With CoolRunner-II CPLD devices, great things come in small packages.

When you’re looking for a simple, low cost, and easy-to-use programmable logic device that incorporates multiple functions, think Xilinx CoolRunner™-II CPLDs. These versatile, nonvolatile devices can save you time and money on your next design by reducing board costs and redesigns.

Cost can be thought of in several different ways, depending on your point of view. For a buyer, it’s the bottom line of a bill of materials. For a design engineer, it’s time invested and looming deadlines.

Engineers also face tradeoffs, such as how fast a product can be designed with a minimal number of board layouts. Your success may rest in the decisions you make while trying to accomplish this goal. When making component choices, it pays to have built-in flexibility; with reprogrammable logic, you get the best dollar value as well as the ability to deliver products ahead of schedule.

Additionally, engineers must consider such factors as single-chip solutions, package size, density, versatility, flexible I/O structures, and the ability to modify pin functionality after placement on the board. By considering these items before parts selection, you can save costs and still maintain flexibility.

Single Chip Integration

If you have unlimited board space, a large stocking warehouse, and inexpensive test and assembly costs, some of these cost factors may not enter into the price equation. But if you’re in a competitive marketplace, usually one or more of these items will be scrutinized:

• Power-efficient board size
• Minimum number of parts and suppliers
• Low assembly costs.

Board Size
Typically, the packaging of your product is defined by board size, which is driven by the number of components you need to get the job done. If you can squeeze out the required functionality and still stay within the power budget, you have met your goal.

Don’t forget that cost can mean board space to some engineers and inexpensive parts to others. With CoolRunner-II you can select small BGA packages such as 56- or 132-ball chip scale packages (CSP) for high integration or flat pack (FP) packages for low-cost solutions. If you are concerned with board size, you may need unique CSP options. Xilinx also offers 0.5 mm to 0.8 mm ball spacing packages that can save you more than 50% when compared to similar I/O count FP package options.

Although the space savings from a 14 mm-by-14 mm, 100-pin FP package to an 8 mm-by-8 mm, 132-ball CSP package may seem trivial, consider the routing involved. With flat packs, all pins typically route outward from the package. With BGA packages, routing can be achieved by running traces between the adjacent solder balls. These packages also offer more options when using denser, multilayer PCBs. This may yield twice the routing efficiency of a comparable FP package, further reducing board space. Thus, the capability of these small packages goes well beyond the “wow factor” of their physical size.

Parts and Suppliers
Lower power consumption can also be achieved through reduced component counts. A single low-power CPLD device improves reliability by reducing the total number chance of cold or weak solder joints that may cause intermittent failures. Heat dissipation may be reduced. And more solder joints also increase the chance of manufacturing problems. The more solder joints, the higher the chances of developing manufacturing problems. Heat dissipation may be reduced through fully utilizing a single part instead of powering multiple parts that may not be fully utilized. These two factors can have a direct impact on customer service and customer reliability ratings.

There are many functions you can squeeze into a single CPLD and still get the low power operation you desire. Maintaining multiple components for specific functions can lead to a nightmare for procurement. By expanding your supplier base, you increase demands on many different departments within your company and thus lengthen your time-to-market. These areas may include accounting, shipping and receiving, or component engineering. If you have a quality department, they may want reports on each individual device.

Furthermore, the more devices you specify, the higher the chance of encountering a production problem. It would be devastating to not be able to ship a multi-million-dollar product due to a $2 part on back order. By using more parts, you also run a higher risk of device obsolescence. This may not cause a delay in shipment, but it typically costs a board re-layout.

Assembly Costs

The more components shipped to your contract manufacturer, the more money you spend in shipping costs. Each electronic component placed on your board reflects a direct assembly charge. If your contract manufacturer charges you to stock devices, this will also add cost to your end product.

By keeping the component count down, you can dramatically reduce both direct expenses and the indirect cost of doing business.

Integration and Flexibility

If you need multiple I/O standards for unique memory devices or CMOS level translation, conversion devices may be necessary. Depending on your application, specialty memory devices may also be required. If your processor does not support HSTL or SSTL memory types, you may need to select voltage referenced to CMOS translators. In high-volume applications, these single-function translators can cost from $4 to $6 in 48-pin packages. The problem is, they only serve one purpose. If you don’t use all of the pins, it’s wasted board space and power. With a single CPLD, you get translation coupled with extra logic capabilities and the freedom to use pins for other purposes besides translation.

Even if you don’t use any specialty memory, what about legacy parts that use different voltage levels than your processor? Again, you have the choice of purchasing a single function device that can cost around $2 for a 48-pin package. A comparable pin count CPLD can cost half as much as this single function device – and again, give you more functionality. So if you need voltage level translation in the range of 1.5V to 3.3V, CoolRunner-II CPLDs can also provide this integrated function.

One specialty function that sometimes is not considered but may prevent board re-spins is input hysteresis. Schmitt trigger inverters can cost from $4 to $8 in 20-pin packages. These devices usually operate from 1.6V to 3.6V, which gives them a wide operating window. CoolRunner-II CPLDs have input hysteresis on every input pin. And because you can configure CoolRunner-II CPLD input buffers to any voltage from 1.5V to 3.3V, they also have a wide range of operation. In a head-to-head comparison, CoolRunner-II CPLDs can cost 75% less than a discrete Schmitt trigger device.

Also, by using a CPLD solution, you can enable the input hysteresis, if required; if not, just leave it disabled. Because you don’t always know if you need hysteresis, this flexibility may save a new board layout.

Moreover, with features such as clock dividers and doublers (DualEDGE flip-flops), you can set up independent clock domains in CoolRunner-II CPLDs, thus eliminating the need for independent oscillators or crystals. The devices can handle fast-running sequential functions such as pulse width modulator, conversion functions (BCD to decimal), and serial communications functions.

Conclusion

Due to their multifunctional nature, Xilinx CPLDs can integrate many applications to save costs in your design. The high-performance, low-power CoolRunner-II CPLDs can reduce the number of board redesigns, minimize the total number of devices, and increase overall flexibility. This will have a direct impact on bringing your product to market faster.

To get you started with CPLDs, Xilinx offers multiple aids, including beginner tutorials with demo boards and reference designs that include detailed application notes with HDL code. Some design examples include SMBus, I 2 C, SPI, and processor interfaces. You can also look at full-up reference designs, such as designing an MP3 player. Whatever your level of experience, Xilinx makes it easy to use reprogrammable logic.

Printable PDF version of this article.  (02/15/03) 180 KB