21 Jul
2006
21 Jul
'06
10:40 p.m.
While not strictly OT, I figure if any crowd would know an answer to this,
this one would. I'm building some electronic toys that need a retro look.
I'm looking for some toggle switches (SPST) with long paddle handles, very
old school. All I can find in the regular catalogs have anemic small
paddles. Any pointers?
Ken
21 Jul
21 Jul
10:49 p.m.
On 7/22/06, Ken Seefried <ken at seefried.com> wrote:
While not strictly OT, I figure if any crowd would know an answer to this,
this one would. I'm building some electronic toys that need a retro look.
I'm looking for some toggle switches (SPST) with long paddle handles, very
old school. All I can find in the regular catalogs have anemic small
paddles. Any pointers?
Do you have any pictures of vintage equipment with the kind you are
looking for? There are C&K switches with wide paddles, but they tend
to run $4-$6 in small (under 100) quantities.
You might start looking at the C&K line. They have detailed part
numbers that specify connection type, paddle length/width, color, etc.
Without knowing what look you are going for, it's hard to recommend a
specific part of specific vendor.
-ethan
11:02 p.m.
Ken Seefried wrote:
While not strictly OT, I figure if any crowd would know an answer to
this, this one would. I'm building some electronic toys that need a
retro look. I'm looking for some toggle switches (SPST) with long
paddle handles, very old school. All I can find in the regular catalogs
have anemic small paddles. Any pointers?
How big is big? Are you looking for the extra long
*plastic* "flat" bats? Or, just the older steel bat
toggles?
NKK makes some of the latter -- bat is maybe 7/8" long
(measured from the outermost portion of the switch body).
They also make some of the flat plastic handles though
only ~3/4" long (not like the 2" long bats). Former have
#6-32 screw terminals, latter are smaller solder eyelets.
Mountain makes one with a bat about 1" long (from the mounting
surface) that's sort of flat.
Marlin had some really cool *lighted* plastic toggles with
reasonably long (flat) bats. But stock there tends to be
variable. (red, green and yellow! :> ) I've seen the same
sw's elsewhere so if interested I'll have to dig deeper.
Note that these tend to be much larger than the C&K minitoggles
so make sure you've planned extra space behind your panel, etc.
11:22 p.m.
See anything here that's right? Carling makes some very long paddle
switches:
http://www.carlingtech.com/news/pdf/pr43_MiniSWCat.pdf
11:53 p.m.
Chuck Guzis wrote:
See anything here that's right? Carling makes
some very long paddle
switches:
http://www.carlingtech.com/news/pdf/pr43_MiniSWCat.pdf
But then who sells the odd ball parts except lots of ... ???
Still if you have IMSAI 8080 you can still get replacement switches.
http://www.imsai.net/
Now I wonder who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
22 Jul
22 Jul
3:05 a.m.
On 7/21/2006 at 9:53 PM woodelf wrote:
But then who sells the odd ball parts except lots of
... ???
Still if you have IMSAI 8080 you can still get replacement switches.
http://www.imsai.net/
Now I wonder who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
Well, the surplus dealers sometimes have some unusual stuff:
http://www.electronicsurplus.com
http://www.meci.com
There's about a zillion of these places around.
Cheers,
Chuck
1:54 p.m.
Chuck Guzis wrote:
The usual surplus places I check at the moment don't have any
good switches. I plan to make a simple EEPROM programer with
a octal keypad and was looking for a cheap way of doing it.
I plan to send out a order to BG Micro for some sockets soon
for the PDP-8 replica from Spare time Gizmos they are selling
as well as a little box for it.
PLUG for Spare Time Gizmos -- only two complete kits left.
http://www.sparetimegizmos.com/
Now I wonder
who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
Well, the surplus dealers sometimes have some unusual stuff:
http://www.electronicsurplus.com
http://www.meci.com
23 Jul
23 Jul
9:52 a.m.
woodelf wrote:
Chuck Guzis wrote:
The usual surplus places I check at the moment don't have any
good switches. I plan to make a simple EEPROM programer with
a octal keypad and was looking for a cheap way of doing it.
If you put a processor in it, why not consider just using
a serial console as your "user I/O"?
I made a *touch tone* phone years ago with a *dial* on it.
