[PIC] Retro computing with PIC processors (COSMAC Elf without 1802)

[trossin]

Back in 1978 I built my first computer from a magazine article in Popular
Electronics. The computer was called the Cosmac Elf and was based on RCA's
CDP1802 microprocessor. There is a users group where the old timers and
some new kids are still active.

http://groups.yahoo.com/group/cosmacelf

Recently, I wanted to build another Elf so that I could keep my existing
machine in it's original shape but thought about using a PIC instead of the
antique processor. The goal was to create a bus accurate full speed version
of the 1802. Also, RCA created the CDP1861 video display processor that
teamed up with the DMA features of the 1802 to produce low resolution video.
I wanted to emulate this chip as well with another PIC.

The results can be found here:

http://www.tedrossin.atbhost.net/Electronics/RCA/RCA.html#ElfClone

It seems that other antique processors could be emulated (6502, 6800, Z80
...) as well as creating new "processors" using the same techique.
Emulating the processors allows new instructions to be added as well as
adding debug features that would not be possible with the original
processors.





--
View this message in context: http://old.nabble.com/Retro-computing-with-PIC-processors-%28COSMAC-Elf-without-1802%29-tp28610048p28610048.html
Sent from the PIC - [PIC] mailing list archive at Nabble.com.

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

--
You received this message because you are subscribed to the Google Groups "piclist_archive" group.
To post to this group, send email to piclist...@googlegroups.com.
To unsubscribe from this group, send email to piclist_archi...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/piclist_archive?hl=en.

[Michael Watterson]

trossin wrote:
> Back in 1978 I built my first computer from a magazine article in Popular
> Electronics. The computer was called the Cosmac Elf and was based on RCA's
> CDP1802 microprocessor. There is a users group where the old timers and
> some new kids are still active.
>
> http://groups.yahoo.com/group/cosmacelf
>
> Recently, I wanted to build another Elf so that I could keep my existing
> machine in it's original shape but thought about using a PIC instead of the
> antique processor. The goal was to create a bus accurate full speed version
> of the 1802. Also, RCA created the CDP1861 video display processor that
> teamed up with the DMA features of the 1802 to produce low resolution video.
> I wanted to emulate this chip as well with another PIC.
>
> The results can be found here:
>
> http://www.tedrossin.atbhost.net/Electronics/RCA/RCA.html#ElfClone
>

I have a CDP1802. It controls spread spectrum secure SSB Military Radio.
1980s vintage designed by Racal. If someone wants some of these boards
it could maybe be arranged.

An 8bit PIC I think would be a poorer choice than an FPGA or ARM. Maybe
a 32F (MIPS) PIC if you feel you have to support microchip.

> It seems that other antique processors could be emulated (6502, 6800, Z80
> ...) as well as creating new "processors" using the same techique.
> Emulating the processors allows new instructions to be added as well as
> adding debug features that would not be possible with the original
> processors.
>
>

Yes. Though you can still buy Z80, 6502, 6800 and chips with them as
higher speed cores. Many of these can be downloaded as VHDL or Verilog
for FPGA.

Why would you do this on other than an FPGA or emulate on a PC if the
goal is to design an improved cpu?

[jim]

All,

I had the same idea a few years back, and I too wrote code to emulate an
1802. I was looking just now for it, but can't find it.
It was probably on the hard drive that crashed a few months or so ago, and
couldn't be recovered.

But my reasoning in response to the question...

An 8bit PIC I think would be a poorer choice than an FPGA or ARM. Maybe
a 32F (MIPS) PIC if you feel you have to support microchip.

Is...

You're probably correct in thinking it is a poorer choice than an FPGA or
ARM. But in my case, I emulated the
the 1802 just because I wanted to see if I could do it. As I found out, I
could, and it worked quite well.


And, my response to the question....

Why would you do this on other than an FPGA or emulate on a PC if the
goal is to design an improved cpu?

Is....

