For digital signals I always use 4th harmonics (cuz main bit of energy is there, but not the signal's form) frequency and trace length of 0.25 of that wave length ( like in old time RF lines) to make that kind of assumption.
Well digital signals will usually have the majority of their energy concentrated in a few frequency components so wouldn't it be possible to average out a corresponding wavelength to atleast get a general idea if the trace acts like a transmission line? Similarly, analog signals are not a single frequency in practice. Information is carried over a certain bandwidth. This is why you characterize a signal with its carrier or center frequency and a bandwidth. When doing calculations you take the center frequency but you should always look at the frequency responce for the whole bandwidth. So i think that this idea can be used for digital signals in a similar manner. Anyways, great video and a great series. Looking forward for more episodes.
Hi, and thanks for watching! So for determining whether the trace acts like a transmission line, you do need to consider some very high limiting frequency. If you set the line length using a high frequency, then all lower frequencies will also not be affected by impedance mismatch. If you have a specific bandwdith, use the high frequency end and not the low frequency end. As far as choosing which frequency, consider this example: Suppose you are designing a high speed channel and your receiver needs at least 2 GHz of bandwidth for a maximum 4 Gbps NRZ/NRZI data rate. Now suppose your rise time is fast enough that you have 90% of your power contained up to 10 GHz. Should you determine critical lengths based on 2 GHz or on 10 GHz? This is a realistic example because, in USB and in many SerDes transmitters for example, the signal rise time could be specified as approximately 100 ps or less, so that would certainly give 10 GHz of signal bandwidth. However, the receiver needs much less bandwidth to properly sample and resolve signals at similar data rates. So how much bandwidth does the channel need to have, meaning how high of a frequency can you use to determine the input impedance? If you don't use a high enough frequency and you set a critical length at a lower frequency, your line's critical length will be long but a lot of power could be reflected at high frequency, which specifically will slow down the edge rate in the receiver.
I don't have an impedance requirement. I can't design to something I don't know! That's what I'm trying to figure out. Also you didn't really define v here. Obviously it's not c. What value do I use?
Nice content Zach, I am improving.
For digital signals I always use 4th harmonics (cuz main bit of energy is there, but not the signal's form) frequency and trace length of 0.25 of that wave length ( like in old time RF lines) to make that kind of assumption.
Resonance matters. So Analog or Digital is maybe irrelevant, the question becomes which wavelength is resonant on the TL in question?
Well digital signals will usually have the majority of their energy concentrated in a few frequency components so wouldn't it be possible to average out a corresponding wavelength to atleast get a general idea if the trace acts like a transmission line?
Similarly, analog signals are not a single frequency in practice. Information is carried over a certain bandwidth. This is why you characterize a signal with its carrier or center frequency and a bandwidth. When doing calculations you take the center frequency but you should always look at the frequency responce for the whole bandwidth. So i think that this idea can be used for digital signals in a similar manner.
Anyways, great video and a great series. Looking forward for more episodes.
Hi, and thanks for watching!
So for determining whether the trace acts like a transmission line, you do need to consider some very high limiting frequency. If you set the line length using a high frequency, then all lower frequencies will also not be affected by impedance mismatch. If you have a specific bandwdith, use the high frequency end and not the low frequency end.
As far as choosing which frequency, consider this example: Suppose you are designing a high speed channel and your receiver needs at least 2 GHz of bandwidth for a maximum 4 Gbps NRZ/NRZI data rate. Now suppose your rise time is fast enough that you have 90% of your power contained up to 10 GHz. Should you determine critical lengths based on 2 GHz or on 10 GHz?
This is a realistic example because, in USB and in many SerDes transmitters for example, the signal rise time could be specified as approximately 100 ps or less, so that would certainly give 10 GHz of signal bandwidth. However, the receiver needs much less bandwidth to properly sample and resolve signals at similar data rates. So how much bandwidth does the channel need to have, meaning how high of a frequency can you use to determine the input impedance? If you don't use a high enough frequency and you set a critical length at a lower frequency, your line's critical length will be long but a lot of power could be reflected at high frequency, which specifically will slow down the edge rate in the receiver.
@@Zachariah-Peterson thanks for the detailed answer
Hey look, Eric Bogatin just says... nhaa, i'm just kidding :D Cool and interesting point of view! Thanks
I don't have an impedance requirement. I can't design to something I don't know! That's what I'm trying to figure out. Also you didn't really define v here. Obviously it's not c. What value do I use?
Ignore all that, they're all wrong 😂