2 Lecture schedule

w4: TSEK38: Radio Frequency • Le1: Introduction (Ch 1) • Le2: Fundamentals of RF system modeling (Ch 2) Transceiver Design • Le3: Superheterodyne TRX design (Ch 3.1) Lecture 6: Receiver Synthesis (I) w6: • Le4: Homodyne TRX design (Ch 3.2) • Le5: Low-IF TRX design (Ch 3.3) Ted Johansson, ISY • Le6: Systematic synthesis (calculations) of RX (Ch 4) [email protected] w7: • Le7: Systematic synthesis (continued) • Le8: Systematic synthesis (calculations) of TX (Ch 5) w8: • Le9: Systematic synthesis (continued)

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

3 4 Lab schedule Repetition of lectures 3 - 5 w6: • We: Lab1a (after Le6): 15-19 (ASGA) • Heterodyne Architecture • Th: Lab1b: 17-21 (SOUT) w7: • Homodyne Architecture • We: Lab1c (after Le8): 15-19 (SOUT) • Low-IF Architecture • Th: Lab1d: 17-21 (EGYP)

Instructions: • One long lab (4 x 4 h). • Lab manual in the Lisam Course Documents/2019/Lab folder. • Supervision (Ted) available at the times above. • To pass: complete and document the exercises in the lab manual, go through with Ted.

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

5 6 Heterodyne with analog IF architecture FDD, one antenna, shared LO1 Heterodyne receiver • Careful frequency planning required (IF frequencies). • Many possible intermodulation effects must be considered. IFA SAW VGA LNA LPF ADC IR IF • Pulling of LO by TX avoided. Filter Filter LPF ADC • Trade-off between sensitivity and selectivity in RX reduced by 2nd stage.

• LO in first stage can be shared, giving different IFTX and IFRX. B I Q LO2R • Trade-off between sensitivity and linearity and power in RX. LO1 VGA in IQ paths

Duplexer avoided • Most of RX gain at IF or BB (after removing blockers). B LO2T • Today heterodyne architecture mostly for TX rather than for RX since IR filters I Q difficult to integrate. LPF DAC RF Filter PA Driver VGA VGA LPF DAC SAW

RF, IF filters and duplexer not integrated → Most of gain at IF (75 %) and RF. matching issues. IF gain is more power efficient

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 7 8 Homodyne receiver (Zero-IF, direct conversion) Low-IF receiver • Fewer components, image filtering avoided – no IR and IF filters • Tradeoff between heterodyne and homodyne • Large DC offset can corrupt weak signal or saturate LNA (LO mixes itself), notch filters or adaptive DC offset cancellation – eg. by DSP baseband • Blocking profile dictates IF = BW/2. control • DC offset and 1/f do not corrupt the signal (cf. homodyne). • IP2 of mixer critical • Image problem reintroduced - close image! • Flicker (1/f) can be difficult to distinguish from signal • Still even-order can result in low-freq. beat: differential • Channel selection with LPF, easy to integrate, noise-linearity-power tradeoffs circuits useful but not sufficient are critical, even-order distortions low-freq. beat: differential circuits useful

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

9 10 Close-image problem IQ imbalance and Image rejection

• IQ imbalance causes image crosstalk and degrades image rejection. Tough requirements for IQ match if image is large, otherwise signal strongly corrupted PSignal IR =10log PImage fIF = ½ BW typical 1+ 2(1+δ )cosε + (1+δ )2 IR =10log 1− 2(1+δ )cosε + (1+δ )2

δ - amplitude (gain) imbalance IR is partly insensitive, ε - phase imbalance e.g. for 0.3d B gain imbalance, IR=30dB keeps up to 2° phase imbalance

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

11 12 Direct conversion transmitter Systematic Receiver Synthesis (1)

• Up-conversion is performed in one step, fLO = fc • 4.1 Introduction • Modulation, e.g. QPSK can be done in the same process • 4.2 Sensitivity, • BPF suppresses harmonics • LO must be shielded to reduce corruption • Receiver Desensitization from transmitter • I and Q paths must be symmetrical and LO in quadrature, otherwise emissions crosstalk • Mismatch between antenna and Rx • 4.3 Intermodulation characteristics • degradation

