Appendix AFormula Sheet
The equations of this tutorial on two pages, in the notation of Chapter 1: \(T_\mathrm{b}\) = 1 UI = 200 ps, \(f_\mathrm{N}=2.5\) GHz, symbols \(a_k=\pm1\), TX levels \(\pm A\) with \(A=0.5\) V, cursors \(p_k\) of a 1 V, 1-UI pulse.
| Formula | Meaning |
|---|---|
| Signals and BER (Chapters 2 and 4) | |
| \(f_\mathrm{b}=1/T_\mathrm{b},\quad f_\mathrm{N}=1/(2T_\mathrm{b})\) | bit rate and Nyquist frequency: 5 Gb/s, 2.5 GHz |
| \(y(t)=A\sum_n a_n\,p(t-nT_\mathrm{b})+n(t)\) | received signal: superposition of shifted pulse responses |
| \(p_k=p(t_\mathrm{s}+kT_\mathrm{b})\) | cursors: \(p_0\) main, \(k<0\) pre-, \(k>0\) post-cursors |
| \(\mathrm{EH}=2A\big(p_0-\sum_{k\neq0}\lvert p_k\rvert\big)\) | peak-distortion (worst-case) inner eye height |
| \(\mathrm{BER}=Q\!\left(\mathrm{EH}/2\sigma_\mathrm{n}\right),\ Q(x)=\tfrac12\operatorname{erfc}(x/\sqrt2)\) | BER for Gaussian noise of rms \(\sigma_\mathrm{n}\) at the slicer |
| \(\mathrm{EH}\ge2Q\sigma_\mathrm{n}=14.07\,\sigma_\mathrm{n}\) | eye needed for \(10^{-12}\) (\(Q=7.03\)) |
| \(S(f)=A^2T_\mathrm{b}\,\mathrm{sinc}^2(fT_\mathrm{b})\) | NRZ power spectral density; first null at \(f_\mathrm{b}\) |
| \(M=\mathrm{EH}_\mathrm{ISI}-2Q\sigma_\mathrm{n}-\Delta V_\mathrm{jit}-2V_\mathrm{os}-V_\mathrm{sens}\) | linear voltage margin at the sampler |
| Channel (Chapter 3) | |
| \(v=c/\sqrt{\varepsilon_r},\quad t_d=\ell/v\) | propagation speed and delay; FR-4: 6.67 ps/mm |
| \(\Gamma=\frac{Z_L-Z_0}{Z_L+Z_0},\quad \mathrm{RL}=-20\log_{10}\lvert S_{11}\rvert\) | reflection coefficient and return loss |
| \(\alpha_D\approx\pi f\sqrt{\varepsilon_r}\tan\delta/c\) | dielectric loss (Np/m), grows as \(f\); skin loss grows as \(\sqrt f\) |
| \(\lvert S_{11}\rvert=x/\sqrt{4+x^2},\ x=\omega Z_0C_p\) | pad capacitance on a doubly terminated node |
| \(f_\mathrm{3dB}=1/(\pi Z_0C_p)\) | pad pole (node sees \(Z_0/2\)) |
| \(\mathrm{IL}(f_\mathrm{N})-\mathrm{IL}(f_\mathrm{N}/2)>9.54\) dB | PAM4 can win over NRZ only above this octave loss |
| Equalization (Chapters 6, 7 and 8) | |
| \(y_n=\sum_k c_k x_{n-k},\quad\sum_k\lvert c_k\rvert=1\) | TX FFE under the peak-swing constraint |
| \(\mathrm{DE}=20\log_{10}\frac{\lvert c_0\rvert+\lvert c_1\rvert}{\lvert c_0\rvert-\lvert c_1\rvert}\) | de-emphasis of a 2-tap FFE (\(c_1=-0.25\): 6 dB) |
| \(H_\mathrm{CTLE}(s)=\frac{1}{A}\frac{1+s/\omega_z}{(1+s/\omega_{p1})(1+s/\omega_{p2})}\) | CTLE model, peaking \(A=\omega_{p1}/\omega_z\) |
| \(A=1+g_\mathrm{m}R_S/2,\ \omega_z=\frac{1}{R_SC_S},\ \omega_{p2}=\frac{1}{R_DC_\mathrm{L}}\) | degenerated differential pair |
