Evaluating individual components of a backplane system is helpful in determining whether or not the components being measured meet a particular performance criterion. However, when the individual components are integrated, the overall performance may behave differently than the sum of the individual components. Measurements are needed to confirm a correlation. If a correlation is established, extrapolation of specific system components can be achieved. The backplane/daughtercard connector, PTH, and board laminate will be analyzed in an integrated fashion. A schematic of the test vehicle is shown in Figure 9.

Figure 9. Gigabit System Test Vehicle
The goal of the analysis was to vary the individual components to determine the overall system performance. Board materials used in the system were Rogers 4000 series, Megtron (a Getek equivalent), and FR4. Trace geometries were 8- and 10-mil line widths with lengths of 10, 20, 30, and 40 inches. The overall thicknesses of the backplanes were 0.250 of an inch. All the backplane and daughtercard traces have nominal 100-Ohm differential impedance. The daughtercards used to launch and receive the signals were 0.125-inch thick with 5-mil line widths/spaces and 2-inch/4-inch stubs. The measurements were taken at 1.25 Gbps, 2.5 Gbps, and 3.3 Gbps. Eye patterns were used as the measure of merit, and a pseudo-random bit pattern was used in the eye-pattern measurements.
The results show that at very long line lengths and fast data rates, there is very little improvement in going from an 8-mil line width to a 10-mil line width (see Figures 10 to 12).

Figure 10. 20-Inch Line Width Comparison 1
Left: 20" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 1.25 Gbps
Right: 20" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps

Figure 11. 20-Inch Line Width Comparison 2
Left: 20" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 2.5 Gbps
Right: 20" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps

Figure 12. 20-Inch Line Width Comparison 3
Left: 20" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 3.3 Gbps
Right: 20" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
For a 40-inch track width at 2.5 Gbps in Megtron/Getek using 10-mil lines, the eye-opening was ~225 ps and the amplitude was ~90 mV. The same test done with an 8-mil line width shows an eye-opening of ~225 ps with an amplitude of ~90 mV. For 40-inch track lengths, the Rogers material shows an improvement over the FR4 and Megtron materials (see Figures 13 to 15).

Figure 13. Material Comparison 1
Left: 40" Rogers trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps
Right: 40" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps
Bottom: 40" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps

Figure 14. Material Comparison 2
Left: 40" Rogers trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps
Right: 40" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps
Bottom: 40" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps

Figure 15. Material Comparison 3
Left: 40" Rogers trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
Right: 40" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
Bottom: 40" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
With a 40-inch track length and 8-mil line widths, the eye-opening in FR4 at 2.5 Gbps was ~175 ps/50 mV. Using the same test scenario but changing the material to Megtron the eye opens up to ~225 ps/90 mV. Again, the same test was performed using the Rogers material and the eye-opening was ~300 ps/160 mV. The material performance is more evident at 3.3 Gbps with the eye-opening for FR4 essentially closed, Megtron at ~100 ps/30 mV and Rogers ~175 ps/90 mV. At 1.25 Gbps with 40-inch track lengths, the three different materials perform essentially the same, thereby verifying that the dielectric loss plays a large role at the very fast data rates (high frequencies), but conductor loss will dominate at the lower frequencies. For shorter backplane track lengths such as 10-inch, line widths and dielectric loss still have a measurable effect in the signal performance. For a 10-inch track length in FR4, 8-mil line widths, and 3.3 Gbps data rate, the eye opening is ~200 ps/110 mV. The same scenario with 10-mil line widths yields an eye opening of ~225 ps/170 mV. At 2.5 Gbps, 10-inch track length, 8-mil line width, in FR4 yields an eye opening of ~320 ps/220 mV. With the same data rate, track length, and material and 10-mil line width, the eye opened up to ~350 ps/250 mV. In a Megtron backplane with 10-inch track length, 3.3 Gbps, and 8-mil lines, the eye opening was ~240 ps/180 V. Repeating the test with 10-mil line widths yields an eye opening of ~230 ps/190 mV. When the data rate was slowed to 2.5 Gbps, the eye opening with Megtron for the 8-mil line width was ~340 ps/250 mV, and with a 10-mil line width the eye opening was ~340 ps/260 mV. This shows that increasing the line width from 8 mil to 10 mil yields a slight improvement in performance but may not offset the penalty of increased board thickness to maintain the impedance (see Figures 16 to 18).

Figure 16. 10-Inch Line Width Comparison 1
Top Left: 10" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 1.25 Gbps
Top Right: 10" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps
Bottom Left: 10" FR4 trace lengths through two HSD connectors
with 10-mil lines and spaces at 1.25 Gbps
Bottom Right: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 1.25 Gbps

Figure 17. 10-Inch Line Width Comparison 2
Top Left: 10" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps
Top Right: 10" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 2.5 Gbps
Bottom Left: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 2.5 Gbps
Bottom Right: 10" FR4 trace lengths through two HSD connectors
with 10-mil lines and spaces at 2.5 Gbps

Figure 18. 10-Inch Line Width Comparison 3
Top Left: 10" Megtron trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
Top Right: 10" Megtron trace lengths through two HSD connectors
with 10-mil lines and spaces at 3.3 Gbps
Bottom Left: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps
Bottom Right: 10" FR4 trace lengths through two HSD connectors
with 10-mil lines and spaces at 3.3 Gbps
A test was devised to evaluate the system effects of varying the launch capacitance. For conventional backplane connectors, a PTH launch is used and can give capacitance values up to 3 pF, depending on the thickness of the backplane. An experiment was done where a 10- and 20-inch backplane was measured with a PTH capacitance of 2.5 pF. To see the effect of the launch capacitance on the system eye-pattern, the launches were counter-bored to leave barrel depths of 50 mils, 100 mils, and 150 mils. The eye patterns were compared at 3.3 Gbps over track lengths of 10 and 20 inches. For the 10-inch track lengths, the trace capacitance with a standard PTH was measured to be 39 pF. The measured trace capacitance with a 50-mil PTH was 36.7 pF; 100-mil PTH was 38.2 pF, and 150-mil PTH was 38.7 pF. The resulting eye patterns shown in Figure 19 show that there is very little improvement in the eye opening, even with a 10 percent reduction in the overall trace capacitance.

Figure 19. 10-Inch PTH Comparison
Top Left: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with standard PTH
Top Right: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 50-mil barrel depth
Bottom Left: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 100-mil barrel depth
Bottom Right: 10" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 150-mil barrel depth
Intuitively, it is expected that a reduced launch capacitance will be more effective for the shorter track lengths than the longer track lengths. Figure 20 shows the resulting eye pattern at 3.3 Gbps in FR4 over a 20-inch track and various PTH barrel depths. It is clear from the plots that there is very little improvement from the standard PTH barrel to one that was reduced to only a 50-mil PTH barrel.

Figure 20. 20-Inch PTH Comparison
Top Left: 20" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with standard PTH
Top Right: 20" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 50-mil barrel depth
Bottom Left: 20" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 150-mil barrel depth
Bottom Right: 20" FR4 trace lengths through two HSD connectors
with 8-mil lines and spaces at 3.3 Gbps with 100-mil barrel depth


