Date   : Thu, 19 Jul 2007 08:54:35 +0100
From   : info@... (Sprow)
Subject: modern BBC remake

In article <4f0484feedinfo@...>,
   Sprow <info@...> wrote:
> I've been chewing it over for at least 5 years under the project
> name "TurboB". But more on that seperately.

Here goes with my thought processes:

Hardware:
 * Motherboard designed to fit as a drop in replacement to a normal B/B+ case
 * Uses a 14MHz 6502 from WDC
   Clock speed selected by *CONFIGURE command.
   Flash ROM device for 12 ROMs, physical sockets for 4.
 * Programmable logic mops up the acres of 7400 series logic currently used
   SRAM instead of DRAM, no refresh circuitry required.
 * 6845/video ULA/Teletext emulated in programmable logic
   New set of registers to allow modes up to 1280x1024x256.
   Shadow screen memory, like a B+/Master.
 * 1770 floppy controller
   Possibly in programmable logic as I've not got many chips left, though
   there seem to be some with brokers which would save a lot of effort.
 * Other peripherals
   1MHz bus/RS423/analogue in.
   Battery backed real time clock stolen from MiniB.
   Speech as standard.
   Sound in programmable logic, as J Kortink has the PRBS white noise worked 
   out.
 * Upgrades
   Space for GoMMC upgrade.
   Space for Econet/Ethernet networking (probably in place of cassette port).
   Internal tube connectors per the Master.
   External tube connector configurable as either an IDE interface or Tube
   via *CONFIGURE command.
 * System/user VIA
   Possibly no printer port, replaced by a USB PIC and/or PS/2 keyboard port.

Software:
 * Based on patched B+ OS
   Software writing takes ages, certainly longer than any hardware design, so
   the hardware would necessarily mimic the original machine faithfully with
   any new or enhanced peripherals supported by a paged 'Utils' ROM or small
   patches applied to the existing OS.

Reasons not to do it:
 * The cost of tooling up for a new run of injection moulded cases/power 
   supply/keyboards would be ?10,000's, so to get it cased would involve
   sacrificing a working machine's case.
 * It'd take about 2 years to complete, by which time any likely interest 
   may have dwindled (what happened to the Econet <> Ethernet bridge?). If
   I had a completed one now today things might be different...
 * Other than the convenience of every known peripheral on one board, it's
   not actually bringing anything new into the world. All of the above can
   be achieved today by one route or another I think.

Reasons to do it:
 * Because we can.

Estimated price:
A bare PCB, 2 layer of that size is about ?30. Depending on the programmable
logic device chosen it may need to be 4 to get the traces out, making about
?40.

The assembly would take a few hours based on the huge number of components
(compared with, say, MiniB). If anticipated volumes were > 50 (say), it'd be
worth paying the ?200 for a paste stencil and having them semi automatically
built and soldered.

We'd only be making 10's off I suspect. I estimate ~?300.