PCB signal integrity can be simply divided into three categories:
1. Timing problem. Timing issues are the key issue, the current designers are basically using the core chip manufacturers off-the-shelf program, so the design of a part of the work is how to ensure that the PCB can meet the chip requirements of the timing.
2. General SI problem. That is to solve the drive problem, the termination resistance or series damping resistance numerical calculation, PCB laminate structure and characteristic impedance calculation, alignment topology analysis.
3. Microwave transmission problem. Known as the GHz SI. The design needs to address the transmission link because the traces, vias and materials and other small size caused by a variety of usually only in the microwave field will be considered. For the general SI problem, Hyperlynx, SQ and ICX can be a good solution. Domestic users have basically mastered how to deal with and analyze the general SI problem. In terms of tools, SQ, Hyperlynx or ICX can be a good solution. Performance, SQ's strengths are itself a PCB layout and routing tool, so it is suitable for the actual PCB layout and wiring performance is better, that is, on-site commissioning (On-The-Fly) performance is better. But SQ has no timing analysis capability, only a simple and limited timing measurement function. Hyperlynx advantage is easy to use, but neither the timing analysis function nor the timing measurement function. In contrast, Hyperlynx is more convenient in EMC forecasting, which is the advantage of Hyperlynx. The complexity of the weaknesses and settings on the ICX's GUI makes it less desirable for users to adopt the tool. At present the general SI problem, mainly PCB layout designers in progress. Schematic designers are basically do not do this work. The main reason is that the use of tools is not familiar and less understanding of SI. In addition, there are a considerable number of domestic users do not understand the basic logic of the work level requirements and impedance concepts, do not understand the Source-Load driver analysis. Which resulted in a lot of early design of the PCB because the replacement of the device manufacturers, causing the system to work abnormalities and redesign. A typical example is the only use of Pericon 16245 and can not use TI's 16245, this example has occurred many times.
Timing problems, the current domestic users basically did not grasp. A small number of SQ users will use the Excel table to prepare timing requirements, later from the SQ measured parameters manually fill in the Excel table to calculate whether the final design meets the timing requirements. Even so, few people can understand the timing, whether it is schematic designers and PCB layout designers are difficult to read from the chip data sheet timing. Many times, due to the wrong constraints caused by the design of repeated times. Timing issues major parallel interface problems. From the figure below, you can see the complexity of the current interface. Because these interfaces are new interfaces, the speed from 50Mhz-500MHz range, the timing requirements are high, the delay is the main problem, the domestic users have not yet grasp these interfaces within a short time, so the latter part of the timing verification is basically Unable to proceed.
The microwave transmission problem is caused by the serial interface, which is particularly important in communication, mainly to solve the multi-board problem, that is, the card core chip - connector - backplane - connector - card core chip link transmission problem. There is no timing problem in the serial link. Only the transmission problem needs to solve the inter-symbol interference problem caused by the signal amplitude caused by the frequency arrival of the microwave segment and the change of the 01 bit stream. This part of the problem must be combined using frequency domain and time domain tools. At present the mainstream SQ, Hyperlynx and ICX to deal with these problems is not very effective. Cadence launched a $ 200,000 MGH tool (SI 630) in SQ to solve these problems. The tool is rarely used, and it does not seem to differ from SQ on the surface, but it adds a way to extract 3D vias Tools and a quick simulation of converting the S parameters back to the table parameters supported by the time domain tool.
Hyperlynx and ICX processing S parameters are done using Eldo, but because of the lack of MGH algorithm in the analysis of long bit stream in the required time is not practical. From a rigorous point of view, these tools are still ineffective in dealing with the microwave segment, since the 3D extraction of the interconnect structure depends only on the 3D field, and the non-ideal effects of the multi-board system must be taken into account, and these tools are not of. Therefore, in fact the user needs to use Ansfoft HFSS and other tools. The part of the SI is the current hot, many users published papers are related to this. Even so, for ordinary system designers, in fact this part of the work is not the main, relatively speaking, chip manufacturers must use these tools to solve the chip Serdes design and system applicability problems. The usual solution is to use HFSS to extract the S-parameters of the vias and traces, and then use the manual or other tools such as Optimal SI Assit to process the extracted S-parameters so that it can be accepted by HSPICE and other time domain tools (almost all of the HFSS parameters are extracted S parameters must be processed before they can be used in the time domain tool). This process is not easy, only a few expert-level engineers have the ability, because the results given by the tool is not intuitive, only a good background theory to understand the results. The reason why domestic users are enthusiastic about this, the main reason is estimated to want to improve their professional level, it is not the immediate needs of the actual work. Therefore, Ansoft is still not enough to compete.
