FFT analysis and vector averaging

[M.GH]

I have AM modulated (Carrier 250 kHz sampling) signal which I send it to USB 6218 DAQ. Then I measure the FFT using vector averaging to increase SNR. The whole program is in sequential form so that the signal generation is done and repeated in a loop and the next sequence FFT is computed. 

If I do measurement on the same DAQ and with the same sampling frequency for input (250 kS/s) then I am getting correct frequency spectrum but if I use two different DAQ, one for signal generation (AO) and second for measurement (AI) then I think I have synchronization problem because the averaging does not work. However, when I am using the same DAQ and I don't do any sync. 

The reason I would use two DAQs is sampling limitation of my card which is 250 kS/s and I have to go beyond that which is 300 kS/s.  I am stimulating with sound and measuring vibration. So the input sampling rate can be far less but when I go with lower sampling I have aliasing problem and again non-sync. 

I would appreciate if you share your experience about it!

Best regards, 

 MGH

[AndreasGareis]

Dear M.GH,

could you list which HW apart from the USB 6218 you have available and maybe post your aquisition code?

Best regards
AndGar

[M.GH]

Hi,

I have attched the program but since I use an excel sheet reading subVI, you might not be able to open this.

I have another card-NI PCI-MIO available!

But the generated signal goes through a buffer amplifier to the resonator (LC-tunued at carrier frequency). Then it is transmitted to the vibrator after demodulation. I have an accelerator and a charge amplifier to measure the vibration. 

[Henrik_Volkers]

If you do vector averaging on FFT values the phase relation is important !

Think of two values of the same magnitude but 180° phase shift -> no magnitude 😄

So your aquisition and generation needs to be synchronized. If you run it on the same board they use the same timebase and are more or less synced by accident 😉

If you use different boards even the small difference in the oscillators can generate errors. ( Can be corrected with some math assuming the carrier is constant)

But in addition the phase of your signal depends on the time of your generation start and aqusition start. 

[AndreasGareis]

Dear M.GH,

thank you for the information. What model is the PCI-MIO Card? Can't you perform the aquisition on that card alone? Otherwise i only see the possibility of synchronization by sharing a sample clock from the usb daq to the pci card via a PFI line. The propagation delay of the trigger signal might become a problem so you'd need to reduce the cable length between the two as much as possible.

Take a look into this white paper (part 3):

Synchronization Explained

Best regards

AndGar

[M.GH]

Thank you  AndGar and Henrik for your reply!

The card is a NI PCI-MIO-16E-4 (NI 6040E) and my intention is to use 6218 daq because I need its isolated property.

kind regadrs,

MGH