Searching \ for '[PIC]: Taking it to an extreme' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/devices.htm?key=pic
Search entire site for: 'Taking it to an extreme'.

Exact match. Not showing close matches.
PICList Thread
'[PIC]: Taking it to an extreme'
2000\06\16@222735 by Bob Ammerman

picon face
I have been deeply involved in the thread about the 'open source emWare like
project'  (in fact I started it :-), but I have been thinking about another
end of the spectrum (I've noticed most spectra end up with many more than
two ends :-).

A 10Mb/second ethernet signal could be directly generated by a PIC. I am
currently generating video at that rate using the USART of a PIC18C452.

What I am thinking is that an application could build a complete ethernet
packet (containing, if desired a UDP datagram embedded in an IP frame) in
RAM. This packet would include the complete ethernet preamble, packet body
and CRC.

The packet could then be sent _without the use of a conventional NIC chip_
onto an Ethernet.

This would require, of course, a PHY (physical) interface of some sort to
the Ethernet, and at least some sort of collision detection (although the
latter would probably be unneeded if we are connected to a switch instead of
a simple repeating hub.

Has anybody tried anything like this?

Anybody know where to go to get the details of the physical level of the
Ethernet?

How about the data link level (ie: preamble, framing bits, CRC computation,
collision resolution, etc)?

Note that this would be a unidirectional link (I am sure receiving a packet
is a _much_ more difficult task), but could be very useful for smart-sensor
type devices (ie: alarm systems, thermostats, etc.).

TIA

Bob Ammerman
RAm Systems
(high performance, high function, low level software)

2000\06\16@231016 by Dale Botkin

flavicon
face
On Fri, 16 Jun 2000, Bob Ammerman wrote:

> I have been deeply involved in the thread about the 'open source emWare like
> project'  (in fact I started it :-), but I have been thinking about another
> end of the spectrum (I've noticed most spectra end up with many more than
> two ends :-).

Ain't it the truth??  8-)

> Anybody know where to go to get the details of the physical level of the
> Ethernet?

> How about the data link level (ie: preamble, framing bits, CRC computation,
> collision resolution, etc)?

www.techfest.com/networking/lan/ethernet.htm
www.techfest.com/networking/lan/ethernet2.htm
http://www.acm.org/classics/apr96/

Dale
---
The most exciting phrase to hear in science, the one that heralds new
discoveries, is not "Eureka!" (I found it!) but "That's funny ..."
               -- Isaac Asimov

2000\06\17@093432 by Scott Dattalo

face
flavicon
face
On Fri, 16 Jun 2000, Bob Ammerman wrote:

> I have been deeply involved in the thread about the 'open source emWare like
> project'  (in fact I started it :-), but I have been thinking about another
> end of the spectrum (I've noticed most spectra end up with many more than
> two ends :-).
>
> A 10Mb/second ethernet signal could be directly generated by a PIC. I am
> currently generating video at that rate using the USART of a PIC18C452.
>
> What I am thinking is that an application could build a complete ethernet
> packet (containing, if desired a UDP datagram embedded in an IP frame) in
> RAM. This packet would include the complete ethernet preamble, packet body
> and CRC.
>
> The packet could then be sent _without the use of a conventional NIC chip_
> onto an Ethernet.

I've been closely following this thread with great interest. However, this stuff
is outside my area of expertise. If you do take this approach, wouldn't require
the PIC to continuously monitor the ethernet traffic and accepting/rejecting
messages? This is probably the biggest benifit of the MAC. The second biggest is
the 8 or 16k of ram buffer that they provide. The third is the CSMA/CD support.
For cost sensitive applications, I could see sacrificing one PIC to bit bang
ethernet. But quite honestly, I don't think it's possible. Perhaps an SX chip...

Anybody got an idea how much some of these 10baseT ethernet controllers
cost? The CS8900 is in the $10-$15 range as I understand. What about the
RTL8019? What other chips could be used? The AMD, Intel, Infineon, etc. chips
are geared toward PCI busses and 10/100 controllers.

Scott

2000\06\17@200628 by Bob Ammerman

picon face
As I mentioned before: receiving on the internet would be quite difficult to
do in a PIC, but sending should be relatively easy. Collision detect should
be rather simple, IMHO.

The idea here is not to substitute a PIC for a NIC (hey  - that's almost
poetic :-)), but rather to eliminate the NIC entirely, using the
one-and-only pic in the system to directly drive the ethernet.

Bob Ammerman
RAm Systems
(high performance, high function, low-level code)

{Original Message removed}

2000\06\17@204219 by Dan Michaels

flavicon
face
Bob Ammerman wrote:
........
>
>The idea here is not to substitute a PIC for a NIC (hey  - that's almost
>poetic :-)), but rather to eliminate the NIC entirely, using the
>one-and-only pic in the system to directly drive the ethernet.
>

So, presumably you're gonna call it the PicNic, so you can
get into trademark problems, too!!!!

2000\06\17@220924 by Bob Ammerman

picon face
----- Original Message -----
From: Dan Michaels <spam_OUToricomTakeThisOuTspamLYNX.SNI.NET>
To: <.....PICLISTKILLspamspam@spam@MITVMA.MIT.EDU>
Sent: Saturday, June 17, 2000 8:41 PM
Subject: Re: [PIC]: Taking it to an extreme


{Quote hidden}

ROTFL

Bob Ammerman

More... (looser matching)
- Last day of these posts
- In 2000 , 2001 only
- Today
- New search...