Date   : Tue, 26 Oct 2010 09:31:39 +0100
From   : jumbos.bazzar@... (Mark Haysman)
Subject: Reversing the Tube ULA (destructively)

>> Yes, I remember, though this could just be a limitation of the
>> programmable logic he used - that adding a larger FIFO would not have
>> provided suitable gains for the work involved?
>
> I think he used a 64 or 128 macrocell CPLD, so it's probably a
> limitation of not enough flipflops, using a small FPGA would have solved
> this though brought it's own problems too, such as needing level
> shifters as most FPGAs operate at 3.3V or less, and needing a
> configuration prom, CPLDs generally don't need this as they are mostly
> flash based.

Well you've been in the discussion on the *. forums and seen the 6502 
Internal Co-Pro I've done for the M128, which uses a Xilinx XC3S50AN. That 
chip doesn't need an external config prom, as it has it built in. The level 
shifters are a pain in the butt, and seem to operate most illogically! I'm 
currently using a 74LVC16T245, which is designed so that either side of the 
buffer can operate at any voltage (between 2.5 and 5), however, operating 
the Beeb side at 5v just causes garbage on the screen, it seems happier 
running it at 3.3v?

> I just wish someone did a CPLD with lots of logic but not using stupid
> numbers of pins :(

I'd prefer a 5v FPGA in a 40 pin DIP package with the internal logic of an 
XC3S200AN! Bliss!

>> I suppose the question is, do we want an EXACT replica, or "something
>> that works"?
>
> That's the difference between an exact clone of the hardware and a
> logical clone. For example if you took my Atom clone, put it in an Atom
> case and screwed it shut you hopefully woudn't know it was a clone, as
> it's logically identical (or near as damn it !), but it's not a literal
> clone of Acorn's circuitry as it uses CPLDs, modent static ram etc to
> simplify things.

Have to agree that an "exact" clone would be preferable. I'm working with Ed 
and Richard on his open source tube code to try and get it 100%, but there 
seem to be odd quirks in there that would be a lot easier to solve if Ed 
could have success with a proper RE of the original chip.

Mark.