Because not everybody that works with, or dabbles with PIC's has the
knowledge, expertise, experience, hardware, software, etc. to work with
FPGA's.
A good number do, but I'd be willing to bet that there are more PIC
experimenters than there are PIC and FPGA experimenters.

Most FPGA's, (Not all), are significantly more expensive than PIC's. And if
you just want to see if something can be done, a PIC is less expensive than
going the FPGA route. In addition, learning VHDL and/or VERILOG is a steep
learning curve compared to PIC asm or hll's. So that would be another
reason.

Anyway, that's my take on the subject.

Regards,

Jim

[trossin]

Right on about the FPGAs. They cost a great deal of money and since they
are not available in "proven through hole technology" making a one off
prototype is even more money.

The goal of the project was to keep the cost way down so other people could
reproduce it and to keep the bus interface the same so that others could get
the feel of retro computing. The PICs I used were $6 and I was able to
plug them into a solderless breadboard. The cheapest FPGA solution I've
seen is well over $100 unless you are willing to create a custom PC board
which again is another $75 but you will need a PC board anyway to implement
the other logic.

View this message in context: http://old.nabble.com/Retro-computing-with-PIC-processors-%28COSMAC-Elf-without-1802%29-tp28610048p28612258.html

Sent from the PIC - [PIC] mailing list archive at Nabble.com.

[trossin]

I forgot to mention that an 8-bit PIC was fast enough to get the "processor"
to run at full speed and respond to interrupts and DMA requests just like
the real 1802. In order to get the video to work, the interrupt and DMA
response times have to match the the real 1802. This was the most timing
critical part of the project. Using the PIC's interrupt mechanism was not
the solution. The DMA in, DMA out and interrupt pins are sampled with
polling before instruction decode.

The 16F series was too slow but the 18F parts did the job (little more clock
rate and more powerful instructions).

View this message in context: http://old.nabble.com/Retro-computing-with-PIC-processors-%28COSMAC-Elf-without-1802%29-tp28610048p28612340.html

[Michael Watterson]

trossin wrote:
> Right on about the FPGAs. They cost a great deal of money and since they
> are not available in "proven through hole technology" making a one off
> prototype is even more money.
>
> The goal of the project was to keep the cost way down so other people could
> reproduce it and to keep the bus interface the same so that others could get
> the feel of retro computing. The PICs I used were $6 and I was able to
> plug them into a solderless breadboard. The cheapest FPGA solution I've
> seen is well over $100 unless you are willing to create a custom PC board
> which again is another $75 but you will need a PC board anyway to implement
> the other logic.
>
>
>

I think Spartan 3E is about $15? Not sure quantity.
If you buy 10 PCBS, a double sided PCB 100mm x 100mm may be under $15

You can buy mini-PCB with FPGA pre-mounted.

A BGA rework station is a once off cost.

You can get PCB adaptor and mount TQFP up to 60 pins using a 2mm tip
soldering iron.

I'd agree it's a fun achievement to use a 18F for this. Given negligible
price difference I'd go for 18F.

VHDL and the Xilinx tools are probably easier than PIC assembler though
:-) But it's not Programming as you know it, Jim. :-D

But 8bit PIC is retro computing. It's a 1977 design. I remember by late
80s you could get 20MHz Z80 as "core" in an ASIC. Hitachi did a Z80
with MMU for 512k and some I/O too?

x86 and ARM are practically Retro now :)

[jim]

FYI,

I did my version with a 16F877 part. I did over clock it somewhat (to about
27Mhz or so IIRC), because it was for my enjoyment only.
If these are to be reproduced, I think your choice to go with a different
PIC is the right way to go.

Also, I didn't try to emulate the video chip as that wasn't what I was
interested in doing. I just wanted to be able to run some of
my simple 1802 programs on a PIC that emulated the 1802, and do it full
speed. At least the speed of the ELF. BTW, I have an ELF II
from Netronics. It still works just fine. I remember when I first got it,
there was a program to display the USS Enterprise on a TV.
My wife and kids thought it was the greatest thing a space ship on a TV from
a circuit I had built.

