13 Mar
2003
13 Mar
'03
4:43 a.m.
>> This is what Eric Smith tells me as well. As
I told
>> him, it doesn't make much sense. The board has a 6502
>> as well as the 6507 with the data lines tied together.
>> I guess they could use opposite phases or something.
>> I was tracing down the sync signals for the video
>> and they seem to be coming from this part??
>> I do have a schematic someplace but I haven't located
>> it yet.
>> Now I have a mystery to solve?
I have an old Voelker-Craig terminal that uses this video
generation scheme. The Design uses two (count em) two
6502s, but only one runs rom code. The other steps through
a few 2114s as video ram. add a latch, shift register - voila video.
Think of the 6502 running on an inverted E clock with some
bits used as a rather expensive (even then) binary counter.
I would have used 74LS393s ;) Using some low order bits
to address ram while using higher ones for H and V sync would work.
Strangely, the whole system was clocked from a 4x colourburst xtal:)
>Possibly it's a variation of the Lancaster TVT
(Cheap Video) design?
Cheap video actually used the CPU and an ISR to generate video, right?
This two CPU design didn't. Somehow they avoided the usual vram muxes-
maybe the 2nd 6502 clocked a read out of vram, and write to the shiftreg ;)
Interestingly, the University of Waterloo Computer Systems Group designed
a little board that had a 6502 ~and a 6809, an LS123 and a switch - plugged
right into the 6502 socket on the PET 8032 mainboard to allow the proto
SuperPet (later SP9000) to run real languages on 96K of 16K Dram and
bank switched roms. Later Superpets had a Synertek 6551 uart and
copyrighted "poems" burned into obscure Harris fuse proms to prevent
easy copying of the hardware. Both CPUS were essentially tied together
(ie A0 on 6502 to A0 on the 6809 etc) - the switch held one or the other
CPU in reset under manual or reg. control. Of course some finagling was
required with ROM selects to allow each CPU to run only it's ROM.
FLEX or OS-9 would have been nice on this instead of Commodore's ugly
"DOS".
Still, with a real baby-PDP-11-ish 6809 CPU with 16 bit index registers and
good addressing modes, the CSG was able to miraculously offer APL, and
structured "Waterloo" flavours of Fortran, Basic, Cobol, Pascal and ASM
written in WSL - a C-like Waterloo Systems Language. These products
were later offered by Watcom for the Cemcorp Icon running on an early
186 based QNX system, and IBM PCs in early 80s. Watcom for quite a
while held speed records for their excellent PC/DOS C and Fortran
compilers which now updated and offered as openware- seeopenwatcom.org.
Watcom was bought by Sybase to capitalize on it's SQL-Anywhere product
and dropped other language compilers/interpreters.
H