[En-Nut-Discussion] WLAN: Some question to the Ethernut community

[Harald Kipp]

Hi Peter,


>Why horsepower/CLPD ??

:-)
You do not know the final CPLD that's on the CF/IDE card.
It's an XC95144XL.

>IOSoft in the UK does connect a WiFi card directly to a PIC processor
>without any glue. I hope I have not to conclude that a PIC processor is a
>'better' product for this task... :)
>http://www.iosoft.co.uk/wlan2.php

Michael uses PCMCIA to get something done, because
we were not able to provide a CF adapter in time.
But finally CF is the goal.

The first problem is, that almost all CF+ cards do
support 3.3V only. So you need at least some kind
of level translation. It may work with the Netgear
MA701 (CF+). Once I accidently plugged it into
Michael's PCMCIA card and forgot to change the 5V
jumper, using a PCMCIA to CF adapter. The card has
survived this torture, but of course this was sheer
luck, not all cards will take this.

The second problem is, that most 8-bit microcontrollers
do not support an external wait signal. Michael's
current CPLD logic uses simple registers to provide
the PCMCIA signals. It is planned to implement a
state machine into the CPLD, which handles complete
read/write cycles. I talked to Louis Beaudoin about
this and in his opinion the XC95144 is oversized.
A XC9572 will provide enough macrocells for the
state machine and many signals can be shared between
IDE and CF.

Third, I'd like to add hot plugging. So the final CF/IDE
card will even use additional busdrivers.

Forth, I want to use a minimum number of port pins.
The card will use up to 3 interrupts and possibly
one pin to reset the CPLD. All registers will be
memory mapped.

It is no problem, to downgrade the hardware later.
But this is our fourth design over more than a year
trying to finish it. I do not want to take any
further risk, like sharing the wrong signals or
running out of macrocells. The board should be
open to all kind of experiments.

I assume, that Jeremy's board is designed towards
a minimal and cheap solution, while we try to create
a most flexible development platform.

Niels, my son, is currently working on the part
placement and may have to change the schematics again for
optimal routing of the traces. As soon as this is
done, I'll publish a first preview.

Regards,
Harald