Post Go back to editing

Questions about ADRV9025 calibrations

Category: Hardware
Product Number: ADRV9025

I have some question when reading UG1727 chapter SYSTEM CONSIDERATIONS FOR INITIAL CALIBRATIONS,staring on page 85.

1. In External Transmitter LO Leakage Initial Calibration,staring on page 86.It says "The purpose of this calibration is to obtain a good estimate (gain/phase) of the external loop channel conditions prior to operation.".

But I still can't understand why external loop has something to do with LO leakage. In my understanding, LO leakage is because of the mixer inside, not external channel, besides, the nonlinear cross-tuning frequency of external PA will not appear at the LO position.

May I ask if my above understanding is wrong? If there is any problem, can you help me point it out? I can understand using the estimate (gain/phase) of the external loop channel to do DPD, but can you help me understand the relation between external loop channel and LO Leakage? 

2. Contrary to “External Transmitter LO Leakage Initial Calibration” mentioned in question 1,the UG does not mention the "External Transmitter QEC Initial Calibration".I take it that,image is because of the mismatch of phase or amplitude of I/Q signals, the process happens internal, so there is no meaning of external calibration.May I ask if my understanding is wrong? 

3. As mentioned several time in UG, such as in page 91, "QEC tracking uses an offset LO on the feedback path during tracking. This ensures that the quadrature errors of the transmitter path are not aligned with those of the observation receiver path.".

I think not only TX has the LO problem, RX and ORX also have DC offset and LOL caused by mixers in RX/ORX channel. I don‘t understand why there is no "Internal/External Receiver/Observation Receiver LO Leakage Initial Calibration"? I understand that since it is RX, we can't generating a signal that can cancelling LOL, but can we do it in the digital domain?Is there already this mechanism, just didn't mention it?

  • Please see below for our comments:

    1. We have 3 different levels of LO Leakage calibrations. One is Internal LO Leakage cal which uses the internal loop back path/ feedback path to get the initial channel conditions

    Two is external LO leakage cal which requires the complete PA path to the feedback path to be estimated. The LO leakage might be higher and can go bad with the additional gain when connected with PA.

    The third level is the tracking cal which ensures that the LO leakage is within the specified level across temperature/ any changes in IQ mixing.

    2. Tx.QEC uses the internal feedback path and the image correction will not require the external path

    3. In the case of Rx and ORX, we do the DC offset correction which is similar to LO leakage correction in TX.

  • Thanks!

    Does these initial calibrations and tracking calibrations all done by the built in ARM M4 processor? Does the data available to the FPGA?

  • All the calibrations are taken care by the inbuilt ARM processor. You can have the RX data and ORX avaialble to the FPGA and process it as required. For eg, you can get the ORX data on JESD to the FPGA and use your own DPD algorithm.

  • OK,thanks!

    We will try our own DPD algorithm,besides,we want to try our own LOL/QEC calibration algorithm.

    You can have the RX data and ORX avaialble to the FPGA and process it as required. For eg, you can get the ORX data on JESD to the FPGA and use your own DPD algorithm.

    It means that I can close all TX(3 different levels)/RX/ORX LOL calibrations and TX/RX/ORX QEC calibrations, and try our own LOL/QEC calibrations algorithm using FPGA?

  • Whats your requirement? ADRV9026 can take care of DPD, LOL, QEC calibrations on its own.

    Is there a specific purpose/advantage of running your LOL, QEC algorithms?

  • It would be great if we can reproduce the similar effect compared with the built-in algorithm of ADRV9025, but it is expected to be unlikely, and we only want to do a principle verification at present. The demand is only for technical research. We purchased the ADRV9025 EVB as a research platform for WIFI, and there is no demand for productization. I'm a noob in this field, so I apologize if the question is too amateurish.