On Wed, 12 Apr 2006 16:43:09 +0000, Slow Code wrote:
"Leland C. Scott" wrote in
:
"Slow Code" wrote in message
ink.net...
"Leland C. Scott" wrote in
:
"Slow Code" wrote in message
nk.net...
"Leland C. Scott" wrote in
:
"Jim Hampton" wrote in message
...
Has gone? Hmmmm ....
Not really
I'm just glad a few folks can get a few posts across with a bit of
sanity -
even if only once in a while 
It's been a load of fun reading the wild posts about me. It's like
sitting back with a cold one and watching a dog chase it's own tail
for hours on end because it's too dumb to know better. From reading
the posts the really dumb ones stand out from the crowd. Better they
left their mouths shut rather than opening it thus relieving all
doubt.
Anyway I've had better things to do the last few weeks, finally got
my TI 320C6713 DSP EVM boad with the development software so I've
been busy with another engineer buddy working on some DSP projects.
Nothing like going back to review all the discrete time system
theory, FFT's and z-transform stuff I learned years ago. It's a lot
more interesting than the silly name calling and baiting to start a
flame war going on in this news group lately and I don't want to
waste my time with such crap either Jim.
That gave me a headache, I got the TAPR 56000 EVM, Put it together.
Tried some 9600 packet, and 400 bps psk telemetry when A0-40 was
still working. I have it collecting dust now, as soundcard software
is easier. Don't even hook up my MFJ-1278 TNC anymore either.
I wouldn't say that the sound card software approach is "easier" at
least from a design standpoint. From a user's point of view you're
right about the ease of use, you can't beat it. The main difference
between a DSP board as opposed to sound card software is performance.
And that is a two part issue. One is raw computational performance and
the second is power consumption. The TI 6713 on my EVM gets barely
warm while cranking out 1200 to 1300 MFLOPS. Then take a look at the
monster sized CPU in your PC and it's heat sink. No contest. A last
point DSP chips don't waste transistors on stuff that isn't needed
like virtual memory, privilege levels etc. that only matters to CPUs
running general applications and need to be protected from other users
on the system.
One of the things that the TI EVM development software does for you is
the complex scheduling of the various routines that have time critical
deadlines to meet. If you do the sound card routine, writing it
yourself that is, you have to not only do the application code but you
have to write your own scheduler routine and maybe with multi-level
interrupts you assign to the different threads running.
Writing code for DSP chips has gotten easier now that most vendors
have "C" compilers you can use. And at least with TI they have a DSP
BIOS that handles the low level hardware crap so you don't have to
using assembly code. Add on hardware often comes with plug-in modules
containing the required low level code so you don't have to write it
yourself. You just call the low level routines from "C" using the
provided function prototypes in the vendor's "xxx.h" files and the
linker finds the code in the vendor supplied library files.
One of the interesting things I've found out is some hard-core
audiophiles are using some of the DSP EVM boards to do some custom
filter and complex frequency-gain adjustments etc. I've seen some
lively discussions among some of them over which EVM system is best to
use or going a roll-your-own approach is better. Some of them were
looking for doing direct digital to audio applications, some audio
equipment has direct digital outputs, with noise reduction etc.
because they don't like the limitations of the commercial gear out
there.
Thanks for the post. It's refreshing to have an intelligent exchange
of messages compared to the infantile crap some others are trying to
get me to waste time on.
http://focus.ti.com/docs/toolsw/fold...k6713.html#top
TMS320C6713 EVM
http://www.dspguide.com/pdfbook.htm Free for down load a text book
on DSP theory
My EVM came with five thick books. I went through part of the first
one and realized to write my own software for it, it was going to take
some time to learn things. So I mainly used software written by others.
Why re-invent the wheel. I still don't have a lot of time to play,
but maybe someday I'll plug it in again and learn a little.
SC
Hey, it's never to late. You stop learning when you're dead in my view.
It sure beats exchanging insults on this news group.
