[En-Nut-Discussion] RFC: Ethernut IDE/CF Interface

[Harald Kipp]

Hi,

Several people suggested an Ethernut IDE/CF interface
during the last months (should I say years).

Once again we thought to have something final now,
but actually new problems appeared and I'd like
to do it right.

Some history:

Michael Fischer came up with a working IDE board
based on two PALs. We decided to switch to a CPLD.

The CPLD version was again done by Michael and I
did the board layout. This already included a
5V CompactFlash interface.
http://www.ethernut.de/en/ide/index.html

Again a new design was started, adding 3.3V
capabilities to the CF interface with 74LVC245
level shifters.

But this interface does support true IDE mode
only and would not work with CF+ cards like
the Netgear MA701.

Recently Michael builds a PCMCIA prototype for
the Ethernut and a 5V MA401 WLAN card. But he
soon realizes, that accessing the card fails
because in port mode and memory mode the card
activates its WAIT line. This problem is similar
to the RTL8019AS IOCHRDY issue. The ATmega128
can't extend memory access cycles on external
signals. I'm sure that this will also occure
with CF cards, the ATmega128 is too fast.

The final idea is now to use one dedicated CPLD
for the CF interface and one for the IDE drives.
The IDE would use a 5V CPLD and the CF interface
would be attached to a 3.3V TTL tolerant type.
But I'm not sure, if this is really required
or if both interface could use a single CPLD.

The CF CPLD should then take care of the CF Wait
Signal and emulate the memory access cycle delays.
But I do not know much about clocked CPLDs. The
one on the Ethernut is fully static.

Any ideas, comments?

Harald