Siglent SDS800X HD Review & Demonstration Thread

[joeqsmith]

Showing the two peaks closes enough that the readouts will no longer resolve it. 

[joeqsmith]

Watched Dave's recent Rigol DHO800 review, the connector SNAFU and the oops retraction.  Part of this video he is comparing with a low end Siglent off and on.  Puts it in a bad light but I assume that was because he was beating the price drum.   

[mawyatt]

Channel 1 & 2 fed with a 1MHz 0dBm sinewave.   Math used to add both channels.  FFT then ran on the math channel.  This is about the limit of the old LeCroy.  It can display the separate peaks but the readouts will only display a single peak.  Again, wrong tool for the job but I am curious how well your scopes can resolve two peaks like this.  Feel free to use an external combiner rather than the math.

For the classic Two Tone IMD test you want to exercise the channel under test with the full 2 tone signals combined which implies summing the 2 tone before sampling the summation, then following with an FFT.

Would expect a better result summing the 2 tones digitally after digitization with the math function from 2 separate digitized channels, than combining the 2 tones, then digitizing the combination followed by an FFT.

Should be a simple test to perform, altho we haven't done the 2 channel version.

Best,

[tautech]

Watched Dave's recent Rigol DHO800 review, the connector SNAFU and the oops retraction.  Part of this video he is comparing with a low end Siglent off and on.  Puts it in a bad light but I assume that was because he was beating the price drum.   

At that time Joe, Dave had zero idea of what was around the corner and coming from Siglent, zip, nada, zero.

Therefore he could only compare it with existing Siglent models at a similar price point and probably had the 4ch X-E in his mind as a comparison.

[2N3055]

Showing the two peaks closes enough that the readouts will no longer resolve it.

SD6052X - Digital combiner, cca 1MHz combined with 1MHz+11,9Hz difference, -125dBm level.
I'm not really sure that some 0.5dB difference is not coming from a foot long attenuator stack...

4M FFT on scope. Von Hann Window

Full disclosure: first and second tone frequency and distance were fine tuned to land in the middle of frequency bin..
So this is as ideal as it can get.

[gf]

Channel 1 & 2 fed with a 1MHz 0dBm sinewave.   Math used to add both channels.  FFT then ran on the math channel.

If both are 1MHz, why do you get two peaks? Is that due to inter-instrument error between two separate signal generators?
[With a 2-channel AWG, I would expect the signals to be phase-locked if the frequency is the same.]

For the classic Two Tone IMD test you want to exercise the channel under test with the full 2 tone signals combined which implies summing the 2 tone before sampling the summation, then following with an FFT.

Would expect a better result summing the 2 tones digitally after digitization with the math function from 2 separate digitized channels, than combining the 2 tones, then digitizing the combination followed by an FFT.

Digital summation with a math function does, of course, not introduce any IMD between the two signals. You just get the sum of the two (complex) spectra, since FFT(x+y) = FFT(x) + FFT(y).

[mawyatt]

Digital summation with a math function does, of course, not introduce any IMD between the two signals. You just get the sum of the two (complex) spectra, since FFT(x+y) = FFT(x) + FFT(y).

Exactly!! The separate channel digitized signals can't "see" each other, thus no IMD should be present and why we mentioned such.

Best,

[pdenisowski]

Regarding the IMD tests:  I'm not sure I completely understand what is being measured / quantified about the scope here.

In my experience (and certainly in the case of spectrum analyzers and VNAs), IMD is most often measured by providing the DUT with an input signal that consists of two tones at different frequencies (F1 and F2) and then measuring the amplitudes of the close-in 3rd order products (2F1-F2 and 2F2-F1) at the DUT output relative to the amplitude of the fundamentals.  This is the used to compute the theoretical intercept of the slope of the fundamental and 3rd order products and yields the third order intercept point (IP3 or TOI).  A higher IP3 is generally desirable because it indicates greater linearity of the DUT, i.e. less IMD created by the DUT.

