MuP21 kits

[jfox@xxxxxxxxxxxxxxxxxxx (Jeff Fox)]

Dear MISC readers,

some of you may have heard my opinion on this subject before:

Dr. Ting has a one man company which he runs in his spare time.  He
generates one to two new products every month and provides documentation.
People complain that he doesn't proofread and his products have poor
quality control. People complained that they could not order his
products easily especially outside of the US because it meant at
least several hundred worth of long distance repeating yourself
to Dr. Ting's elderly Chinese father who runs the mail order line
almost doesn't speak English and knows nothing at all about computers.

People complain that there were 12 versions of the documenation on the
kits and they are all full of different errors and none of them even
have version numbers.  These are the issues that took up a lot of
bandwidth on the MISC mail list in the old days.  Until enough
people answered each other's questions and did  all the debugging
and engineering work that they needed to do to get their kits working.

Dr. Ting thinks the fun part is solving all those extra puzzles and
building and programming a computer from scratch with instructions
that don't work. really.  he is a wierd guy.  Nice though.  In fact
it is so much easier than the stuff that he does that I guess he
just thinks of it as a introduction into the real world of computers.
Namely there is little documenation and most of it is wrong. You have
to use your noodle a lot to figure out which page is right and
which page is wrong.

Part of this comes from the fact that 8 mup21 chips were prototyped
so 8 sets of documenatation had been accumulated before chips were
manufactured. Some things might have once been true for p21a or
were just wrong in every version of the documenation because none
of it was ever proofread.

>I have a MuP21 kit. When I first tried to connect to the board, I would
connect 

I forget.  Is this one you built yourself and debugged step by step or was
it an assembled kit?  You are responsible for testing unassembled kits.
To tell the truth I don't think Ting checks the CPU he puts into unassembled
kits and rumor is that yield after sitting for a year in oxygen and then
being packaged in the plcc resulted in 20-30% of the chips now being somewhat
flakey.  The people who got several chips reported better results. 
Generally flakyness is inversely proportional to voltage too.

Some people doing commercial projects actually did extensive testing
and characterization of the chips.  They considered it proprietary and
didn't publish much.  They did have real engineers working on it
though thinking that they needed to substitute a few weeks or months
of engineering effort rather than rely that Dr. Ting had done
everything they needed to do in a stare ten minutes some time ago.

Anyway P21 is not my product. I offer Ting's boards in my store but
I don't make money selling them.  It is a service to anyone who wants
to get the best tutorial program my P21Forth.  I can't imagine how
anyone can do anything more than flash an led with Ting's software.
He intends it to be that way because he wants people to enjoy all
the fun of getting it to the point where they can figure out something
as complex as that and get it running.  It really is quite easy if
they have twenty years experience with Forth, love to solve puzzles,
have time, don't need documentation etc.  ie like Ting or Chuck.

The whole point of P21Forth was so Ting could send people assembled
and tested and documented end user quality software so that more than
10% would give up on trying to start with everything in the most
primitive form possible.  For someone like Chuck or Ting, that IS
where you want to start.  You would never use an ANS Forth like
P21Forth.  I recommend it for beginners.  Get the kit with P21Forth
tested and certified to have done a graphic demo and performed
50billion tasks switches before shipping without an error.

>the serial line to the host pc, and press reset on the board, I would see the 
>eForth banner and could sometimes type a few characters before is hung. I found
>that if i was quick, i could type SLOW and the board would seem more stable,
>it would somtimes maintain the connection for a few minutes before hanging.
Then
>one day I pulled out the video clock chip, and now it runs fine, for the most

If you removide the video clock (?) (oscillator?) it will sort of run but
only on programs where you provide dram refresh in software.  The video
coprocessor instruction set includes the REFRESH opcode.  You throw a
few of those into the video frame to keep memory warm.  If you don't
you had better visit every 1K page you want so often or it will die.
Of course Ting's totaly minimal software is about 1K so things will
almost work until you forget what you did and try to use memory
randomly.  You can use a slower clock and run a program with only
refresh and jump in it to get the maximum cpu bandwidth in dram if
you are not generating video.  But these are all very basic things
about P21 that are very important.  It sounds like you might have
problems due to having almost no documentation.  

How many issues of the 28 issues of More on Forth engines did you get.
(50% of the docs on the subject) How many of my docs on P21 did you
download and print?  Did you get a copy of P21Forth when it was free.
I always said it is the only documented program and the only one
supported by me.  If you didn't get all that stuff and just got one
manual you have got about 10% of the docs and have the ones that
are the oldest and most full of errors. I would ask you the version
number but it might not help.

