William \"Chops\" Westfield wrote:

> There is all this hype at the moment about 32bit CPUs replacing 8bit
> cpus, as the cost of the CPU and memory goes down to 8bit levels, and
> chip manufacturers target the market.
>
> NXP in particular has the LPC1102, an ARM Cortex M0 with 16 pins, 32K
> of flash, 8K of RAM, and the usual set of 8bit-like features, and a
> more-or less 8bit price.
>
> HOWEVER, this thing comes in a WLCSP Package (Wafer Level Chip Scale
> Package), which is essentially a 4x4 Ball Grid Array with 0.5mm
> spacing.  Not hobbyist friendly at all :-(
>
> Now, while I myself am primarily a hobbyist when it comes to building
> hardware, it has always been my perception that there are a large
> number of "real companies" out there making "real products" that are
> never-the-less not in an economic position to commit the sort of
> resources needed to create a design based on such a part (4-layer
> extra fine-pitch PCB and probably outsourced assembly), especially
> during initial development.  And for that matter aren't really
> creating boards in high enough volumes to justify that sort of
> manufacturing for their final product, either.  In other words, a lot
> of the remaining 8bit applications exist no so much because there
> isn't a more powerful 32bit replacement, but because the 8bit cpus
> remain much cheaper and easier to "develop and manufacture, for some
> large classes of products, compared to the more modern wonder-chips.
>
> In other words, the LPC1102 ought to be doomed to failure.
>
> Am I completely off base here?  Has the sort of PCB technology needed,
> and the automated manufacturing, become so available that the initial
> "hand assembly" phases is completely gone?  (This is certainly true at
> my real job, where I (nominally) write software for boards with many
> layers and many chips that cost big $$$$$ and have very long design
> cycles...  We've got rework techs and even an Xray machine for looking
> at those BGA joints, but they're all for adjusting things AFTER the
> boards come back fully manufactured.)  Are the "small businesses"
> doing micro-based product design beyond the "we build the first few by
> hand, perhaps painfully" or not?  Are there limits on manufacturing
> technology used by/for that class of company, or is that a thing of
> the past?

It took a long time for any company to find a market for the Cortex M0
for most of the reasons that you have outlined. NXP has been the first to
identify a market where they think Cortex M0 can be applied.

The death of 8 bit micros has been announced many times in the last
decade. The very fact that companies like Microchip are still doing
fundament development on 8 bit micros suggests that these premature
announcements are far from over. The extended mid range parts are
a good example of this. The additional instructions, linear memory,
additional index register and modernized memory management and
better implementation all make this part family competitive for
some time. This is a recent investment.

You have raised some important points. There is always been a
market for small run specialized applications using the unique
technology of small companies. These companies don't have the
resources to put the tools in place for multi thousand piece product.
Probably what is more important is setup costs are going to dictate
that for small run products a solution that is lower tech will probably
cost less and be more effective.

History is on our side on this. One of the earliest tool sets we
developed was for a special purpose 4 bit processor. 31 years
later it is still in production.

We are betting that 8 bit micros will be alive for some time.

Regards,


w..
--
Walter Banks
Byte Craft Limited
http://www.bytecraft.com




--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.