98 news

[jfox@xxxxxxxx (Jeff Fox)]

Dear MISC readers:

Hi,  It has been a while since I commented here.  I was ready to 
jump in on the discussion of oversampling with a history of the
proposed and implemented designed with some estimates of
expected performance on F21 but I thought the subject was
covered pretty well.  I think it will be easier now to explain
my thoughts after the discussions have explained some of the
background picture.

As for the questions and comments in the last MISC-d Digest (V98 #8)

Dr. Ting's new board uses a separate 1.8Mhz clock counted down to get
the 7200 baud...  If you simply replace it with a 2.4.. the same
ciruit will run at 9600 baud.

This board uses the MuP21h in the 44 pin plcc and has a 82C51
serial I/O chip.  The board has a connector for a PC serial keyboard
connector to one port on the 82C51 and a to an RS-232 connector for
serial I/O.  Dr. Ting never debugged the PC keyboard interface, 
but most of the hardware is there but not the software.

P21Forth 1.01 and 1.02 were written for the previous kit with
the MuP21 DIP and a parallel port of 3 cmos chips.  These versions
could do solid parallel I/O and marginal bit bang serial.  As
the temp or voltage drifted the serial could flake out.

I have ported Dr. Ting's drivers for the new P21h kit serial I/O
chip to P21Forth and it works very with the new serial hardware.
I have run it on a direct connection to this PC and also over
my radio modem and telnet over the internet to Australia and 
to a P21Forth 1.04 running there.

So in addition to doing email and browsing the WEB on MISC boards
at work I can telnet around the planet and debug a P21Forth on
a MISC board any time. ;-)
 
I want to make one more change to 1.04 before releasing it.  At the
moment one can use the MuP21 eForth 2.08 that Dr. Ting offers
for this board, but it is very very limited and restricted.
I think the board is more fun with P21Forth.

As Dave said:

>A 2.4576 MHz oscillator can be used in place of the 1.8432 MHz to make
>the kit run at 9600 baud.

>> I recently put together the mup21h kit
>> ... 'Mup21 Eforth 2.08' 
>
>Is the kit a forth engine?  who makes it, what is it?

There is lots of material about MuP21 (P21) and the company that makes
it.  Since Offete Enterprises Inc. doesn't have a web site I often
get requests for orders for Offete.  If you can get to the web you
will find the last product list that Dr. Ting sent to me posted
there.  Since only a few thousand chips were produced and no
specmarks were delivered the chips are not listed in most chip
directories on the internet. 

I may offer some Offete boards and chips and documentation via 
internet shopping at my web site soon if there is any interest
(or maybe even if there is not. ;-)

Since P21 came out in 10/94 (whew) it clearly demonstrated that
Chuck could produce custom VLSI with his own CAD software that
was quite remarkable.  The stats on P21 are really quite
remarkable. 

However P21 is sort of an odd duck since it was designed to make
a 7 chip VLSI CAD workstation based on a chip that could be 
manfactured for about $1 in large quantity.  It was seriously
stipped down in terms of I/O having video out only on chip 
and requiring polling of external I/O hardware mapped into its
memory space.  It has a number of odd things, 20 (+) bits of
addressing and a 20 bit data bus, an odd representation on
the bus hardware, stack machine design etc.

>Is there a forth PIC - or a true x-tra low-cost single chip forth
>solution?  
>
>thanks
>
>arpad benares

Of course you can use Forth on a PIC and other low cost single chip
devices.

I don't really consider a microcontroller with Forth in ROM to be
a Forth chip, but some people do.  There are of course single chip
microconrollers with Forth in ROM.

I think there may be microcontrollers with architectures that 
closely resemble the Forth virtual machine, but I really don't
know of any available.

Those of use working with MISC realize that the very tiny chip
size will allow us to go on chip with some combination of
ROM/FLASH/PROM/SRAM/DRAM etc to get very cheap very tiny very
high speed solutions.  It will happen, but it isn't available now.

I have seen one of Chuck's single chip designs a while back for
a particular problem and it was so tiny it was really cute.  Our
chips tend to be sized by the I/O pad requirements so if you have
ROM and RAM on chip and don't provide external pins you can a
design like that old P8 with 8 pins (or less) and make the
thing so that several fit on 1 square mm of die, the idea
was a manufacture cost of well under ten cents.

We all would like to see a 1G .1$ single chip solution
available for some problems. (or maybe not) ;-)

We will also like the more expensive chips that will have
more memory on them.  But we are not there yet.  There was
some confusion by those who read the iTV web page as to
whether we had 1M byte of ROM on i21 at this time.

Jeff Fox

Jeff Fox              jfox@dnai.com    
Ultra Technology Inc. http://www.dnai.com/~jfox/