Signetics 2650 Info
Edwin Parsons
on early Signetics 2650 development kits.
I have come across a very small board which
is marked as a SIGNETICS 2650 PC3000 ITC.
It contains a 2650 microprocessor, 82S115 ROM
and a couple of 2606 (RAM? chips) plus a
handfull of support components.
If anyone could shed any light on what this
is, or could provide any interesting information
on the 2650 series of processors it would be
appreciated.
regards
Falstaff
If I remember correctly, the 2650 has only 15 address bits.
Frank (Well, I thought it was interesting...)
--
"Mutual respect, even if we disagree."
------------------------------------------------------------------------
Frank A. Vorstenbosch +31-(70)-355 5241 fals...@xs4all.nl
Bruce Simpson
>ed...@peninsula.apana.org.au (Edwin Parsons) writes:
>>I am wondering if anyone has any information
>>on early Signetics 2650 development kits.
>>If anyone could shed any light on what this
>>is, or could provide any interesting information
>>on the 2650 series of processors it would be
>>appreciated.
>If I remember correctly, the 2650 has only 15 address bits.
Correct. If my "aging" memory can be relied on, it was one of the very
first microprocessors. I got my own hands dirty in about 1977 using a 2650
along with the PIPBUG ROM and 2Kbytes of static RAM. I think that the 16th
bit of the address line was stolen for some other purpose (can't remember
what).
For its day, the 2650 had a *very* advanced instruction-set, more akin to
that of a minicomputer of they day than a microprocessor.
It had:
- a fully internal stack (no push/pop instructions or external RAM used)
- heaps of addressing modes (immediate, absolute, relative, indirect
(most of which had the option of pre/post increment/decrement)
- two banks of general purpose registers
I guess everyone looks back at the past through rose-tinted glasses but it
seems I sure got a lot more of a kick out of hand-coding 2650 code and
entering it through a hex-keypad then than I do now using C++ on a 486
processor and high-res graphics display :-(
*==[FREE-FAX GATEWAY INTO THE AUCKLAND NZ TOLL-FREE AREA]==*
* FREE listings in NZ's largest WWW Business Directory *
* http://www.actrix.gen.nz/biz/faxmail/faxmail.htm *
*==========================================================*
R & D Googe
>I am wondering if anyone has any information
>on early Signetics 2650 development kits.
>I have come across a very small board which
>is marked as a SIGNETICS 2650 PC3000 ITC.
>It contains a 2650 microprocessor, 82S115 ROM
>and a couple of 2606 (RAM? chips) plus a
>handfull of support components.
>If anyone could shed any light on what this
>is, or could provide any interesting information
>on the 2650 series of processors it would be
>appreciated.
>regards
You could try looking into Electronics Australia back issues (way
back!) around 1976-77 when EA had a project on the 2650!!
- Rob
S 27 36'58
E 152 45'29
Russell Lang
>fals...@xs4all.nl (Falstaff) wrote:
>>ed...@peninsula.apana.org.au (Edwin Parsons) writes:
>>>I am wondering if anyone has any information
>>>on early Signetics 2650 development kits.
>>If I remember correctly, the 2650 has only 15 address bits.
>Correct. If my "aging" memory can be relied on, it was one of the very
>first microprocessors. I got my own hands dirty in about 1977 using a 2650
>along with the PIPBUG ROM and 2Kbytes of static RAM. I think that the 16th
>bit of the address line was stolen for some other purpose (can't remember
>what).
The top address bit was for indirect addressing.
That was a lot of ram. I thought the early kits had 1kbytes of ROM
(PIPBUG) and 256bytes of static RAM.
>For its day, the 2650 had a *very* advanced instruction-set, more akin to
>that of a minicomputer of they day than a microprocessor.
But just try writing a cross-compiler for it. With only 8-bit
registers, writing address indexing code was awkward. I think
it would have been possible to do it with indirect addressing.
It was best programmed in assembly.
--
Russell Lang Email: r...@eng.monash.edu.au Phone: +61 3 9905 5708
Department of Electrical and Computer Systems Engineering
Monash University, Australia
Russell Lang
>I am wondering if anyone has any information
>on early Signetics 2650 development kits.
>I have come across a very small board which
>is marked as a SIGNETICS 2650 PC3000 ITC.
>It contains a 2650 microprocessor, 82S115 ROM
>and a couple of 2606 (RAM? chips) plus a
>handfull of support components.
