High Speed Characterization Report

Table of Contents Introduction...1 Product Description...1 Results Summary...3 Time Domain Data...3 Impedance...3 NEXT...3 FEXT...4 Insertion Loss...5 Return Loss...6 Near End Crosstalk...7 Test Procedures...9 Fixturing:...9 Time Domain Testing...11 Impedance:...11 NEXT and FEXT:...11 Frequency Domain Testing...11 Return Loss:...12 Equipment...13 Time Domain Testing...13 Samtec, Inc. 2005 Page: ii All Rights Reserved

Introduction This testing was performed to evaluate the electrical performance of the HDLSP series of High Density/High Speed IO Cable Assemblies. Time domain and frequency domain measurements were made. Time domain measurements included impedance and crosstalk. Frequency domain measurements were performed using Tektronix s IConnect and Measurement XTractor software (Version 3.6.0) and included insertion loss (IL), return loss (RL), near end crosstalk (NEXT) and far end crosstalk (FEXT). All measurements were made utilizing test boards specifically designed for this project and are referred to as test board in this report. The test boards were identified as SI Kit 0.635mm 100 ohms Revision A 8/10/2010 Product Description The test sample consists of twenty-four individual low skew pair cables (#32 AWG 100-Ω twinax). At each end of the assembly there is a Hypertransport edge card connector with two rows of 12 signal pairs each. Each cable end provides for termination of the twinax cables. The Hyptertransport connector contacts are on a 0.635mm (0.025 ) pitch. Assembly length is 2 meters. The assembly is wired to facilitate a Pin A1 to Pin B1 mapping between the cable ends. The HDLSP assembly was tested by mating it to a HDI6-035-01-RA-TR 4-Row, 0.635mm Socket with HDC-035-01 Edge Mount PCB Cage at each end. One sample of the assembly was tested. The actual part number that was tested is shown in Table 1, which also identifies End 1 and End 2 of the assembly; a relative sample picture is shown in Figure 1. Four differential signal pairs, one from each row of the mating socket, were tested. End 1 is at the left with the assembly stretched out and viewed so that the cable label containing the assembly part number can be read in normal fashion form left to right. Length Part Number End 1 End 2 Hypertransport Edge Hypertransport Edge 2000mm HDLSP-035-2.00 Card Connector Card Connector Table 1: Sample Description Samtec, Inc. 2005 Page: 1 All Rights Reserved

Results Summary Time Domain Data Impedance Impedance measurements were performed using a filtered risetime of 100 ps. Note that all measurements were performed with the assembly mated to the respective connector/test board. Data was measured at the cable connector and at 200 ps into the cable. Assembly HDLSP-035-2.00 Path (Row) End Option Cable End 1 End 1 End 1 Z Min (Ω) Z Max (Ω) Z Nom (Ω) A 80.5 104.6 101.4 B 80.4 101.5 100.3 C 79.6 103.9 100.9 D 81.7 102.2 101.3 Table 2: Impedance Measurements NEXT The near end crosstalk was measured in the time domain and converted to a percentage and reported below in Table 3. The incident pulse amplitude from the TDR was 464 mv. The acquired data was measured using a filtered rise time of 100 ps. The End 1 heading in Table 3 represents the near-end of the assembly, i.e. the source end. All NEXT measurements were performed with the assembly mated to the respective connector/test board. Since most of the crosstalk occurs in the connectors, the values in Table 3 represent the crosstalk that occurs in the near-end mated assembly and the test board connectors. END1 Assembly Config. Path NEXT (mv) HDLSP-035-2.00 NEXT (%) In Row A 4.8 1.0 Between Rows A B 2.8 0.6 In Row C 4.0 0.9 Between Rows C D 2.2 0.5 Table 3: % NEXT Samtec, Inc. 2005 Page: 3 All Rights Reserved

FEXT The far end crosstalk was measured in the time domain and converted to a percentage and reported below in Table 4. The incident pulse amplitude from the TDR was 464 mv. The acquired data was measured using a filtered rise time of 100 ps. The End 1 heading in Table 4 represents the near-end cable assembly connector, i.e. the source end. All FEXT measurements were performed with the cable assembly mated to the respective connector/test board. The values in Table 4 represent the crosstalk measured at the far end of the assembly. END1 Assembly Config. Path FEXT (mv) HDLSP-035-2.00 FEXT (%) In Row A 4.2 0.9 Between Rows A B 1.8 0.4 In Row C 3.6 0.8 Between Rows C D 0.6 0.1 Table 4: % FEXT Samtec, Inc. 2005 Page: 4 All Rights Reserved

Test Procedures Fixturing: All measurements were performed using the test boards that have nominal trace lengths of 1.26 inches plus 6-inch long BullsEye 50-Ω coax cable assemblies with SMA jacks for interconnection to the HDLSP cable assembly. The test board provides a pair of THRU reference traces. Figure 10 below shows how the THRU reference trace pair was utilized to compensate for the losses due to the coaxial test cables and the test board coax cables and traces during testing. R Reference Traces Coax Cables Coax Cables TDT TDR Tektronix 11801B SD26 Ch3,4 BullsEye SMA Jacks Tektronix 11801B SD24 Ch1,2 Figure 10: Test setup for Thru Reference Acquisition Measurements were then performed using the test boards as shown in Figure 11. A picture of the test board and cable is shown in Figure 12. Coax Cables Cable under test Coax Cables TDT TDR Tektronix 11801B SD26 Ch3,4 BullsEye SMA Jacks Tektronix 11801B SD24 Ch1,2 Figure 11: Characterization test setup Samtec, Inc. 2005 Page: 9 All Rights Reserved