| \(z_n=y_n-\sum_k w_k\hat d_{n-k},\quad w_k=A\,p_k\) | DFE; zero-forcing taps equal the post-cursors |
| \(t_\mathrm{CK\to Q}+t_\mathrm{settle}+t_\mathrm{setup}\le T_\mathrm{b}\) | DFE tap-1 timing (direct feedback) |
| \(w_k\leftarrow w_k+\mu\,\operatorname{sgn}(e_n)\,\operatorname{sgn}(\hat d_{n-k})\) | sign-sign LMS adaptation |
| Formula | Meaning |
|---|---|
| Circuits (Chapters 9 and 10) | |
| \(\Delta V=I_TR_\mathrm{L},\ \tau=R_\mathrm{L}C_\mathrm{L},\ P=V_\mathrm{CC}I_T\) | CML gate: swing, time constant, static power |
| \(V_{d,\mathrm{pp}}=I_TZ_0\) (CML), \(I=V_{d,\mathrm{pp}}/4Z_0\) (SST) | driver current, doubly terminated |
| \(g_\mathrm{m}=I_C/V_T\); full switching at \(\approx4V_T\) | HBT pair: 38.7 mS/mA, about 100 mV |
| \(\tau=C/g_\mathrm{m}\) (StrongARM), \(\tau=\frac{R_\mathrm{L}C}{g_\mathrm{m}R_\mathrm{L}-1}\) (CML) | latch regeneration time constant |
| \(P_\mathrm{meta}=\frac{2V_L}{A\,\Delta V}\,e^{-T/\tau}\) | metastability probability per decision |
| \(\sigma_{\Delta V_t}=A_{VT}/\sqrt{WL}\) | Pelgrom mismatch (offset) |
| \(V_\mathrm{sens}\approx2Q\sigma_s+V_\mathrm{min}+2\lvert V_\mathrm{os,res}\rvert\) | sampler sensitivity |
| \(\overline{v_n^2}=4kTR\,\Delta f,\quad v_{n,\mathrm{rms}}=\sqrt{kT/C}\) | resistor and sampled (\(kT/C\)) noise |
| Jitter, PLL and CDR (Chapters 11, 12 and 13) | |
| \(\mathrm{TJ}=\mathrm{DJ}_{\delta\delta}+2Q\,\sigma_\mathrm{RJ}\) | total jitter at a BER (dual-Dirac) |
| \(\sigma_\phi^2=2\!\int\!\mathcal{L}(f)\,df,\quad\sigma_t=\sigma_\phi/(2\pi f_0)\) | phase noise to rms jitter |
| \(\omega_n=\sqrt{\frac{I_\mathrm{CP}K_\mathrm{VCO}}{2\pi NC_1}},\ \zeta=\frac{R}{2}\sqrt{\frac{I_\mathrm{CP}K_\mathrm{VCO}C_1}{2\pi N}}\) | CP-PLL natural frequency and damping |
| \(f_0=\frac{1}{2\pi\sqrt{LC}},\ R_p=Q\omega_0L,\ g_\mathrm{m}R_p/2\ge1\) | LC tank and start-up |
| \(K_\mathrm{BB}=\mathrm{TD}\sqrt{2/\pi}/\sigma_\mathrm{j}\) | bang-bang gain, linearized by jitter |
| \(S=K_P\cdot\mathrm{LSB}/T_\mathrm{upd},\ \mathrm{JTOL}\approx\mathrm{TM}+2S/\omega\) | BB-CDR slew rate and slew-limited JTOL |
| \(\Delta t_\mathrm{PI}=T_\mathrm{clk}/N_\mathrm{steps},\ \text{dither}\approx2S\,T_\mathrm{lat}\) | PI step (6.25 ps) and limit-cycle dither |
| Flow and test (Chapters 14 and 15) | |
| \(E_b=P/f_\mathrm{b}\) | energy per bit (216 mW \(\to\) 43 pJ/b) |
| \(N_\mathrm{bits}\ge3/\mathrm{BER}\) | error-free bits for a BER claim at 95 % confidence |
| \(N\ge10\,\ell f_\mathrm{max}/v,\quad\Delta\hat\sigma/\sigma\approx1/\sqrt{2N}\) | lumped-line sections; Monte Carlo accuracy |
Generated from book/tutorial/chapters/A_appendices.tex. Figure, table and equation numbers match the PDF edition.