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 13 Introduction 14 4.1 Introduction

• Many different ways to realize the receiver. • Key design parameters, RX: Somewhat fewer for transmitter. • sensitivity: overall noise figure (4.2) • TDD half-duplex: GSM, WLAN, DECT, …, 5G, • intermodulation: 3rd order , IP3 (4.3) also IP2 • FDD full-duplex: WCDMA, LTE, … • single-tone desensitization (4.4, not!) • Usually different frequency bands for RX and TX. • adjacent/alternate channel selectivity: channel filter, • The TX signal may be 120 dB stronger than the phase noise, (4.5) RX signal. • interference blocking: channel filter, phase noise, (4.5) • FDD receivers are generally trickier to implement • : AGC, ADC (4.6). because of TX leakage into RX, VCO pulling from LO, ... • Book: Full-duplex, but can be applied half-duplex (some not relevant for half-duplex).

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

15 16 4.2 Sensitivity and NF AWGN

• Sensitivity = weakest detectable signal to obtain a Additive white (AWGN) is a basic noise minimum SNR for achieving required BER. model to mimic the effect of many random processes that • Simple model: noise at input is . Directly occur in nature: related to the overall NF of the RX. • Additive because it is added to any noise that might be intrinsic to the information system. • Varies because: signal modulation and • White refers to the idea that it has uniform power across characteristics, signal propagation in the channel, the frequency band for the information system. It is an external noise level. analogy to the color white which has uniform emissions at • In this book: Carrier-to-Noise Ratio (CNR) for the all frequencies in the visible spectrum. analog signal. SNR is for the digital BB. But we will • Gaussian because it has a normal distribution in the time use S and SNR! domain with an average time domain value of zero. • Calculation of sensitivity and NF: pp. 230-232. [Wikipedia]

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

17 18

Eb/N0 vs. Signal-to-Noise Ratio BER vs SNR SNR BER • A better measure of signal-to-noise ratio for digital data is the ratio of energy per bit transmitted (Eb) to the noise • BER depends on SNR of the power density (N0). received signal, i.e. the signal after the receiver block. • SNR (a quantity which can be measured) is related to E /N (an artificial quantity used in comparisons) by • More complicated modulation b 0 schemes require higher SNR for SignalPower E R the same error (trade off between SNR = = b b BW and BER) NoisePower N0 B • It may be possible to correct errors with advanced Forward Error Correction (FEC) Coding Rb is the (bitrate / bandwidth). (reduce BER for the same SNR) B

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 19 20

BER versus SNR in demodulation 2 BER vs. Eb/N0

Dependent on modulation and demodulator implementation.

10-3

(S/N) = Psig/N = (EbRb)/(N0B) (S/N) = Eb/N0 * Rb/B Rb/B ≈ 0.5 - 1.5 (typically)

QPSK Eb = energy per transmitted byte N0= noise power density Rb = bit rate B = channel bandwidth p. 92

For spread-spectrum (e.g. CDMA) SNRmin = (Eb/N0)dB + 10log(Rb/B) Rb/B<<1 => SNRmin < 0

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

21 22

4.2.1 Reference sensitivity and NF 1 Reference sensitivity and NF

Input signal: Min (S/N)o required for BER for the sensitivity level = (S/N)min = SNRmin. Smin: Sensitivity => Sensitivity = Smin[dB] = 10log(PS,min) = -174 + 10log(B) + NFRX + SNRmin (4.2.4) PNi (Nref): integrated thermal noise power within Rx BW. SNRin (CNRin): signal-to-noise => NFRX [dB] = SNRin - SNRout = SNRin - SNRmin = Smin - (-174 dBm/Hz + 10log(B)) - SNRmin (4.2.7) Output signal: Receiver ADC To demodulator Front-End SNR > SNRmin Calculation flow: 1. BER requirement. Reference noise or Receiver noise factor: FRX = (S/N)i / (S/N)o = (PS/PNi) / (S/N)o (4.2.1) 2. Use graph to convert to Eb/N0. 3. Calculate SNRmin.

