Differences between E01 and E01S
| FileStore E01
| FileStore E01S |
| |
| The Winchester disc interface is essentially a 1MHz bus running at 2MHz;
the SCSI controller resides within the E20 expansion.
| The harddisc interface is included on the main board, providing a
SASI/SCSI interface on the rear of the server.
The E40S and E60S harddisc upgrades are little more than Rodime SCSI
harddiscs and a power supply in a box... |
| |
| As a consequence of the above, it is usual that E01 servers only
have one harddisc expansion. More are possible, but it is tricky.
| As a consequence of the above, expanding the FileStore is possible by
simply adding harddiscs to the daisy-chain arrangement, up to a maximum of four
units, in any order (a potential maximum storage of 241.2Mb (60Mb × 4 + 1.2Mb (2
× 600Kb floppies)). |
| |
| Earlier versions of the server code used the old-style date
(see here for details) and may fail on dates after 1996.
| As far as I am aware, there are no date problems with the FileStore
until the end of time in 2108. |
| |
Various internal differences:
- No on-board SCSI interface - but you already knew that.
- The firmware is split into two ROMs - one holds the server and the other holds the
OS.
- The front flap sensor is 'read' using the VIA.
- VIA port B is hooked to a set of links to provide a
Net ID.
| And the E01 differences are...
- On-board SASI/SCSI interface.
- The firmware is contained in a single 64Kb ROM.
- The front flap sensor is 'read' using a latch mapped into memory.
- VIA port B is unused. There is no Net ID in the E01S.
|
The above may not be correct for all versions of the
E01,
it seems the schematic in the service manual does not match some of them; namely SW3 and flap sensor are accessed by a memory location and not the VIA, plus no NetID on VIA port B. |
| |
The internal network clock is fundamentally broken. Namely, the C+ and C- lines are reversed so it provides the opposite mark/space ratio to a conventional clock box. This severely effects the usefulness of the on board clock and generally for a network of any length an external clock should be used.
Note that you can't just reverse the clock lines at the Econet socket as the internal Econet interface will still use the inverted ones.
| The internal clock is wired up correctly, but like the E01, it can sometimes step in if it fails to correctly detect that there is an existing clock box on the network. Leaving the timing links open effectively disables the built-in clock. |
| |
| The command set is similar to the Level 3 server; i.e.
MaxUser and Report; but only
if you have an older version of the firmware. It appears as if later firmware makes the
commands E01S-like.
| The command set has been rationalised into something more specific
to the server (less likely to clash with extension ROMs, etc) - generally by prefixing
"FS" in front of things; i.e. FSMaxUser and
FSReport. |
There is a modification "ICmod" to the E01S (E01 also?) which prevents the possibility of spurious writes to NVRAM by fixing if/when/how the !CE line is synced to chip access. More specifically, !CE is gated off of the rising edge of AS (address strobe) so that !CE is only low when AS goes high; as opposed to being low as and when determined by the misc. functions latch (which will be somewhat longer than necessary as the processor will be fetching and executing code to access the device after setting the latch).
| If you can think of any other differences, please email me. | |
With thanks to Alan Williams for additional information on the clock circuit screwup.
|