I looked at this a few times. Has anybody written and tested a software program for a Z80 with SD cards to actully read and write sectors. Different SD cards have a complex protocol interface. I have been using an SPI module in a Cyclone IV FPGA with its own internal Z80. Works fine but like the above proposed simple hardware inherface

Assembly Interface
Bytes are transferred through the interface with this simple assembly code:
transfer_byte:
out $00
receive_byte:
out $01
in $00
ret
First the byte to be transmitted is written to port 0, then a write to port 1 causes the
SPI transfer to take place. Finally, a read from port 0 retrieves the received byte.
My notes indicate that no "nop" instructions are required before the "in"
instruction, but I was probably using one of those old 4 MHz Z80 chips at the time.
A 10 MHz chip probably requires two "nop" instructions and a 20 MHz chip
probably requires six, unless you use a faster clock for the SD transfer. The limit on
the SPI clock is probably going to be the counter chip you use in the circuit below.
The SD cards themselves are already designed to transfer data at insane speeds.

Anyway, this sets up port 0 to read/write the shift registers. A write to port 1 then
starts the SPI transfer. Ports 2 & 3 when written to toggle the /CS on the SD card,
with port 2 lowering /CS and port 3 raising it. Ports 2 & 3 when read from similarly
toggle a pull-up / pull-down resistor attached to the SD card's data output pin,
which can be used to detect if a card is present by changing the polarity and noting
whether the bits read from the interface continue to match the pull-up / pull-down,
so that an empty card slot looks different than an unresposive but present card.

As an extra note, a bulk and a bypass cap should be provided right at the card socket pins on the power/GND supply. Older cards were much more demanding.

It's a shame there is no 3D model for the SD Card footprint, at least that I've been able to find. It is the one missing piece of the 3D render

I too had problems with multiple SD cards (4) sharing the SPI data line. Cost me a week screwing around! Wish I had known that. The fix was to have each SD card have its own FPGA data pins. Good to know it's not just me!

Curious about how many clock cycles to be safe.
John

Thanks cluso99. I did things differently. I just send 8 clocks and stop everything at once.
John