I can see why this situation is frustrating. "Suffice it to say that, with its influence, the semiconductor industry could have made considerable headway into enabling sensible design exchange decades ago." Perhaps as TTM pressures continue to increase (is that even possible?) they will look to this as a method to improve things. In the meantime, do you think such an exchange would open up the EDA market to more vendors?
One recent trend is to have reference designs done by 3rd parties, not the semi company. This can make sense when the semi company doesn't have the applications skills and experience to quickly create a 'real' design. Everything from Video processing, Wireless Internet and Motor control reference designs are being done by 3rd parties. When these are done well, they become a 'manufacturing kit' that just needs some 'value added software' poured into. A contract manufacturer can then just crank out the design.
This trend may work against us as the leverage is dissipated from the semi companies to a host of smaller third parties. If these 3rd parties could get together and begin a standards process however, maybe that would make the difference...
@JanineLove: do you think such an exchange would open up the EDA market to more vendors?
The exchange mechanism I'm suggesting is focused on enabling the ability to import the fundamental aspects of reference designs - namely schematics and PCB design data with little if any net or component properties added. The mechanism, while eliminating much manual re-entry, would require the user to add more complete properties to the design elements in order to take full advantage of their EDA tool suite. Exporting the updated design, for vendor independent deployment, would once again output only the fundamental design data.
To the extent that full and open data exchange is a barrier of entry to new EDA companies, I believe more complete data exchange would be required. This is beyond the scope of my suggested reference-design-centric data exchange mechanism.
One way out of this lockup could be throught something like circuits.io which offers a library of design and design data , created in the open source way, and a unique concept: modules include definitions of both schematics and pcb layout.
This saves a lot fo work and place for errors and complexity. This is also very good for reference designs - it lets just use the reference design as a building block in a short time.
And the open source part might be a way to build large design libraries.
DrFPGA, The move to use 3rd parties makes sense for the reasons you cite. However, these designs still need to be published in a manner that encourages the widest OEM adoption and consumption of the funding semiconductor company's devices. I believe the rate of return on reference design investments made by device vendors would increase, if there was an intelligent, portable reference design format. They are the greatest beneficiary and, as such, their influence should be the highest.
Just adding software and committing to manufacturing will work for some segments. However, many OEMs will need to pull these externally developed reference designs into their environment to leverage internal standards for aspects such as documentation and supply chain management. In most cases, the reference design will become part of the OEM's IP and should be treated as such.
Perhaps a useful distinction is a 'Manufacturing kit' vs. a Reference Design. My definition of a manufacturing kit is a design that is sufficient for an OEM to go to market with (by just changing the logo on the PCB and perhaps changing the software a bit). The hardware pretty much is ready to send to a CM. 3rd parties (perhaps design service organizations allied with the CM) can charge a fee to the licensing OEM and perhaps even get a mdest fee from the CM for 'enabled' business. The Manufacturing Kit would seem to be the highest value to the OEM and the CM.
Device suppliers would try and convince the Manufacturing Kit creator to use their devices, but they have much less 'leverage' than when they do a reference design. Does the Manufacturing Kit concept change the company dynamics much with respect to standardization?
Vince, apparently this has been a bigger problem for some OEMs than for others, and I'm curious as to whether you think the problem is more due to lack of standards for data formats, OEMs and semiconductor vendors using different PCB CAD tools, or some other issue.
The reason I ask is because it seems to me that most if not all semiconductor vendors not only have reference designs (hardware & software) to help sell their products, but that those reference designs also include electronic copies of the schematics in the format of whatever CAD tool was used to create those schematics (a .dsn file for example) and Gerber files for the PCB layers, which can also be imported into the OEM's PCB CAD tool, as well as a detailed BOM, layout guidelines for those who wish to incorporate portions of the reference design into a larger PCB, and so on.
I'd like to hear your thoughts on what the big hangup is in transferring reference design data from an IC vendor to an OEM.
@Frank Eory: I'd like to hear your thoughts on what the big hangup is in transferring reference design data from an IC vendor to an OEM.
Frank, I agree that some semiconductor companies publish their reference designs in the native CAD formats that they use. Take, for instance, the DM3730 BeagleBoard-xM - schematics are available in PDF and OrCAD. The PCB is delivered in Gerber or Allegro. While users of the Allegro flow would find the delivery acceptable, those using other EDA tools would find the delivery difficult to utilize without time consuming translation or re-entry. Most PCB systems can import Gerber layers, but associating this data with the schematic so that design re-purposing can take place is not trivial. If I'm not using Allegro, and decide the need to re-purpose the BeagleBoard design into another form factor, I have quite a bit of re-work to do on the reference design delivery before being able to re-place and re-route the PCB. This re-work would be greatly reduced if a standard means of intelligent data exchange was available.
Next, consider this app note from Maxim. The schematic is available as a JPEG or as part of a PDF. This delivery method is relatively useless, per my first column on the subject. This is a typical reference design format for a variety of analog building blocks across a variety of device vendors. The elimination of the above two scenarios is my "big hangup."
I'm advocating an EDA vendor independent means of exchange (similar to STEP used for mechanical design), which is substantially more efficient than any of the means used today by semiconductor vendors to distribute reference designs.