Post Go back to editing

IQ Offset characteristic change when using tx attunuator (Carrier Leakage)

Category: Hardware
Product Number: ADRV9002
hi,

We are implementing 5GNR V2X using ADRV9002.

When the TX Attenuator value changes, the IQ Offset also changes.

Local Leakage appears to be changing.

And another question. 
After initial calibration, when ATTEN is 0, the expected IQ Offset is -60dBc but the measured value is -40dBc.

measurement data:


Thank you,



Edit contents
[edited by: benjamin405 at 5:04 AM (GMT -4) on 31 Oct 2023]
  • Hi benjamin405,

    I was able to replicate your result of the IQ offset changing with the internal attenuation, but not the initial calibration issue. I will follow up internally to see what the root cause is.

    Kind Regards,
    Michał

  • hi M_Bugajski,

    On all boards we manufacture, the IQ offset measures greater than -40dB after ADRV9002 initialization.
    Is there a way to improve the IQ offset below -60dB?
    (As you can see in the previous question, it appears to be a local leakage problem.)

    Best Regards,
    benjamin405
  • hi,

    This is the result measured at the adrv9002 output stage.
         - Frequency: 5860MHz
         - Bandwidth: 10MHz(5G V2X)
         - TX Attenuator: 0
    After EVM measurement (Figure 1), the data was deleted and only the carrier was measured (Figure 2).
    -55dBm was measured.
    Looking at the specifications (Figure 3), Carrier Leakage is listed as -56dBm at 5800mhz.
    So, considering the slight error, I believe that the -55dBm I measured satisfies the specifications.
    Does it meet the specifications?

    If it meets the specifications, is there a way to improve Carrier Leakage?
    for example,
    1. Additional offset is given externally.
    2. Use products with better specifications.
    (recommendation required)

    Best Regards,
    benjamin405

  • Hi benjamin405,

    It looks like it does meet the specifications. I'm afraid the only way to improve that is to enable tracking calibrations, which may not work in your case.

    Another way would be shift your signal away from DC, but that would require a higher sample rate.

    Kind Regards,
    Michał