DS92LV2421: Relationship of transmission rate and eye pattern

Hi,

It appears that you are calculating based on the eye width. However, due to the presence of some jitter affecting the width of the eye crossing, you can see that the measured eye width is much smaller than the serial bit unit interval (or period). If you use the eye width and try to calculate a frequency, it will result in a higher bit rate than the actual serial bit stream, since you are using a time interval that is smaller than 1 UI.

Your calculation for Figure 39 is correct based on the previous thread, because 20 MHz x 28 bits/CLK = 560 Mbps (1.78 ns per UI) on the serial line when CLK = 20 MHz. If we assume that the time interval for Figure 39 is approximately 0.3 ns/DIV, you can see that 1 UI is close to 6 time divisions. This results in about 1.8 ns for 1 UI, which is close to the 1.78 ns you are expecting. Likewise, if we measure the eye width manually by looking at the time divisions, there are approximately 4.5 time intervals between L_Open and R_Open. This results in about 1.35 ns, just as the measurement shows.

I've illustrated the difference in the attachment below:

Thanks,

Michael