The sequel is here! Is it okay to have no reference? Let me explain it all at once!
[Copy link]
Author: Huang Gang, member of YiBo Technology Expressway
(, downloads: 0)
From the simulation results of the previous article, a pair of differential lines can indeed control impedance well by returning to each other. So why do all design rules, even from our cognition, feel that a reference ground is needed to form a complete differential line transmission structure? For many engineers, they may not be able to tell the specific reason, but the concept of ground has been deeply rooted in their hearts. So today, Mr. High Speed hopes to explain it to you clearly!
(, downloads: 0)
First of all, let's take the model established in the previous article and continue to expand on it. Why can impedance be controlled without ground? Theory and formulas tell us that impedance is proportional to the distance from the reference plane. At the same time, we also know that if the distance to the reference plane (H parameter in the figure below) is too close, the impedance is low under the same circumstances. Therefore, let's first verify that if the reference plane is from relatively close to gradually far away, what kind of change will the impedance of the differential line present?
(, downloads: 0)
Sure enough, when we scan the parameter H, we will find that as H increases, the impedance does increase slowly. However, the most important point to note is that when H increases to a certain value, the impedance increase becomes very small. . .
(, downloads: 0)
Therefore, if H increases further, it will be very close to the 100 ohms in the previous article without ground. Of course, since this model in this article also adds a pair of adjacent traces with a gap of 3W, it will also be slightly affected by the adjacent traces, and the impedance will be slightly lower. As for why this article adds a pair of adjacent traces, you will know later!
(, downloads: 0)
We know that if it is a single-ended signal, it must have a return path, which is what we commonly call a reference plane, in order to control the impedance and ensure the transmission of the signal. The differential signal can be a return path for each other to control the impedance and transmit forward. For a single-ended signal, it will only have impedance to the "ground", but as a differential line, it has two forms: differential impedance and common-mode impedance. The so-called differential impedance is what we usually call the impedance of the differential line, which is the impedance of the differential lines P and N in the case of differential mode transmission. However, the differential line also has common-mode impedance, which is the impedance of the differential lines P and N to the "ground" respectively, which is used to transmit the common-mode signal of the differential pair. Normally, there is no common-mode signal in an ideal differential line, but this is only in ideal conditions. In non-ideal conditions, common-mode signals will be generated more or less, so the common-mode signal will be transmitted along with the differential pair and crosstalk to other signals.
(, downloads: 0)
After talking about so many theories, are there any specific risks if there is no ground? Let's continue to analyze it based on the above model. In addition to the differential impedance of the differential pair, let's take a look at its common mode impedance. If the differential impedance is normally controlled at 100 ohms, the ideal common mode impedance is 25 ohms (that is, the single-ended 50 ohm parallel impedance to the ground). From the figure below, we can see that as the distance from the reference plane becomes farther and farther, the common mode impedance increases significantly. What's more important is that the common mode impedance is not very close to that of the ground distance of 40 mil, which seems to be quite far, and the difference is still quite large.
(, downloads: 0)
Then the question is, what is the effect of the increase in common mode impedance? This is why our model has built two pairs of differential lines. It can be seen that as the distance from the ground plane increases, the common mode crosstalk between the two pairs of lines becomes very serious. This is the biggest problem when there is no ground plane. After the common mode impedance increases, the reflection becomes serious. In addition, due to the lack of the ground plane, the common mode energy can easily spread around, so the crosstalk will increase sharply.
(, downloads: 0)
To summarize, first of all, in the absence of a ground plane, the differential impedance of the differential line can indeed be controlled by the return current between the differential lines, but the differential line will have a common mode impedance, which requires a ground plane to control. In an ideal situation, there is indeed no common mode impedance in the differential line, but in a non-ideal situation, once a common mode component is generated, it can easily spread to other signals on the board or even to space, forming serious crosstalk and EMI radiation to other signals. This is a parameter that many fans may not pay attention to. Therefore, as the rate increases, the ground plane is basically an indispensable component of the differential line. Don't try this special design easily.
(, downloads: 0)
Here comes the question
Everyone can think about it. Under what circumstances will the differential line generate a common mode signal to interfere with others?
|