Microcontroller and flash memory integration

Electronic News, Feb 3, 1997 by John Bryant

Nine years ago, flash memory became available, and designers relished the concept of updating their program code without removing the memory chip from the system board. However, at the time, the higher cost of flash compared to ROM or EPROM outweighed the perceived benefits of in-system updateability.

As has been widely reported, the flash market has grown very large since then, and in turn, flash has replaced EPROMs in many cases. Furthermore, as microcontroller and DSP vendors have improved their process technologies to obtain higher integration, flash has become one of the features that has made it on-chip.

Whether on-chip or off, an OEM realizes many benefits from incorporating flash memory into a system design. During the product development stages, flash allows the software programmer to change the prototype operating code in a matter of seconds. Flash accelerates product testing, allowing the easy substitution of test vectors for operating code during the final assembly stages.

What's more, product scrapping due to misprogramming or bugs is no longer an issue, as every device is reprogrammable. This is especially important after the flash memory has been soldered to the circuit board.

Flash also enables just-in-time(JIT) manufacturing, minimizing an OEM's product inventory by making many model options available with a simple code or data modification. Similarly, JIT manufacturing makes it easy for an OEM to offer customers user-specific features. Last, but perhaps most important, flash makes it painless to remotely make code changes without the need to disassemble the product in the field.

The ability to make remote updates allows an OEM to add new features as well as keep the equipment operating at optimum performance levels. The system circuitry can reprogram the flash memory, changing a product's operation through adaptive control to adjust for environmental conditions, operator characteristics or equipment wear.

Integrating memory onto a microcontroller is not a novelty. For many years, MCU vendors have been adding EPROM, ROM, EEPROM, and RAM to their devices. However, vendors faced significant challenges when they tried to include flash with their microcontrollers.

In 1993 Atmel set an industry precedent when it figured out the recipe for mixing a flash memory array with standard microcontroller logic. For example, the company's AT89x series of microcontrollers contain between 1 and 20 kilobytes of flash memory. Although the AT89x MCUs are socket and function compatible with the industry-standard 80C52, the integrated flash memory greatly increases the capabilities of these MCUs.

And Atmel is certainly not alone. Companies such as Hitachi, Motorola, Microchip, NEC, and Toshiba have put flash into their standard MCUs. On-chip flash provides an OEM with a host of benefits including lower overall system cost, less board real estate, faster memory access times, lower power consumption, and higher reliability.

Although the flash-based microcontroller is marginally more expensive than a non-flash microcontroller, it is still less expensive than buying discrete microcontroller and flash devices. Furthermore, the most cost effective flash devices are in the 8- to 16-megabit range. This implies that the cost of flash devices below 1M is better leveraged when the memory is integrated with the logic of a microcontroller or DSP.

An OEM will realize additional cost savings with flash by not having to use a socket to attach EPROM- or ROM-based devices to the circuit board. The elimination of a separate flash device also reduces board space requirements by over 50 percent. Specifically, a single 40-pin AT89C52 replaces a standard 40-pin 80C52 plus a 32-pin flash device and two or three 16-pin interface circuits. The system uses the interface circuits to demultiplex the memory bus and replace the I/O ports lost to the memory bus.

The savings associated with board-space, board-cost, and the potential reduction of board layers by eliminating the board traces required to tie the controller and flash devices together must also be considered. Another way of looking at this benefit, is that the I/O that would have been used for the address and data lines to external memory is regained.

In addition, system performance is also enhanced by using on-chip flash. The microcontroller performs faster memory accesses to integrated flash, making bus rates above 40MHz possible because the timing margins of the control signals are significantly increased by putting the logic functions on the same chip.

Integrating the flash memory on-chip significantly lowers the system power consumption and reduces the capacitive loading on the external address and data buses. Eliminating the need to drive 16 or more high-speed output buffers is a big deal in a minimal chip system. Power savings may also be achieved by lowering system voltage.

Generally, a silicon vendor will use a more advanced and lower voltage process technology on its processors than on memory devices, enabling integrated flash to ride this process curve. And as clock rates continue to climb, reducing off-chip accesses helps to lower EMI levels and in turn increase product reliability.