If I wanted to quantify the IP3 of a scope's front end, I would simply generate two tones (ideally using completely separate sources and circulators) with slightly (~1 MHz or so) different frequencies, combine them externally, and then feed that combined input signal into a single scope channel.  Then I would perform an FFT of the signal and measure the difference in amplitude between the fundamentals and above-mentioned 3rd order products and do the math (IP3 = Ptone + deltaP/2).

I can understand why we might want to quantify the linearity of a scope's front end (vertical amplifier, etc.) but I don't think I understand the methodologies / results I'm seeing here.    Sorry if I'm being dense

[egonotto]

Digital summation with a math function does, of course, not introduce any IMD between the two signals. You just get the sum of the two (complex) spectra, since FFT(x+y) = FFT(x) + FFT(y).

Exactly!! The separate channel digitized signals can't "see" each other, thus no IMD should be present and why we mentioned such.

Best,

Hello,

If I understand correctly, the combined signal is fed into C1 and the FFT would already show the interference of the input circuit and the ADC.

Best regards
egonotto

[mawyatt]

Hello,

If I understand correctly, the combined signal is fed into C1 and the FFT would already show the interference of the input circuit and the ADC.

Best regards
egonotto

Yes

However one must be careful when combining the signals to not introduce IMD due to interaction between the two signal sources before they enter the DUT, usually at smaller levels and with sufficient source isolation (>6dB) this isn't an issue.

Best,

[Performa01]

The first thing I noted was the seemingly random usage of window functions. Hanning and Gaussian are not comparable and for a meaningful test we should stick with the Flattop window on both instruments.

-snip

I was aware about that but the focus was on spurious so I spent very little time on that so I chose the window type on the base of the peak shape well knowing to be in "uncal" condition, but to be honest ... i do not know how to select flattop on SA ... you sure it's possible with my model ? (SSA3021X)

I understand. Flattop is the default window on every general purpose SA, because amplitude accuracy and dynamic range (side lobe suppression) are the most important criteria. Most have some additional window (Gaussian, Nutall) and a few have even more to choose from. But for continuous signal analysis, where we can acquire plenty of data, the Flattop window is the obvious choice.

Watched Dave's recent Rigol DHO800 review, the connector SNAFU and the oops retraction.  Part of this video he is comparing with a low end Siglent off and on.  Puts it in a bad light but I assume that was because he was beating the price drum.   

Yeah, some time ago I’ve analysed the FFT part and thoroughly disproved any unwarranted claims about the inferiority of the old little Siglent (reply #200):

https://www.eevblog.com/forum/testgear/siglent-sds1104x-e-in-depth-review/200/

Channel 1 & 2 fed with a 1MHz 0dBm sinewave.   Math used to add both channels.  FFT then ran on the math channel.  This is about the limit of the old LeCroy.  It can display the separate peaks but the readouts will only display a single peak.  Again, wrong tool for the job but I am curious how well your scopes can resolve two peaks like this.  Feel free to use an external combiner rather than the math. 

I assume you meant two sine waves with close to 1 MHz frequency. Using the math channel for adding the two signals is a great way to simplify the cabling and avoid the 6 dB insertion loss.

The SDS9824X HD does a reasonable job; its Marker Table doesn’t have the necessary frequency resolution and accuracy is generally poor because of the cheap 25 ppm quartz timebase (and it shows), but here you go:


SDS824X HD_FFT_1MHz_Span500Hz_d5Hz

In the above screenshot, a 999995 Hz sine was combined with a 1 MHz sine, hence the difference between the two signals is 5 Hz. The two signals are just distinguishable on the screen, and peaks search had to be configured for 12 dB excursion (instead of the 20 dB default) in order to find the two separate peaks. As mentioned before, Frequency resolution is insufficient, but at least the level accuracy is nothing to complain about.