Regards,

Jim

[YES NOPE9]

> trossin wrote:
>>
> I think Spartan 3E is about $15? Not sure quantity.

> If you buy 10 PCBS, a double sided PCB 100mm x 100mm may be under $15
>
> You can buy mini-PCB with FPGA pre-mounted.
>
> A BGA rework station is a once off cost.
>
> You can get PCB adaptor and mount TQFP up to 60 pins using a 2mm tip
> soldering iron.
>

> VHDL and the Xilinx tools are probably easier than PIC assembler
> though
> :-) But it's not Programming as you know it, Jim. :-D
>

[Mike Harrison]

On Wed, 19 May 2010 14:49:09 -0600, you wrote:

>> trossin wrote:
>>>
>> I think Spartan 3E is about $15? Not sure quantity.
>
>> If you buy 10 PCBS, a double sided PCB 100mm x 100mm may be under $15
>>
>> You can buy mini-PCB with FPGA pre-mounted.
>>
>> A BGA rework station is a once off cost.
>>
>> You can get PCB adaptor and mount TQFP up to 60 pins using a 2mm tip
>> soldering iron.
>>
>> VHDL and the Xilinx tools are probably easier than PIC assembler
>> though
>> :-) But it's not Programming as you know it, Jim. :-D
>>

enterpoint and Trenz do a number of FPGA modules in DIL format
http://www.enterpoint.co.uk/
http://www.trenz-electronic.de/products/fpga-boards/oho-elektronik.html

[jim]

All,

The cost of the boards made by the suppliers specified support my claim for
the expense of using FPGA's as an experimenter / hobbyist.
They're just too expensive to first, even purchase, and second, to justify
the cost if one did buy one.

Regards,

Jim


-----Original Message-----
From: piclist...@mit.edu [mailto:piclist...@mit.edu] On Behalf Of

[Michael Watterson]

Mike Harrison wrote:
> On Wed, 19 May 2010 14:49:09 -0600, you wrote:
>
>
>
> enterpoint and Trenz do a number of FPGA modules in DIL format
> http://www.enterpoint.co.uk/
> http://www.trenz-electronic.de/products/fpga-boards/oho-elektronik.html
>
>
>

Trenz is where I bought my Spartan 3E starter kit. It has eth phy, 16x2
LCD, USB bootloader (no JTAG needed unless you want to program the
coolrunner CLPD too).

I use scilab to make coefficients for DSP (instead of Matlab and Xilinx
DSP tools) and wrote a simple scilab script to convert results to xilinx
format

[Sean Breheny]

This project looks awesome! Thanks for sharing it!

Sean

[William "Chops" Westfield]

>> I too wrote code to emulate an 1802.

Y'all are just after that SEX instruction...

Here we have a Z80 CP/M system, implemented on an AVR (and it contains
some interesting hints on implementing an emulator in small code
spaces.) http://spritesmods.com/?art=avrcpm

BillW

[Tamas Rudnai]

On Thu, May 20, 2010 at 6:13 AM, William "Chops" Westfield
<wes...@mac.com>wrote:

> Y'all are just after that SEX instruction...
>

LOL!

> Here we have a Z80 CP/M system, implemented on an AVR (and it contains
> some interesting hints on implementing an emulator in small code
> spaces.) http://spritesmods.com/?art=avrcpm

Nice one! So you are basically saying, that it is impossible to write a Z80
emulation in gcc (to fit in the MCU), or only that existing project you have
tried was written in a way that was unusable in a small device?

With the 128 vs. 512 sector size: Can you implement some kind of caching so
you do not need to write back too often the entire block?

Thanks,
Tamas

>
>
> BillW
>
> --
> http://www.piclist.com PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>



--

int main() { char *a,*s,*q; printf(s="int main() { char *a,*s,*q;
printf(s=%s%s%s, q=%s%s%s%s,s,q,q,a=%s%s%s%s,q,q,q,a,a,q); }",
q="\"",s,q,q,a="\\",q,q,q,a,a,q); }

[William "Chops" Westfield]

On May 19, 2010, at 10:43 PM, Tamas Rudnai wrote:

>> http://spritesmods.com/?art=avrcpm
>
>
> Nice one! So you are basically saying, that it is impossible to
> write a Z80 emulation in gcc (to fit in the MCU), or only that
> existing project you have tried was written in a way that was
> unusable in a small device?

