Optimizing PCB New Product Introduction Using ODB++
Julian Coates, ODB++ Solutions Alliance
A PCB design is not successful until it has been manufactured on time, according to the designer’s intentions, and at an acceptable total cost. The traditional approach of outputting multiple legacy file formats and sending them to the manufacturer on the assumption that he will fix any manufacturability problems carries a high indirect cost in terms of supply chain and cycle time risk.
And, in the end, the designer pays that cost. Design and manufacturing must be seen as one business process integrated by intelligent data. This article describes the business advantages of ODB++ intelligent data, along with some comparisons and contrasts with alternative formats.
At ELMATICA ODB++ is the preferred format for all aspects within PCB / PWB
Intelligent vs. Non-intelligent Data
In general, there are two types of manufacturing data: intelligent and non-intelligent. Non-intelligent data are collections of files in vector formats such as Gerber, HPGL or Excellon drill files, which use a sequence of single-entity definitions composing a single layer of information. An external source is needed to determine the use of such a single layer. Sometimes this is the filename or a readme.txt document. For a manufacturer to make sense of these files, they must reintegrate them, reconnecting the disjointed information, and then get a view of what the complete PCB looks like – sort of like turning applesauce back into apples.
Non-intelligent data must be imported into a DFM or CAM tool; however, an engineer must use functions to identify the single layers – such as what represents a signal layer, or what represents the solder mask – using a graphics editor. In Figure 1, a set of Gerber files has been imported and we can see that the layer buildup is in alphabetic filename order. In the figure, the general characteristics of the PCB are visible – pads and traces on this layer – but other information such as physical characteristics, the solder mask, parts outline, etc., are not available. The inset shows information for U37, pin 24 (in yellow) on the top layer (filename top.gbr).
Figure 1: With non-intelligent data, all that is displayed with the selected element is the information that it is an oval pad at an XY location.
On the other hand, intelligent data comes from the EDA system which already contains objects such as components, pad stacks, traces, and even net names. In the late 1990s, ODB++ was developed by Valor Computerized Systems (acquired in 2010 by Mentor Graphics) as an open, vendor-neutral, data format to address the need for an intelligent data set that could be assembled from PCB design data and manufacturing process rules and data, which then provide the vehicle for DFM analysis and manufacturing process preparation. A few years after the introduction of ODB++, PCB design companies, fabricators and contract manufacturers found ODB++ allowed for an easy interchange of data between them.
ODB++ Delivers Results in the Real World
ODB++ is a directory structure with an index, thus avoiding the need for a single huge file. This allows the CAM or DFM tools to load only the data needed for any specific processing application or task – a very useful characteristic in these times when PCB design data density is outpacing the processing capabilities of typical CAD/CAM hardware platforms.
With ODB++, the manufacturing engineer can, without having to first rebuild the model of the PCB assembly from multiple disjointed files, see the same type of objects that the PCB designer used, thus allowing much easier analysis. For example, DFM tools can import data from a wide range of EDA systems, either by importing native CAD formats or by importing the ODB++ that was directly created in the EDA system.
Once an ODB++ model of the product has been created, an engineer can display a layer from the buildup, for example, the “components top,” and then select a component and see not just an outline of the component pad as with non-intelligent data, but specific information (intelligence) such as the RefDes, part number, CAD package, number of pins, XY location, length and width, height, rotation and pin pitch. These are all critical data to the manufacturing process and simply not available without manual intervention when using non-intelligent data. Selecting a component pad and requesting information displays the full information, as shown in Figure 2.
Figure 2: With the intelligent data in ODB++, a full set of information is available. In this figure the RefDes, part number, CAD package, number of pins, XY location, length and width, height, rotation and pin pitch are all shown, and pin number 24 of component U37 is identified as connected to netname +5V.
Essential to the PCB new product introduction (NPI) step is the ability to use data sets assembled during the PCB design phase and to relate that data to the intended manufacturing process. Until fairly recently, data sets used in the PCB design were often incomplete, in terms of not being able to fully support the manufacturing hand-off process. Frequently, a great deal of manual translation was required to do any kind of DFM analysis or manufacturing process preparation. That need gave rise to the invention of the ODB++ format.
ODB++ is the most intelligent proven CAD/CAM format available today, capturing the complete PCB fabrication, assembly and test knowledge in a single, unified database. ODB++ is an ASCII, fully expandable open format that can capture and store all the information needed for the manufacturing and assembly of a printed circuit board, imported directly from the CAD database and other sources such as PLM and DFM systems. This information includes layer graphics like pads and drills, test points, fiducials, components, netlists, and even any additional documents that may be needed. Data representation is fully WYSIWYG, so displayed objects are exactly the same as they will appear in the manufacturing process.
In addition to the information imported from the CAD system, ODB++ can also store data generated by the DFM application itself, such as the parts list, analysis results or a DFM report. Once stored in the ODB++ file, a standard function from within a graphical editor can be used to view the specific or variant related parts list. It is even possible to store some design specific datasheets or documents in the “user” section of the ODB++ file by using simple drag and drop in a file browser. Thus, the need for parallel files, such as drawings, is eliminated; the risk of mistakes is reduced because users set themselves up for a “right first time” hand-off to manufacturing.
Since initial development in the mid-1990s, ODB++ has become the mainstream solution for design-to-manufacturing hand-off for many of the world’s leading electronics OEMs, with thousands of PCB designs being processed in the format every year. Leading manufacturers report up to 80% of their incoming work arriving in ODB++ format, enabling them to focus on the added-value tasks of DFM validation and manufacturing preparation, instead of first having to decode and reintegrate all the legacy files that represent the only proven alternative.
ODB++ is available as a format open to the industry in general, via the ODB++ Solutions Alliance (www.odb-sa.com). Dozens of Solutions Development Partners have supported the format for years, and the alliance has thousands of members who stay connected with its evolution. Since 1995, ODB++ has been under the stewardship of Valor, benefiting from a large multi-vendor user base and continuous investment in improvements to the format and the community of tools that use the data content. This aggregated investment over 15+ years by designers, multiple CAD/CAM software vendors, PCB fabrication and assembly manufacturers will continue as before to deliver benefits to the industry as the format and tools evolve.
It can accurately be stated that the industry in general has voted for ODB++ as the intelligent PCB manufacturing data format that really works and can be relied upon to deliver time, cost and quality advantages in practical daily reality. From time to time, new formats are suggested, but so far they have not gained critical mass in the market due to not providing significant marginal technical advantage, or requiring an industry-wide investment that cannot be justified, or some combination of the two.
From the viewpoint of the ODB++ Solutions Alliance, the different formats could be compared as follows:
Quantified data available at www.odb-sa.com illustrates the main barrier that stands in the way of progress in moving from the legacy formats such as Gerber, component placement lists, and Excellon, to intelligent data such as ODB++: A lack of awareness of the benefits at the point of generation. While the largest OEM designers have high levels of ODB++ implementation achieved over many years, a large diversity of PCB designers have only a basic notion of the format’s benefits, and they need more information in order to move from their current process into intelligent data.
The industry does not need any more data transfer formats. On the contrary, the maximum efficiency gains can be achieved at the minimum total cost to everyone by implementing existing industry-proven solutions.
Julian Coates is the current director of the ODB++ Solutions Alliance. Julian has held a number of marketing management positions. He holds a degree in engineering science from Exeter University (1979), UK.