Re: MISC-d Digest V97 #7

[jfox@xxxxxxxx (Jeff Fox)]

Dear MISC readers:

>RE: Running P21 at 5+ volts
>
>    Most chips are rated for 10% power supply operation, so 5.5 volts should
>    be no problem.  I don't have my P21 set up right now, but if memory serves
>    the video is glitchy at even 5.5 volts.  I think I was running upwards
>    of 6 volts.  This will make the DRAM run hot and probably reduce its
>    life.
>
>    Maybe Jeff Fox can chime in here... I think the max, do not exceed voltage
>    is 7 volts.
>
>-Dave

Chuck said that the DRAM spec says 5.5 is the highest recommended voltage
from the manufacture.  Yes running above that makes em hot.  Some brands
will run faster as the voltage is increased.  I tried some Hitachi chips
and they ran considerably faster as voltage was increased and leveled 
off at about 6.2 volts above which they didn't get faster.

P21 video is pretty solid at 5.5 volts except when doing slow I/O.  The
problem is simply that the video has real time requirement to read the
next video word and display the next video pixel of about 600ns.  If
the CPU spends too much time when it gets its turn the video can be too
late when it gets it turn at memory and displays the next pixel.  It
is almost never a problem with very onpage cpu code in which case the
memory accesses are only 50ns.  This is why OK can run with solid video
below 5V.  P21Forth is 16k of code and does lots of offpage access.
Since these accesses take 150ns a higher voltage is needed to assure that
video remains solid.

ROM access or access to the slow i/o address space (really slow RAM space
but used as slow i/o space on all of Dr. Ting's boards) is 250ns.  So
When you get an offpage CPU instruction access (caused by branch, a
preceeding offpage data access, or a preceeding video coprocessor offpage
access) and then do slow i/o and then try to do an offpage video access
it very possible to get a video glitch.  

At about 5.8V (with the Hitachi drams I have) the video is solid even
with lots of offpage access and slow i/o.  However the Toshiba drams
that I have will not run at 5.8V.

However according to the DRAM specs you do risk damaging the DRAM chips
by running them above 5.5.   Personally I have never had any problems
and usually run them at about 6V.

>From: Christophe Lavarenne <Christophe.Lavarenne@inria.fr>
>To: misc
>Subject: on chip RAM

Excellent explanation as always.  Thanks from everyone again.

>From: Penio Penev <penev@venezia>
>Subject: Re: power supply flakiness of MuP21 boards

And don't connect a 6V nicad camcorder battery up to your P21 board
with the connections backwards unless you want the chips to really hot!
(would you believe they all survived?)

>From: Ray Gardiner <ray@netspace.net.au>
>Subject: RTX2000 successor question

>We have been using the RTX2000 for a number of years in a
>real time imaging system, and are now starting to look about
>for a suitable successor. My first thought was to look at the
>P21 and possibly the F21 in future?  I know very little of the
>current generation forth engines.
>
>Now for the question: Has anyone had experience with moving from
>RTX based applications to P21?  Are there other choices I should
>consider?

There are other stack engine chips like the SC32 that should be
considered.

RTX has lots of great features for critically timed real time 
apps.  It has completely predictable timing, external interrupt,
external clock, etc.  You can get stuff like that with SC32 but
not P21.

P21 has a free running clock and a video output coprocessor to make
memory access timing hard to predict at any specific moment.  It has
no external clock except for the video coprocessor. It has no interrupt
either from a coprocessor or external.  There is a multiply instruction
on some RTX but not on P21.  These factors make P21 quite a different 
beast than RTX.

    Jeff Fox
    jfox@dnai.com              Ultra Technology Inc.
    jeff@itvcorp.com           the iTV Corporation
    http://www.dnai.com/~jfox/