1) not my project. I wrote an emulator once, a long time ago, and
wasn't very happy with how much code it took for a "logical"
instruction decode.
2) The original emulator was from a Propeller, which has even less
code space (per cog) than the average AVR. The implementation is
almost hardware-like (microcode-ish), with load, operate and store
phases. I was impressed how much this simplified things...

BillW

--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

--

[Russell McMahon]

> It seems that other antique processors could be emulated (6502, 6800, Z80
> ...)  as well as creating new "processors" using the same techique.
> Emulating the processors allows new instructions to be added as well as
> adding debug features that would not be possible with the original
> processors.

On the 6800 you could make the processor emit REAL smoke and flames on
instruction $dd = 'Halt & Catch Fire".
(Also $4d mayhaps.)

Hmm.
How far back does an old brain reach?

Loop
86 00
4C
B7 80 02
4C
B7 80 02
26 02
20 F4
DD

Assumes
- B7 .. .. does not touch flags, which is almost certainly correct.
- PIA pre-setup.

Maybe not that far?
If it does work it does so crudely coz some related memory is
definitely in write only mode.

Once upon a time I could power up a D2 and hex-pad key an alpha (of
sorts) moving message display program for the 7 segment numeric
display, complete with message, from memory or BOTB calculation Those
days have gone :-) (At least for a 6800)

The 6800 seemed to have been made for hand or in-head assembly - equal
or better structured and with field bits grouped than anything else
I've seen. You really could create simple programs in your head in
machine language and key them in.



R

[Veronica Merryfield]

> On the 6800 you could make the processor emit REAL smoke and flames on
> instruction $dd = 'Halt & Catch Fire".
> (Also $4d mayhaps.)

I worked with an early transputer that had that instruction. I can't remember the actual instruction now but they had missed a transistor in a totem pole hence connecting the power rails, which stopped the processor. Then it was just a matter of time before the fire.

Veronica

[M.L.]

On Wed, May 19, 2010 at 6:45 PM, Michael Watterson <mi...@radioway.org>wrote:

>
> I use scilab to make coefficients for DSP (instead of Matlab and Xilinx
> DSP tools) and wrote a simple scilab script to convert results to xilinx
> format
>
>

Sparkfun has a few FPGA breakout boards. I'm a bit skeptical of the
usefulness of a 208+ pin part on a .1" breakout board though.

By the way: what is "xilinx format" ?

--
Martin K.

[Michael Watterson]

M.L. wrote:
> On Wed, May 19, 2010 at 6:45 PM, Michael Watterson <mi...@radioway.org>wrote:
>
>
>> I use scilab to make coefficients for DSP (instead of Matlab and Xilinx
>> DSP tools) and wrote a simple scilab script to convert results to xilinx
>> format
>>
>>
>>
> Sparkfun has a few FPGA breakout boards. I'm a bit skeptical of the
> usefulness of a 208+ pin part on a .1" breakout board though.
>
> By the way: what is "xilinx format" ?
>
>

Maybe other tools use it too

*.coe text file with DSP coefficients

Example content:
Radix = 10;
Coefficient_Width = 16;
CoefData =-37,
0,
-41,
0,
-50,
0,
... etc for each coeffient

My poor scilab script (I'm very inexpert at this) an *.sce file
n = 170;
coefs = round(hilb(n,"hm")*32768);
fd=mopen("J:/hilbert170.coe","w");
mfprintf(fd,"Radix = 10;\nCoefficient_Width = 16;\nCoefData =");
m=n/2;
for i= 1:m,
mfprintf(fd,"%i,\n",coefs(i));
end;
mfprintf(fd,"%i,\n",0);
m=m+1;
for i= m:n-1,
mfprintf(fd,"%i,\n",coefs(i));
end;
mfprintf(fd,"%i;\n",coefs(n));
mclose(fd);