As for the PI on the PCB, that is, power integrity, there is no practical tool available. Ansoft's SIwave and Sigrity SPEED2000, PowerSI are not practical use, the algorithm is no special point, the claim that the thing is actually invalid. Only Optimal tool is conceptually better, but now only Beta. Its Beta version and the PCB interface is very poor, not a short time will be accepted by the user.
From the above situation, it is clear that the combination of timing analysis and general SI analysis is the most marketable and the main SI market. Mentor can place the main market on it, and the new ICX performance should be provided. Can attract SQ users to such tools. Mentor's ICX + Tau conceptually has this ability, but its current performance is not ideal, the improvement is too slow. And recently another company Sisoft introduced the tool Quantum-SI has such a function, the tool has just launched, but very much in line with the current needs of users.
Comparison of PCB SI Tool Functions
Obviously, an ideal mainstream PCB SI tool should be a combination of timing analysis and SI analysis. The following analysis, to Sisoft's Quantum-SI as the benchmark, the purpose of doing so is conducive to understanding SQ, Hyperlynx and ICX / Tau. About Sisoft's background see its website related tools. But this does not affect the comparison of SQ, Hyperlynx and ICX / Tau.
price
Hyperlynx price is not low, as a popular SI tool price is relatively high. Quanum-SI price in the 9000-40000 US dollars. Hyperlynx is 5000-50000, the following is Sisoft comparison table: Mentor Tau need 35000-45000 dollars. In contrast, Hyperlynx has no price advantage. ICX + Tau price? Estimated at $ 150,000, higher than the SQ of $ 1,200. But Cadence 630 is more expensive, $ 200 million. The problem is that if the user reports the Mentor program and the Cadence program, it is possible that the management will choose a low price.
Ease of use
ICX / Tau users are more difficult to master, SQ is easier, Hyperlynx is the easiest. If the user is not superstitious SQ, it will be very easy to accept Hyerlynx. And now V7.5 version of the display performance improved, on the contrary SQ due to add a lot of features, it seems more complex settings.
Timing analysis
The current timing analysis is particularly important. High-speed PCB layout design of the essential problem is to solve the timing problem. Using Excel tools for timing calculations is difficult and requires the user to understand the timing requirements of a variety of new complex parallel interfaces. Tau is conceptually very good, especially with ICX to solve the front and back timing analysis. SQ basically can be said that there is no timing analysis function (with a simple timing measurement function), Hyperlynx is no time analysis and measurement function. But Tau is also a problem. Tau and Mentor push the DxDesigner no direct interface, but with the Design Architect and Design View interface; Tau itself Schematic View performance is too bad, so EDIF format third-party schematic data can not actually use; Tau itself lack of model support, Need to write their own model, but also the need for third-party tools TimeDesigner with. If you can build some Tau models, training users to master the creation of Tau model, then it will have a clear competitive advantage. This is Cadence can not catch up in the short term. Unless Cadence acquires TimeDesigner to create a new board-level timing analysis tool. Quantum-SI is stronger than ICX + Tau in this area. Its GUI, commonly used interface design package (in fact, very few to be launched later), SI measurement is very characteristic.
Synchronous switching noise (SSO) and crosstalk analysis
Hyperlyx, SQ and ICX can not analyze SSO, crosstalk analysis is also quite limited. This Sigrity has left the market opportunity. Quantum-SI can be complex multi-board system analysis, is said to support SSO, as shown below. But as far as I can expect, it can only do the same level as ICX or SQ.
Simulation settings
Hyperlynx the most simple, ICX more inconvenient, SQ than ICX to be convenient. But Hyperlyxn also has its shortcomings, that is, almost no fixed on the measurement settings. This is better Quantum-SI, as shown below, you can set up a number of networks at the same time, while the embedded 45 kinds of complex measurement mode. In this regard, SQ and ICX are not as good as Quantum.
Therefore, from the current situation, the user most need is a time analysis and SI integration of one tool, and the interface to optimize, set to be simple, and need to include Design KIT. ICX + Tau If you can improve performance like Quantum-SI, it will be welcomed by users. Since Mentor has a design front and rear end, the advantage of ICX + Tau is that other tools can not be replaced.
