Siglent SDG6000X series 200-500 MHz AWG's

[Detlev]

I mean, the R&S is in noise figures any better than the used SSA2K plus, also the close in noise figures any different.

can you say 100% that it is the spectrum analyzer or could it also be the SDG?

[hpw]

I mean, the R&S is in noise figures any better than the used SSA2K plus, also the close in noise figures any different.

can you say 100% that it is the spectrum analyzer or could it also be the SDG?

The Siglent SSA has this limitations, even you connect an 10MHz OXCO as a reference, the used build in freq. synthesizer as walking with close-in spurs variations, do not gets any better. Already posted SSA results on this forum.

In other words, getting what you payed for... as it goes for the the SSA & SDG2K. So my interest are on SDG6K and SDG7K how they performs.

So my current conclusion is now, building a table by simulations and than compare the results with the HW gears.

Hp

[tautech]

What do you need to see chaps ?
SDG6022X and analysers to 8.5 GHz here.

[Detlev]

I mean, the R&S is in noise figures any better than the used SSA2K plus, also the close in noise figures any different.

I'm not sure what you're getting at in the discussion right now. Regarding the phase noise, I posted a screenshot above, attached again here. that with a pure sine without modulation.
You can also do this with the SSA2K plus, then we would have a comparison between the SDG6022X with the R&S FPL and the SDG2000X and the SSA2K plus.

What is wrong with the modulation for you, do you have concrete examples of what you would expect? Maybe it's me that I just don't understand you

Greetings and Happy Easter
Detlev

[hpw]

You can also do this with the SSA2K plus, then we would have a comparison between the SDG6022X with the R&S FPL and the SDG2000X and the SSA2K plus.

What is wrong with the modulation for you, do you have concrete examples of what you would expect? Maybe it's me that I just don't understand you

Greetings and Happy Easter
Detlev

Look, I did many measurements using SSA2K Plus and SDG2K and a KS AWG. Also source from 10MHz Sine OXCO as -120dbc@1Hz. Complained Siglent about false Square wave FM & PM. At the end even the new FM & PM is in questions. So I asked for your SDG6K figures and they differ from my KS measurements.

The basic question is how the made the Square wave ... by math or using a sine to square wave circuit.

In addition the performance of the SSA is important, as low PN & Jitter & close-in noise & other spurs & low over all noise contributions.

I come now to the conclusions about the FM / PM as sine & square waves, just at first, to simulate various with high floating point precision as may the precision on the HW gear is may the limited factor.

My 2 cents

Hp

[Detlev]

If they differ from your KS measurements, how would you rate that? Good, bad, or just different? I find such discussions very interesting.

The FPL1000 has a phase noise of -105dBc (-108dBc typ.) at 1GHz/10kHz. At 1Mhz/100Hz it is better at -124dBc and at 1MHz/1kHz at around -132dBc. That's why I made my measurement at 1MHz.

At 10MHz and 100Hz offset it should be around -115 to -120dBc, that can only be estimated from the data sheet. These are values from the data sheet and are therefore better than the Siglent SAs.

Don't confuse the FPL1000 with the FPC1500, which has already been discussed here in the forum. I would also like an even better phase noise, but I didn't want to sell my house for an SA

I personally think the SDG6022X is very good for the price. A comparison with KS using specific examples would be nice.

Best regards
Detlev

[hpw]

If they differ from your KS measurements, how would you rate that? Good, bad, or just different? I find such discussions very interesting.

A comparison with KS using specific examples would be nice.

As it gets now, any different 

So first to do the homework with math and see about the beef... as looking about a dog 

As with egg and hen questions, as some did forget the rooster as only with them gets any valued eggs.

Hp

[2N3055]

You can also do this with the SSA2K plus, then we would have a comparison between the SDG6022X with the R&S FPL and the SDG2000X and the SSA2K plus.

What is wrong with the modulation for you, do you have concrete examples of what you would expect? Maybe it's me that I just don't understand you

Greetings and Happy Easter
Detlev

Look, I did many measurements using SSA2K Plus and SDG2K and a KS AWG. Also source from 10MHz Sine OXCO as -120dbc@1Hz. Complained Siglent about false Square wave FM & PM. At the end even the new FM & PM is in questions. So I asked for your SDG6K figures and they differ from my KS measurements.

The basic question is how the made the Square wave ... by math or using a sine to square wave circuit.

In addition the performance of the SSA is important, as low PN & Jitter & close-in noise & other spurs & low over all noise contributions.

I come now to the conclusions about the FM / PM as sine & square waves, just at first, to simulate various with high floating point precision as may the precision on the HW gear is may the limited factor.

My 2 cents

Hp

Again you are on your crusade to prove something (which is ok as a principle) but nobody understands what exactly you are trying to say.
Sorry, I'm just being honest and a bit direct.

