confuse the unwary, the documentation gives the clear impression that the /NASIO
circuitry must be added by the user, and the manual goes on to suggest a method
of implementing this (using extra chips, of course). To really confuse matters,
the actual circuitry used (it IS supplied) lives on the prototyping area and uses
a completely different circuit to that suggested!! The DBDR circuitry (for
Nascom-1 owners) also lives on the prototyping area.
The circuit itself uses the Z80-SIO to provide the necessary parallel to
serial conversion. TTL and CMOS logic provide the phase-encoding and the
interface to the drives. The system uses phase-encoding to store bits of
information on the tape. Briefly, this entails a change of flux on the tape
representing a 1 or a 0, depending on whether this happens on a positive or
negative-going clock edge. This ensures that the tape magnetisation changes at
least once every clock cycle. Having seen the source listing of the CFS system
(which uses the PIO) I must say I think that using the SIO was the best approach,
since it avoids the machinations required to generate the serial data stream,
generate checksums, sync. characters, etc., and this is reflected in the command
set available, which is certainly much more extensive than that on CFS (more
about the software later). It also allows an easy upgrade to DMA.
The data rate used is 6000 bits per second, which, because no start/stop
bits are used, is the equivalent of 7500 bps (e.g. using CUTS, if you could get
it to work that fast!). Data is recorded in blocks of 2K bytes, and the
catalogue, which is at the start of the tape, also uses a 2K block. This gives a
capacity of 56K bytes per side, excluding the directory. Any data less than 2K in
length uses a ‘padded out’ block. Each data block consists of sync. characters
(which serve not only to indicate the start of a block but also to synchronise
the read electronics), the load address, a length word, the data stream itself,
followed by the CRC characters. Thus, say, a 10 byte ‘file’ can be loaded without
the other 2038 bytes ‘above’ it in RAM being corrupted.
The hardware falls down in one very important point: whereas the drive
buffers use ports £F8-£FB, the SIO uses ports £FC-£FF. This is bad news indeed to
those using page mode on their RAM/ROM cards, since the page mode circuitry also
uses port £FF (and port £FE on the MAP 256K RAM and Gemini GM813 CPU-I/O-64K RAM
boards). Use TOS, and you’ll inevitably ‘load’ a file into ROM. Switch a RAM card
back into the system & you’ll be reprogramming the SIO!!! The solution is to
reconfigure the SIO at a different I/O address (and change the software to suit).
The port decoding is hard-wired, but luckily there are a couple of unused
inverters on the board, so it is just a matter of breaking a track or two and
inverting an address line to reconfigure the port decoding. This is something I
have yet to do (spot the RAM-A user!). Neither are the address/data lines
buffered on the card: some of the lines have several LS-TTL loads attached to
them. Buffering would then make it easier (neater) to mount the 2708s/4118s on
the card if your Nascom (& RAM-A) card are already chock full of ZEAP, NASDIS,
DEBUG, NASPEN, etc., or has been reconfigured for 2716/6116’s.
The drives themselves (of which no information is supplied) are very
compact, being about 4" cubed in size. They require a 12 volt power supply, which
in this case is drawn from the NASBUS. The manual says that if more than 800mA is
already being drawn from the 12V line, then an extra power supply may be needed.
The cassettes are very small, being the same type as those used in ‘Dictaphone’
machines, only the ones used are certified free from drop-outs, which they do
indeed seem to be (unlike ‘computer quality’ C10 audio cassettes). This means
that Read/Write errors are definitely a thing of the past! Uncertified cassttes
may also be used, but on your own head be it!
The operating software (called ‘TOS’ – Tape Operating System) occupies 2K
bytes and is supplied in two 2708 EPROMS. It resides at location £D0000, and uses
2K bytes of workspace, which ‘sits’ on top of TOS at £D800. Two 4118s are