Integrated thermal noise power within Rx BW: PNi = k * T0 * B (4.2.2) 4. Calculate Smin (sensitivity). Called Reference noise. Ground floor for noise. 5. Calculate NF.

=> RX input signal power: Ps = kT0 * B * FRX * (S/N)o (4.2.3)

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

23 24 Reference sensitivity and NF Reference sensitivity and NF 3

GSM (calc or from the table):

Smin= -102 dBm, Nref = -121 dBm, BERmax= 4×10-2 (4 %) p. 104 => SNRmin = 6.5 + 10*log(271/250) ≈ 6.8 dB (book 8 dB)

=> NFRx = -102 - (-174 - 54) - 6.8 = 11 dB (total max noise)

Specify NFRx = 7 dB: - 3 dB margin because of ~3 dB loss in TDD switch Examples of PNi (reference noise): and RF filter, GSM: -174+10log(200 kHz) = -121 dBm → Vn = 6.3 μV BT: -174+10log(1 MHz) = -114 dBm → Vn = 14.1 μV - 1 dB additional margin. WCDMA: -174+10log(5 MHz) = -107 dBm → Vn = 31.6 μV

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 25 26 4.2.2 Cascaded Noise Figure Noise Figure of Cascaded Stages

Gain is power gain, which depends on the impedance of each stage.

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

27 28 Cascaded Noise Figure: matching 4.2.3 Receiver desensitization, TX leakage • We want to find the receiver desensitization caused by TX leakage into the receiver band for an FDD TRX. • This emission noise passes through the duplexer. • It can be modelled as: • Approach #1: Equivalent Duplexer Noise Figure, • Approach #2: Equivalent Antenna Temperature. • In our calculations, we will assume matching

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

29 30

4.2.3.1 Duplexer noise figure Duplexer noise figure 4

Tx channel

Rx channel Tx noise emission

Antenna port: (4.2.24) BW Noise at Rx port:

PN,TX_ant,RxBand = excess gant_Rx = antenna insertion gain transmitter noise...

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 31 32 Duplexer noise figure 4.2.3.2 Antenna Temperature

Noise factor from antenna to Rx (4.2.24 / 4.2.25): • Different antenna temp compared to TRX temp

effective

Noise figure from antenna to Rx:

• (4.2.32b) is basically the same as (4.2.26b)

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

33 34

Duplexer noise figure 5 Comparison: NF of TDD and FDD Rx 6 • NF of the duplexer can be largely degraded by TX leakage. TDD: NFBPF+Switch = LBPF+Switch= -GBPF+Switch FDD: NFDup = LDup = －GDup

Example: emission noise density in the RX band N 130dBm/Hz, A 44dB attenuation in the duplexer Tx dB = − = NF 8dB, L 2.5dB TDD: Rx = BPF +Switch dB = NFF-E =5.5 dB G = −2.5dB NFRx = LBPF Switch + NFF E Dup dB + dB −

−GDup N − A+174dBm/Hz ⎛ dB Tx dB ⎞ NF =10log⎜10 10 +10 10 ⎟ = 4.4dB NF = 8dB, L = 2.5dB Dup ⎜ ⎟ Rx Dup dB FDD: ⎝ ⎠ NFDup = 4dB NFF-E =5 dB FF −E −1 • If Tx is off, then NFDup = 2.5 dB FRx = FDup + GDup

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

35 36

4.2.4 Mismatch between antenna and Rx Mismatch between antenna and Rx 7

• VSWR = Voltage standing wave ratio Antenna Noisy Rx 2 2 SNR V R I 2 in ( n + a n ) VRa V • Impedance mismatch F = =1+ Ra n Rx 2 SNRout V • Ra Noiseless Causes standing waves and high voltage peaks 2 2 2 Rx V 2 In out • Two waves, forward and reflected: Vn + Ra In FRx =1+ 4kTRa when uncorrelated

2 2 2 Ropt = Vn In = Rn Gn • Varies between 1.5 and 6 may (lower is better). FRx,min −1 ⎛ Ra Ropt ⎞ FRx =1+ ×⎜ + ⎟ FRx,min =1+ 2 Rn Ropt 2 ⎜ R R ⎟ • Special case when load RL is purely resistive: ⎝ opt a ⎠