An SA would be better suited for such extreme use cases, but if it has to be a DSO, then why do I even try to compete with an upper midrange instrument from LeCroy by using a low-end entry-level device like the SDS800X HD? (Answer: because this is a thread about the SDS800X HD)

The vintage of a LeCroy isn’t really important in this context. The extreme applications demonstrated here certainly exceed the capabilities of many other expensive scopes – but not for a midrange LeCroy, no matter how old it is.

With all that said, I pull out a worthy contender for a change:


SDS7404A H12_FFT_1MHz_Span100Hz_d0.5Hz

In the above screenshot, a 999999.5 Hz sine was combined with a 1.0000000 MHz sine, hence the difference between the two signals is 0.5 Hz or 0.5 ppm. The two peaks are just distinguishable on the screen, and excursion for peaks search could be more than 50 dB. Displayed frequency resolution is sufficient here, but the actual resolution is not, as the 0.5 Hz is displayed as 0.79 Hz. Frequency accuracy is in the realm of 1 ppm and amplitude accuracy better than 0.15 dB in this scenario.

[joeqsmith]

SD6052X - Digital combiner, cca 1MHz combined with 1MHz+11,9Hz difference, -125dBm level.
I'm not really sure that some 0.5dB difference is not coming from a foot long attenuator stack...

4M FFT on scope. Von Hann Window

Full disclosure: first and second tone frequency and distance were fine tuned to land in the middle of frequency bin..
So this is as ideal as it can get.

Thanks for checking.  Very nice.   Here using an external combiner.  The LeCroy has trouble resolving the two peaks once you get below 10Hz separation.   

As a side note, I recently installed the latest firmware available.   While running this test, the scope threw some internal software error and had to restart.    I've owned this scope for several years and have never had the firmware crash.  Not impressed.

[mawyatt]

Regarding the IMD tests:  I'm not sure I completely understand what is being measured / quantified about the scope here.

In my experience (and certainly in the case of spectrum analyzers and VNAs), IMD is most often measured by providing the DUT with an input signal that consists of two tones at different frequencies (F1 and F2) and then measuring the amplitudes of the close-in 3rd order products (2F1-F2 and 2F2-F1) at the DUT output relative to the amplitude of the fundamentals.  This is the used to compute the theoretical intercept of the slope of the fundamental and 3rd order products and yields the third order intercept point (IP3 or TOI).  A higher IP3 is generally desirable because it indicates greater linearity of the DUT, i.e. less IMD created by the DUT.

If I wanted to quantify the IP3 of a scope's front end, I would simply generate two tones (ideally using completely separate sources and circulators) with slightly (~1 MHz or so) different frequencies, combine them externally, and then feed that combined input signal into a single scope channel.  Then I would perform an FFT of the signal and measure the difference in amplitude between the fundamentals and above-mentioned 3rd order products and do the math (IP3 = Ptone + deltaP/2).

I can understand why we might want to quantify the linearity of a scope's front end (vertical amplifier, etc.) but I don't think I understand the methodologies / results I'm seeing here.    Sorry if I'm being dense

Probably a little guilty here as we "thought" the two tones were for evaluating the DSOs linearity as with chassis classic IMD test. Rereading the post that started this discussion, now not sure that was the intent as it could be the separation of the 2 tones in the FFT display and how well the DSO could resolve such.

Your description of the 2 Tone IMD test is spot on and much easier to implement than sweeping the input signal level and plotting the 3rd harmonic wrt fundamental to find the TOI, altho we've done both!!

We've found that scope linearity is often overlooked by folks, as they seem more intrigued with number of bits, BW and noise levels rather than linearity.

Best,

[pdenisowski]

I understand. Flattop is the default window on every general purpose SA, because amplitude accuracy and dynamic range (side lobe suppression) are the most important criteria. Most have some additional window (Gaussian, Nutall) and a few have even more to choose from.

At least on R&S spectrum analyzers, flattop is not necessarily the default window, depending on measurement mode.

