Satellite stuff (Was: Re: BBC printer port (ever used for

[Tony Duell]

> Yep, the Astrid receiver I have is exactly like that - there's some TTL
> in it which is used to drive the computer's serial port, plus sockets
> for hook-up to a tape unit for recording.
> 
> That's what makes this other box that I have odd - the Astrid receiver
> includes the necessary logic for interfacing to the CPU, and this other
> box doesn't seem to have enough I/O lines to be for complex data
> aquisition (such as imagery) - but I can't imagine what it's doing that
> couldn't have just been done in software on the host computer. 

Yes, but rememebr back then that there were many incompatible host 
computers (Beebs, Amstrad CPCs, Spectrums, C64setc, etc, etc) and that it 
was common to keep the software to a minimum to make it easier to use the 
device with just about every machine out there.

> > Alas not. I got the I2S driver tape I think, but nothing more :-(
> 
> That's a shame. There must be a lot of big and complex systems that
> would make interesting museum displays which have gone the same way due
> to lack of software preservation :-(

My I2S image displays are not going near any museum. They're far too 
interesting for that!

THey are complex. Very. Crates of 30 boards each, stuffed with TTL and 
DRAM chips. The oldest model I have, a 70/E, has 2 crates of memory + 
controllers, the RAMs being 4K bit DRAMs (somewhere over 3000 such chips 
in the complete system). Later models used 16K and 64K DRAMs and got the 
whole machine into 1 crate.


The basic idea is to have several byteplates (512*512*8 bits) which are 
combined by lookup tables, adders, and more lookup tables to produce the 
RGB signals to the monitor (either using 8 bit or 10 bit DACs). The 
manual points out you can subtract byteplanes by programming a 2's 
complement lookup table, multiply them by programming log/antilog tables, 
and so on.

There's also a cursor genreation board, controlled either by a 
Summagraphic tablet or a trackball (I have 3 machines, 3 trackballs and 2 
tablets IIRC).

And a histogram board which produes a histogram of the pixel values in 
memory. It akes 1 frame time to do this (of course).

And the freeback ALU, which treats planes 0 and 1 as a 16 bit 
'accumulator' plane and lets you do accumulator = accuumulator {*} 
any_other_plane where (*) is any operation a 74181 ALU chip can perform 
(add, subtract, AND, OR, XOR, etc). Again one frame time for a complete 
update.

The latest model I have (model 75) has something called the transform 
sequencer which cotnains a 2910 sequencer chip + microcode PROMs + 
control store RAM. It appears you can program sequences of operations 
into it to be repeated automatically. Alas while I have scheamtics of 
that card, I don'd have any info on how to use it (or an explanation of 
the contents of the PROMs).

-tony