100Gb/s C2M Channel Simulation

Transcription

1 HSI-PR1-1 1Gb/s C2M Channel Simulation IEEE82.3ck July, 218 Hiroki Miyaoka, Toshiaki Sakai Socionext Inc. sakai_3ck_1a_718 1

2 Agenda HSI-PR Overview 2. C2M Channel Simulation 2.1 Simulation Conditions 2.2 Channel for Simulation 2.3 Simulation Results 3. Conclusion sakai_3ck_1a_718 2

3 1. Overview HSI-PR1-1 Feasibility of 1Gb/s C2M transmission with additional channels and equalizer configurations. Comparison with COM (test version) results Simple IL (insertion loss) discussion is not enough. channels simulated # Contributor module IL (nopkg) SDD22 PSXT note 7 Mellitz * QSFP -9.5dB -18.3dB -36.4dB 9dB_BC 8-1.8dB -1.6dB -33.8dB 1dB_WC dB -2.6dB -37.9dB 11dB_BC dB -16.6dB -34.5dB 12dB_WC dB -18.9dB -4.dB 13dB_BC dB -13.2dB -38.dB 14dB_WC * sakai_3ck_1a_718 3

4 HSI-PR C2M Channel simulation 2.1 Simulation Conditions sakai_3ck_1a_718 4

5 2.1.1 Simulation Model HSI-PR1-1 victim aggressor symbol description Av, Afe, Ane Tx output amplitude : victim, aggressor (FEXT, NEXT) Rd single ended termination resistor Cd single ended device capacitance HFFE(f) Tx FFE transfer function S (tp), S (rp) PKG model Tx/Rx S () channel under test H Rx (f) Rx noise filter H CTLE (f) Rx CTLE transfer function η one-sided noise spec Add Dual-Dirac jitter, peak to peak Tx : before FFE Rx : considered as margin sakai_3ck_1a_718 5 σrj random jitter, RMS Tx : before FFE Rx : before FFE/DFE

6 2.1.2 Simulation Set Up Static Channel Model Simulation Behavior model using MatLab PAM4 at 58.Gbd to see margin and other applications. (conservative) Jitter, noise and crosstalk are considered. Tx jitter, (DJ, RJ) included RJrms is the same as 5G-PAM4 (conservative). Rx/CDR jitter (DJ) are considered as opening margin. (EW5, EH5) Crosstalk noise in channel S-parameter Impedance are nominal : 5-ohm single ended T-spaced FFE Rx FFE parameters are set to minimize ISI. CTLE coefficients are optimized for each channel. (See back up slides.) PKG model is based on current design and material (GZ41). sakai_3ck_1a_718 6 item value unit modulation pattern PAM4 PRBS13Q baud rate 58 Gbd DJ_Tx 6, mui HSI-PR1-1 RJ_Tx 1, mui_rms EOJ_Tx UI SNR_Tx 32.5, none db Rt_Tx 5 ohm Tx_FFE 4/2 tap/pre Cd_Tx/Rx 1 ff Cp Tx/Rx extracted ff Rx FFE + DFE Rx fr 5/ + 1/ + 1 3/4 fb FFE tap/pre + DFE tap Av.8 Vppd AVx 1.2 Vppd BER 1.E-5 eta 8.2E-9, V 2 /GHz DJ_Rx UI RJ_Rx 1, mui_rms Rt_Rx 5 ohm

7 HSI-PR Channel simulation 2.2 Channel for Simulation sakai_3ck_1a_718 7

8 Insertion Loss [db] 2.2 C2M : channel insertion loss HSI-PR1-1 5 Frequency [GHz] CAUI-4 C2M 5GAUI-1 C2M "1GAUI-1 C2M" CAUI-4 C2M (NRZ) Gbd PAM4 x2 scaling 1GAUI-2 C2M (PAM4) Gbd scaled 1GAUI-1 C2M (PAM4) Gbd If PAM4 modulation is selected, 1dB~12dB target insertion loss at Nyquist frequency ( GHz, scaled frequency) is an option. sakai_3ck_1a_718 8