The example if run in Scilab, creates a file for dsp to generate a + and
- 45 degree shift for Communications bandwidth audio. You then can use
those as I & Q for a SSB external pair of mixers that are fed with 0 and
90 VHF or UHF and cancel the unwanted sideband.
Or what ever else needs I and Q audio in the FPGA, maybe an FPGA
implemented FM/AM/SSB or digital Audio of some codec and final output
controlling FPGA logic that drives a VHF polar amp using FPGA NCO and PWM.

[William "Chops" Westfield]

Digilentinc seems to be a low-cost "educational kit" provider in the
US. They've got a couple of boards with "lots of stuff" for less than
or near $100.

This one looks interesting:
http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,400,790&Prod=BASYS2

BillW

[Dario Greggio]

Veronica Merryfield ha scritto:

>> On the 6800 you could make the processor emit REAL smoke and flames on
>> instruction $dd = 'Halt & Catch Fire".
>> (Also $4d mayhaps.)
> I worked with an early transputer that had that instruction. I can't remember the actual instruction now but they had missed a transistor in a totem pole hence connecting the power rails, which stopped the processor. Then it was just a matter of time before the fire.

hey but this is wonderful :))
I probably only heard rumors of something alike, but this is spectacular :)


--

Ciao, Dario
--
Cyberdyne

[Walter Banks]

Dario Greggio wrote:

> Veronica Merryfield ha scritto:
> >> On the 6800 you could make the processor emit REAL smoke and flames on
> >> instruction $dd = 'Halt & Catch Fire".
> >> (Also $4d mayhaps.)
> > I worked with an early transputer that had that instruction. I can't remember the actual instruction now but they had missed a transistor in a totem pole hence connecting the power rails, which stopped the processor. Then it was just a matter of time before the fire.
>
> hey but this is wonderful :))
> I probably only heard rumors of something alike, but this is spectacular :)

The HCF on the the M6800 actually just cycled the address bus of the processor and did not silicon harm, it could only be stopped with a reset. Two opcodes would do that $DD and $4D. This was one of Gerry Wheelers midnight to morning projects that he later wrote up for
Byte Magazine along with some other gems he found like store immediate and more useful logical operations between the primary registers. The later was most likely just missing documentation.. Most of these including HCF were don't care conditions on the processor's
implementation. They did prove useful in proving at least one M6800 clone was actually a copy.

Regards,


Walter..
--
Walter Banks
Byte Craft Limited
http://www.bytecraft.com

[Dario Greggio]

Walter Banks ha scritto:

> They did prove useful in proving at least one M6800 clone was actually a copy.

Yeah, I agree: nice one this one :)

--

Ciao, Dario
--
Cyberdyne

[Philip Pemberton]

On 19/05/10 23:07, jim wrote:
> The cost of the boards made by the suppliers specified support my claim for
> the expense of using FPGA's as an experimenter / hobbyist.
> They're just too expensive to first, even purchase, and second, to justify
> the cost if one did buy one.

Poppycock, there are plenty of "cheap" FPGA dev boards!

- AVNET Spartan3 development board. About 40 Euros. Onboard USB UART,
programmer and a ton of other stuff. Cheap, no frills, but still
powerful enough to run a CPU on.

- Altera Cyclone II Starter Board (also sold as the Terasic DE1).
Roughly US$150. FPGA, SDRAM, parallel Flash, Wolfson WM8731 audio CODEC,
VGA, a bunch of switches, LEDs and a pair of I/O headers.

- Enterpoint Drigmorn2. Xilinx Spartan3A (700kgate on the standard
board), SPI Flash, an x32 SDRAM, HD44780 LCD, a couple of LEDs, switches
and a pair of 20pin I/O headers. Cost about £150 with programming cable
included. They also do a USB programming cable which is a *really* low
cost version of the Xilinx Platform cable, but can only be bought with
one of their dev boards. Which is fine, because even the DM2 and the
cable comes out at less than half the price of a Xilinx JTAG cable.