(Amusing to see the reactions from people using it. :> )
I you expect to just use octal for data entry, the same
scheme could apply (imagine a classic 500/2500 series
station set with a ZIF socket up under the handset
cradle and a small LED display above the dial... :> )
Definitely a geek conversation piece!
Now I
wonder who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
Well, the surplus dealers sometimes have some unusual stuff:
http://www.electronicsurplus.com
http://www.meci.com
12:02 p.m.
If you're looking for an octal keypad, why not start with this one?
http://www.electronicsurplus.com/commerce/catalog/product.jsp?product_id=760
50&czuid=1153670221149
Sounds pretty spiffy to me--and large enough that you possibly could build
your entire project in the enclosure.
Cheers,
Chuck
On 7/23/2006 at 6:52 AM Don Y wrote:
woodelf wrote:
Chuck Guzis wrote:
The usual surplus places I check at the moment don't have any
good switches. I plan to make a simple EEPROM programer with
a octal keypad and was looking for a cheap way of doing it.
If you put a processor in it, why not consider just using
a serial console as your "user I/O"?
I made a *touch tone* phone years ago with a *dial* on it.
(Amusing to see the reactions from people using it. :> )
I you expect to just use octal for data entry, the same
scheme could apply (imagine a classic 500/2500 series
station set with a ZIF socket up under the handset
cradle and a small LED display above the dial... :> )
Definitely a geek conversation piece! Now I
wonder who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
Well, the surplus dealers sometimes have some unusual stuff:
http://www.electronicsurplus.com
http://www.meci.com
2:41 p.m.
Don Y wrote:
If you put a processor in it, why not consider just using
a serial console as your "user I/O"?
<Sharp reply>
I'm using the EEPROM for the CPU micro-code I'm building!!!
</sharp reply>
I want this to be a stand alone project- no hookups off
a PC printer port.What I really need is a PROM programer but I have
not even heard of them since EEPROM's have come out.
I made a *touch tone* phone years ago with a *dial* on
it.
(Amusing to see the reactions from people using it. :> )
I you expect to just use octal for data entry, the same
scheme could apply (imagine a classic 500/2500 series
station set with a ZIF socket up under the handset
cradle and a small LED display above the dial... :> )
Definitely a geek conversation piece!
<grumble>
Why does low cost development software - schematic layout
and PCB development and prom/eprom burners mean 1000's
of $$$.
</grumble>
3:04 p.m.
woodelf wrote:
I believe there are some "free" tools out there.
Some are "crippleware" (i.e. layout a PCB with fewer
than N pads). Some are "abandonware" (end of the
line for that product). Some are "free software"
(e.g., gEDA) but may require a bit more computer
savvy than folks may have.
I understand the complaint, though. I could buy
another car with what I've spent on software! :<
At some point, you just say, "Enough! I'll live
with this set of bugs/features..."
Don Y wrote:
Most of the DataI/O boxes will do bipolar PROMs.
You can hunt down a "19" or a "9" (?). They
*won't* go forward beyond their generation, though
(e.g., you aren't going to find them supporting
huge EPROMs, newer programmable logic, etc.)
If you put a processor in it, why not consider just using
a serial console as your "user I/O"?
<Sharp reply>
I'm using the EEPROM for the CPU micro-code I'm building!!!
</sharp reply>
I want this to be a stand alone project- no hookups off
a PC printer port.What I really need is a PROM programer but I have
not even heard of them since EEPROM's have come out. I made a
*touch tone* phone years ago with a *dial* on it.
(Amusing to see the reactions from people using it. :> )
I you expect to just use octal for data entry, the same
scheme could apply (imagine a classic 500/2500 series
station set with a ZIF socket up under the handset
cradle and a small LED display above the dial... :> )
Definitely a geek conversation piece!
<grumble>
Why does low cost development software - schematic layout
and PCB development and prom/eprom burners mean 1000's
of $$$.
</grumble>
3:06 p.m.
On 7/23/2006 at 12:41 PM woodelf wrote:
Why does low cost development software - schematic
layout
and PCB development and prom/eprom burners mean 1000's
of $$$.
Huh? I don't think I ever paid more than about $100 for any of my PROM
programmers. All were-hook-it-up-to-a-PC type and they work just fine. I
still have the first one--connected to a PC (won't work on anything much
faster, but will do 2708s).
