHz BW...
In fact, at some point they might start "shooting through" where they start capacitivelly coupling to the output, decreasing attenuation..

[gf]

The "peak values" as you wrote are the amplitudes of the spectral lines, but they have got a "phase" as well, so math with two ffts have to be done with complex representations of each spectral bin, imho. So the measurement you want to do is basically a Bode, that math has to be done with complex numbers..

The FFT trace shows only the magnitude (or log magnitude, depending on the selected scale) of the complex FFT result. It does not expose phase. But keep in mind that abs(a/b) = abs(a) / abs(b). So dividing the (real-valued) magnitudes is equivalent to taking the magnitude of the complex quotient. And with log magnitudes (decibel scale), the division turns into a subtraction, since log(x/y) = log(x) - log(y). So it is fine as long as you don't need the phase response as well.

[eTobey]

Is there an indication, when the opamp is beeing overdriven? Cant find anything about it in the manual.

[2N3055]

Is there an indication, when the opamp is beeing overdriven? Cant find anything about it in the manual.

?
Wrong place, maybe?

[eTobey]

No, just a lack of terms...

Since it has quite an impact on measurement, when a signal goes off the screen (above or below / but not necessarly inside the horizontal limits of the screen), it would be good to see that with an indication, as it could be missed, when moving around (the visible portion) on a signal.

[mawyatt]

On another SDS800 thread (see #340) Bill was investigating probe & DSO effects.

https://www.eevblog.com/forum/testgear/sds800x-hd-bug-reports-firmware/msg5757649/#msg5757649

This got our curiosity and we did a couple quick measurement setups wrt the probe performance an it's "effect on the DUT".

Setup is crude (not technically correct and calibrated) but good enough to show the effects. An AWG signal generator (SDG6022X) is set to 0dBm and swept from 1MHz to 300MHz. For the AWG 50Ω termination we utilized the input of a Spectrum Analyzer (SSA 3121X +) so we could "see" the probe effects on the SA since it's now the DUT for the DSO probe. A BNC "T" was used to allow the DSO Probe BNC Probe Adapter to be inserted and the various test probes inserted into the adapter.

The SA was set to 1MHz to 300MHz range, peak detect, and +2dBm Reference level with 1dB/div. A reference was captured to show the results without the DSO probe and should show ~0dBm across the 300MHz span, see PNG0.

The SDS800X HD was setup with an FFT of similar parameters, 1MHz -300MHz Span, +2dBm reference and 1dB/div Scale. Other FFT parameters, Flat Top Window and 1us/div TimeBase. CH1 was set to 10X for probes and 100mv/div since 0dBm equates to ~316mv peak (note: don't saturate the DSO Input).

PNG112 shows SDS800 result with PP510 Probe and PNG1 shows SA result.

PNG113 is with SP3020A Probe, SA result is PNG2.

PNG 114 shows Rigol PVP3150 Probe with PNG3 SA result.

Note the "Effect" the probes have on the DUT (Spectrum Analyzer acting as DUT) which is different than what's displayed on the DSO as the back end of the probes and DSO characteristics influence the measurement at this stage as shown by the DSO.

One important point we should mention. Our SA (SSA 3121X +) has a Tracking Generator (TG) output which one might consider utilizing with these measurements as the source instead of the AWG. This particular SA "Normalizes" the SA result from the TG by taking a "Reference Sweep" which is used to Compensate for the TG amplitude variations across the frequency sweep. The result shows on the SA Display as a nice level amplitude across the entire sweep implying a constant level from the TG!! This is NOT the case here as the TG hasn't been leveled or compensated in any way, just the SA Display is Normalized!! One can easily be fooled by this and think the SA TG has a constant level output

Dedicated RF/MW sweep generators usually have a proper leveled output, and suspect more expensive SA Tracking Generators also have a leveled output. Leveling involved actual measurement of the generator output and feeding back the result for corrections, AWGs generally don't do this and just rely on an accurate generation of the waveform without output feedback assuming a prefect load (either 50Ω or infinite).

Anyway, just thought we should mention this in case some folks might decide to use the SA TG as a source for DSO bandwidth measurements, here small amplitude variations shouldn't matter much since the DSO bandwidth is so highly influenced by the selected probe and the DUT characteristics.

Best and Happy Holidays

[mawyatt]

Added PCBite Sensepeek SQ200 10X DSO Probe.

Merry Christmas and Happy Holidays.

Best

[KungFuJosh]

Added PCBite Sensepeek SQ200 10X DSO Probe.

How do you like their probes? I've been thinking about grabbing the SQ500 for the hell of it, but I'm curious if it's worth having or not.

Thanks, and Happy Holidays!

[mawyatt]

The PCBite concept has served us well and is worth considering if you are probing complex PCBs and stuff, the magnetic Holding Discs work well as do the PCB Holders allowing hands off probing.

The DSO probes are nice, we only have the SQ200s tho.

Happy Holidays

Best

[mawyatt]

An important aspect of what's shown in #1838~9 is the effect the DSO probe has on DUT under measurement and shown by the Spectrum Analyzer, which happens to be the DUT in this case.

Note that the SA is showing a result which indicates the probe represents a more complex "load" to the DUT (SA in this case) than a simple shunt capacitance. This may cause the DUT to behave differently than expected and one should also realize that the result displayed on the DSO screen is not exactly what the DUT is experiencing due to the probe loading. In addition, the frequency response behavior can't be just related to a clock rate at the DUT for example, since the clock edges contain significant energy well beyond the clock rate and this must be considered

Edit: Starting a thread related to DSO probe effects at the DUT.

Happy Holidays,

Best

[eTobey]

It has LAN, it has internal storage, it stores a lot in it, but it cant capture mask tests screenshots in the internal storage? 

I consulted the manual, but nothing could help there.

[Furna]

Indeed Page 208 of the manual read
"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".

Didn't test yet, but I guess you can only save to internal what is recallable => Setup and Reference (table at page 209).
Also Page 211 read
"Save the screenshot to the path “U-disk0\sds800xhd\” in the format of BMP
First, Insert the U disk."
IMHO worth to report to Siglent the error in documentation if this is confirmed.

@eTobey: you already setup your client PC as an NTP server ... consider the idea to set it up as an SMB server (just enable File Sharing feature of Windows).