Myth # 1: Is a signal above GHz rate a high-speed signal?
When layout mentions “high-speed signal”, it is necessary to define what “high-speed” is first. Is the signal at MHz rate high-speed or the signal at GHz rate high-speed?
Traditional SI theory has a classic definition of “high-speed signal”.
Si: signal integrity, that is, signal integrity.
SI theory understands the signal transmission behavior of PCB interconnection lines, and the signal edge rate almost completely determines the maximum frequency component of the signal. Usually, when the signal edge time is less than 4~6 times of the interconnection transmission delay, the signal interconnection path will be treated as a distributed parameter model, and the SI behavior needs to be considered.
The so-called “high speed” refers to “interconnection transmission delay with signal edge time less than 4~6 times”. It can be seen whether the signal transmitted by the circuit board is “high speed” depends not only on the edge rate of the signal, but also on the path length of the circuit board. When there is a certain proportional relationship between the two, the signal should be processed according to the “high speed signal”.
Myth # 2: With the simulation software platform, you can design a high-speed PCB layout?
EDA design software platform integrates the function of high-speed signal simulation, which is of great help to the formulation and implementation of high-speed PCB design rules, signal quality simulation and evaluation.
However, in the actual design process of PCB, sometimes the simulation results show that the signal quality is good, but the signal quality in the actual test is very poor, which does not meet the signal test standard.
In fact, simulation and testing are inseparable. Take IBIS model as an example, which is usually called “behavior-level model”. This kind of simulation model is also established through the V and I test curves of the chip under different working conditions, so there is a problem. If we don’t pay attention to which working condition of the chip model is selected during simulation, the simulation will be inaccurate, such as Slow, Typical and Fast.
As can be seen from the above example, the “simulation model base” is very important to the simulation results, and it is necessary to pass the simulation test of actual products and the actual comparison, and the revised simulation model can be counted as an “accurate simulation model”.
Having a good simulation design platform can’t solve all the problems, but also requires an “accurate simulation model”. In addition, it is also necessary to consider the application scenarios of actual product projects. A simulated signal network will be affected by factors such as power noise and other signal crosstalk, which will also cause the difference between the test results and the simulation results.
Myth # 3: The PCB routing “transmission line model” in layout design simulation software is very accurate?
The PCB traces in the simulation software, whether microstrip or stripline, can be modeled by simulation tools. This model is based on the size of the laminate and the actual traces, and usually can meet the accuracy requirements. However, if it is “very accurate”, there are still some gaps, which need to be analyzed from the following aspects:
(1) the roughening/browning process of 1)PCB copper has an impact on signal quality.
In the process of PCB processing, in order to improve the bonding strength between the copper layer and the dielectric layer of PCB and reduce the risk of PCB delamination, there will always be a roughening/browning process, that is, the surface of the copper layer will be roughened by polishing or corrosion.
In the transmission of high-speed signals in conductors, there is “skin effect”, which means that when high-frequency signals are transmitted, the current flowing in the conductors will migrate to the periphery or the “skin” of the conductors.
The rough surface of PCB will affect the loss on the one hand, and the signal transmission delay on the other. This is easy to understand, just as it will take more time for a car to drive on a rugged mountain road than on an asphalt road.
(2) The dielectric constant Dk and tangent loss angle Df of 2)PCB dielectric change with frequency.
The Dk and Df of PCB media in simulation tools are usually constant, but from the point of view of actual signal transmission, Dk/Df changes with frequency.
Dk/Df will change with the transmission signal rate, so if these two parameters are treated as constants in simulation tools, the simulation accuracy of transmission line model will be affected, and the higher the signal transmission rate, the greater the impact.
(3) “Anisotropy” influence of 3)PCB
PCB is usually a woven structure of “epoxy resin+glass cloth”. The arrangement direction of glass cloth is divided into “warp” and “weft”. At the same time, according to the thickness and spacing of glass fibers, it can be divided into different types of PCB, such as: 1080, 2116, etc. When different types of glass cloth are used for PCB board, the proportion of glass cloth and resin in the board is different.
The Dk/Df values of glass cloth and resin material are quite different. When the relative position of normal PCB routing and glass cloth is different, the Dk/Df values of reference media will be different, and the impedance and loss of signals will be different, as shown in the following figure, which is why some projects require that the routing direction of the whole PCB should be 10.
Myth 4: One simulation software platform can solve all signal simulation problems.
At present, there is no unified simulation software platform that can be applied to all signal simulation scenarios. Cadence SPB SigXplorer is used for behavioral signal quality simulation, Synopsys HSPIe is used for transistor simulation, Ansoft HFSS is used for three-dimensional electromagnetic field modeling, and Ansoft ADS is used for time-domain and frequency-domain hybrid simulation.
