CommVerge - Home

The 1994 EDN DSP-Chip Directory is a keeper: Not only have we revised and updated last year's chip specs, but we've also added new entries, including Zoran's 16-bit fixed-point ZR38000, the Texas Instruments' 32-bit MVP, and Analog Devices' SHARC, a 32-bit floating-point DSP. The 16-bit fixed-point devices continue to dominate low-cost applications. As new variations of each DSP µP hit the market, each family fills a greater variety of applications. The largest markets for these 16-bit speedsters include telecommunications, digital cellular telephones, and disk drives. Floating-point DSPs are being designed in by the cluster. Often you'll see products that, instead of having just a couple of DSPs within, have many. For a number of years, designers generally thought floating-point DSPs would be used for prototyping fixed-point DSP-based

products. Even though floating-point devices are easier to use, most developers aren't opting to use them as a crutch. As a result, floating-point devices end up in applications where the three most important design criteria are performance, performance, and performance. Designing with DSPs continues to get easier. You can now buy world-class software-development tools. And in many cases, you don't even need to write the code—you can just buy the algorithms premade. For example, TI has created the TMS320 Software Cooperative, which offers a collection of algorithm data sheets. You select the algorithm you want by paging through the packet. Getting a DSP-based board to work with Microsoft's Windows is also getting easier. In conjunction with DSP real-time operating-system (RTOS) vendors, Microsoft is busy creating a standard method for applications to call for DSP functions. Any DSP-based board that works within the guidelines of the standard can be plugged into the PC and used with any application. Another year of above-average growth for DSPs hasn't been affected by a marketing myth that superfast CISC or RISC µPs will begin to replace DSPs. In many cases, the replacement trend is working the other way. The supremely fast DSP is using surplus cycles to replace slower microcontrollers. As you flip through the following pages, you'll find a comprehensive listing of the major DSP chips, ranging from stripped-down, 16-bit, fixed-point processors to full-blown, 32-bit floating-point processors. Each chip family has its own page full of architectural information and a quick run-down of variations, peripherals, and price/packaging information. If your time's limited, check out the handy reference tables in the upper-right corner of each page—we've noted important feature and performance specs in an easy-to-read format. Keep a copy of the directory as a reference for your design work throughout the coming year. —Additional reporting by Contributing Editors Ray Weiss and David Shear DSP-Chip Directory Index Supplier     DSP chip    Type     Page    Circle No. Analog Devices    ADSP-2100 Family    16-bit fixed-point    79    484 ADSP-21020    32-bit floating-point    80    485 ADSP-2106x    32-bit floating-point    83    486 AT&T    ADSP16xx    16-bit fixed-point    84    487 ADSP32C/3210    32-bit floating-point    95    488 DSP Group, Semiconductor Division    Oak/Pine Core    16-bit fixed-point    96    489 Motorola    DSP56156/166    16-bit fixed-point    99    490 DSP56001/2/4    24-bit fixed-point    100    491 DSP96002    32-bit floating-point    105    492 NEC    µPD77C25    16-bit fixed-point    106    493 µPD77017    16-bit fixed-point    109    494 µPD77220    24-bit fixed-point    110    495 SGS-Thomson    ST18 Family    16-bit fixed-point    113    496 Star Semiconductor    SPROC-1x00 Family    24-bit multiprocessor    114    497 Texas Instruments    TMS320C1x    16-bit fixed-point    117    498, 499 TMS320C2x    16-bit fixed-point    120    500 TMS320C3x    32-bit floating-point    123    501 TMS320C4x    32-bit floating-point    124    502 TMS320C5x    16-bit fixed-point    126    503 TMS320C80    32-bit multiprocessor    131    504 Zilog     Z89Cxx    16-bit fixed-point    132    505 Zoran Corp    ZR38000    16-bit fixed-point    135    506 Key to abbreviations in schematicsAB—combined program-and-data address bus ACC—accumulator ADC/DAC—analog-to-digital and digital-to-analog converters ADDR GEN—address generator ALU—arithmetic logic unit BIT MANIP—bit manipulation BS—barrel shifter CDB—control data bus CM—cache memory CPUB—CPU bus DAB—data address bus DB—combined program-and-data data bus DDB—data data bus DM—memory for data only    DMAAB—DMA address bus DMADB—DMA data bus DMAC—DMA controller FP—floating point FX—fixed point GDB—global data bus HOST INTER—host interface IDB—instruction data bus INT—external interrupt MAC—multiply/accumulate MULT—multiplier PAB—program address bus PDB—program data bus P/DM—program and data memory PIO—parallel I/O PM—memory for program only PPCP—parallel processor communications port    PRAB—peripheral address bus PRDB—peripheral data bus REG—register REGB—register bus SIO—serial I/O TIM—timer XAB—external address bus XDB—external data bus XDAB—external data address bus XDDB—external data data bus XIOAB—external I/O address bus XIODB—external I/O data bus XPAB—external program address bus XPDB—external program data bus