In this blog post, we’ll focus on the calibration of I/Q mismatch for direct conversion transmitters. The most common problems caused by these transmitters are gain and phase mismatch (imbalance) between the I/Q rails and the local oscillator (LO), and leakage due to finite isolation between the LO and RF ports of the modulator.

In practice, analog front-end components such as DAC, low-pass filters, and mixers also contribute to the overall I/Q mismatch of the transmitter. Analog components usually don’t have the perfect characteristics you would expect (phase exactly at 90 degrees, equal gains, and so on). This problem causes a finite sideband image of the transmitter signal that degrades its quality. This issue can be solved by choosing very high quality and precisely matched analog components. However, this solution leads to a very costly transmitter whose stability is hard to maintain over the course of its lifetime.

Several low cost and flexible methods exist that use digital signal processing (DSP) to compensate for I/Q imbalance effects. They are based mainly on two approaches: predistortion filtering and direct calibration.

 

Predistortion Filtering

 

Predistortion filtering consists of some techniques such as: block least-square (LS) (sometimes called widely-linear LS), recursive LS

[1, 2, 3, 4], and post-inverse estimation [5].

The figure 1 here shows a block diagram of the signal processing chain of the block LS approach. The feedback signal processes and updates the predistortion filter coefficients, similar to the principle of adaptive filtering.

The post-inverse calibration technique consists of finding the inverse path in the modulator, as illustrated in figure 3.10 here.

The predistortion filter approach requires significant processing, such as: generating a baseband modulated reference transmission signal in order to estimate I/Q phase and gain imbalance; matrix processing; adaptive filtering; and compensating for the transmission path in order to maximize the image rejection ratio (IRR). Furthermore, LO leakage needs to be mitigated prior to performing predistortion filter estimations. Synchronizing the transmitted data and the feedback data is mandatory for this approach. Bad synchronization introduces inter-symbol interference (ISI) to the feedback data, leading to incorrect estimation of the predistortion filter coefficients.

 

Direct Calibration

 

In the direct calibration approach, a simple sinusoid reference tone is transmitted with the signal; no synchronization between the transmission and feedback is required. This approach uses a simple gain and phase compensation with a few multipliers and adders in the transmission path, and an error measurement circuit in the feedback path.

LO leakage can also be estimated using the same error measurement circuit as in the feedback receiver in Figure 4, then the DC offset can be adjusted via the DAC inside the RF transmitter, or digitally compensated in baseband signal. The LMS6002D transceiver from Lime Microsystems has an integrated DAC per analog I/Q rail in order to compensate for the transmitted DC offset [6].

Nutaq’s Radio420X FPGA mezzanine card (FMC) product line includes an enhanced version of the Tx I/Q mismatch and LO leakage calibration techniques proposed by Lime Microsystems, which improve the IRR and reduce the LO leakage. An example of the transmitter I/Q imbalance and LO leakage before and after calibration in the Radio420X is shown in the following figures.

Transmitter I-Q imbalance and LO leakage before calibration

Transmitter I/Q imbalance and LO leakage before calibration

 

Transmitter I/Q imbalance and LO leakage after calibration

Transmitter I/Q imbalance and LO leakage after calibration

In these figures, the transmitter center frequency is at 1952.5 MHz, the 15 dBm (marker R) transmission tone is at 1953 MHz, and the sideband image is at 1952 MHz. The measured sideband IRR (marker 1) and LO leakage (marker 2) values after calibration are about 48 dBc and 46 dBc, respectively. Another transmitter I/Q imbalance and LO leakage calibration example can be found in our previous blog post.

It turns out that the low-cost wideband radio transceiver based on the LMS6002DFN is capable of achieving good transmitter RF performance without using and high cost analog components or calibration algorithms power processing hungry. Furthermore, the Radio420X package includes the transmitter I/Q imbalance and LO leakage calibrations, so no further calibration is required on the user’s end.

References

  1. Myllari, Olli, Lauri Anttila, and Mikko Valkama. 2010. “Digital Transmitter I/Q Imbalance Calibration: Real-Time Prototype Implementation and Performance Measurement.” European Signal Processing Conference. Aalborg, Denmark. http://www.eurasip.org/Proceedings/Eusipco/Eusipco2010/Contents/papers/1569290640.pdf
  2. Anttila, L., M. Vlkama, and M Renfors. 2008. “Frequency-Selective I/Q Mismatch Calibration of Wideband Direct-Conversion Transmitters.” IEEE Transactions on Circuits and Systems II: Express Briefs 55(4): 359-363, 2008. doi: 10.1109/TCSII.2008.919500
  3. Picinbono, B., and P. Chevalier. 1995. “Widely Linear Estimation with Complex Data.” IEEE Transactions on Signal Processing 43(8): 2030-2033. doi: 10.1109/78.403373
  4. Kiayani, Adnan Qamar. 2009. “DSP Based Transmitter I/Q Imbalance Calibration: Implementation and Performance Measurements.” Tampere University of Technology, Tampere. http://dspace.cc.tut.fi/dpub/bitstream/handle/123456789/6686/kiayani.pdf?sequence=3
  5. Ding, Lei, Zhengxiang. Ma, D. R. Morgan, M. Zierdt, and G. T. Zhou. 2008 “Compensation of Frequency-Dependent Gain/Phase Imbalance in Predistortion Linearization Systems.” IEEE Transactions on Circuits and Systems I: Regular Papers 55: 390-397. doi: 10.1109/TCSI.2007.910545
  6. Lime Microsystems. 2012. LMS6002DFN Multi-band Multi-standard Transceiver – Programming and Calibration Guide.