Above 10 GHz, surface rounghness can more than double conductor loss. This is 20 Gbps. For > 15 Gbps definitely consider the copper roughness and using low loss laminates.
It doesnât. Stub resonances are stub resonance. The resonant frequency depends on the length of the stub and the Dk of the laminate. Keep stubs short regardless of the materials used.
all the usual methods: use low loss laminates, keep traces short, use wide traces, use smooth copper.
I think the impedance of the paths is more important than a target impedance in the frequency domain. Using controlled impedance interconnects is more about keeping the cross-talk low.
If you need controlled impedance traces, use the same methods as in signal integrity. see my book on Signal and Power integrity simplified.
Sorry- I am not as familiar with rf testing methods.
sorry, donât know.
no idea what you are referring to.
a rough rule of thumb is a trace length in inches equal to the rise time of the signal in nsec. For a 1 nsec rise time signal, traces shorter than 1 inch do not need termination.
1 Like
Yes. Terminate a different pair in its differential impedance. terminate a signal-ended line it is characteristic impedance. read my book on signal integrity.
use SMT resistors or on-die resistors.
Not on boards with return planes.
same methods as digital boards. Just manage the cross talk with signal spacing.
keep the two lines that make up the differential pair symmetric. This means same length and same cross section.
two most important terms are the vertical collapse or eye height opening and the horizontal collapse which is jitter.
I use Ansys electronic desktop, Hyperlynx, ADS and Polar
Do you have a course curriculum designed around this book? The local university and trade college is interested in adding some courses on the subject as there currently are none. Iâd love to chat some with you if youâre willing to share some advice on setting up some introductory and intermediate courses on PCB design.
contact me off line to discuss: eric.bogatin@colorado.edu
1 Like
A good place to start is my textbook on prototype design.
I am working on posting a series of videos related to the labs I have created.
Send me a direct email note to discuss this further.
âeric
1 Like
Thanks for the answer Eric. The reason why I ask this question is because my products are not avionics instrumentation and do not interact with any of that, but are designed to operate other systems in the aircraft and they power from the standard 28V available in the aircraft.
I love this rule of thumb!
But it implies that a rise time in microseconds wonât need termination on a single board, let alone on a single chip. Until recent speeds, was termination theory mostly about not knowing how long the transmission line might be?
1 Like