The most obvious advantages of differential signals compared to common single-ended signal traces are reflected in the following three aspects:
a. Strong anti-interference ability, because the coupling between two differential traces is very good, when there is noise interference in the outside world, it is almost simultaneously coupled to two lines, and the receiver only cares about the difference between the two signals. Therefore, the external common mode noise can be completely canceled.
b. Effectively suppresses EMI, the same reason, because the polarities of the two signals are reversed, the electromagnetic fields that they radiate to each other can cancel each other out. The closer the coupling, the less the electromagnetic energy that is discharged to the outside world.
c. accurate timing positioning, because the differential signal switching is located at the intersection of the two signals, rather than the ordinary single-ended signal depends on the level of the two threshold voltage to determine, and therefore affected by the process, the temperature is small, can reduce the timing error At the same time, it is also more suitable for low-amplitude signal circuits.
The current popular LVD (ow voltage differential signaling) refers to this small-amplitude differential signal technology. For PCB engineers, the most important thing is to ensure that these advantages of differential routing can be fully utilized in actual routing. Maybe as long as people who come into contact with Layout understand the general requirements of differential routing, it is "equal length, isometric." The equal length is to ensure that the two differential signals keep the opposite polarity at all times, reducing the common mode component; the equal distance is mainly to ensure the consistency of the two differential impedances and reduce the reflection. "As close as possible to principle" is sometimes one of the requirements for differential routing. However, all of these rules are not intended to be applied mechanically. Many engineers do not seem to understand the nature of high-speed differential signaling.
The following focuses on some common misunderstandings in PCB differential signal design.
#1: Think of the differential signal as not requiring a ground plane as a return path, or thinking that the differential traces provide a return path for each other.
The reason for this misunderstanding is that it is confused by superficial phenomena, or the understanding of the mechanism of high-speed signal transmission is not deep enough. From the structure of the receiving end, it can be seen that the emitter currents of the transistors Q3 and Q4 are equal and reversed. Their currents at the ground just cancel each other out (I1=0), and thus the differential circuit may be present for similar ground bombs and others. Noise signals on the power and ground planes are insensitive. The partial return offset of the ground plane does not mean that the differential circuit does not use the reference plane as the signal return path. In fact, in the signal reflow analysis, the mechanism of the differential trace and the common single-ended trace is the same, that is, the high-frequency signal is always The most important difference is that the differential line carries out the re-flow along the smallest loop of the inductor. In addition to the coupling to the ground, the differential line is also coupled with each other. Which type of coupling is strong is the main return path.
In the PCB circuit design, the coupling between the differential traces is generally small, often only accounting for 10% to 20% of the coupling degree, and more still is the coupling to the ground, so the main return path of differential traces still exists in the ground plane . When there is a discontinuity in the local plane, there is no reference plane area, and the coupling between the differential traces will provide the main return path. Although the influence of the discontinuity of the reference plane on the differential traces is not serious for ordinary single-ended traces, it will still reduce the quality of the differential signal and increase the EMI, which should be avoided as much as possible. Some designers believe that the reference plane below the differential trace can be removed to suppress some of the common-mode signals in the differential transmission. However, in theory, this approach is undesirable, and how is the impedance controlled? Not providing a common-mode signal ground impedance loop is bound to cause EMI radiation.
#2: I think it's more important to keep equal spacing than match line length.
In actual PCB layout, it is often not possible to meet the requirements of differential design at the same time. Due to factors such as pin distribution, vias, and trace space, the matching of the line length must be achieved through proper routing. However, the result must be that parts of the differential pair cannot be parallel. How do we do this? Choose it? Before we come to the conclusion, let's take a look at the next simulation result. From the above simulation results, the waveforms of scheme 1 and scheme 2 are almost coincident, that is, the impact caused by the unequal spacing is negligible. In comparison, the mismatch of the wire length has a much greater impact on the timing. (Option 3). From a theoretical point of view, even though the inconsistent spacing will cause the differential impedance to change, because the coupling between the differential pairs is not significant, the impedance variation range is also very small, usually within 10%, which is equivalent to only one The reflection caused by the hole does not have a significant effect on signal transmission. However, once the line length does not match, a shift occurs in timing, and a common mode component is introduced into the differential signal to reduce the signal quality and increase the EMI. It can be said that the most important rule in the design of PCb differential routing is the length of the matching line. Other rules can be flexibly handled according to design requirements and practical applications.
#3: Think that the differential route must be very close.
Making differential traces closer to nothing more than to enhance their coupling can both improve the immunity to noise and also make full use of the opposite polarity of the magnetic field to cancel the electromagnetic interference to the outside world. Although this approach is very favorable in most cases, it is not absolute. If we can ensure that they are adequately shielded from outside interference, then we do not need to let the strong coupling through each other achieve anti-interference. And the purpose of suppressing EMI. How can we ensure that differential traces have good isolation and shielding? Increasing the spacing from other signal traces is one of the most basic approaches. The electromagnetic field energy decreases with the distance in square relationship. When the line spacing is more than 4 times the width, the interference between them is extremely weak. Basically ignored. In addition, the ground plane isolation can also play a good shielding effect, this structure is often used in high-frequency (10g and above) ic package pcb design, known as cpw structure, can ensure strict differential impedance Control (2z0). < span="">
Differential traces can also walk in different signal layers, but this method is generally not recommended because different layers such as impedance and via difference will destroy the effect of differential mode transmission and introduce common mode noise. In addition, if the coupling between the two adjacent layers is not tight enough, the ability of the differential traces to resist noise will be reduced, but crosstalk is not a problem if the proper spacing from the surrounding traces can be maintained. In general frequency (below GHz), EMI will not be a serious problem. Experiments have shown that 500 Mils of differential traces, the attenuation of radiation energy beyond 3 meters has reached 60 dB, enough to meet the FCC's electromagnetic radiation standards, so Designers do not have to worry too much about differential coupling due to insufficient electromagnetic compatibility.