9 HSI-PR Channel Configuration #7~12, Mellitz* Flyover cable 3 Meg6 like PCB (tand ~.4) 2 Tachyon like PCB (tand ~.25) 3 Meg6 like PCB (tand ~.4) 12mm 42mm 34AWG twinax cable 1.5 Meg6 like PCB 1.5 Tachyon like PCB FQSFP Line card BGA PKG breakout - With vias/footprint - 14-layer, 93mil thick No on-module AC coupling capacitors. No PKG breakout added PKG model host (Tx) : 27mm PKG module (Rx) : 12mm PKG both design extracted sakai_3ck_1a_718 worst case base case * : 9

10 S-parameter [db] Channel Characteristics (#7,9dB_BC) SDD dB SDD dB HSI-PR SDD dB -6 PSXT -36.4dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 1

11 S-parameter [db] Channel Characteristics (#8,1dB_WC) SDD21-1.8dB HSI-PR SDD22-1.6dB -6 SDD dB PSXT -33.8dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 11

12 S-parameter [db] Channel Characteristics (#9,11dB_BC) SDD dB SDD22-2.6dB HSI-PR SDD dB -6 PSXT -37.9dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 12

13 S-parameter [db] Channel Characteristics (#1,12dB_WC) SDD dB HSI-PR SDD dB -6 SDD dB PSXT -34.5dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 13

14 S-parameter [db] Channel Characteristics (#11,13dB_BC) SDD dB SDD dB HSI-PR SDD dB PSXT -4.dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 14

15 S-parameter [db] Channel Characteristics (#12,14dB_WC) SDD dB HSI-PR SDD dB -6 SDD dB PSXT -38.dB scaled 1GAUI-1 C2M SDD Frequency [GHz] sakai_3ck_1a_718 15

16 S-parameter [db] Channel Characteristics : IL (SDD21) SDD21 vs frequency best case 9dB 11dB 13dB HSI-PR worst case 1dB 12dB 14dB Frequency [GHz] sakai_3ck_1a_718 16

17 S-parameter [db] Channel Characteristics : crosstalk crosstalk (PSXT) vs frequency HSI-PR worst case 1dB 12dB 14dB -4-6 best case 9dB 11dB 13dB Frequency [GHz] sakai_3ck_1a_718 17

18 2.2.2 PKG Characteristics HSI-PR1-1 Designed PKG models are used. S-parameter [db] SDD21: SDD22: SDD22 SDD11 28GHz SDD21 Frequency [GHz] simulated 27mm-trace S-parameter [db] SDD21: SDD22: SDD22 SDD11 SDD21 Frequency [GHz] simulated 12mm-trace -5 28GHz note : No xtalk is considered in these PKG models. 18 sakai_3ck_1a_718 18

19 HSI-PR Channel simulation 2.3 Simulation Results sakai_3ck_1a_718 19

20 EW5 [mui] EW5/EH5 vs IL EH5 [mv] HSI-PR EW5 vs IL 8 EH5 vs IL 12 best case 7 best case EW5_FFE_BC EW5_FFE_WC criteria EW5 > 1mUI Similar IL, but large EW5 difference EH5_FFE_BC EH5_FFE_WC Similar IL, but large EH5 difference. criteria EH5 > 2mV 2 sakai_3ck_1a_718 worst case IL, no PKG [db] 2 1 worst case IL, no PKG [db] item unit #221 #223 #225 #227 #229 #231 channel type 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 14dB_WC no-pkg db upp EW5 mid mui low upp EH5 mid mv low channel

21 EW5 [mui] EW5/EH5 vs IL (with PKG, Cd effect) EH5 [mv] HSI-PR EW5 vs IL 9dB 11dB 13dB criteria EW5>1mUI SFP Mellitz BC dB EH5 vs IL 13dB 11dB SFP Mellitz BC Jane Tracy Rx6 1dB 12dB 14dB Mellitz WC Tracy Rx5 14dB 1dB 12dB IL, w/ PKG and Cd effect [db] criteria EH5>2mV Jane Mellitz WC 1dB 12dB 14dB Tracy Rx6 1dB 12dB 14dB Tracy Rx IL, w/ PKG and Cd effect [db] - The effect of IL (insertion loss) is not so obvious. - Other factors, like crosstalk, ILD and/or reflection affect. - It may be useful to utilize COM as a reference to design channels, even if it is as informative spec. sakai_3ck_1a_ Mellitz : Tracy : Jane : SFP : 21