Here's one for $79 that even programs the old 2716 high-Vpp EPROMs:
http://www.futurlec.com/EPROMProg.shtml
Of course, it's limited to 27xxx EPROMS, but there are ones in the same
price range that will do other types.
Quite frankly, the flexibility one gets from using a PC as a programming
controller really beats the standalone keypad-type.
Cheers,
Chuck
3:29 p.m.
Chuck Guzis wrote:
Quite frankly, the flexibility one gets from using a
PC as a programming
controller really beats the standalone keypad-type.
Well I do most of my hobby development work at 3am
and don't want to turn on the PC. :)
Since I plan to do 512x8 proms stand alone is the better
idea for me.
Looking at the 2816 data sheets you have listed 10^4 or 10^5
cycles -- I take that to be write cycles and any number
of read cycles is that correct the way I read it?
Cheers,
Chuck
3:43 p.m.
woodelf wrote:
Chuck Guzis wrote:
Most of the cheap 27xx style programmers won't do bipolar PROMs
(which your 512x8's undoubtedly are -- 7641 comes to mind?)
Quite frankly, the flexibility one gets from
using a PC as a programming
controller really beats the standalone keypad-type.
Well I do most of my hobby development work at 3am
and don't want to turn on the PC. :)
Since I plan to do 512x8 proms stand alone is the better
idea for me. Looking at the 2816 data sheets you have listed 10^4
or 10^5
cycles -- I take that to be write cycles and any number
of read cycles is that correct the way I read it?
Yes.
3:49 p.m.
woodelf wrote:
Well I do most of my hobby development work at 3am
and don't want to turn on the PC. :)
Since I plan to do 512x8 proms stand alone is the better
idea for me.
Looking at the 2816 data sheets you have listed 10^4 or 10^5
cycles -- I take that to be write cycles and any number
of read cycles is that correct the way I read it?
If you are expecting to turn the crank on the microcode
development more than a few times, consider designing
the interface to the microstore in such a way that you
can piggyback a small SRAM in place of the PROM. Use
latches with tristate outputs to drive the address to
the microstore (assuming you aren't pressed for speed
in the final design).
Then, when you want to twiddle the contents of the SRAM,
tristate the latch outputs and enable address buffers
on your "piggyback SRAM board" so that you can force the
address to whatever suits you. Drive the data you want
into the RAM and generate a write cycle.
I.e. make a PROM emulator.
When done "programming", disable the address buffers on
your piggyback board and reenable the tristate latches
*in* the processor.
<shrug> If you are the type that likes to do iterative
designs (instead of getting it nailed down on paper
ahead of time), this can be a real time saver. And,
bipolar PROMs get expensive if you're burning a new
set every hour...
5:02 p.m.
On 7/23/2006 at 1:29 PM woodelf wrote:
Since I plan to do 512x8 proms stand alone is the
better
idea for me.
Bipolar, as in 82s141 type? I'd forgotten that this was a "vintage" list,
sorry. I had my share of grief programming bipolars back in the 80's--lots
of "oh damn, that doesn't work, toss it in the trash and try again"
irritation.
Still, my inclination would be to use a PeeCee if I had to do the job.
Somehow, keying in and checking 1536 dgits (you did say "octal" didn't
you?) for every EPROM doesn't appeal to the lazy person in me.
Cheers,
Chuck
6:14 p.m.
Chuck Guzis wrote:
On 7/23/2006 at 1:29 PM woodelf wrote:
Bipolar, as in 82s141 type? I'd forgotten that
this was a "vintage" list,
sorry. I had my share of grief programming bipolars back in the 80's--lots
of "oh damn, that doesn't work, toss it in the trash and try again"
irritation.
I could I guess if I really need one programmed I could find
somebody on the list. Prom burning is still needed ( not food
burning like my supper here) to repair older machines.
Still, my inclination would be to use a PeeCee if I
had to do the job.
Somehow, keying in and checking 1536 dgits (you did say "octal" didn't
you?) for every EPROM doesn't appeal to the lazy person in me.
But all the low cost EPROM programmers use a PeeCee not todays
CRAP!I was looking at the low cost one and DAMIT software
will not run under the windows I have. Stupid software protection.
Having a printer port is NOT my idea of open hardware.
