Jesse Lackey wrote:
> Ethernet PHY... I agree, a microcontroller with MAC but no PHY is
> nearly pointless.  Adding the PHY chip is cost, boardspace, and
> hassle.

I agree.  That's why I like the 18F67J60 for simple networking projects.

Our network stack is working quite nicely, and can drive the on-chip MAC/PH=
Y
of that PIC directly, among others.  A simple demonstration project uses
only 15% of the program memory space, and includes the device independent
packet layer implementation for the built in MAC/PHY, ARP server and client=
,
IP, enough of ICMP to reply to PINGs, UDP, DHCP client, and TCP.  This has
now been well beat on by several different folks, including subjecting it t=
o
PING floods, deliberately flaky networks, etc.  Everything appears to work
correctly.  In one case our stack stood up all weekend to a torture test,
while the previous version based on the Microchip stack had a average up
time of 600ms.  Yes, milliseconds.  To be fair, that was a old version of
the stack, but I still don't have much confidence in Microchip code.

Most of the code has been quietly included in our PIC Development Tools
release for a while now, although it hasn't been tied together into a
demonstration project with proper documentation yet.  We'll probably make
another ReadyBoard targeted to ethernet projects, and will include a workin=
g
example using this network stack.  Things have been really busy for a while
now, so such side projects aren't moving very quickly.  (A MiWi ReadyBoard
is further along in the works, stay tuned).

If anyone wants to use the network stack, let me know.  It's all available
for free and there is no restriction on commercial use.

> I would imagine a 10/100 PHY is not a simple thing and may require
> more expensive/alternative/additional steps chip manufacturing
> process (for
> the analog) than is needed otherwise.

Possibly, so skip the 100 part.  That's pointless anyway on a small
microcontroller and with any modern ethernet switch.  You'd be hard pressed
to find a ethernet switch nowadays that doesn't adjust its speed per port.
That means you can connect the small embedded system at 10 Mbit/s, which is
more than it can handle sustained anyway, but not slow down the rest of the
network.  Or put another way, you're already paying for 10 to 100 conversio=
n
capability in the switch.  You might as well use that to make the small
embedded system simpler.  Note that there is no loss of throughput since th=
e
small system can't keep up with 10 Mbit/s in the long run anyway.

The whole fixation on 100 Mbit/s by Microchip is silly marketing at its
worst, actually hurting real capability if it prevents a integrated MAC and
PHY on the same chip as the PIC.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.