22 EW5 [mui] EW5 [mui] EW5/EH5 vs IL (no PKG) : XT, noise/jitter effect noise and jitter : similar degradation This difference comes from ILD? EW5 vs IL EW5_FFE_BC EW5_XT_BC EW5_NJ_BC EW5_FFE_WC EW5_XT_WC EW5_NJ_WC This difference comes from ILD? EH5 vs IL HSI-PR1-1 EH5_FFE_BC EH5_XT_BC EH5_NJ_BC EH5_FFE_WC EH5_XT_WC EH5_NJ_WC noise and jitter : similar degradation 15 XT : small degradation (BC) 8 XT : small degradation (BC) 1 5 criteria EW5 > 1mUI XT : large degradation (WC) XT : large degradation (WC) criteria EH5 > 2mV X,NJ : with crosstalk and noise/jitter NJ : no crosstalk, with noise/jitter NO : no crosstalk, no noise/jitter sakai_3ck_1a_ item unit #221 #233 #234 #223 #235 #236 #225 #237 #238 #227 #239 #24 #229 #241 #242 #231 #243 #244 channel (IL, db) 9dB_BC (9.5) 1dB_WC (1.8) 11dB_BC (11.7) 12dB_WC (12.1) 13dB_BC (13.9) 14dB_WC (14.2) XT, Noise/Jitter X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO EW5 EH5 IL, no PKG [db] IL, no PKG [db] upp mid mui low upp mid mv low channel

23 EW5 [mui] EW5/EH5 vs COM EH5 [mv] HSI-PR1-1 EW5 (mid-) vs EH5 (mid-) vs 14 Mellitz BC criteria EW5 > 1mUI Tracy Rx5 Tracy Rx6 12dB 14dB Mellitz WC 1dB 11dB 13dB 9dB SFP y = x R² = Tracy Rx5 Tracy Rx6 12dB 14dB Mellitz WC 1dB 11dB Mellitz BC 13dB 9dB SFP y = 22.93x R² =.855 criteria EH5 > 2mV COM [db] COM condition : 58Gbd - Rx-FFE (5-tap, 4-post) included (test version COM 2.34 by Mellitz) - CTLE pole and zero, fb/2.5(fp1,fz), fb(fp2), fb/8(lf) sakai_3ck_1a_ COM [db] COM values are well correlated to EW5/EH5. - For C2M channel estimation, COM can a good tool. - These COM values for C2M channels are not direct criteria of good or bad channels, for now. Mellitz : Tracy : SFP :

24 HSI-PR Conclusion sakai_3ck_1a_718 24

25 3. Conclusion HSI-PR C2M channel discussion is often based on IL (insertion loss) only. - However, as this result shows, other factors affect opening significantly. - Using COM may be a solution, even as informative. 2. The degradation of opening depends on XT and nonnoise(n)/jitter(j) factors. - Without XT and N/J, there still difference between BC channel and WC channel with same IL. - It may come from ILD or higher than Nyquist IL suck. Need further investigation. 3. Further investigation needed. - Comparison with COM including Rx FFE - Equalization - PKG parameters (COM PKG model or standard PKG S-parameter) - Other Channel models, parameters 25 sakai_3ck_1a_718 25

26 Thank you!

27 HSI-PR1-1 backup slides 27 sakai_3ck_1a_718 27

28 A1 Simulation Summary (Mellitz *, C2M 9dB~14dB) HSI-PR1-1 item unit #221 #222 #223 #224 #225 #226 #227 #228 #229 #23 #231 #232 type 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 14dB_WC Channel IL nopkg [db] PSXT [db] c(-2) Tx FFE c(-1) c() c(1) p1_hf p2_hf CTLE z_hf p_lf z_lf Rx ffe() ffe(1) FFE ffe(2) ffe(3) ffe(4) DFE tap EW5 EH5 upp mid mui low upp mid mv low sakai_3ck_1a_718 channel 28