Cheers,
Chuck
Back to the design ... any market for a 20 bit cpu. :)
Buy 3 or more and I'll throw in a few extra bits too.
6:26 p.m.
Still, my
inclination would be to use a PeeCee if I had to do the job.
Somehow, keying in and checking 1536 dgits (you did say "octal" didn't
you?) for every EPROM doesn't appeal to the lazy person in me.
But all the low
cost EPROM programmers use a PeeCee not todays
CRAP!I was looking at the low cost one and DAMIT software
will not run under the windows I have. Stupid software protection.
Having a printer port is NOT my idea of open hardware.
6:43 p.m.
Alexandre Souza wrote:
This is not crap, this is not stupid, this is not open hardware. This is
a low cost eprom programmer. A QUICK LOOK --- eprom programers for DOS and ISA
card less that $200
Printer port $300. New for windows/TODAY $500-$100
Still, my inclination would be to use a PeeCee if I had
to do the job.
Somehow, keying in and checking 1536 dgits (you did say "octal" didn't
you?) for every EPROM doesn't appeal to the lazy person in me.
But all the low cost EPROM programmers use a PeeCee not todays
CRAP!I was looking at the low cost one and DAMIT software
will not run under the windows I have. Stupid software protection.
Having a printer port is NOT my idea of open hardware. You may think whatever you want. But there is no
such thing as a free
lunch. If you want a wildely-compatible programmer, it HAS to have some kind
of intelligence. If you want a CHEAP program, the intelligence needs to be
elsewhere. Of course, better options are always welcomed, Willem did his
part, why don't you do yours? His programmer is "open hardware", only the
software isn't updated anymore, but still useable for many many many parts.
The good thing on computers is that every unhappy person is free to
create a better device to make people and himself happy.
See ya
Alexandre
.
6:47 p.m.
A QUICK LOOK --- eprom programers for DOS and ISA
card less that $200
Printer port $300. New for windows/TODAY $500-$100
Willem - $50 (built). Eprommer (www.ustr.net) - $90 (built). Willem can
be used with DOS, Eprommer too, if I'm not mistaken. And what is the problem
of having an old PC just for EPROM programming?
Have you ever seen www.willem.org???
6:49 p.m.
woodelf wrote:
Chuck Guzis wrote:
Trick is to burn only those parts of the microstore pertinent to
the opcode you are debugging. You can then learn from those
mistakes and "do it right" for the next opcode in a virgin portion
of the PROM. Cuts down on the number of wasted parts.
On 7/23/2006 at 1:29 PM woodelf wrote:
> Bipolar, as in 82s141 type? I'd forgotten that this was a "vintage"
> list,
> sorry. I had my share of grief programming bipolars back in the
> 80's--lots
> of "oh damn, that doesn't work, toss it in the trash and try again"
> irritation. I could I guess if I really need one programmed I
could find
somebody on the list. Prom burning is still needed ( not food
burning like my supper here) to repair older machines.
Boot to DOS. Or, keep an old machine handy for just this
reason (I keep my Compaq Portable 3 for just such things)
Still, my inclination would be to use a PeeCee if
I had to do the job.
Somehow, keying in and checking 1536 dgits (you did say "octal" didn't
you?) for every EPROM doesn't appeal to the lazy person in me.
But all the low cost EPROM programmers use a PeeCee not todays
CRAP!I was looking at the low cost one and DAMIT software
will not run under the windows I have. Stupid software protection.
Having a printer port is NOT my idea of open hardware. Back to the design ... any market for a 20 bit cpu.
:)
Buy 3 or more and I'll throw in a few extra bits too.
Aim for 36 bits (?) and run MULTICS! (or have I
misremembered that...?)
7:03 p.m.
On 7/23/2006 at 3:49 PM Don Y wrote:
Boot to DOS. Or, keep an old machine handy for just
this
reason (I keep my Compaq Portable 3 for just such things)
I just came away from using a PC XT (clone) that has a SysGen
omnibridge+Microsolutions Matchpoint+Central Point Deluxe Option board.
Has a 30MB CMI hard disk run on an OMTI RLL controller for a roomy 45 MB.
Actually, it's pretty amazing how an 8088 at 8MHz is entirely usable under
DOS (or CP/M-86).
