Siglent SDG1032X - Sweep does steps of frequency

[eTobey]

On the picture, you can see, that in the beginning, the frequency is constant for about 1s. (notice the constant artefacts, that show a constant frequency from '-1s' to '0s')

It is a 3s sweep from 100Hz to 30MHz.

Is this a normal behaviour of such devices? I would have expected the frequency to rise without steps.

(Yes, i know, its a noobish question, but i accidently posted it in the wrong section, and cant delete it.)

[2N3055]

On the picture, you can see, that in the beginning, the frequency is constant for about 1s. (notice the constant artefacts, that show a constant frequency from '-1s' to '0s')

It is a 3s sweep from 100Hz to 30MHz.

Is this a normal behaviour of such devices? I would have expected the frequency to rise without steps.

(Yes, i know, its a noobish question, but i accidently posted it in the wrong section, and cant delete it.)

I have an advice for you.
Every time you have a problem like this, ask yourself what did you expect to happen. And how would you do it if you were to make a such device.

Let's turn the thing upside down. From 100Hz to 30MHz in 3 seconds....
Is it log or linear sweep?
If linear, that is roughly 10MHz/s sweep.

At time:
zero +1s you are at 10MHz (100ns period)
zero +100ms you are at 1MHz (1µs period)
zero +10ms you are at 100Khz....

Wait...
At zero+10ms you are still at first period of 100Hz .... hm...

How can it beeee... Maybe we cannot ask AWG to generate mathematically/physically  impossible frequencies....

[eTobey]

Its a log sweep.

I dont understand what you try to explain. I can not imagine impossible frequencys, apart from 0 or negative ones.

[nctnico]

Its a log sweep.

Log sweeps are typically implemented as a series of steps on AWGs and thus skipping frequencies. This can lead to rather odd results like partial sine waves at low frequencies. But still, I would expect the image to show a more constant amplitude. Did you turn peak-detect on on the scope and use a BNC cable?

[2N3055]

Its a log sweep.

I dont understand what you try to explain. I can not imagine impossible frequencys, apart from 0 or negative ones.

Are you one of those people that think that if you can write something in a mathematical formula then it has to exist in physical world?

Please create me 100Hz signal that lasts 100ns.

[nctnico]

Its a log sweep.

I dont understand what you try to explain. I can not imagine impossible frequencys, apart from 0 or negative ones.

Are you one of those people that think that if you can write something in a mathematical formula then it has to exist in physical world?

Please create me 100Hz signal that lasts 100ns.

You are wrong here. You can easely create part of a 100Hz sine wave which lasts 100ns. You just won't get a full cycle (as math dictates). And that is exactly what an AWG will do when instructed to make a sweep. During linear sweep, the frequency will be varied continuously (resulting in creating only small fractions of sine waves at low frequencies; but still mathematically correct). Log sweeps are more difficult to implement on an AWG and many will output frequency bursts (containing partial waves for low frequencies) to step through the frequencies and thus skipping frequencies.

[Joe1vm]

Hi.
It looks familiar..... https://www.eevblog.com/forum/testgear/siglent-sdg1000x-waveform-generators/msg5288380/#msg5288380

But I still can not understand such amplitude variation across frequency range....

[eTobey]

Are you one of those people that think that if you can write something in a mathematical formula then it has to exist in physical world?

Please create me 100Hz signal that lasts 100ns.

Thought about just creating it for you, but then thought, na, you are not worth it...

But I still can not understand such amplitude variation across frequency range....

This variation is coming from a little circuit, but thats of no relevance.

[KungFuJosh]

In case you don't know, this is the thread for that AWG where all the topics actually belong: https://www.eevblog.com/forum/testgear/siglent-sdg1000x-waveform-generators/

[2N3055]

Are you one of those people that think that if you can write something in a mathematical formula then it has to exist in physical world?

Please create me 100Hz signal that lasts 100ns.

Thought about just creating it for you, but then thought, na, you are not worth it...

 
Familiar ??

Again both Nico and you didn't understand a single thing, but it never stopped you (or him) from being wrong...

[gf]

No, it really is you who is wrong all the time.

A sweep is a simple formula: output = amplitude * sin (f * t * ω)
t is incremented for each sample interval as a fraction of ω, f is incremented for each sample interval to create the sweep from the start to end frequency. If you look carefully at the formula, you'll notice that you won't get a full cycle for any of the freqencies at all when f is not constant.

Exactly. For details, see also https://en.wikipedia.org/wiki/Chirp

[2N3055]

No, it really is you who is wrong all the time.

A sweep is a simple formula: output = amplitude * sin (f * t * ω)
t is incremented for each sample interval as a fraction of ω, f is incremented for each sample interval to create the sweep from the start to end frequency. If you look carefully at the formula, you'll notice that you won't get a full cycle for any of the freqencies at all when f is not constant.