2 FRx = FRx,min + Rn Ra (Ga − Gopt ) For Ra/Ropt = 2 (3) ⎛ R R ⎞ =1+ R G ⎜ a + opt ⎟ NF = 3 dB, NF = 3.5 dB (4.3 dB) n n ⎜ R R ⎟ min Rx ⎝ opt a ⎠ NFmin= 6 dB, NFRx= 6.8 dB (7.8 dB)

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 37 38 4.3 Intermodulation characteristics Third-order intermodulation product

4.3.1 Intermodulation products and intercept points

General expression:

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

39 40 Intermodulation characteristics Intermodulation characteristics (4.3.3)

IM3 is the main problem, close to the carrier. 2×I • The RX linearity (cascaded IIP for the whole receiver) is the in main cause of intermodulation distortion + LO phase noise + IIP = (3I - IM3)/2 = Iin + ∆3/2 (4.3.10) IP3 test 3 in Δf Δf …

using ∆3 = Iin - IM3 (see prev. slide)

fRx IM3 = 3Iin - 2IIP3 • Allowed maximum degradation Dmax,in caused by noise and Δf < BW/2 interference: IM2 is a minor problem, except for the IP2 test Dmax,in = Sd,i - SNRmin direction conversion receiver. f f Rx int Sd,i = receiver input desired signal = (usually) = Smin_ref + 3 IIP2 = 2Iin - IM2 [dBm] (4.3.9) dB = reference sensitivity level + 3 dB.

IM2 = 2Iin - IIP2,Rx SNRmin = minimum signal to noise/interference ratio

4.3.2 Cascaded IP… Can be rather complicated when frequency selectivity and matching are considered.

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

41 42 4.3.3.1 Degradation because of inherent noise Alternative calculations (not in the book)

• Nnf = receiver inherent noise (from NF, converted to the input port): • Calculations of SNRmin when degraded by noise, linearity, etc.

Nnf = -174 + 10log(B) + NFRX. • PIM# is the cascaded IIPRX (mostly IIP3,RX and IIP2,RX).

• Nnf reduces the Dmax,in: • Nother = other noise sources, usually assumed <= -100 dBm

Pin (S / N)min = , Pin > Pin,min (+3dB) Nin + Nref (FRx -1)

Nin = Nref + PIM # + NOther

Headroom for PIM# limited by SNRmin and • Example: assume GSM with Smin = reference sensitivity = -102 dBm => Pin PIM # = - FRx Nref - NOther NOther esp. LO phase noise + spurious Sd,i = -99 dBm, SNRmin = 8 dB, NFRX = 10 dB, B = 250 kHz. (S / N)min Additionally PIM2 limited by self-mixing of the => Da = 10log(1.9953E-11 - 9.9527E-12) = -110 dBm (-107 with Dm only) test signal (interferer) as discussed for Zero- (Dm=10^((-99-8)/10);Nf=10^((-174+10+10*log10(250E3))/10);Da=10*log10(Dm-Nf)) Rx noise floor IF Rx

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 43 44

Nin estimation 8 IP3/IP2 requirement estimation 8 P (S / N) = in , P > P (+3dB) min N + N (F -1) in in,min in ref Rx N I = -38 dBm ⎛ Sin −SNRmin −174+10log BW +NFRx Other dBm ⎞ P in P = 10log⎜10 10 −10 10 −10 10 ⎟ in IM # dBm ⎜ Nin = Nref + PIM # + NOther P = - F N - N ⎟ IM # Rx ref Other Sin = -86 dBm ⎝ ⎠ (S / N)min BW = 1.25 MHz N = −100dBm ⇒ P ≅ −96dBm Other dBm IM # dBm SNRmin = 8 dB P N IM # dBm Other dBm ⎛ ⎞ NFRx = 10 dB When all distortion 10log⎜10 10 +10 10 ⎟ 3(−38) − (−96) ⎜ ⎟ P = -30 dBm IIP = = −9dBm is attributed to IP3. ⎝ ⎠ bl 3,Rx 2 Iin = -38 dBm S = -86 dBm ⎛ Sin −SNRmin −174+10log BW +NFRx ⎞ in = 10log⎜10 10 −10 10 ⎟ ≅ −94.5dBm 3(−38) − (−99) ⎜ ⎟ IIP 7.5dBm When distortion is BW = 1.25 MHz ⎝ ⎠ 3,Rx = = − 2 attributed both to SNRmin = 8 dB IIP2,Rx = 2(Pbl − 3) − (−99) = 33dBm IP3 and IP2 by half NFRx = 10 dB + headroom for extra each. noise and distortion