Does anyone know if MS Network can be run on an XT or does it have
286-specific code? I can't recall.
Cheers,
Chuck
6:55 p.m.
On 7/23/2006 at 4:14 PM woodelf wrote:
Back to the design ... any market for a 20 bit cpu. :)
Buy 3 or more and I'll throw in a few extra bits too.
Maybe 24 (nice multiple of 2, 3, 4, 6, 8 and 12 bit words); leads to a
usably large address space and roomy enough for a 3 address architecture.
20 feels a bit tight to me. Also allows one to use off-the-shelf 8 bit
parts.
Cheers,
Chuck
24 Jul
24 Jul
1:52 a.m.
Chuck Guzis wrote:
Maybe 24 (nice multiple of 2, 3, 4, 6, 8 and 12 bit
words); leads to a
usably large address space and roomy enough for a 3 address architecture.
20 feels a bit tight to me. Also allows one to use off-the-shelf 8 bit
parts.
What off the shelf parts??? Buffers and a few FF's or latches are 8 bit wide.
Cheers,
Chuck
I am trying to keep this design to 1 ALU PCB and 1 Control PCB
and 24 bits is just a little too large I think to fit because
glue logic can add up. Also with 24 bits I would like 12 bit
bytes and that is even more complexity.
2:34 a.m.
On 7/23/2006 at 11:52 PM woodelf wrote:
What off the shelf parts??? Buffers and a few FF's
or latches are 8 bit
wide.
...and a lot of SRAM and EPROM is also.
You didn't mention how "vintage" your ALU is going to be (I'm assuming
this
is a vintage design and not an FPGA). Are you using something like the
74S481 4-bit slice or are you going for something from the 29xx family?
Cheers,
Chuck
5:02 a.m.
Chuck Guzis wrote:
On 7/23/2006 at 11:52 PM woodelf wrote:
You didn't mention how "vintage" your
ALU is going to be (I'm assuming this
is a vintage design and not an FPGA). Are you using something like the
74S481 4-bit slice or are you going for something from the 29xx family?
I was planning to use generic 2901's and LS parts other than PROM.
Ripple carry for the ALU is used since this is only 5 slices. Since I am
using a 6809 style clock I hope to get about 1.5 Mhz with the system
with a non-pipelined ROM.
Cheers,
Chuck
5:20 a.m.
woodelf schrieb:
Chuck Guzis wrote:
The cost
for two additional 74s182 (usable instead of am2902) is not
very high,
but there is a price in glitches and reduced clock frequency involved
when propagating
a ripple carry over 5 units (=20bit). You won't find the ripple carry
effect unless you
feed the ALU with certain pathological patterns.
You might also find the Mick&Brick book worthwhile for your enterprise.
On 7/23/2006 at 11:52 PM woodelf wrote:
You didn't mention how "vintage"
your ALU is going to be (I'm
assuming this
is a vintage design and not an FPGA). Are you using something like the
74S481 4-bit slice or are you going for something from the 29xx family?
I was planning to use generic 2901's and LS parts other than PROM.
Ripple carry for the ALU is used since this is only 5 slices. Since I am using a 6809 style clock I hope to get about 1.5 Mhz
with the system
with a non-pipelined ROM.
Holger
7:31 p.m.
Holger Veit wrote:
woodelf schrieb:
The cost for two additional 74s182 (usable instead of
am2902) is not
very high,
but there is a price in glitches and reduced clock frequency involved
when propagating
a ripple carry over 5 units (=20bit). You won't find the ripple carry
effect unless you
feed the ALU with certain pathological patterns.
I am guessing 25 ns for ripple
carry per slice ... 125 ns delay.
24 bits would be a different story.
You might also find the Mick&Brick book worthwhile
for your enterprise.
I am guessing would be if I was using 'typical' computer archtecture that
favors register to register operations.
> using a 6809 style clock I hope to get about 1.5
Mhz with the system
> with a non-pipelined ROM.
This is using early 70's chips so a 1 MHZ memory
cycle is still fast.
Holger
11:15 a.m.
On 7/24/2006 at 3:02 AM woodelf wrote:
I was planning to use generic 2901's and LS parts
other than PROM.
Ripple carry for the ALU is used since this is only 5 slices. Since I am
using a 6809 style clock I hope to get about 1.5 Mhz with the system
with a non-pipelined ROM.