Create that chirp (it is not a sweep) and feed it into scope, do a FFT and tell me what you see?

There is a reason when we do sweep that we actually specify dwell time at frequency to enable time for equipment to detect the frequency in full amplitude.

Frequency chirp are usefull but for other things like radars...

[eTobey]

I asked a question here, that you did not really answer. Your answer was more offending than satisfying.

Create that chirp (it is not a sweep) and feed it into scope, do a FFT and tell me what you see?

There is a reason when we do sweep that we actually specify dwell time at frequency to enable time for equipment to detect the frequency in full amplitude.

I see no contribution here from you, so you can either contribute by giving explanations or answers in a useful and understandable way, or just dont write anything at all.

[gf]

Create that chirp (it is not a sweep) and feed it into scope, do a FFT and tell me what you see?
There is a reason when we do sweep that we actually specify dwell time at frequency to enable time for equipment to detect the frequency in full amplitude.

A traditional continuous Glide-Sweep is basically the same as a Chirp.
However, many modern spectrum analyzers do a Step-Sweep, measuring a finite number of discrete frequency points.

Since you cannot configure the number of discrete frequency points, I would expect that the SDG1000X generates a continuous Glide-Sweep. And where can you configure a dwell time in the SDG1000X? I could not find anything in the manual. IMO that's only relevant for a step-sweeping SA + TG (or for a NA). If you do a Bode Plot, then I do not think that the scope configures the SDG to sweep, but more likely it tunes the SDG to each frequency point individually.

If you manage to fit the whole interval of the chirp exactly into the FFT window then you should see a Chirp Spectrum. Although the waveform has a constant peak amplitude in the time domain, the spectrum is not flat in the frequency domain, but shows Fresnel ripples.

EDIT: If you would do a STFT instead, then you could see the frequency as a function of time in the spectrogram.

[gf]

On the picture, you can see, that in the beginning, the frequency is constant for about 1s. (notice the constant artefacts, that show a constant frequency from '-1s' to '0s')

It is a 3s sweep from 100Hz to 30MHz.

Is this a normal behaviour of such devices? I would have expected the frequency to rise without steps.

(Yes, i know, its a noobish question, but i accidently posted it in the wrong section, and cant delete it.)

1) You sweep up to 30 Mhz and use a sample rate of only 20 MSa/s? The signal is undersampled.

2) Don't expect the math function Freq() to give reliable results with a chirp/sweep. My understanding is that it does not measure instantaneous frequency, but it attempts to find full periods in the time domain and reports 1/duration_of_period.

[2N3055]

Create that chirp (it is not a sweep) and feed it into scope, do a FFT and tell me what you see?
There is a reason when we do sweep that we actually specify dwell time at frequency to enable time for equipment to detect the frequency in full amplitude.

A traditional continuous Glide-Sweep is basically the same as a Chirp.
However, many modern spectrum analyzers do a Step-Sweep, measuring a finite number of discrete frequency points.

Since you cannot configure the number of discrete frequency points, I would expect that the SDG1000X generates a continuous Glide-Sweep. And where can you configure a dwell time in the SDG1000X? I could not find anything in the manual. IMO that's only relevant for a step-sweeping SA + TG (or for a NA). If you do a Bode Plot, then I do not think that the scope configures the SDG to sweep, but more likely it tunes the SDG to each frequency point individually.

If you manage to fit the whole interval of the chirp exactly into the FFT window then you should see a Chirp Spectrum. Although the waveform has a constant peak amplitude in the time domain, the spectrum is not flat in the frequency domain, but shows Fresnel ripples.

EDIT: If you would do a STFT instead, then you could see the frequency as a function of time in the spectrogram.

1000X cannot do dwell.

Fast continuous glide sweep will create linear chirp.

In Bode Plot there are discrete frequencies stepping, you are right.

IF you manage to fit whole chirp exactly into window you WILL get Chirp spectrum.
You will not get same result as people expect to see with slow sweep and FFT in peak hold mode.

That is why I said not to confuse chirp burst with frequency sweep. I wanted to point out that our equipment need time to detect frequency with sufficient amplitude to be actually detected, and to have enough time to achieve full amplitude for us to be able to verify the right amplitude. Which is what you do when you sweep. So it is either dwell or slow enough so window can grab enough   energy in its bin to do the good job. That is if you use FFT. You can simply use long capture and look at it in time domain. Where also you need at least one full period in all frequencies in question to be able to see what was it's P-P amplitude at that frequency.

And yes, on 3000xHD and 7000A by using Gated Acquisition you can traverse chirp and see how frequency changes.
You can also see frequency change with track plot, but you need to sample properly...