At Look at Embedded Technologies in 2000
O.K., it's the year 2000. What can
we expect for Embedded Systems? Here's my predictions:
Embedded Internet
Everyone's looking for new applications, new systems
"Internet Appliances" to embed a TCP/IP stack. The non-news of
2000 is that the fastest growing Embedded Internet market segment is going
to be the traditional high-growth embedded systems applications -
Industrial Automation, security systems, motor control, industrial
control. Why look for new market segments when there are existing
customers out there already for this stuff?
What this means also is that
the companies that didn't "get it" before, that don't know how
to service microcontroller customers (good
docs, good support, great tools for all customers big and small)
will not be able to partake of the Embedded Internet market. Look for Motorola
to extend their reach with TCP/IP stacks for their CPU32. NETsilicon
is going to methodically (and successfully) go after Industrial Control
networks on the factory floor. National Semiconductor is going to
implement a carefully planned strategy with their easy to use CompactRISC
products. US Software, with their staff of 15+ tech support
engineers, is probably going to be the winner in TCP/IP with RTOS
solutions. And expect ZiLOG to surprise everybody with some
revolutionary products that will add ease of use of embedded internet for
existing embedded systems developers.
Note that all these solutions include real TCP/IP
stacks - not proprietary serial protocols masquerading as TCP/IP stacks,
such as the iPIC.
Personal Internet Appliances
A few innovative Personal Internet Appliances (PIA) are being introduced
to the marketplace in the form of portable mail stations. There will be
lots of fanfare surrounding each product introduction. They will initially
see some acceptance in the marketplace, but will see serious competition
as the price of PDAs come down, similar to the way WebTV couldn't compete
with low-priced PCs. In the end, a segment will be carved out for
under-$100 devices based on 16-bit microcontrollers with on-chip flash.
Why? Because x86 processors require too much OS overhead (read: memory
& clock speed requirements) that make them too expensive for these
applications. And the price of external Flash memory chips have been
fluctuating wildly, making their use too volatile for cost-efficient
systems.
Embedded Internet in Home Appliances
It will flop. None of the home appliance manufacturers I've contacted are
seriously working on this strategy. I haven't seen any surveys that say
consumers will purchase these items. Introducing networked home appliances
for the purposes of pre-failure diagnostics is a Marketing 101 mistake,
similar to why airlines don't promote their safety records - it introduces
the specter of product failure in the mind of the consumer.
8-bit Microcontrollers
Despite predictions to the contrary, the venerable 8051 will not die this year. It
will never die. There are too many tools, too many systems, too many
engineers out there programming it. May it live forever.
National Semiconductor will revive the COP8 with
some surprising new products with a faster clock speed.
Philips Semiconductor, once again, will do nothing
innovative with their microcontrollers, except maybe a Controller Area Network
(CAN) strategy to expand their handful of existing CAN 8-bit
microcontrollers.
Microchip will add new flavors to its PIC12, PIC16,
etc. products, but it's the PIC18 that will take center stage. Five, maybe
six new PIC18s in 2000 (with different package options), sporting CAN
interfaces and other networking peripherals will debut.
Motorola, to no one's surprise, won't be doing much
with its 68HC05 and 68HC11 products, instead choosing to focus on their
flash-based 68HC908 products.
Scenix will continue to focus on applications in
Southeast Asia because of their CEO's extensive experience in that market,
but don't expect much headway in the U.S. and Europe due to their lack of
an adequate distributor network.
Onward and Upward - 16-bits+
Motorola will introduce a next generation to the M*Core that will be low
power, low cost, and have plenty of Flash. Hitachi
will retain its position as the #1 supplier of 16-bit microcontrollers,
but only because Infineon has not yet been able to address the high power
consumption of their terrific C167 16-bit microcontroller.
Texas
Instruments may grab more 16-bit market share with their innovative
and low power MSP430 if they can just divert their PR focus from DSPs.
Infineon's 16-bit C167 will see steady gains, mostly due to their
excellent development tool support and their U.S. based Spacetools
3rd party tools program, and Willert
Software Tools in Germany.
Flash Memory
Expect anyone licensing Flash from TSMC or SST to have reliable,
functional Flash microcontrollers of 16K or more.
Digital Signal Processors
Texas Instruments will maintain and hold their lead in the DSP market.
They've got too much momentum for anyone to even come close, although
Analog Devices' 32-bit SHARC is planning a fight with some cost-effective
new products for 2000. ADI's easy to use 16-bit DSP, the ADSP-2100, will
languish, waiting for the joint ADI-Intel register-based 16-bit DSP to
obsolete it. The question is, will it make it to market before TI's 16-bit
register-based DSP?
Look for Motorola to make some advances with their
56K DSP architecture. Now that they've bought Metrowerks they've got
in-house development tools like Texas Instruments (Go-DSP,
probably the best DSP development environment available) and Analog
Devices (White Mountain DSP, Edinburgh
Portable Compilers).
Development Tools
Harsh reality: the good will get better, and the bad will fail. With the
expansion of personal contacts of embedded systems engineers via email,
newsgroups, and websites, word of mouth will case attrition in the
development tools marketplace. Witness the disappearance of the in-circuit
emulator company Dr. Krohn & Stiller
(web site down) - their products were highly unstable, and thanks to email
their problems became universally known.
For programming language, there will always be C and
assembly, while C++ will gain a few more points of market share. Embedded
java will continue to miss forecaster's prediction of acceptance
In Conclusion
Expect a new blitz of products coming as you from all directions. Despite
any and all product claims, flexibility and
practicality will win out, as they always have. Solutions
looking for a problem will disappear faster then fuzzy logic or embedded
java. Products with clear documentation, easy to use development tools,
and superior online support will quickly find their way into emerging
markets. These will be driven by companies with people that understand the
embedded market. Companies composed of people that have been writing
Windows programs, with marketing done by their design department will
delight and astound the press, but in the end, most Embedded Systems
Developers will make the practical choices for the emerging markets.
Happy 2000
In closing, Happy New Year to you all, and my personal Thanks to everyone
that has supported my efforts to promote and maintain Microcontroller.com.
All a webmaster wants is to know that their efforts are appreciated and
that they are producing content for an appreciative audience. February
2000 will mark the Fourth Anniversary of Microcontroller.com - that's four
years of changing web hosts and Gratuitous Capitalization. The site has
gone from a 486 to a server that sports a dual Pentium III 500 with I
forget how much RAM. To everyone who has emailed me suggestions and
comments on this website, and to my friends in Michigan and Indianapolis
that have supported my efforts and helped promote this site - an extra
Happy New Year. In the words of Alan Borghard, may your emitters be
grounded and your diodes forward.
Bill Giovino is Executive Editor of
Microcontroller.com.
|
Internet