Figure 12: Test PCBs and HDLSP cable assembly. The cable assembly terminations had a particular signal/ground line configuration. The respective End 1 signal line numbers that were made available as test ports and that were used during the testing are shown in Table 5 below. The cable assembly had 2 rows and the mating test board connector had 4 rows of signal contacts. The cable assembly mated to either rows A and B or rows C and D of the test board connector during testing. G A4 A5 G A7 A8 G A10 A11 G G B4 B5 G B7 B8 G B10 B11 G G C4 C5 G C7 C8 G C10 C11 G G D4 D5 G D7 D8 G D10 D11 G Table 5: Respective signal line numbers as viewed from End 1 Table 6 below shows the signal line numbers corresponding to signal paths for the different configurations tested. The test board jack numbers correspond to the assembly line numbers. Path Assembly Jack Row End1 End2 A A4,A5 B4,B5 B B10,B11 A10,A11 HDLSP-035-2.00 C C10,C11 D10,D11 D D4,D5 C4,C5 Table 6: Signal Path Row Signal Line Numbers Samtec, Inc. 2005 Page: 10 All Rights Reserved

Time Domain Testing Impedance: The Tektronix 11801B oscilloscope was set up in TDR (time domain reflectometry) mode using a 100-pS filtered risetime and 16 averages. The horizontal setup of the TDR used a 512 point record length and a horizontal scale of 200 ps/div to allow the near end connector and a portion of the cable to be displayed. All connector impedance measurements were made at the near-end connector. Cable impedance was measured 200 ps into the cable after the connector. NEXT and FEXT: Near end crosstalk (NEXT) and far end crosstalk (FEXT) measurements were made using the Tektronix 11801B oscilloscope. A thru reference of the coaxial test cables, BullsEye cables and test board reference traces was performed to determine the pulse amplitude of the TDR generator (see Figure 10). To acquire NEXT, a signal was applied using the oscilloscope pulse generator. NEXT was measured on an adjacent signal line pair at the near end (see Figure 13). To acquire FEXT, a line pair was driven with the oscilloscope pulse generator. FEXT was measured on an adjacent line pair at the far end (see Figure 14). All adjacent lines were terminated, at both ends, with 50Ω SMA loads; refer to Figures 13 and 14. Frequency Domain Testing All frequency domain measurements were made using the Tektronix 11801B oscilloscope. Testing was performed using a risetime of 35 ps. The horizontal scale was set to 5 ns/div, the record length was set to 5120 points and the number of averages was set to 128. These values were selected to ensure the ratio between the number of points and the window length was long enough to capture the highest frequencies and still yield a small enough frequency step to gain adequate resolution. End 1 of the assembly was the source end for all frequency domain measurements. All adjacent lines were terminated at both ends with 50Ω SMA loads; refer to Figures 13 and 14. Attenuation: Insertion Loss test setup losses were compensated for by acquiring a thru measurement (reference output pulse) of the coaxial test cables, BullsEye cables and the test board reference traces (see Figure 10). A thru measurement of an assembly was taken and then post processed by using Tektronix IConnect software. The result is the insertion loss of the cable assembly. Samtec, Inc. 2005 Page: 11 All Rights Reserved

Return Loss: An open circuit reference measurement was taken using a signal line pair on the test boards. A matched reflection waveform of the cable assembly, i.e. with the assembly terminated in 50-Ω SMA loads on the far end test board, was acquired and then post processed by using Tektronix IConnect software. The result is the return loss of the cable assembly. Near and Far End Crosstalk: NEXT and FEXT were measured in the time domain using the oscilloscope and then converted to frequency domain data using Tektronix IConnect software. Initially a thru reference measurement of the coaxial test cables, BullsEye cables and test board reference traces was performed to compensate for the test setup losses (see Figure 12). To acquire NEXT a line pair was driven using the oscilloscope pulse generator. NEXT was measured, in the time domain, on an adjacent line pair (see Figure 13). NEXT was then post processed using Tektronix s IConnect software to generate the NEXT of the cable assembly in the frequency domain. To acquire FEXT a line pair was driven using the oscilloscope pulse generator. FEXT was measured in the time domain on an adjacent line pair at the far end (see Figure 14). FEXT was post processed using Tektronix s IConnect software to generate the FEXT of the cable assembly in the frequency domain. HDLSP Sample Ch1, 2 Ch3, 4 Tektronix 11801B BullsEye SMAs NEXT test FEXT test 50 Ω Termination Figure 13: NEXT Measurement Setup. Samtec, Inc. 2005 Page: 12 All Rights Reserved

Coax HDLSP Sample Coax TDT TDR Tektronix 11801B SD26 Ch3, 4 BullsEye SMAs 50 Ω Termination Figure 14: FEXT Measurement Setup Tektronix 11801B SD24 Ch1, 2 Equipment Time Domain Testing Tektronix 11801B Oscilloscope Tektronix SD24 TDR/Sampling Head Tektronix SD26 Sampling Head Samtec, Inc. 2005 Page: 13 All Rights Reserved