Design Approach

Co-location Deployment Considerations for Direct RF Sampling Transceivers

Document ID
Release Date
1.0 English

A general design approach for handling the 16 dBm blocker is outlined in this section.

  1. The antenna BPF must filter the blocker to a level such that:
    1. It does not saturate/overdrive the first gain stage (LNA) of the AFE.
    2. The HD2 and IMD2 products generated by the gain stages from the blocker are well below the power of the wanted signal so that the 3GPP required 95% throughput is met. As illustrated in Figure 1 and Figure 2, the HD2 and IMD2 products from signals in the PCS band fall directly on the C-band. Although the specification only uses CW for the blocker, the multi-carrier scenario should also be considered in the analysis as the IMD2 generated is typically 6 dB higher than the HD2 product. This can be seen from a measurement of an LNA performance in Figure 3.
      Figure 1. HD2 and HD3 Products of a Continuous Wave in the PCS Band
      Figure 2. Harmonic Products of Two Wireless Carriers in the PCS Band
      Figure 3. Actual IMD2/HD2 Measurement on Qorvo QPL9058 (VDD=5.0V) LNA at 0 dBm output. OIP2 = 33 dBm at 5VDD; OIP2 = 29 dBm at 3.3VDD.
  2. The features of the optional BPF between gain stages are as follows:
    1. The purpose of the optional BPF shown in Figure 1 and Figure 2 is to provide additional filtering of the blocker in case the antenna filter does not have enough rejection to prevent the second amplifier from generating IMD2/HD2 products that can desensitize the receiver.
    2. The cost and footprint of this BPF is rather negligible and can be leveraged to relax the antenna filter requirement, and in turn to lower the cost and weight of the system. As in the TX lineup, the Johanson 3750BP14D0900 [REF 2] is a very good candidate with a 0603 size footprint.
  3. An anti-alias BPF is needed for both RF sampling and ZIF receivers to get the best system NF, and it serves the following purposes:
    1. Provides additional filtering of the blocker so that it does not saturate/overdrive the ADC input.
    2. Provides last stage filtering of the blocker so that if the ADC alias of the blocker falls on the wanted C-band, it will be lower than the wanted signal to avoid desensitization. The filter rejection amount needed for this purpose is typically high. This scenario is common for a convenient RF-ADC sampling rate of 2949.12 MSPS. However, more 5G optimal devices, such as the Zynq UltraScale+ RFSoC DFE, provide additional options for the designer to frequency plan such an adverse scenario, and, consequently, remove the need for high-rejection filtering.
    3. Provides filtering for out-of-band noise and spurs that would otherwise alias/fold onto the first Nyquist zone at the ADC output.
    4. For a ZIF receiver, this anti-alias BPF also serves as a cleanup filter prior to the complex mixer in the demodulator. The current state of ZIF RFICs typically has active mixer and gain stages that feed the ADC input. M x N products of significant level from the mixer can be detrimental to the receiver sensitivity.
  4. The digital filters are used as follows:
    1. Advanced devices, such as the RFSoC DFE, have hardened, low-power digital filters throughout the datapath to further remove remnants of the interferer. With GHz’s of Nyquist bandwidth and flexible frequency planning, the designer can place the blocker alias outside of the C-band. In this case, the digital filter is used to remove the blocker prior to the symbol demodulator, which eases the analog filtering.