Are you doing this to explore old 4-bit bipolar slice logic or to design
your own instruction set? The later 29Cxxx logic seems to be easier to
work with--e.g. the 29C116 is a 16-bit slice that in addition to being
CMOS, reduces the package count considerably.
Just curious. :)
Cheers,
Chuck
12:53 p.m.
Chuck Guzis wrote:
On 7/24/2006 at 3:02 AM woodelf wrote:
Pick up a copy of Mick & Brick, regardless. A good read for
this type of undertaking!
I was planning to use generic 2901's and LS
parts other than PROM.
Ripple carry for the ALU is used since this is only 5 slices. Since I am
using a 6809 style clock I hope to get about 1.5 Mhz with the system
with a non-pipelined ROM.
Are you doing this to explore old 4-bit bipolar slice logic or to design
your own instruction set? The later 29Cxxx logic seems to be easier to
work with--e.g. the 29C116 is a 16-bit slice that in addition to being
CMOS, reduces the package count considerably. 
5:29 p.m.
Pick up a copy of Mick & Brick, regardless. A
good read for
this type of undertaking!
Seconded ! Anyone who is designing a microcoded processor, particularly
one using the AMD2900 seires chips, and who hasn't read Mick and Brick is
making life difficult for themselves :-)
I am told paper copies are not easy to find, although isn't it on the web
somewhere now? I don't know, I have a paper version...
-tony
6 p.m.
Tony Duell wrote:
Dunno. Mine is a hardbound book (yellow/orange dust jacket
makes it real easy to find! :> )
AMD also had innumerable small databook-lets (lettes?)
covering various aspects of bitslice design and bitslice
components. Mine are so old the *covers* are faded :-(
Pick up a copy
of Mick & Brick, regardless. A good read for
this type of undertaking!
Seconded ! Anyone who is designing a microcoded processor, particularly
one using the AMD2900 seires chips, and who hasn't read Mick and Brick is
making life difficult for themselves :-)
I am told paper copies are not easy to find, although isn't it on the web
somewhere now? I don't know, I have a paper version...
25 Jul
25 Jul
3:20 a.m.
Tony Duell schrieb:
Numerous used copies exist for cheap at Amazon. Therefore, it is
probably not worth photocopying it,
particularly because it contains many small two-sheet schematics. I
haven't seen it online yet.
--
Dr.-Ing. Holger Veit
Head IT-Management
Fraunhofer IAIS
Institute for Intelligent Analysis & Information Systems
Schloss Birlinghoven
D-53757 Sankt Augustin
e-mail: holger.veit at iais.fraunhofer.de
Tel. +49 2241 14 2448
Fax. +49 2241 14 2342
Pick up a copy
of Mick & Brick, regardless. A good read for
this type of undertaking!
Seconded ! Anyone who is designing a microcoded processor, particularly
one using the AMD2900 seires chips, and who hasn't read Mick and Brick is
making life difficult for themselves :-)
I am told paper copies are not easy to find, although isn't it on the web
somewhere now? I don't know, I have a paper version...
-tony
24 Jul
24 Jul
5:26 p.m.
Are you doing this to explore old 4-bit bipolar slice
logic or to design
your own instruction set? The later 29Cxxx logic seems to be easier to
work with--e.g. the 29C116 is a 16-bit slice that in addition to being
CMOS, reduces the package count considerably.
Now, I've not really used the 29116, I've only come across it in the PERQ
AGW3300 graphics processor, but it strikes me from the limited data that
I have on it, that if you use this chip you're limited to a 16 bit data
path. It's not really a bitslice chip in that you can't stick <n> of them
togehter (like you can with the 2901).
There were, IIRC, CMOS versions of the 2901 which would cut the power
consumption, although not the package count.
-tony
5:55 p.m.
On 7/24/2006 at 10:26 PM ard at p850ug1.demon.co.uk wrote:
Now, I've not really used the 29116, I've only
come across it in the PERQ
AGW3300 graphics processor, but it strikes me from the limited data that
I have on it, that if you use this chip you're limited to a 16 bit data
path. It's not really a bitslice chip in that you can't stick <n> of them
togehter (like you can with the 2901).
