Hello,
Below is the frequency response I obtain when connecting Out1 to In1 with a BNC cable and performing a frequency sweep. The phase response is far from 0°, even at moderate frequencies, well below the amplitude cutoff freequency.
could you please explain what part of this phase response comes from the analog front-end? the ADC? the DAC ? Could you please provide complete specifications (i.e. including phase response) for these blocks?
it seems to me that the phasemeter does not compensate for this systematic phase error. Could you please confirm?
Best regards,
JJ
Hello Dylan,
As mentioned in the post : the device I am using to perform the sweep is a Moku:Go. The sweep is performed with a BNC cable connecting Out1 to In1.
Best regards,
JJ
Hello JJ,
It looks like you are using the FRA in Multi-Instrument Mode (MiM). In MiM, there is no compensation for the delay between the DACs and ADCs, so that is why you are seeing the phase wrapping at lower frequencies. In single instrument, the Moku does indeed compensate for this delay. Hope this helps!
Hello JJ,
You can also use the ‘Normalization Trace’ on the bottom right of the screen. This will essentially measure the latency and then apply compensation. If you are measuring a real circuit/DUT, you will want to normalize before connecting your DUT since the normalization will compensate for the delay in your circuit and skew your results.
I was aware MIM instruments had some limitations compared to standalone ones, but I had not identified the delay being compensated or not as one.
Do we have the same issue for the phasemeter, then?
This is really a crucial point for me, as I work in the field of oscillators: I typically use the FRA to sweep a resonator’s response close to resonance and determine the phase \phi_0 at resonance. Then the phasemeter is used as a PLL to “close the loop” around the resonator, and setting a phase shift of -\phi_0 between the phasemeter’s output and its input.
So, your explanation makes it very clear what to expect from the FRA in terms of phase error (and I thank you again for that). But I must also understand in which case(s) the ADC/DAC delay is compensated for when operating the phasemeter’s PLL, either in MIM or standalone mode.
Hello,
With the phasemeter you will not experience this delay. The phasemeter doesn’t interpret the delay as a change in phase since it is phase locked. The phasemeter will just phase lock onto the input signal and measure the phase, frequency, and amplitude from that signal. I hope this helps!
Hello,
I think my previous request was not very clear. I will perform additional measurements and come back at you. Thank you for your time.
Best regards,
JJ
Hello,
I have experimentally checked that the FRA behaves as you explained. The overall phase response is much flatter in standalone mode, with the phase-shift @68kHz dropping from (approx.) -30° in multi-instrument mode to -1° in standalone mode.
I have also verified the delay issue can be observed with the phasemeter in MIM (this was expected) but also in standalone mode (unexpected).
For the MIM case: in the example below, I have connected In1 to Out1 with a BNC, and also In2 to Out2. The scope generates a fixed 68kHz sine wave on Out1 and the phasemeter locks to it. I use the “lock to input signal” feature to generate on Out2 a signal with the same frequency as the input, and set the phase shift to 0°. Yet the two signals displayed on the Scope have a -28° phase shift, which corresponds to the In2/Out2 delay, in accordance with the FRA measurements and your explanations.
However, I was expecting the ADC/DAC delay to be compensated in standalone mode, as in the FRA case. But clearly it is not: I set the Moku to work with the phasemeter in standalone mode, connected In 1 to a Tektronix AFG delivering 100mVpp @ 68kHz, Out1 to a Tektronix scope. As previously, I lock the phasemeter to In1, then set Out1 to have 0° phase-shift wrt to In1. Both the In1 and Out1 signals are visualized on the Scope in the picture below (respectively blue and yellow signals). You can see there is a -29° phase shift between the two, in spite of setting phase shift to 0° and operating in standalone mode.
This is very counterintuitive… What’s more, the fact that the delay is compensated in some cases (FRA), and is uncompensated in other cases (phasemeter) is very confusing. I understand this is not an issue when working only in the digital domain (e.g. the tutorial https://www.youtube.com/watch?v=u0XjQxB3t5c, which avoids using the ADC/DACs to demonstrate the PLL’s operation), but it is a real pain when trying to interface the Moku with the outside world.
Could you then provide some technical information about the ADC and DAC delays (none of which are mentioned in the Moku’s specifications)? And hopefully add some functionality to the Moku for choosing to compensate or not for delay, at least for the phasemeter operation ?
Best regards
JJ
Hello,
For the PLL, there isn’t necessarily supposed to be 0 degrees of phase shift between the input and output, but rather the phase difference should remain constant (-29 degrees in your case). If you need your input and output to be in-phase, you can simply add a phase shift to the output and the PLL will keep this constant in-phase relationship and thus compromise for the ‘delay’. As far as the delay between the DACs and ADCs, I measured this to be about 1 microsecond. If you need anything else please let me know!
!