As to answer to one of your questions I understand, both Keysight you are mentioning and Siglent AWG in question are AWG: Arbitrary waveform generators, that numerically construct signals from clocked digital values through ADC   DAC  to output amplifier. They are not modulated physical sources (like VCOs, disciplined or not) or pulse forming sources that form output signals through physical pulse forming networks...

[2N3055]

If they differ from your KS measurements, how would you rate that? Good, bad, or just different? I find such discussions very interesting.

A comparison with KS using specific examples would be nice.

As it gets now, any different 

So first to do the homework with math and see about the beef... as looking about a dog 

As with egg and hen questions, as some did forget the rooster as only with them gets any valued eggs.

Hp

I don't get the whole animal farm thing...

[hpw]

Again you are on your crusade to prove something (which is ok as a principle) but nobody understands what exactly you are trying to say.
Sorry, I'm just being honest and a bit direct.

As to answer to one of your questions I understand, both Keysight you are mentioning and Siglent AWG in question are AWG: Arbitrary waveform generators, that numerically construct signals from clocked digital values through ADC to output amplifier. They are not modulated physical sources (like VCOs, disciplined or not) or pulse forming sources that form output signals through physical pulse forming networks...

You mean DAC or not?

1) The AWG needs some math functions to fill up the internal buffer based on the user setup.
   Than this buffer is played in a cycle. Or they calculate the sample word on runtime, but I guess not.

2) How they build the FM/PM is for me unknown. As for sine, there exit clear math formulas.

3) For the square wave FM/PM the formulas are currently unknown to me. Or they use Sine FM/PM modulation and than convert to Square wave using any converter.

4) To compare each gear ends in a disaster...

5) better is to simulate first and compare the results (= the reference) with the various gear.

my 2 cents

[nctnico]

I guess the actual question is: how do you know the generator is calculating the waveforms correctly? From my experience with the SDG2000 series generator and judging from this thread, Siglent doesn't know the answer to this question for their own gear so it would be nice to be able to check for yourself what formulas are being used and which rounding errors there are.

[maxwell3e10]

A bit off-topic, but I thought it would be interesting how this spectrum and distortion compares to old-school instruments: HP 3326A and HP 89410A. It's not so bad, the harmonic and intermodulation distortions are much better, that is what HP3326 is designed for. The close-in phase noise is also better, while wideband noise is somewhat worse, about -114dBc/Hz (partly limited by 89410). HP3326 can of course do phase modulation of the square wave since it is generated from a sine.

[Detlev]

Hello everyone,

Today I did an FM modulation with a 100Hz sine wave. First I looked at what sidebands I would expect. To do this, I used the Bessel calculator

https://keisan.casio.com/exec/system/1180573474

assumed a modulation index of 0.01 and read the values for the 1-4 harmonics and converted them to dB.

So I have the following starting values:
Carrier: 10MHz / 0dBm
FM Frequency:  sine 100Hz
frequency Dev.: 1Hz
Mod index: 0.01

Expected amplitudes:
1. 0dBm
2. -46.02dB
3. -98.06dB
4. -153dB

Measured amplitudes:
1. -0.33dBm
2. -46.08dB
3. -98.12dB
4. gets lost in the noise

I think the SDG6022X is doing very well here.

Attached are the unmodulated and modulated measurements and a photo of the settings of the SGD6022X

have a nice day
Detlev

[Detlev]

Hi guys,

today I started with the modulation with a square wave. The parameters are identical to those of the sine modulation.

What would I expect?

If you now modulate with the square wave, then you modulate, if you use Fourier, with the fundamental wave and then with the harmonics of the square wave.

So I would await for the values of the fundamental from above and additionally the value of the 3rd harmonic of the square wave (I restrict to the 3rd harmonic since the next one would no longer be in the spectrum).

the 3rd harmonic is at 300Hz and has 1/3 of the amplitude of the fundamental of the square wave.

I enter this into the Bessel Calculator and get:

Order: 1
Initial value: 0
Increments: 0.0033333
Jv(x): 0.001666647685

with 1/3 the amplitude you get an amplitude of 1/3 Jv(x) and then you get 0.55555 x10^-3

Then convert to dB:
20log(0.55555x10^-3)= -65.11dB

So I would expect:

1. 0dBm
2. -46.08dB
3. -98.12dB
4. -65.11dB

Measured amplitudes:

1. 0.02dBm
2. -43.96dB
3. -92.67dB
4. -63.14dB

Here the measured values deviate from the expected ones. All of them are a bit too high.

If someone has an explanation for this, I would be very grateful. It's also possible that I've made a mistake in my reasoning, but I'm relatively sure that it fits.

It seems that the SDG does not calculate exactly during modulation and only interpolates with sufficient accuracy. This fits with sine but no longer exactly with square.

Attached is a photo of the SDG settings and a screenshot of the spectrum.

