AD9082 for 802.11ax

Hello!

I have certain experience with direct conversion systems. Looking at 802.11ax requirements, particularly 5.925-7.125 GHz frequency band I have a feeling making discrete receiver based on either direct conversion or IF structure might be prohibitively complex, so we are looking for kind of integrated solution, relieving such a challenges. Direct conversion transceivers like ADRV9009 is what I roughly understand, however, their frequency range definitely won't cover above 6GHz. This we we came to AD9082, however, I can't say I truly understand its RF sampling technique. I am under impression that sampling with at most 6GSps we can't digitize signal from mentioned 5.925-7.125 GHz frequency band.

Could you please point me to proper reading so I could better understand this chip operation principle.

Thanks.

Top Replies

  • +1
    •  Analog Employees 
    on Jan 19, 2021 10:02 PM

    Hello,

    The link for the AD9082 is as follows:   AD9082 (Rev. A) (analog.com)

    The link for the latest User Guide, UG-1578, describing in detail the IC operation is as follows:  AD9081/AD9082 User Guide (Rev. PrC) (analog.com)

    For the particular application you describe above, one could consider the following transceiver architecture.

    1)  Operate the DAC at 9.8304 GSPS and the ADC at 4.915 GSPS such that the center of the ADC's 3rd nyquist zone is positioned between  4.915 GHz to 7.3728 GHz  allowing the ADC to directly sample the full 802.11ax band situated between 5.925-7.125 GHz.

    2) IQ data rate could be set to 1.2288 GSPS to realize 1 GHz of instantaneous BW with DAC datapath set to 8x interpolation and ADC datapath set to 4x interpolation.   Alternatively, one can select higher decimation/interpolation factors using channelizer stages allowing for multi-band operation with perhaps IQ rates set to 307.2 MSPS in support of 160 MHz wide channels that may exist within the 1 GHz of instantaneous BW. 

    3) The ADC can directly sample the spectrum with perhaps an RF line-up consisting of a 2 or 3 gain stages w/ DSA and bandpass filtering inserted.

    4) While the DAC can also produce a signal in this frequency range falling in its 2nd Nyquist zone, the frequency response from DAC does roll-off in this region due to the DAC's inherent Sinc response as well as rf matching challenges.  However., this roll-off would be fairly stable for a given LTCC balun type such that it may be possible to compensate for any passband "tilt" using a digital equalizer filter.  The RF line-up would still need to remove DAC images in the 1st Nyquist zone as well as 3rd, 4th, ext as well as amplify the desired band region falling the 2nd Nyquist zone.
    6530BP44A1190_062920.xlsx (johansontechnology.com)

    An alternative approach resulting in perhaps better passband flatness (thus avoiding digital equalizer) would be to use a mixer like the  LTC5553 whose LO also operates at the same frequency of the DAC (i.e. 9.8304 GHz) such that a single low phase noise PLL can generate both the AD9082's CLKIN and LTC5553 LO frequency.   The DAC would generate signals falling between 2.704 to 3.904 GHz that it could be upconverted to 5.925-7.125 GHz.  Mini-circuits LTCC filter following the DAC balun can be used to suppress DAC images before going into the LTC5553 mixer.
    Mini-Circuits
    LTC5553 Datasheet and Product Info | Analog Devices

  • Hello,

    Thank you so much for detailed and comprehensive response. I really appreciate describing possible options. I've missed the trick with N-th Nyquist zone, but apparently not only this one. We have to work more on the signal chain design.