--
Phil.
pic...@philpem.me.uk
http://www.philpem.me.uk/

[jim]

Phil,

If you're fortunate enough to be able to spend more than about $50.00USD on
a board that you may use once, you have my envy.
Of course, if I were going to spend that kind of money, I'd find a way to
use it more than once. But, that still leaves the
fact that most experimenters can't do that though.

However, I choose not to argue about it. I made a statement, and I stand
behind that statement. If you, or anyone else,
want to refute it, go right ahead. It matters not to me.

Regards,

Jim

-----Original Message-----
From: piclist...@mit.edu [mailto:piclist...@mit.edu] On Behalf Of
Philip Pemberton
Sent: Thursday, May 20, 2010 8:15 PM
To: Microcontroller discussion list - Public.
Subject: Re: [EE]: Where can you buy mini-PCB with FPGA mounted ?

[Russell McMahon]

> The HCF on the the M6800 actually just cycled the address bus of the processor
> and did not silicon harm, it could only be stopped with a reset.
> Two opcodes would do that $DD and $4D.

<Uphill both ways in a cardboard box in the snow with no shoes mode on>

HCF was immensely useful for hardware testing. As Walter said, the
processor cycled the address lines sequentially & endlessly - at
about max clock rate AFAIR. All address decoders etc were accessed in
turn and could be checked with a scope.

HCF could be implemented as a trap if desired but run away programs
often seemed to find it by themselves. Filling unused memory with dd
made it more likely. That way, if it is in HCF you know something has
gone wrong as opposed to it sitting in some obscure loop without what
is happening being evident. This was when emulators and their ilk did
exist but were not something liable to be affordable or available.
Programs were usually run from RAM during development to both cut down
EPROM erase cycle time (30 minutes typical AFAIR with a small UV lamp,
5 minutes with a powerful enough one, a day in the sun), save precious
EPROM program/erase cycles and avoid use of costly EPROMS at all when
avoidable. >>$100 for 1k x 8 triple-supply 2708 AFAIR.

I have a friend who started with an 8008 and the horrendously nasty
expensive and largely unavailable 1702 EPROMS. He managed a
spectacular destruction event at one stage and AFAIR it took months to
get new ICs from the US.


Those were the daze!


R

[cdb]

I notice that Mikroelectronika now have a Pic32 development board.

One of the demo programs for that is a Sinclair Spectrum emulator,
that'd be a Timex something or other for those across the lake, I
believe.

Colin
--
cdb, co...@btech-online.co.uk on 21/05/2010

Web presence: www.btech-online.co.uk

Hosted by: www.1and1.co.uk/?k_id=7988359

[Peter]

I assume you people know about this gem? it's a COSMAC ELF smulator written in
JavaScript. Runs in your browser. Fully functional. Made in Poland:

http://www.cosmac.szyc.org/

Looking forward for a MCS51 and a PIC emulator in JavaScript ... Adding AVR
to the bestiary would not be bad either :D

-- Peter

[Olin Lathrop]

Peter wrote:
> Looking forward for a MCS51 and a PIC emulator in JavaScript ...
> Adding AVR to the bestiary would not be bad either :D

Why? What are you going to do with those if you had them?


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.

[Scott Dattalo]

Olin wrote:
> Peter wrote:
>> Looking forward for a MCS51 and a PIC emulator in JavaScript ...
>> Adding AVR to the bestiary would not be bad either :D
>
> Why? What are you going to do with those if you had them?

The same thing you'd do with any bestiary.

[Olin Lathrop]

Scott Dattalo wrote:
>>> Looking forward for a MCS51 and a PIC emulator in JavaScript ...
>>> Adding AVR to the bestiary would not be bad either :D
>>
>> Why? What are you going to do with those if you had them?
>
> The same thing you'd do with any bestiary.

What a non-answer!

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.