I intend using CF cards in the range 32 - 128 Mb as mass storage with a Z80-based system, and would prefer to use an 8-bit interface to avoid using extra latches etc. So, to save re-inventing the wheel I've been poking around to see what others have done.
Zilog's Advice Note AN015404-0608 demonstrates the use of a 'universal' CF interface (using latches and glue logic - pity) which uses either memory or true-IDE. 8-bit mode in both cases. Their example uses a 64Mb CF card - in which ALL of the memory is accessible in both modes.
Another reference to the successful use of an 8-bit IDE interface is:
8-Bit IDE/ATA Interface (1802 processor) "This circuit provides an 8-bit interface to IDE/ATA devices. This circuit will only allow half of an IDE hard drive to be utilized, unless the device supports the 8-bit mode. CF requires that 8-bit mode be supported. therefore CF cards used on this interface will utilize the full capabiliities of the card."
Which sounds promising ... but I have found 2 references to where this isn't happening in practice:
Simple 8 bit IDE Interface with Compact Flash card. on Flickr - Photo Sharing! (SPECTRUM)
Uses 8-bit IDE - only half of the memory can be accessed
ZX CF interface (SPECTRUM) "On schematic is chip select logic. It is practically same as by 8-bit simple IDE interface. But main advance here is that we work with CF card in memory mode, where data transfer is 8-bit wide, so no capacity loss. Full capacity of CF card is available, while interface remains simple."
So - is there anyone with first-hand experience of using CF cards in 8-bit IDE mode who could advise where the Spectrum guys are perhaps going wrong with 8-bit IDE ? It would be useful to learn from the mistakes (assumption) of others.
'best, Colin