Signal behavior with function generator

[rvalente]

Hello Mates,

I own a FY6800 (feeltech), when the frequency is changed it has this characteristic to add a discontinuity in the signal phase.
Here a 5Vpp was changed from 500 to 1500Hz.

Is this expected or does Feeltech handle poorly the frequency generation?

How other function generators handle this? I may buy a DG821 very soon if it has a better continuity while changing freqs.

Greetings

[David Hess]

It just was not designed to be phase coherent.  That has to be done deliberately.  In most applications it is irrelevant.

[Anthocyanina]

I've set the oscilloscope as close to yours as i could from the information on the screenshot. The trigger details that I can't see in your screenshot have been set as shown in this screenshot. Then i set a list sweep on my generator(EDU33212A), it has 500 and 1500 Hz on the list, and plays each frequency for 1ms each. It clearly changes between each frequency and the phase is continuous. Hope this helps! 

[tautech]

They all do this when adjusted manually but not for a sweep which is what really matters.
Learn to properly use the feature set instead of focussing on something that was never part of the design brief.

[Anthocyanina]

I realised the frequencies shown in your screenshot were 2500 and 1500 hz, I've set the oscilloscope for a pulse trigger, single shot, the generator was outputting the lower frequency continuously, then i manually entered the higher frequency and got this single shot capture. The phase is also continuous when doing it manually in my case. I also did that first with the 500 and 1500 hz sines, and got the same result with the phase being continuous

[TimFox]

The "DDS" circuit (direct digital synthesis) uses an internal register that corresponds to phase.
That register is then mapped through an appropriate look-up table and DAC to obtain the output.
At constant frequency, each internal clock step adds a digital value from a phase-increment register (determined by the output frequency) to step the phase register through its range.
How the change in frequency controls the change in these registers is a function of the firmware in the system.
I don't understand why the firmware would automatically reset the phase register itself on a discrete frequency change, but that seems to be the design in the FY6800.

[rvalente]

They all do this when adjusted manually but not for a sweep which is what really matters.
Learn to properly use the feature set instead of focussing on something that was never part of the design brief.

No really this way, Im using the AWG to sweep the frequency manually, the ECU which reads the signal has its on implementation of frequency input and often throws and error.

Could you test how siglent handles this?

[2N3055]

My SDG600X does same as Keysight shown: Frequency Sweep is fully continuous and manual change preserves phase and continues with different frequency without any discontinuity...

I believe SDG2000X should behave same... I don't know about SDG1000X but I suspect that one too, but cannot try.

[Kleinstein]

DDs generators usually have a contineous phase. A few models may have separate ranges for very low or very high frequency with different output filters that can cause a discontinuity. The phase jump is strange. If only for manual steps is usually not a problem. For Frequency scans a continous phase can be an advantage.

[tautech]

They all do this when adjusted manually but not for a sweep which is what really matters.
Learn to properly use the feature set instead of focussing on something that was never part of the design brief.

No really this way, Im using the AWG to sweep the frequency manually, the ECU which reads the signal has its on implementation of frequency input and often throws and error.

Could you test how siglent handles this?

Manual sweeps are always a problem.
To get a proper clean output you need to select the Sweep function.

[TimFox]

In the example of phase discontinuity shown, that looks more like a "reset" of phase than a "jump".

[tautech]

In the example of phase discontinuity shown, that looks more like a "reset" of phase than a "jump".

Yes well it's controlled by adjusting an encoder, not a potentiometer so of course it's a reset.

[2N3055]

In the example of phase discontinuity shown, that looks more like a "reset" of phase than a "jump".

Yes well it's controlled by adjusting an encoder, not a potentiometer so of course it's a reset.

On my SDG600X there is no jump or discontinuity, no matter if you type in frequency or move encoder. You can see where it changes but no reset or any kind of jump.
But agree, sweep is fully blended though, pretty much perfect.
I believe SDG2000X should be same.

[rvalente]

In the example of phase discontinuity shown, that looks more like a "reset" of phase than a "jump".

Yes, it does looks like reset, at least on my fy6800 the change of freq in ch1 reflects a similar behavior in ch2.

In the example of phase discontinuity shown, that looks more like a "reset" of phase than a "jump".

Yes well it's controlled by adjusting an encoder, not a potentiometer so of course it's a reset.

I believe the phase could be soft tunned in firmware, right? Definitely not easy to do but seams doable

[2N3055]

...
I believe the phase could be soft tunned in firmware, right? Definitely not easy to do but seams doable

Like David said, manufacturer has to put in effort to make it so. It seems they didn't. Large difference in price is not always so much difference in components, but lots of time spend polishing details.

These are incredibly cheap AWG that are surprisingly useful for such a simple devices. But they have built in limitations in way they were designed.

[Martin72]

Hi,
Tried to recreate it with 1.5Khz and 10khz, manually setting via keypad (later via jogshuttle)...
Maybe the triggersetting is wrong, used "pulse" (pics) and "intervall".
When sweeping (on my 1062X), there are no issues to see.
Used internal AWG and 1062X.

