System Link budget

The gain budget of the beamformer is formulated by the link budget of the whole module which the beamformer be involved. For the whole module, the link budget can be planned into three parts: receiving path, transmitting path, and air channel.

A. Receiving path

The receiver required sensitivity is depended on the modulation and coding scheme (MCS) of signal, which affect the required signal-to-noise ratio (SNR) for demodulation in baseband circuit. In this system, the design target is 4.6-Gbps throughput over 3-meter distance point-to-point transmission. The data rate is supported by using π/2-16QAM modulation with 3/4 coding rate of single-carrier physical layer (SC PHY) [42]. The practical throughput depends on the packet error ratio (PER) of the demodulated signal [49]:

 

Throughput 1 PER bit rate (4.1)

The relation between packet error ratio (PER) and symbol error rate (SER) [49]:

1 (1 ) where n is the symbols in a packetⁿ,

PER  SER (4.2)

Thus,

 

Throughput 1 SER ⁿbit rate (4.3)

Note that the 16QAM have 16 symbols in its constellation and carries 4 bits in a symbol, and there are 448 symbols in a packet that defined in the 802.11ad specification [42].

The throughput can be confirmed to exceed 89.4% of bit rate while the SER is below

5× 10^-3. The SER versus SNR among modulation scheme in additive white Gaussian noise (AWGN) channel as shown in Fig. 4.6. It shows that the SER is below 5× 10^-3 while the SNR is higher than 12 dB. Thus we use 12 dB of SNR for the receiving path link budget calculation.

The link budget of receiver path can be calculated from the sensitivity. Behind the baseband, the noise figure is determined by the BDA with its front-end circuits due to the 16-dB gain is adequate to isolate its back-end loss in the noise figure calculation. The BDAs have a noise figure of 7 dB, the front-end SPDT switch has 2-dB insertion loss, and the transition and routing loss between the chip and antenna is estimated as 2 dB. As the results, the 11-dB noise figure is estimated for each antenna path. In the principle of phased array [50], an N-element array can improve the SNR by N times. In other words, the SNR is improved by 6 dB for the 4-element array.

The sensitivity(S_min)can be derived from noise floor(kT , channel bandwidth (B), ₀) noise figure of front-end circuit (NF), and minimal required SNR of baseband





[51]:

Note that the bandwidth is 1.76 GHz for each channel of 802.11ad band.

Fig. 4.6. SER versus SNR [52].

B. Antenna and air channel

The antennas do not offer real power gain, but focus the main beam of radiation pattern to a specific direction. In fact, the high antenna gain can avoid the power dissipation in the unwanted radiation direction. At 60 GHz, the 4-element antenna array demonstrates above 10-dBi gain [53]. In order to cover bigger beamwidth and achieve the tolerance of excited phase error, a conservative 8-dBi gain for each 4-element antenna array is expected. Note that the small-range excited phase error will cause the degradation of the directivity, but the wider beam width can cover a larger radiation angle.

The propagation loss of microwave can be estimated by Friis free space equation [54]:

The channel loss is depended on the operated frequency and path distance. The proposed

module is designed for application of 3-meter distance with 60-GHz carrier frequency.

From the Friis equation, the path loss (PL) is 77.6 dB.

C. Transmitting path

The transmitter budget planning is the last step because the EIRP of transmitter is determined by the receiver sensitivity and the path loss. The N-element phased-array improves the EIRP by Ntimes [55], thus the 4-element array can offer extra 6-dB gain.

The maximum signal amplitude is 3 times of the minimum signal amplitude in the 16QAM modulation. In other words, there are up to around 5-dB variation of power while transmit the 16QAM signal. The link budget must confirm that the lowest Tx power is adequate to cover the sensitivity of Rx through the air channel. Thus a back-off 5 dB from Tx OP1dB is considered in the calculation.

The BDA have 6-dBm OP1dB, thus the operated output power is 1-dBm after a 5-dB back-off. By adding the 4-5-dB loss produced by front-end SPDT switch and transition, the operating power is -3 dBm for each Tx antenna. Thus the received power of each Rx antenna is:

which is equal to the Rx sensitivity. From the above calculation, the link budget of the whole module as shown in Fig. 4.7.

In the whole module, the beamformer is placed between antenna arrays and transceiver. We suppose that the transceiver have the same noise figure and OP1dB as the BDA, it is a conservative assumption due to the separated PA and LNA usually have the better performance. From the budget of front-end and back-end sides, the gain budget of the beamformer is estimated. The Butler matrix have the loss of 3 dB, the switches for T/R port selection with array group selection have the loss of 5 dB, the front-end switch used for selecting the antenna array have the 2-dB loss. For the packaging concern, the flip-chip and routing loss for the I/O of the beamformer is estimated as 2 dB for each transition. Finally, as the only active components, the BDAs are expected to offer a bidirectional 16-dB gain to compensate the loss of passive components. As the results, the gain budget of the beamformer is shown in Fig. 4.8.

Note that the beamformer is operating in Rx mode in the receiving path, and operating in Tx mode in the transmitting path. The design of the functional blocks and the whole beamformer are presented in the sections 4.2 and 4.3, respectively.

BM

Fig. 4.7. Link budget of the whole module.

SPDT

Fig. 4.8. Gain budget of the beamformer.