I wrote my first computer program on the 2650. It didn't run
because I couldn't find the unconditional branch instruction.
Turns out the instruction I needed was BCTA/UN - Branch on
Condition True Absolute/UNconditional. The instruction set
is quite regular and easy to understand once you have worked
with it a bit. Features I remember are:
4 registers.
4 branch conditions. Neg/Zero/Pos/Unconditional or </=/>/unconditional.
Absolute or Relative Branches.
Indirect memory addressing (used instead of 16-bit index registers).
The ROM is likely to contain the PIPBUG monitor in 1k bytes.
Don't expect online help :-)
Try hooking up a 110bit/s 20ma current loop serial interface
(think TeleType ASR-33). With the help of a diode or two you
can probably drive it with RS-232 if you know some electronics.
There was a kit design in Electronics Australia, about 1977.
Look for the 2650 baby-micro. I think this was a cheaper version
of the commercial development kit.
Bruce Simpson
>br...@faxmail.co.nz (Bruce Simpson) writes:
>>Correct. If my "aging" memory can be relied on, it was one of the very
>>first microprocessors. I got my own hands dirty in about 1977 using a 2650
>>along with the PIPBUG ROM and 2Kbytes of static RAM. I think that the 16th
>>bit of the address line was stolen for some other purpose (can't remember
>>what).
>The top address bit was for indirect addressing.
>That was a lot of ram. I thought the early kits had 1kbytes of ROM
>(PIPBUG) and 256bytes of static RAM.
I was using some of the very first 2114 static RAM chips (1024x8) or
something. It was a tremendous advance over the 2102's that I'd used
earlier :-)
>>For its day, the 2650 had a *very* advanced instruction-set, more akin to
>>that of a minicomputer of they day than a microprocessor.
>But just try writing a cross-compiler for it. With only 8-bit
>registers, writing address indexing code was awkward. I think
>it would have been possible to do it with indirect addressing.
>It was best programmed in assembly.
There was some smart software for the 2560 around Australasia in the late
'70s. Electronics australia published a line-assembler that required only
1Kbyte or RAM or so and there was a BASIC interpreter that ran in just
2Kbytes or so.
Actually writing this sort of thing was real easy (to start with) until you
hit the +-64byte range limit of auto-inc/dec indexed relative addressing
and ran out of stack space :-(
Bruce Simpson
>Features I remember are:
> 4 registers.
Weren't there more than that? Wasn't there a common register (R0, the
accumulator) and two sets of general-purpose registers (R1-R4) that could
be selected alternately (but not simultaneously). Mind you, I could be
getting confused here.
> 4 branch conditions. Neg/Zero/Pos/Unconditional or </=/>/unconditional.
> Absolute or Relative Branches.
Yeah, after learning to cut uP code on the 2650 I was really disappointed
with the lack of orthagonality and addressing modes in the 8080 :-(
> Indirect memory addressing (used instead of 16-bit index registers).
But 1 bit was stolen so the range wasn't +-32Kbytes.
>The ROM is likely to contain the PIPBUG monitor in 1k bytes.
>Don't expect online help :-)
>Try hooking up a 110bit/s 20ma current loop serial interface
>(think TeleType ASR-33). With the help of a diode or two you
>can probably drive it with RS-232 if you know some electronics.
Yep, I used an old glass-TTY at 110bps with mine (gee, now I think that
28.8Kbps is slow :-)
Mark Zenier
: br...@faxmail.co.nz (Bruce Simpson) writes:
: >that of a minicomputer of they day than a microprocessor.
: But just try writing a cross-compiler for it. With only 8-bit
: registers, writing address indexing code was awkward. I think
: it would have been possible to do it with indirect addressing.
: It was best programmed in assembly.
You're all leaving out the worst part. In most of the absolute
addressing instructions, the address field was only 13 bits, (only the
branch had 15 bits) so any system bigger than 8k bytes ended up using a
horrible kludge memory banked scheme. The page was the two high order
bits of instruction counter, and registers were only 8 bits long and
couldn't act as data pointers. If you wanted access data address space
from more than one page, you either had to map the data ram in every
page of the memory map, or set up indirect thunks for all your data in
each page of prom.
I never programmed it but after I wrote a cross assembler for it,
and having to deal with the strange quirks (like what's a valid
address for Zero Branch Relative instruction) it didn't look so
neat.
Mark Zenier mze...@eskimo.com mze...@netcom.com