1 Synchronisation issues for digital measurements » History » Version 2
Version 1 (JANVIER, Thibault, 12/15/2015 08:55 AM) → Version 2/4 (JANVIER, Thibault, 12/15/2015 08:59 AM)
h3. 1. Synchronisation issues for digital measurements
Usually, the USRP does not abide the “IQ sampling rate” that has been entered as an input parameter, and forces a value that is close to it. This “actual IQ sampling rate” is displayed on the user interface. As a result, there is a continuous drift of the sampling time instant regarding the optimal sampling time instant: at a given time, the optimal sampling time instant is reached and the constellation appears, later, the sampling time instant falls between two adjacent symbols and the constellation vanished. This phenomenon can be witnessed on the following video (to be downloaded). This is a hardware issue that we cannot fix. The error measurements derived from the constellation are only relevant when the constellation is well displayed. video.
!20151209_194518.mp4!
+Video: Illustration of the continuous drift of the sampling time instant regarding the optimal sampling time instant+
On the other hand, assuming the actual IQ sampling rate is equal to the desired sampling rate,
Synchronisation of the sampling time instants: could be done manually by controlling a command that would shift the sampling time instant from 0 to Ts/2 => the aim is to minimize the ISI and the EVM while controlling the time shift. The sampling frequency is assumed to be equal to the symbol rate. Make drawings to explain
Usually, the USRP does not abide the “IQ sampling rate” that has been entered as an input parameter, and forces a value that is close to it. This “actual IQ sampling rate” is displayed on the user interface. As a result, there is a continuous drift of the sampling time instant regarding the optimal sampling time instant: at a given time, the optimal sampling time instant is reached and the constellation appears, later, the sampling time instant falls between two adjacent symbols and the constellation vanished. This phenomenon can be witnessed on the following video (to be downloaded). This is a hardware issue that we cannot fix. The error measurements derived from the constellation are only relevant when the constellation is well displayed. video.
!20151209_194518.mp4!
+Video: Illustration of the continuous drift of the sampling time instant regarding the optimal sampling time instant+
On the other hand, assuming the actual IQ sampling rate is equal to the desired sampling rate,
Synchronisation of the sampling time instants: could be done manually by controlling a command that would shift the sampling time instant from 0 to Ts/2 => the aim is to minimize the ISI and the EVM while controlling the time shift. The sampling frequency is assumed to be equal to the symbol rate. Make drawings to explain