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

45 46 IIP2 of RX 4.3.3.3 Phase noise and spur degradation

• Phase noise (PN) and spurious emissions (SP) from LO contaminate the signal. Will increase the IIPx,min.

• Intermodulation interference tones will mix with the LO PN (NPN = average phase noise density) and/or spurios (NSP at ) and generate in-band noise and spurs, which degrades the CNR.

• Covered by Nother: NPN + NSp <= Nother

Iin Iin

PN( f) = PN(2 f)+6dB Δ Δ -6 dB/oct Δf = n × BW

Δf Δf

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

47 48 Phase noise and spur degradation Phase noise estimation in two-tone test

9 • PN(∆f) = average PN density over Rx BW at offset ∆f PN (Δf )+Iin +10log BW PN (2Δf )+Iin +10log BW 10 10 N PN + NSp ≈ 2×10 + 2×10 NSp ≈ NPN PN(Δf ) = N PN (Δf ) −10log BW − Iin [dBc/Hz] PN (Δf )+Iin +10log BW ⎛ −6dB ⎞ 2 10 10 ⎜1 10 10 ⎟ PN(Δf) = PN(2Δf)+6dB = × ⎜ + ⎟ ⎝ ⎠ PPN|dBm/Hz (4.3.37) ⎛ PN (Δf )+Iin +10log BW ⎞ N =10log⎜2.5×10 10 ⎟ PN +Sp dBm ⎜ ⎟ PN (Δf )+Iin +10log BW PN (2Δf )+Iin +10log BW ⎝ ⎠ N =10 10 +10 10 PN PN( f ) I 10log BW 4dB N I = Δ + in + + ≤ Other dBm NSp (Δf )+Iin NSp (2Δf )+Iin in Iin 10 10 NSp =10 +10 PN(Δf ) ≤ N − I −10log BW − 4dB -6dB/oct Other dBm in • In practice, we set NSp ≈ NPN = −100 + 36 − 61− 4 PN(Δf ) ≤ −129dBc/Hz → −135dBc/Hz @Δf with 6 dB margin PN(Δf) = PN(2Δf)+6dB Δf Δf Δf = n × BW From this, the impact on IIP3 can be calculated (eq 4.3.51).

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson 49 50 Phase noise estimation, IP2/Blocking test IIP2 of RX additional calculations

• SNR budget under IP2 test (homodyne Rx).

• NPN+SP + NTX,SM ≤ Nother • The contribution by Tx leakage and self modulation is estimated to -107 dBm in this example (FDD). 2×-30dBm = -27dBm

Iin Iin • +6 dB is equivalent to 2x2 Pn (PN of 2 tones and 2x for LO spurs).

NOther 107dBm ⎛ dBm − ⎞ N ≤ 10log⎜10 10 −10 10 ⎟ = −101dBm PN +Sp dBm ⎜ ⎟ ⎝ ⎠

foff PN( foff ) + (Iin + 3dB) +10log BW + 3dB ≤ −101dBm

PN( foff ) ≤ −138dBm

PN ( foff )+Iin +3dB+10logBW N + N ≈ 2 ×10 10 For foff >> Δf the PN approx. the same PN Sp

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

additional calculations 51 52 Estimation of IIP2 additional calculations Estimation of Tx leakage vs IIP2

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson

53 54 Estimation of Tx leakage additional calculations Estimation of Tx leakage additional calculations

TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson TSEK38 Radio Frequency Transceiver Design 2019/Ted Johansson www.liu.se