True, but there was IIRC, a 32-bit version also. AMD calls 'em
microprocessors, but without so much as a P-counter, it's hard to justify
the name. Still, I wonder if a 5-package 2901 version without the carry
lookahead would be any faster than 2 trips (microprogrammed) through the
16-bit CMOS unit.
Cheers,
Chuck
7:06 p.m.
On 7/24/2006 at 10:26 PM ard at p850ug1.demon.co.uk wrote:
YEs, but the OP wants to make a 20 bit processor...
Now, I've not really used the 29116, I've
only come across it in the PERQ
AGW3300 graphics processor, but it strikes me from the limited data that
I have on it, that if you use this chip you're limited to a 16 bit data
path. It's not really a bitslice chip in that you can't stick <n> of them
togehter (like you can with the 2901).
True, but there was IIRC, a 32-bit version also. AMD calls 'em microprocessors, but without so much as a P-counter,
it's hard to justify
the name. Still, I wonder if a 5-package 2901 version without the carry
lookahead would be any faster than 2 trips (microprogrammed) through the
16-bit CMOS unit.
Yeah, and it would probably be even faster to simulate the thing on a
Pentium<n> for suitable values of <n>. But that's not the point, surely.
-tony
5:58 p.m.
Tony Duell wrote:
Exactly. It's a one-trick pony.
Are you doing
this to explore old 4-bit bipolar slice logic or to design
your own instruction set? The later 29Cxxx logic seems to be easier to
work with--e.g. the 29C116 is a 16-bit slice that in addition to being
CMOS, reduces the package count considerably.
Now, I've not really used the 29116, I've only come across it in the PERQ
AGW3300 graphics processor, but it strikes me from the limited data that
I have on it, that if you use this chip you're limited to a 16 bit data
path. It's not really a bitslice chip in that you can't stick <n> of them
togehter (like you can with the 2901). There were, IIRC, CMOS versions of the 2901 which
would cut the power
consumption, although not the package count.
5:23 p.m.
Chuck Guzis wrote:
Hmmm... I always felt the 2901 constrained you rather too much, and that
some useful signals were not brought out. (OK, not as bad as the 2910
sequencer, which rapidly becomes a pain if you need more than 12 bits of
microcode address [1]).
[1] OK, I am biased. The PERQ 1 CPU used a 2910 and 4K of control store.
All later PERQs had 216K of control store, but kept the 2910 for
compatability. There was a circuit, commonly knwon as the '2 bit kludge'
(pun totally intentional) to extend the address from 12 bits to 14 bits.
And a kludge it is...
On 7/23/2006 at 11:52 PM woodelf wrote:
You didn't mention how "vintage"
your ALU is going to be (I'm assuming this
is a vintage design and not an FPGA). Are you using something like the
74S481 4-bit slice or are you going for something from the 29xx family?
I was planning to use generic 2901's and LS parts other than PROM. Ripple carry for the ALU is used since this is only 5
slices. Since I am
A 74S182 is only a 16 pin DIL package, and would give you lookahead
carry. Take a look at the PERQ 2 schematics on bitsavers -- the 16K CPU
board diagrams are there. And look at the PERQ CPU Techref also on
bitsavers. The PERQ is a 20 bit machine using 74S181 ALUs, and the carry
circuit should be applicable to you.
-tony
9:09 p.m.
Tony Duell wrote:
A 74S182 is only a 16 pin DIL package, and would give you lookahead
carry. Take a look at the PERQ 2 schematics on bitsavers -- the 16K CPU
board diagrams are there. And look at the PERQ CPU Techref also on
bitsavers. The PERQ is a 20 bit machine using 74S181 ALUs, and the carry
circuit should be applicable to you.
If it's interesting, the MIT CADR has a 32 bit alu using the 74181 and
uses the '182 for carry. The schematics are also on bitsavers. It's a
reasonably general microcoded 32 bit cpu. Not bad for it's day.
Also, I have verilog for the 181 and 182 if you go that way :-) I like
to simulate... (I have verilog for the whole cadr to be honest, but that
seems off topic)
-brad
8:41 p.m.
Also, I have verilog for the 181 and 182 if you go
that way :-) I like
to simulate... (I have verilog for the whole cadr to be honest, but that
FWIW, the Texas Instruments TTL databook has gate level schematics for
the '181 and '182. and gives the logic equations for the '182 (which are
pretty obvious when you see them :-)). If you can't find a '182, it
wouldn't be hard to program some kind of PLD or PROM to do the same function.
