Post Go back to editing

adrv9364-z7020 noise figure problems

Category: Hardware

Hello,

I'm using adrv9364-z7020 + ADRV1CRR-BOB EVB and performed some noise measurements

by sending 1 MHz CW at very low power, -80 dBm and changing the LO.

The first thing that I noticed is that at high frequency for LO, 2.4GHz, the noise level is -171 dB/Hz that is

a very close value to noise figure reported in AD9634 transceiver datasheet.

Lowering the LO frequency, the noise level start to become worser and at 160MHz I measured

-164 dB/Hz. This value is very far respect to AD9634 datasheet that say a 2dB of figure noise at

maximum rx gain at 800MHz.

I would have expected a better value at 160MHz for noise level.


Also I would like to show a strange spectrum (figure below) that I obtain within those parameters for the tranceiver in IIO Oscilloscope:
RX LO Frequency: 800MHz
Sampling Rate: 8 MHz
RF Bandwidth: 4 MHz
RX Hardware Gain: fast_attack



That hole in the spectrum is very strange and can't explain where it come from. 


I would like to ask why I obtain such bad performance.

Parents
  • -164 dB/Hz. This value is very far respect to AD9634 datasheet that say a 2dB of figure noise at

    how are you measuring the noise level at a particular LO freq with Matlab or IIO? is the Gain index is same at all the LO frequencies.

    That hole in the spectrum is very strange and can't explain where it come from. 

    output spectrum seems strange, are you seeing the same response with all channels, is this response without any signal at the Rx channel.

Reply
  • -164 dB/Hz. This value is very far respect to AD9634 datasheet that say a 2dB of figure noise at

    how are you measuring the noise level at a particular LO freq with Matlab or IIO? is the Gain index is same at all the LO frequencies.

    That hole in the spectrum is very strange and can't explain where it come from. 

    output spectrum seems strange, are you seeing the same response with all channels, is this response without any signal at the Rx channel.

Children
  • Thanks for your reply and sorry for late response. 

    For noise measurement we performed the following way:

    We produced a sine wave of 161 MHz at power level of -15dBm using a singnal generator (Keysight N5173B) and attenuate using a programmable attenuator (mini-circuits RUDAT-6000-90).
    The signal measured with spectrum analyzer is -76.1 dBm. 

    Than we feed the ADRV9634-Z7020 and using IIO Oscilloscope with following setup:
    - RF Bandwidth: 4 MHz
    - Samp Rate: 8 MSPS
    - RX LO: 160 MHz
    - RF Port: A_BALANCED
    - Gain Control Modes: slow_attack

    Within this configuration the HW Gain is automatic setted to 73 dB and than we measure the peak dBFS
    and the background noise close to che peak using IIO Oscilloscope at high avarage values (70-100) and FFT size of 1024.
    The measured peak value is -10.22 dBFS while the noise level close to peak value around -62 dBFS.
    Than we calculate the noise level with the following formula:

    -76.1 - (-62+10.22) - 10*log10(8MSPS/1024) = -166.8 dB/Hz

    This result is quite high compared to thermal noise of -173.9 dB/Hz but we expected a better value as AD9361 datasheet shows. 

    The test was replicated sending 2.401GHz at measured level of -82.5 dBm and evaluated the noise level in same way and calculation. Peak value -16.65 dBFS, background noise at -66 dBFS with high avarage and calculation show:

    -82.5 - (-66 + 16.65) - 10*log10(8MSPS/1024) = -170.87 dB/Hz. 

    This value is 3dB over thermal noise and shows a good result compatilve with AD9361 datasheet at 2.4GHz.



    output spectrum seems strange, are you seeing the same response with all channels, is this response without any signal at the Rx channel

    Yes, we saw same response in RXA and RXB without any signal and with a 50Ohm termination to the pigtail. 
    We have two boards in laboratory and saw same response on both.

  • AD936X device Rx ports have different impedance across the frequency band, and around 2.4 GHz impedance is ~50 ohm and that is why you are able to achieve better noise figure and impedance is ~ 200 ohms at 160 MHz, so need to do impedance matching to attain the better noise figure. 

    Yes, we saw same response in RXA and RXB without any signal and with a 50Ohm termination to the pigtail. 

    I will check on this and get back you to soon.

    Thanks.

  • AD936X device Rx ports have different impedance across the frequency band, and around 2.4 GHz impedance is ~50 ohm and that is why you are able to achieve better noise figure and impedance is ~ 200 ohms at 160 MHz, so need to do impedance matching to attain the better noise figure. 

    We can't find this information on AD9361 datasheet but i looked in UG-570 reference manual document and checked it. Thanks for this reply, I will perform a new measurement with the right matching impedence. 

  • Refer RF Port Interface section in the user guide of AD9364 for impedance information of Rx and Tx ports.