[2N3055]

SDG6000X 10kHz -> 50kHz and back..
And nice showcase of use of Qualified triggering....

[mawyatt]

We took a look at our SDG2000X Plus and 6000X Plus. Started with 1Hz on the AWG and with scope sweep of 200ms/div, then manually switched AWG to 2Hz and screen captured the phase jump/reset to zero. Then did same for jump from 2Hz back to 1Hz manually.

First two screens are from SDG2000X Plus and last two from SDG6000X Plus, both AWGs "enhanced".

Seems these are always resetting to zero phase for a manually entered new frequency, to start the new frequency from phase zero, so phase is discontinuous with manual frequency changes.

Best,

Edit: This is somehow rearranging the entries

Now it's OK!!

[mawyatt]

Another interesting result when the dial is quickly spun from 1Hz to 1.4Hz. Note how the phase resets to zero on each frequency update from the dial as it transitions from 1Hz to 1.4Hz.

First {27} is from SDG2000X Plus, second {28} from SDG6000X Plus.

Best,

[2N3055]

I looked at what Martin and Mike has shown, and couldn't reproduce it.

Then I went and checked: In Utility/Phase mode I had it set for INDEPENDENT phase mode. It perfectly transitions from frequency to frequency without discontinuity.

But...
If I set it for DEPENDENT phase mode it resets phase every time , because it has to start both CH from same phase..... There is no other good simple way to keep phase relationship for different channels (with different frequency and waveform) but to reset from change. 

Thanks Mike for noticing "phase resets to zero on each frequency update" which made me look there...

Best,

[Circlotron]

DG1022Z with two sine waves locked together in frequency and phase. Was using it to drive two audio power amplifiers powering a small induction motor. Made life difficult to say the least.

https://youtu.be/Qg6sqLdAIs0

[mawyatt]

I looked at what Martin and Mike has shown, and couldn't reproduce it.

Then I went and checked: In Utility/Phase mode I had it set for INDEPENDENT phase mode. It perfectly transitions from frequency to frequency without discontinuity.

But...
If I set it for DEPENDENT phase mode it resets phase every time , because it has to start both CH from same phase..... There is no other good simple way to keep phase relationship for different channels (with different frequency and waveform) but to reset from change.

Thanks Mike for noticing "phase resets to zero on each frequency update" which made me look there...

Best,

That makes prefect sense with what you've described. When changed to "Independent" in the Phase Mode under Utility vs. "Phase-Locked" we get the phase continuous as you've shown earlier 

Thanks for fretting this out and enlightening us as usual 

Best,

[mawyatt]

So why does this discontinuous phase matter?

Normally it doesn't but one scenario comes mind, I'm sure there are others as well.

Consider making a Bode Plot of a Closed Loop System to retrieve the Open Loop Response. If the system under study is active and has resonances, or highly non-linear behavior, then this discontinuous nature could induce the resonance or invoke the non-linearities, all of which may have very long time constants to recovery of "Steady State". When the measurements are acquired during the Bode Plot if these effects are not allowed to die down to noise levels, then corrupted measurements are likely. Since the Bode function is generally not a true frequency "Sweep" but "Stepped" either linearly or logarithmically in frequency, this Phase Discontinuous behavior would occur at even step along the way potentially corrupting the results.

If one envisions doing Bode Plots utilizing a given AWG, then it may be wise to investigate wether the Phase Continious Mode/Feature/Behavor is available.

We found this discontinuous behavior "unsettling" at first (pun intended), but thanks to 2N3055 we are now less concerned!!

Best,

[2N3055]

We found this discontinuous behavior "unsettling" at first 

LOL!!

[2N3055]

So why does this discontinuous phase matter?

Normally it doesn't but one scenario comes mind, I'm sure there are others as well.

Consider making a Bode Plot of a Closed Loop System to retrieve the Open Loop Response. If the system under study is active and has resonances, or highly non-linear behavior, then this discontinuous nature could induce the resonance or invoke the non-linearities, all of which may have very long time constants to recovery of "Steady State". When the measurements are acquired during the Bode Plot if these effects are not allowed to die down to noise levels, then corrupted measurements are likely. Since the Bode function is generally not a true frequency "Sweep" but "Stepped" either linearly or logarithmically in frequency, this Phase Discontinuous behavior would occur at even step along the way potentially corrupting the results.

If one envisions doing Bode Plots utilizing a given AWG, then it may be wise to investigate wether the Phase Continious Mode/Feature/Behavor is available.

We found this discontinuous behavior "unsettling" at first (pun intended), but thanks to 2N3055 we are now less concerned!!

Best,

I put up a setup where one scope monitors frequency transitions while other scope performs FRA.. Just out of curiosity.

Internal gen in SDS2000X HD performs seamless transition. Same will be for SDS2000X+.  

SDG6000X needs to be in Phase independent mode.. Then transitions are seamless.

If in Phase dependent mode it creates short pause between frequences, i.e. signal goes to zero and restarts exactly 1ms later with next frequency at phase 0° . 

FRA II Siglent code waits for signal to settle (that includes a DUT too) a bit anyways..
It seems like someone did their homework.