1
0
mirror of https://github.com/hsoft/collapseos.git synced 2024-11-15 17:38:05 +11:00
collapseos/doc/hw/z80/spi.txt
Virgil Dupras 7001446212 Complete overhaul of recipes
Recipes contain bits and pieces of hardware-related knowledge, but
these bits feel sparse. I've been wanting to consolidate hardware-
related documentation for a while, but always fell at odds with the
recipes organisation.

We don't have recipes anymore, just a /doc/hw section that contains
hardware-related documentation which often translate to precise
instructions to run Collapse OS on a specific machine.

With this new organisation, I hope to end up with a better, more
solid documentation.
2020-10-30 20:39:39 -04:00

75 lines
2.8 KiB
Plaintext

# Building a SPI relay for the z80
In this recipe, we build a SPI relay (see /doc/hw/spi.txt) for
a RC2014.
# Gathering parts
* A RC2014 Classic
* A proto board + header pins with 39 positions so we can make
a RC2014 card.
* Diodes, resistors and stuff
* 40106 (Inverter gates)
* 74xx138 (Decoder)
* 74xx375 (Latches)
* 74xx125 (Buffer)
* 74xx161 (Binary counter)
* 74xx165 (Parallel input shift register)
* 74xx595 (Shift register)
# Building the SPI relay
The schematic (img/spirelay.jpg) works well with the SD Card
subsystem (B420). Of course, it's not the only possible design
that works, but I think it's one of the most straighforwards.
This relay communicates through the z80 bus with 2 ports, DATA
and CTL and allows up to 4 devices to be connected to it at
once, although only one device can ever be active at once. This
schema only has 2 (and the real prototype I've built from it),
but the '375 has room for 4. In this schema, DATA is port 4, CTL
is port 5.
We activate a device by sending a bitmask to CTL, this will end
up in the '375 latches and activate the SS pin of one of the
device, or deactivate them all if 0 is sent.
You then initiate a SPI exchange by sending a byte to send to
the DATA port. This byte will end up in the '165 and the '161
counter will be activated, triggering a clock for the SPI
exchange. At each clock, a bit is sent to MOSI from the '161 and
received from MISO into the '595, which is the byte sent to the
z80 bus when we read from DATA.
When the '161 is wired to the system clock, as it is in the
schema, two NOPs are a sufficient delay between your DATA write
and subsequent DATA read.
However, if you build yourself some kind of clock override and
run the '161 at something slower than the system clock, those 2
NOPs will be too quick. That's where that '125 comes into play.
When reading CTL, it spits RUNNING into D0. This allows you to
know when the result of the SPI exchange is ready to be fetched.
Make sure you AND away other bits, because they'll be garbage.
The '138 is to determine our current IORQ mode (DATA/CTL and
WR/RO), the '106 is to provide for those NOTs sprinkled around.
Please note that this design is inspired by
https://www.ecstaticlyrics.com/electronics/SPI/fast_z80_interface.html
Advice 1: Make SCK polarity configurable at all 3 endpoints (the
595, the 165 and SPI connector). Those jumpers will be useful
when you need to mess with polarity in your many tinkering
sessions to come.
Advice 2: Make input CLK override-able. SD cards are plenty fast
enough for us to use the system clock, but you might want to
interact with devices that require a slower clock.
# Driving the relay
There is a provider for the SPI protocol (doc/protocol.txt) that
work with this device in B418. It needs SPI_DATA and SPI_CTL
constants which in this case are 4 and 5 respectively.