>part, I've had it running programs on the MuP21 for hours, with no problems.
>Still though, if I haven't typed anything for a while and the board is just
>idleing, it will cease to respond and i have to press reset. Has anyone run
into 

could be a flakey chip, flakey board, flakey memory, bad line, slow
memories. Ting would say "what is wrong? it is your job to write the
programs to determine exactly what isn't working.  We did it with
no documenation and you already have lots! What is wrong?  This is
why you bought that product!"

>this? I think I heard somewhere about running the cpu at different voltages.
>Is there anything else I can do to make it more stable?

I have written a lot of stuff many times about all the voltage issues
and performance issues on P21.  I have also explained it many times
to many people.  I got tired of explaining the same dozen questions
to each of two hundred people in email so we setup the MISC mail
list so that I would not have to start each day with 12 hours of
email.  Even with the mail list I still do that, or have been for
the last week.  I have no phone but still have internet access and
email, it is sort of the last plug and I have hoping that they
would not pull that one too. Try 5.5 or 5.8 if it doesn't get too hot
but you void your dram warrenty if you go above 5.5

>Doesn't it send an OK to the video display on power up? I bought an RF 

I don't think so.  I didn't write ting's software and have never used it myself.
I think it comes up with the video processor turned on to
do refresh but with I/O vectored to serial I/O only.  That is the
way it is setup as I recall. Isn't that what your documentation says?

>modulator,
>and was going to try to run the video through that so I can see it on channel
>3 or 4. Any suggestions? ( I did have the video clock chip in during these
tests)

Step by step debugging. Maybe new components, mem, cpu, pcb.  Maybe a
different voltage.  being totally minimal tings code will run marginally
at far lower voltages that P21Forth that is 16 times larger.  Ting
puts a voltage regulator on the board and runs it at a very low
voltage where is software runs but if you added a few lines of code
it might not. hey that is Ting's idea of fun.  

>the cpu seems to be running fine. There is just a circuit trace from the cpu
>video line to the video socket connector, so I think the connection is ok.
I would
>really like to play with the video cpu on the MuP21. if anyone has some ideas,
>let me know.
>
>TIA,
>
>Lonnie

How fast are the drams?  80 is marginal and not good. Some brands work better,
we also tweaked the amount of freshed for the brand of memory.  I know how do
you tweak the video if it you don't have it to get the dram working when you
can't get video with the dram working?  You read and reread the documenation
and do a LOT of experimentation.  You figure out if the problem is that the
brand of dram you have requires that you rewrite the video coprocessor driver
and recompile a new system.

I do not think that it is fun pulling yourself up by your bootstraps when
your boots have been minimized to where you can't find them.  I think P21Forth
is a reasonable powerful and small subset for either a beginner or a pro
for many things.  My evidence is that there is lots of good documentation 
and some demos for beginners and I consulted on a commerical application
where we ported an application into P21Forth.  I have mentioned it before
it made it possible to port their application program from polyForth to
P21 easily and quickly including the GUI functions in Polyforth and add
all the device drivers that they needed for the LCD, serial ports, touch
screen, and microphone.  (cute very smart handheld and they had some clever
code)
It was fun porting a 1.2 megabyte application from the PC and putting it
into the 16K word p21forth in place of it's demos (ie a couple of k in there)
The job was done in a few days while Chuck and I visited there.  But enough
about why not to start without anything.

I have always said that unless your idea of fun is the same as
Dr. Ting you should get an assembled and tested P21 kit with P21Forth
and manual.  You should run all the dozens of demos that are included
and sit there and be amazed for a few days before trying to get
beyond reading the well written documentation.  They you can write
some cool multiwindow multitasking demos or 3D animation, compile it
and save the image to some blocks like I did with the old 3D demo
and instantly pop it back up from ROM any time. Or do whatever
you want to experiment at hacking.  There are tools for ANS Forth
and machine Forth and lots of documentation.  To me that just seems
like the same thing to do unless you think hacking everything at
once from the most primitive state including untested chips with
no or eronious documentation fun.

It was kind of funny.  that was my opinion but Dr. Ting completely
disagreed with me. I gave him P21Forth so he could sell it for $25
like all his other products and assumed that he would include a ROM
and cable when he sold it.  Instead he removed the ROM image from
the files.  This now meant that the user had to figure out how to
modify the metacompiler and replace the code for the prom programmer
that I had attached to mine and change it to support new hardware
then generate a new metacomopiler with no documentation. 
Then you could use your modified meta compiler to
generate a new ROM image and burn it.  Then you could figure out
how to build a cable with no instructions and write a terminal
program.  IE. he converted the Plug-and-Play P21Forth ANSI Forth
for dummies into the most advanced hacker's kit of all time. Not
a single person every got anywhere when they purchased it from Ting.

Eventually I offered the product with some quality assurance from
my store. I still recommend a tested board unless you really like 
hardware fun.

the Fox