Yup, reusing code written by others is a good idea, why reinvent the
wheel unless you're doing it for self pedagogical reasons. In fact
that's one aim for good software engineering practice; code reuse. Was
that EVM board a stand alone system or was it part of some kit it was to
be used in? I assume you're referring to the Motorola 56000 DSP chip you
mentioned in another post.
Funny talking about DSP in this group. I normally come here to stir a
little when someone says their CB broke. LOL. I know that's mean.
What the heck, you have to play with the kids some time. 8-))
Mine has the motorola DSP56002EVM. TAPR made a kit to use it in. You
can connect a couple radio's to it. It has status lights on the front
you can program to indicate various states like what you would see on a
regular multi-mode TNC. On the back is a programming port and a
communications port. My soldering iron was probably only hot for about
an hour and a half putting it together.
It's really quite neat setup, after you load the modem software you
want
you can use it without having to wire every up again, or switch cables.
It's nice to have two computers hooked to it. One to use as a terminal
for what ever you're doing. PSK, Packet, etc, and the other computer to
load the modem software and program it. If you only have one computer
you have to switch back and forth. One thing I didn't do but wanted to
was burn an Eprom with the different modems on it to switch between
modes faster. What I had to do each time I used it was load what ever
software I wanted to use at the time. There was another way you could
load two different modems in it at the same time but I never did that.
That sounds like a nice kit they put together. They supply any block
diagrams or other documentation explaining how the code worked?
If you can program yours in a higher language and compile it down to the
EVM, that's a lot better than using an assembler like I got with mine,
though some members of TAPR wrote some apps to make things easier.
If you look around you may find a compiler released under the public GNU
license agreement that would work for your DSP chip. If you get the bug
again it would be worth the time to look.
Like I said, I'm not doing much with it right now. Time is the
problem. It's easiest to use the soundcard.
I've looked at some of the FFT routines used on some of the software
defined radios being sold. The one I looked at wasn't documented very
well. I wish that people who write code for public release did a better
job at documenting what they did with more in-line comments and extensive
cometary in the header files.
If you develope a new mode, you might have to port it to the PC so
everyone else can use it.
What I'm working on, with another engineer buddy from work, is sort of a
midnight engineering project that's work related but not necessary for
work if you get my meaning. Many companies like to swipe work related
stuff people do and then paten it under the company's name with no
compensation to the employee and that's even if they did all the work on
their own time and at their own expense. There was a magazine published at
one time I've read called "Midnight Engineering" that covered exactly
these sorts of issues with engineers developing ideas on their own at
home. I don't know if the magazine is still published or not.
There are a number of people in Amsat that
are working with DSP for the satellites. Look around for Phil Karn,
he's a good one to bounce ideas off of, and knows who else is
experimenting.
I've seen Phil's name mentioned often.
Also browse TAPR, I don't know what they're doing DSP
wise these days.
They are in to the new thing; software defined radios. They sell a kit
with a home grown DSP board to use with it. Even the professional
community is in to it big time as well.
I was never a member. I only really got into the EVM
& DSP when you couldn't buy an AEA DSP.
I thought that AEA still made DSP modems for packet. At least that's what
I thought I saw at last years Hamvention. They usually set up in a
hallway between a couple of the larger rooms. The rage now is using
commercial WiFi 802.11b gear for extended range. One company I checked out
on the Internet is selling a 2 watt bi-directional 2.4 GHz amp. To get one
you must supply documentation to them proving you're a licensed Ham before
they'll ship one out to you. No proof no amp. As long as the WiFi stuff is
operated on the correct channels the frequencies used fall in to the 2.4
GHz allocation Hams can use so we can legally run much higher power with
gain antennas that ordinary WiFi users can't. And here's the kicker
operating that way, under Part 97, any interference ordinary users suffer
they can't do much about it since the commercial WiFi stuff is licensed
under Part 15.
Regards,
Leland C. Scott
KC8LDO