Greetings and good night
Detlev

[gf]

The FFT spectrum of a numerically simulated FM gives me numbers close to your measured values.
(Used only a 10kHz carrier for convenience, but the result can be shifted to any other carrier frequency)

Code: [Select]

fc = 10000    % carrier freq Hz (sine wave)
df = 1        % deviation Hz
fs = 100000   % sample rate Sa/s
fmod = 100
dp1 = 2*pi*(fc+df)/fs   % phase increment per sample for fc+df
dp2 = 2*pi*(fc-df)/fs   % phase increment per sample for fc-df
dp = [ones(1,fs/2/fmod)*dp1 ones(1,fs/2/fmod)*dp2];
dp = repmat(dp, 1, 10);
signal = sin(cumsum(dp));
% place frequency scale origin at fc
plot((0:length(signal)-1)*fs/length(signal)-fc,20*log10(abs(fft(signal)/length(signal)*2)))
grid on
ylim([-100 0])
xlim([-2000 2000])


[Detlev]

Then I'm probably not quite right with my approach of looking at the fundamental wave of the square-wave signal and its 3rd harmonic separately.

Do you also have decimal values of the simulation?

Unfortunately I don't have a way to simulate this

[gf]

Then I'm probably not quite right with my approach of looking at the fundamental wave of the square-wave signal and its 3rd harmonic separately.

Do you also have decimal values of the simulation?

Peaks:

0: -3.5717e-04
100: -43.922
200: -92.042
300: -63.006
400: -104.09
500: -71.875
600: -111.13
700: -77.713
800: -116.14
900: -82.069
1000: -120.02
...

Unfortunately I don't have a way to simulate this

GNU Octave?

[Detlev]

Your simulation matches my measurements perfectly. Thanks for the simulation.   

Then the SDG6022X calculates correctly.

I didn't know GNU Octave until now. Let's see if I can figure it out. I haven't been in this area yet.

[Detlev]

Peaks:

0:     -3.5717e-04
100: -43.922
200: -92.042
300: -63.006
400: -104.09
500: -71.875
600: -111.13
700: -77.713
800: -116.14
900: -82.069
1000: -120.02

I extended the measurement to the 11th harmonic and the values match your simulation almost perfectly

I lowered the carrier frequency to 1MHz because the setup is a little less noisy and you can see low values better. The screenshot is attached.

The SDG6022X shows its sunny side

It's great to see that the simulation goes so well with the measurement

Now it would be interesting to see comparative values to other AWGs.

I wish you a nice day
Detlev

[tv84]

It's great to see that the simulation goes so well with the measurement

Almost too good to be true!

[Detlev]

Almost too good to be true!

The structure consists only of the Siglent SDG6022X, the Rohde & Schwarz FPL1000 and the Sucotest cable.

There's not much that can go wrong   

[mawyatt]

It seems these SDG6000X AWGs are pretty good as shown in images #29 and #30.

Same goes for the little brother SDG2000X AWG as shown in #31 screen image.

Results are from a SSA3021X Plus, certainly not in the same class as the R&S FLP1000, but still respectable.

Edit: Added Image #32 which is for the SDG2000X AWG at 10MHz.

Best,

[Detlev]

It seems these SDG6000X AWGs are pretty good as shown in images #29 and #30.

Same goes for the little brother SDG2000X AWG as shown in #31 screen image.

Results are from a SSA3021X Plus, certainly not in the same class as the R&S FLP1000, but still respectable.

Edit: Added Image #32 which is for the SDG2000X AWG at 10MHz.

Best,

Thanks for your measurements. You confirm the simulation and also my measurements.

Siglent did a good job there.

Your measurements are also interesting for me because they give me a comparison between the FPL1000 and the SSA3021X. I also shortlisted the SSA3021X. In the end, it ended up being the FPL1000, which gave me many sleepless nights       I don't always spend so much money on a measuring device 

[tautech]

It seems these SDG6000X AWGs are pretty good as shown in images #29 and #30.

Same goes for the little brother SDG2000X AWG as shown in #31 screen image.

Results are from a SSA3021X Plus, certainly not in the same class as the R&S FLP1000, but still respectable.

Edit: Added Image #32 which is for the SDG2000X AWG at 10MHz.

Best,

Thanks for your measurements. You confirm the simulation and also my measurements.

Siglent did a good job there.

Your measurements are also interesting for me because they give me a comparison between the FPL1000 and the SSA3021X. I also shortlisted the SSA3021X. In the end, it ended up being the FPL1000, which gave me many sleepless nights       I don't always spend so much money on a measuring device 

Please ensure you select the Plus version.....more features, later improved UI and touch screen and mouse support for just a few $ more.

[tautech]

Customer wanted to see the performance boundaries of SDG6022X in Pulse mode.....things get hard to nicely capture at mV levels with noise mucking with the frequency counter.....it's @ 80 MHz like the 1V p-p pulse shows.