-tony
10:20 p.m.
On 7/24/2006 at 9:09 PM Brad Parker wrote:
If it's interesting, the MIT CADR has a 32 bit alu
using the 74181 and
uses the '182 for carry. The schematics are also on bitsavers. It's a
reasonably general microcoded 32 bit cpu. Not bad for it's day.
If it's a 20 bit CPU that the OP wants, perhaps someone has a set of prints
for the PERQ.
I've always been fascinated with the TI 74S481--a 4 bit slice in a 40 pin
package, but with a lot of functionality inside. I don't know if they're
unobtainium nowadays. The 1978 TI patent on it is here:
http://www.freepatentsonline.com/4224676.html. One curious aspect is the
"position" input pin: if the slice is the most significant, you put a
voltage higher than 3.6v on it; if the least significant, less than 0.8
volts; if in the middle, between 1.8 and 3 volts.
Cheers,
Chuck
9:05 p.m.
On 7/24/2006 at 9:09 PM Brad Parker wrote:
I do, and I am sure they're on Bitsavers...
Be warned that the PERQ is not a simple CPU (the 16K control store
version, which uses PALs for some of the logic, and which manages to
implement more microinstructions as a result, contains over 260 ICs). And
it's somewhat tied to the PERQ memory circuitry too. And it's got a
graphics accelerator that you probably won't want.
Even so, look at the diagrams (and the tech manual). There may well be
some stuff there you can use..
If it's interesting, the MIT CADR has a 32 bit
alu using the 74181 and
uses the '182 for carry. The schematics are also on bitsavers. It's a
reasonably general microcoded 32 bit cpu. Not bad for it's day.
If it's a 20 bit CPU that the OP wants, perhaps someone has a set of prints
for the PERQ.
I've always been fascinated with the TI 74S481--a 4 bit slice in a 40 pin
I whould have a data sheet on that.
package, but with a lot of functionality inside. I
don't know if they're
unobtainium nowadays. The 1978 TI patent on it is here:
http://www.freepatentsonline.com/4224676.html. One curious aspect is the
"position" input pin: if the slice is the most significant, you put a
voltage higher than 3.6v on it; if the least significant, less than 0.8
volts; if in the middle, between 1.8 and 3 volts.
Odd!
As an aside, the Philips P851 uses a custom chip that's a 4-bit slice.
It's called SPALU (Scratch Pad (registers) and Arithmetic Logic Unit) in
the service manual. There's probably enough info in said manual to
recreate the chip if desparate (which I will be if I don't have 4 working
ones out of the 8 I have (4 in the P851, 4 on a spare board).
-tony
10:43 p.m.
Chuck Guzis wrote:
http://www.freepatentsonline.com/4224676.html. One
curious aspect is the
"position" input pin: if the slice is the most significant, you put a
voltage higher than 3.6v on it; if the least significant, less than 0.8
volts; if in the middle, between 1.8 and 3 volts.
I recall some QBus(?) interface chips with similar strapping
options
9:44 a.m.
woodelf wrote:
Chuck Guzis wrote:
Grayhill make$ $ome really nice pu$hbutton matrixe$.
But, they are $tiff like tho$e on a WE touchtone telephone.
Oh, and did I mention they are expen$ive? :>
See anything here that's right? Carling
makes some very long paddle
switches:
http://www.carlingtech.com/news/pdf/pr43_MiniSWCat.pdf
But then who sells the odd ball parts except lots of ... ???
Still if you have IMSAI 8080 you can still get replacement switches.
http://www.imsai.net/
Now I wonder who makes a nice key pad switches? I want real
stuff keys not some 5 cent plastic switch matrix.
6577
days inactive
6580
days old
44 comments
9 participants
participants (9)
-
alexandre-listas＠e-secure.com.br -
ard＠p850ug1.demon.co.uk -
bfranchuk＠jetnet.ab.ca -
brad＠heeltoe.com -
cclist＠sydex.com -
dgy＠DakotaCom.Net -
ethan.dicks＠gmail.com -
holger.veit＠iais.fraunhofer.de -
ken＠seefried.com