Magnitude of FID signal affected by demodulation?

[h3lg3]

Analyzing simple FID experiments "rf90 - adc" revealed unexpected signal time courses of the magnitude of the FID signal. The signal at the beginning of the ADC is modulated, such that early samples show a reduced amplitude. I would expect the FID to be continuously decreasing in magnitude. This might result from the demodulation process and might be relevant for readouts starting at the center of k-space.

[Toreil]

The rapid fluctuation at the start is due to the decimation/downsampling used, the non-exponential decay looks very typical of a saturated preamplifier. Could you try with a smaller sample or a longer wait between the excitation pulse and the ADC window. I assume the time axis is incorrect here, right?

[h3lg3]

Thanks for the quick reply. Time axis is in ms not seconds, but rest is good. Here a picture of the full signal decay.
BW was 20 kHz with 1000 samples. I already added a 1 ms gap between RF and ADC event to avoid issues. Would you propose to add an even longer gap? Using a dedicated ADC_dead_time in PyPulseq of 100us did not help.

[Toreil]

Longer than 1ms just to see if its a signal amplitude issue or something else.

I think ADC_dead_time is not considered at the moment, can you raise that as a separate issue?

[h3lg3]

Alright, I will give an update once I repeated the scan later the week. The adc_dead_time issue is now adressed in PR #145

[Toreil]

If you make it like 5 ms, then the signal should be at a point where it is low enough to avoid saturation effects, so in that case it should be exponential with the high frequency mess at the start which can be reduced using different saturation effects, at a cost of loss of flatness in the sensitivity in the frequency domain

[h3lg3]

@Toreil I repeated the acquisition with a 5 ms delay between rf90 excitation and ADC readout. It can be seen that the FID amplitude reduced to ~6 mV in comparison the the 1 ms delay with an amplitude of ~30 mV. However, the fluctuations remain similar.

[Toreil]

This is weird, we don't see this at all so I'm not sure it's related to the decimation method. I also dont understand why, if it's data processing related, the effect of it would be so different for the 1ms delay and 5ms delay. Do you know if there is some switching going on in your hardware? I.e. do your preamps or TR switch need an active control signal?

[Toreil]

Actually this data looks normal, no? You have the wobble at the start from the decimation and then the FID which is not exponential because the B0 is not uniform

[h3lg3]

Actually this data looks normal, no? You have the wobble at the start from the decimation and then the FID which is not exponential because the B0 is not uniform

Yes, I think the signal decay of the right image is fine. At first I was also a bit confused, but if you check the left image at ~5ms, you see that there is a plateau in the decay and thats where we start sampling for the right image.

[aTrotier]

I think it is also something I observed with UTE sequences on Bruker. I have to double check that