29 A2 Simulation Summary (Mellitz C2M, XT, noise/jitter) HSI-PR1-1 item unit #221 #233 #234 #223 #235 #236 #225 #237 #238 #227 #239 #24 #229 #241 #242 #231 #243 #244 type 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 14dB_WC IL nopkg [db] Channel XT yes no yes no yes no yes no yes no yes no SNDR [db] 32.5 N/A 32.5 N/A 32.5 N/A 32.5 N/A 32.5 N/A 32.5 N/A Tx RJrms [mui] DJ [mui] Rx eta [v^2/ghz] 8.2E-9 8.2E-9 8.2E-9 8.2E-9 8.2E-9 8.2E-9 RJrms [mui] c(-2) Tx FFE c(-1) c() c(1) Rx p1_hf p2_hf CTLE z_hf p_lf z_lf ffe() ffe(1) FFE ffe(2) ffe(3) ffe(4) DFE tap upp EW5 mid mui low upp EH5 mid mv low sakai_3ck_1a_718 channel 29

30 A3.1 Simulation Summary (EW5, Mellitz* 9dB~14dB) HSI-PR1-1 EW5 [mui] criteria EW5 > 1mUI dB_BC 5-FFE 1-DFE EW5 11dB_BC 5-FFE 1-DFE 13dB_BC 5-FFE 1-DFE EW5_U EW5_M EW5_L Rx 1-DFE is slightly worse than 5-FFE. - 1-DFE power is lower than 5-FFE dB_WC 5-FFE 1-DFE 12dB_WC 5-FFE 1-DFE 14dB_WC 5-FFE 1-DFE 2 item unit #221 #222 #223 #224 #225 #226 #227 #228 #229 #23 #231 #232 channel Samtec 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 14dB_WC Rx_FFE pre/post /4 / /4 / /4 / /4 / /4 / /4 / Rx_DFE tap upp EW5 mid mui low upp EH5 mid mv low sakai_3ck_1a_718 * 3

31 A2.2 Simulation Summary (EH5, Mellitz* 9dB~14dB) EH5 [mv] dB_BC 5-FFE 1-DFE EH5 11dB_BC 13dB_BC 5-FFE 1-DFE 5-FFE 1-DFE HSI-PR1-1 EH5_U EH5_M EH5_L 5 4 1dB_WC 12dB_WC 14dB_WC criteria EH5 > 2mV FFE 1-DFE 5-FFE 1-DFE 5-FFE 1-DFE 1 item unit #221 #222 #223 #224 #225 #226 #227 #228 #229 #23 #231 #232 channel Samtec 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 14dB_WC Rx_FFE pre/post /4 / /4 / /4 / /4 / /4 / /4 / Rx_DFE tap upp EW5 mid mui low upp EH5 mid mv low sakai_3ck_1a_718 * 31

32 A4.1 Simulation Summary (channel, EW5 vs XT,Noise/Jitter) HSI-PR1-1 EW5 [mui] dB_BC 11dB_BC 13dB_BC 1dB_WC 12dB_WC EW5_U EW5_M EW5_L 14dB_WC 2 small XT effect for BC channel small XT effect for BC channel small XT effect for BC channel criteria EW5 > 1mUI 15 1 larger XT effect for WC channel larger XT effect for WC channel larger XT effect for WC channel 5 item unit #221 #233 #234 #223 #235 #236 #225 #237 #238 #227 #239 #24 #229 #241 #242 #231 #243 #244 #221 #233 #234 #223 #235 #236 #225 #237 #238 #227 #239 #24 #229 #241 #242 #231 #243 #244 channel (IL, db) 9dB_BC (9.5) 1dB_WC (1.8) 11dB_BC (11.7) 12dB_WC (12.1) 13dB_BC (13.9) 14dB_WC (14.2) XT, Noise/Jitter X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO upp mid mui EW5 low upp mid mv EH5 low X,NJ : with crosstalk and noise/jitter NJ : no crosstalk, with noise/jitter NO : no crosstalk, no noise/jitter sakai_3ck_1a_718 channel : 32