Also, I would expect any spectrum analyzer or oscilloscope FFT implementation to have a certain minimum set of user-selectable windows: rectangular ("no window"), flattop, Blackman(-Harris), and Gaussian.  Hann, Hamming, Exponential, and Kaiser are also not uncommon.

[2N3055]

I understand. Flattop is the default window on every general purpose SA, because amplitude accuracy and dynamic range (side lobe suppression) are the most important criteria. Most have some additional window (Gaussian, Nutall) and a few have even more to choose from.

At least on R&S spectrum analyzers, flattop is not necessarily the default window, depending on measurement mode.

Also, I would expect any spectrum analyzer or oscilloscope FFT implementation to have a certain minimum set of user-selectable windows: rectangular ("no window"), flattop, Blackman(-Harris), and Gaussian.  Hann, Hamming, Exponential, and Kaiser are also not uncommon.

Rectangular ("no window"), flattop, Blackman,  Von Hann and Hamming are on all Siglent scopes.

[pdenisowski]

Probably a little guilty here as we "thought" the two tones were for evaluating the DSOs linearity as with chassis IMD test. Rereading the post that started this discussion, now not sure that was the intent as it could be the separation of the 2 tones in the FFT display and how well the DSO could resolve such.

Thanks - sometimes the hardest part of T&M is people using different methodologies / terminologies to describe the same thing

I would assume the separating (or "resolving") two adjacent tones in spectrum is simply a function of the FFT resolution bandwidth.

Your description of the 2 Tone IMD test is spot on and much easier to implement than sweeping the input signal level and plotting the 3rd harmonic wrt fundamental to find the TOI, altho we've done both!!

Ouch   The sweeping methodology sounds painful.

We've found that scope linearity is often overlooked by folks, as they seem more intrigued with number of bits, BW and noise levels rather than linearity.

Indeed it is. The other thing that I think often gets overlooked is spurious emissions (spurs) generated within the instrument itself.

[mawyatt]

Hello,

If I understand correctly, the combined signal is fed into C1 and the FFT would already show the interference of the input circuit and the ADC.

Best regards
egonotto

Yes

However one must be careful when combining the signals to not introduce IMD due to interaction between the two signal sources before they enter the DUT, usually at smaller levels and with sufficient source isolation (>6dB) this isn't an issue.

Best,

Maybe of interest is some AWGs have the ability to "combine" the 2 channels internally, this could be done in "Digital Domain" before the DACs and channel amps/attenuators, or after the DACs in the "Analog Domain".

Each would have advantages/disadvantages but both techniques could introduce IMD! Even in the "Digital Domain" combining since the combined digital summed signal then undergoes Digital to Analog conversion and subsequent amplification/attenuation before reaching the AWG output.

Best,

[2N3055]

Hello,

If I understand correctly, the combined signal is fed into C1 and the FFT would already show the interference of the input circuit and the ADC.

Best regards
egonotto

Yes

However one must be careful when combining the signals to not introduce IMD due to interaction between the two signal sources before they enter the DUT, usually at smaller levels and with sufficient source isolation (>6dB) this isn't an issue.

Best,

Maybe of interest is some AWGs have the ability to "combine" the 2 channels internally, this could be done in "Digital Domain" before the DACs and channel amps/attenuators, or after the DACs in the "Analog Domain".

Each would have advantages/disadvantages but both techniques could introduce IMD! Even in the "Digital Domain" combining since the combined digital summed signal then undergoes Digital to Analog conversion and subsequent amplification/attenuation before reaching the AWG output.

Best,

And what I have shown was exactly that: digital Channel Combine in SDG6000X.
It is not as good as it would be with good resistive combiner, but plenty good enough for many uses.
All you need to know and be aware of is what are the limits...

[mawyatt]

Ouch   The sweeping methodology sounds painful.

Not so bad if you have Grad Students do all the work


"Indeed it is. The other thing that I think often gets overlooked is spurious emissions (spurs) generated within the instrument itself."

