EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: ker2x on September 03, 2014, 01:43:19 pm
-
Hello again :)
I have a super simple question which probably require a complex answer :
What are thoses pcb with controlled impedence ?
it was mentionned a few times in EEVblog video but i don't understand.
It looks like it's related to carefully designed free space between the pcb line and ground plane but... perhaps it's not.
How does it works ? Why ? :-//
Thank you
-
This is important as you move into changing signals, instead of DC. The frequency components of a signal can be quite high, even for a simple chip, if the transistor size is very small. It has to do with the speed at which the voltage changes. A transistor made with very small sized fab can switch very fast. Impedance is important once your distance between devices is long enough to make the signals not change on each side of a trace at the same time. For high speed signals this can be only 40mm away, for example.
When you have frequencies moving around wires or PCB traces, they are very sensitive to changes in impedance.
Think about some coax for example. Typical TV antenna coax is 75 ohm. Most RF based coax for communication is 50 ohm. What this means is that a signal flowing down that line sees 50 or 75 ohms as it moves. If you have something at the end that is also a matching 50 or 75 ohms, it flows directly into it perfectly.
If you an impedance mismatch, because you have an open cable end (nearly infinite impedance) or it is shorted (nearly zero impedance), there is energy in that signal that needs to go somewhere. It gets reflected back down the cable.
This is important, because in a digital system you are concerned with levels and transitions. If you are trying to send a clock that is toggling between 0 and 3.3V, but have an impedance mismatch, you might get something that is anywhere and unpredictable.
For this reason, most high speed buses have an impedance standard. This is the output impedance of one device matches the input impedance of the other. It is best to also design the PCB trace, such that the impedance along the way also matches. This reduces the reflected interference and helps signal quality on the board.
You can look at how the thickness of the board, traces, etc influence impedance with the Saturn PCB Toolkit (free) http://www.saturnpcb.com/pcb_toolkit.htm (http://www.saturnpcb.com/pcb_toolkit.htm)
As you mentioned in your post, it is ALL about the distance between the trace and ground plane and the thickness of the trace. High speed signals are sent along the space between traces as transverse (90 degree from each other) electromagnetic fields. Even in a differential signal on a pair of traces, where at DC you would expect the signal to go down one end and come back up the other, the signal electromagnetically couples to the closest piece of copper that generate an opposite current flow, and makes a current flow there. It is strange to think of signal propagating along the "void" space in a board, but this is the only way you can understand all of the consequences of routing decisions.
It is also the reason why the material of the board insulating layers (ex. FR4) is what determines the speed of the signal in the board. In FR4 a signal moves around 6" (150mm) per nanosecond.
Not having a good ground plane directly under a conductor, or having a break in that plane, will cause the electromagnetic field to jump somewhere else. This can cause a huge signal integrity issue, due to the massive impedance change. It also is a place where EMI radiation spikes and could be a reason why your device would not pass FCC certification.
OK, I'm going to wrap this up, because this subject can and has filled up many books without being fully covered.
-
Thank you very much for the clear explanation (and :-+ for avoiding math here).
I knew that, at high frequency, pcb trace "became" transmission line... but it didn't help much.
However, the "pair of trace" example is an excellent one and it really make sense :)
And not it's not weird to think of signal propagating in the void, that's how radio works ^-^