33 A4.2 Simulation Summary (channel, EH5 vs XT,Noise/Jitter) HSI-PR1-1 EH5 [mv] dB_BC 11dB_BC 13dB_BC EH5_U EH5_M EH5_L small XT effect for BC channel 1dB_WC larger XT effect for WC channel small XT effect for BC channel 12dB_WC larger XT effect for WC channel small XT effect for BC channel 14dB_WC larger XT effect for WC channel criteria EH5 > 2mV 4 2 item unit #221 #233 #234 #223 #235 #236 #225 #237 #238 #227 #239 #24 #229 #241 #242 #231 #243 #244 channel (IL, db) 9dB_BC (9.5) 1dB_WC (1.8) 11dB_BC (11.7) 12dB_WC (12.1) 13dB_BC (13.9) 14dB_WC (14.2) XT, Noise/Jitter X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO X,NJ NJ NO upp mid mui EW5 low upp mid mv EH5 low X,NJ : with crosstalk and noise/jitter NJ : no crosstalk, with noise/jitter NO : no crosstalk, no noise/jitter sakai_3ck_1a_718 channel 33

34 A4.3 Simulation Summary (channel, EW5 vs XT,Noise/Jitter) HSI-PR1-1 EW5 difference [mui] No significant difference between BC and WC channels for noise/jitter EW5_U EW5_M EW5_L 9dB_BC 11dB_BC 13dB_BC 1dB_WC 12dB_WC 14dB_WC Small XT effect for BC channels Larger XT effect for WC channels 4 2 item unit #233-#221 9dB_XT #234-#233 9dB_NJ #235-#223 1dB_XT #236-#235 1dB_NJ #237-#225 11dB_XT #238-#237 11dB_NJ #239-#227 12dB_XT #24-#239 12dB_NJ #241-#229 13dB_XT #242-#241 13dB_NJ #243-#231 14dB_XT #244-#243 14dB_NJ channel (IL, db) 9dB_BC (9.5) 1dB_WC (1.8) 11dB_BC (11.7) 12dB_WC (12.1) 13dB_BC (11.7) 14dB_WC (12.1) XT, Noise/Jitter XT NJ XT NJ XT NJ XT NJ XT NJ XT NJ upp EW5 mid mui diff low upp EH5 mid mv diff low XT : difference between with and without crosstalk NJ : difference between with and without noise/jitter channel sakai_3ck_1a_718 34

35 A4.4 Simulation Summary (channel, EW5 vs XT,Noise/Jitter) HSI-PR1-1 EWH difference [mv] EH5_U EH5_M EH5_L 9dB_BC 11dB_BC 13dB_BC 1dB_WC 12dB_WC 14dB_WC Small XT effect for BC channels No significant difference between BC and WC channels for noise/jitter. Large XT effect for WC channels dB_XT 9dB_NJ 1dB_XT 1dB_NJ 11dB_XT 11dB_NJ 12dB_XT 12dB_NJ 13dB_XT 13dB_NJ 14dB_XT 14dB_NJ item unit #233-#221 #234-#233 #235-#223 #236-#235 #237-#225 #238-#237 #239-#227 #24-#239 #241-#229 #242-#241 #243-#231 #244-#243 channel (IL, db) 9dB_BC (9.5) 1dB_WC (1.8) 11dB_BC (11.7) 12dB_WC (12.1) 13dB_BC (11.7) 14dB_WC (12.1) XT, Noise/Jitter XT NJ XT NJ XT NJ XT NJ XT NJ XT NJ upp EW5 mid mui diff low upp EH5 mid mv diff low XT : difference between with and without crosstalk NJ : difference between with and without noise/jitter sakai_3ck_1a_718 channel 35