Agree, spurs are bad in any instrument, especially SA, however since these DSO are being utilized as a low frequency SA they are particularly bad here as with SA.

The absolute levels being shown and discussed are remarkable for any DSO much less at this price point, and this low level begins to reveal all that's underneath the Time Domain noise floor that's generally not viewable with typical scope usage!!

Best,

[joeqsmith]

Probably a little guilty here as we "thought" the two tones were for evaluating the DSOs linearity as with chassis classic IMD test. Rereading the post that started this discussion, now not sure that was the intent as it could be the separation of the 2 tones in the FFT display and how well the DSO could resolve such.

Sorry it wasn't clear but I was only curious about the scopes ability to resolve the two peaks.

We've found that scope linearity is often overlooked by folks, as they seem more intrigued with number of bits, BW and noise levels rather than linearity.

I had attempted to use my scope to measure AC line power by oversampling and decimation.  In this case I drove the channels with a precision DAC and ran a least squares.  Used the error to create a model for each channel.  Again, not the best tool for the job but may provide some idea how the measurement could possibly be improved.   First 7 minutes covered the setup.

Guessing the new scopes would do a much better job of making these measurements.

[joeqsmith]

.... why do I even try to compete with an upper midrange instrument from LeCroy by using a low-end entry-level device like the SDS800X HD? (Answer: because this is a thread about the SDS800X HD)

I show the LeCroy to better explain what I am asking to see.  Some of these low cost scopes are very impressive. 

The vintage of a LeCroy isn’t really important in this context. The extreme applications demonstrated here certainly exceed the capabilities of many other expensive scopes – but not for a midrange LeCroy, no matter how old it is.

With all that said, I pull out a worthy contender for a change:
...
SDS7404A
...
In the above screenshot, a 999999.5 Hz sine was combined with a 1.0000000 MHz sine, hence the difference between the two signals is 0.5 Hz or 0.5 ppm. ...

The 7404 is very tempting except it's not good enough to replace my old WaveMaster and it's way overkill for what I am looking for.   After looking at what they offer, I am starting to lean toward the SDS2354X HD (?).   It appears to offer some improvements, but I also give up in other areas.   I would like to see where someone had ran this thing using the Ethernet interface.  I've not found anyone showing any custom applications for it.   Have you written custom software to interface with any of your Siglent products?   If so, I would be interested in hearing how it went.

[Martin72]

Today, at work...
Lecroy Waverunner 9054(8bit), HDO6034A(12bit).
FFT with 100khz span, source Siglent SDG1062X, 50Mhz, 0dBm, 50Ohm.

Maybe we should have the last pages separated and moved to a new thread, title suggestion "FFT comparison and settings".

[joeqsmith]

Maybe we should have the last pages separated and moved to a new thread, title suggestion "FFT comparison and settings".

That's fine.  Personally I was really just asking how the low cost Siglent dealt with these cases I have brought up.

[Martin72]

This is also a very exciting topic that "deserves" its own thread.
Let's see if a moderator reacts, then he can also take this post with him.
Again 50Mhz 0dBm, Siglent 824X HD, 50Ohm termination.
Span 10khz, so far so clear.
I have combined 2 tracks, the current one is at a vertical setting of 110mV/div (C1), the REF B track(purple) is at 100mV/div.(C1)

[2N3055]

Today, at work...
Lecroy Waverunner 9054(8bit), HDO6034A(12bit).
FFT with 100khz span, source Siglent SDG1062X, 50Mhz, 0dBm, 50Ohm.

Maybe we should have the last pages separated and moved to a new thread, title suggestion "FFT comparison and settings".

Sorry Martin,

just a last one. 
Similar to LeCroys: 50 MHz, 0dBm,20dB/div, 100kHz span, 83Hz RBW , FlatTop W.
200MHz BW limit on (HDO6034 is not 1GHz)
I think it looks very good. Noise floor is good 10 dB better.