36 A5.1 Simulation Summary (COM sim, fb and DFE effect) HSI-PR1-1 COM [db] G, 5-FFE/-DFE 53G, 5-FFE/-DFE 58G, 5-FFE/1-DFE 53G, 5-FFE/1-DFE -1.5 TE_Rx5 TE_Rx6 YM Mellitz_ Samtec_9 Mellitz_ Samtec_1 Mellitz_ Samtec_11 Mellitz_ Samtec_12 Mellitz_ Samtec_13 Mellitz_ Samtec_14 COM [db] Tracy_Rx5 Tracy_Rx6 SFP 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC 29GHz IL 26GHz fb DFE [Gbd] With 5-FFE + 1-DFE, COM values become better, however, it requires more power and area. sakai_3ck_1a_718 Mellitz : Tracy : SFP : 36

37 A5.2 Simulation Summary (COM sim, fb and DFE effect) delta-com [db] HSI-PR G->53G, -DFE +DFE 58G->53G, 1-DFE +DFE TE_Rx5 TE_Rx6 YM Samtec_9 Mellitz_ Samtec_1 Mellitz_ Samtec_11 Mellitz_ Samtec_12 Mellitz_ Samtec_13 Mellitz_ Samtec_14 Mellitz_ delta COM [db] Tracy_Rx5 Tracy_Rx6 SFP 9dB_BC 1dB_WC 11dB_BC 12dB_WC 13dB_BC (.5) IL G->53G, -DFE G->53G, 1-DFE DFE DFE With 5-FFE + 1-DFE, COM values become better, however, it requires more power and area. And the effect is channel dependent. sakai_3ck_1a_718 Mellitz : Tracy : SFP : 37

38 Vdiff [mv] B- Rx FFE/DFE output log 1 (BER) (time) BER bathtub curve log1(ber) HSI-PR1-1 - simulation results 4 voltage bathtub curve log 1 (BER) (voltage) Colors are corresponding to each line BER time [UI] 1UI = sakai_3ck_1a_718 38

39 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-1 Simulation Result (#221/#222) log 1 (BER) #221 9dB_BC, FFE -1 time [UI] 1 log 1 (BER) #222 9dB_BC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #221 #222 #225 #226 #229 #23 type 9dB_BC 11dB_BC 13dB_BC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low upp mid mv low

40 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-2 Simulation Result (#225/#226) log 1 (BER) #225 11dB_BC, FFE -1 time [UI] 1 log 1 (BER) #226 11dB_BC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #221 #222 #225 #226 #229 #23 type 9dB_BC 11dB_BC 13dB_BC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low upp mid mv low

41 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-3 Simulation Result (#229/#23) log 1 (BER) #229 13dB_BC, FFE -1 time [UI] 1 log 1 (BER) #23 13dB_BC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #221 #222 #225 #226 #229 #23 type 9dB_BC 11dB_BC 13dB_BC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low upp mid mv low

42 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-4 Simulation Result (#223/#224) log 1 (BER) #223 1dB_WC, FFE -1 time [UI] 1 log 1 (BER) #224 1dB_WC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #223 #224 #227 #228 #231 #232 type 1dB_WC 12dB_WC 14dB_WC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low 3 35 upp mid mv low

43 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-5 Simulation Result (#227/#228) log 1 (BER) #227 12dB_WC, FFE -1 time [UI] 1 log 1 (BER) #228 12dB_WC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #223 #224 #227 #228 #231 #232 type 1dB_WC 12dB_WC 14dB_WC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low 3 35 upp mid mv low

44 log1(ber) Vdiff [mv] log 1 (BER) Vdiff [mv] log 1 (BER) B-6 Simulation Result (#231/#232) log 1 (BER) #231 14dB_WC, FFE -1 time [UI] 1 log 1 (BER) #232 14dB_WC, DFE time [UI] sakai_3ck_1a_ criteria EW5 > 1mUI EH5 > 2mV * : No PKG HSI-PR1-1 item unit #223 #224 #227 #228 #231 #232 type 1dB_WC 12dB_WC 14dB_WC ch IL [db]* PSXT [db] Rx FFE pre/post /4 / /4 / /4 / Rx DFE tap EW5 EH5 upp mid mui low 3 35 upp mid mv low

45