[Lf] Fwd: [RX320] Announcing open320

Tue Jul 8 20:45:12 CDT 2003

Hi Frank,

When I first got my RX-320, I thought about something like this but didn't persue it too far 
because:

(a)  I already had a life, 
(b)  the dsp involved is, well, small, slow and long in the tooth, and
(c)  by and large the receiver worked pretty well anyway.

It seemed then the effort would have been better put to developing on a meatier device, 
using a cheap EVM from whichever device manufacturer but retaining the 'hard part', i.e. 
the analogue RF bits of the 320.

Don't get me wrong at all:  the 320 is amazing value for money and I have nothing but 
admiration for the cramming in of such a lot of engineering in one teensy dsp, with such 
excellent results.  Also, I can see merit in possible upgrades to a receiver that Ten-Tec 
would certainly redesign nowadays rather than upgrade or attempt to extend.

The chap who's doing this has already done the hard work, though he doesn't seem to 
think it (divining initialisation, parsing, i/o etc.  The filtering is so trivial that it'd be easy to 
blink and miss the three lines of code;  similarly the detection.  The actual signal 
processing code will be quite small  -  most of the 5k in the rom will be filter coefficient 
tables and possibly a sine lookup table.  Figuring out how they're indexed might be fun, 
and the AGC (see later) will undoubtedly be a bit gruesome to reverse-engineer.

I think you're right  -  there probably isn't too much to spare in dsp cycles;  it's probably 
running maxed out.  However, something to consider is that the version of the dsp that's 
in the radio isn't the fastest of its type;  there is a 25MHz version, which could either run 
off the existing convertor clock, or could probably be crankable up to 30+MHz without 
undue effect and so would just about double the available processing time.  Now that'd 
be worthwhile.  Changing the chip wouldn't be that big a deal, neither would be adding 
an appropriate clock-block (probably too fast for a fundamental xtal).  The latter at worst 
would be on a little add-on board, or splayed-leggy.

My own wish list, should this all happen, is headed by a handle on the AGC attack time;  
as it stands it is so fast that any transient captures it and punches a big hole.

Long ago and far away I did your trifilar mixer input transformer mod, which made a 
tremendous difference, of course.  Additionally, removing the 140kHz-ish trap (!) of L26, 
C108 and R89 by lifting one leg of the resistor, and bypassing C106/C107 with a 0.1uF 
brings an almost embarassment of sensitivity at LF;  external filtering for LF and/or 
broadcast becomes necessary.  But frankly, with any size antenna, it was anyway.

        All the best,

                Steve  

http://www.w3eee.com

7/8/2003 7:45:26 AM, Frank Gentges KØBRA <fgentges at mindspring.com> wrote:

>
>AMRADers,
>
>There is an activity to help us reprogram the RX320 firmware.  See 
>below.  I'm sure we could improve it for LF and other uses.  I have the 
>feeling the current processor is pretty much loaded already.  Along with 
>firmware, a faster processor might be in order to make space to do some 
>nifty things.  This may only take a chip and/or a crystal.    I spoke to 
>TenTec about getting the source code under an NDA and they did not see 
>their way clear to do this.
>
>We went on to use the PC and narrowband signal display software like ARGO 
>to make up for the shortcomings of the RX320.  The combination worked out 
>well.
>
>Any AMRADers interested should speak up to help and encourage this fine 
>effort.   We may be able to take on the hardware side of speeding up the 
>processor.   About all I can do for now is stand on the sidelines and cheer.
>
>Frank K0BRA
>
>>X-eGroups-Return: 
>>sentto-141764-4068-1057640700-
fgentges=mindspring.com at returns.groups.yahoo.com
>>X-Sender: piss_pistofferson at yahoo.com
>>X-Apparently-To: RX320 at yahoogroups.com
>>To: RX320 at yahoogroups.com
>>From: Aaron <piss_pistofferson at yahoo.com>
>>X-Yahoo-Profile: piss_pistofferson
>>Mailing-List: list RX320 at yahoogroups.com; contact RX320-owner at yahoogroups.com
>>Delivered-To: mailing list RX320 at yahoogroups.com
>>List-Unsubscribe: <mailto:RX320-unsubscribe at yahoogroups.com>
>>Date: Mon, 7 Jul 2003 22:04:59 -0700 (PDT)
>>Subject: [RX320] Announcing open320
>>Reply-To: RX320 at yahoogroups.com
>>
>>Hi all,
>>
>>As you may know, I expressed interest a few weeks ago in figuring out
>>how the RX320 firmware works in the hopes of possibly providing a way
>>for users to improve or customize the firmware.  Well, we're still a
>>long way away from that dream, but to that end, I've submitted a
>>project proposal to Sourceforge.net for an open source, GPLed RX320
>>firmware replacement.  The project, should they approve it, will be
>>called open320.  For those interested, I've attached the proposed
>>project description below the body of this message.
>>
>>I've been working on reverse engineering the EPROM dump and I've got
>>the first 640 or so words fully worked out, with other bits and pieces
>>in various stages of understanding.  I've been taking the ADSP-2101
>>disassembly and trying to format it into human-readable pseudo-C code
>>with a generous amount of comments for greater understanding.  So far,
>>I've got the initialization, command processing, low-level RS232
>>control, and some of the codec and PLL control figured out.  (There's
>>actually an undocumented serial command, but space prevents me from
>>elaborating.)  Most of what remains is the hard part--the math-heavy
>>routines that do the I/Q demodulation, the FIR filtering and so on.
>>
>>I will post another message once the Sourceforge project page is up and
>>will then take further discussions of the project to a separate mailing
>>list so as to avoid clutter.
>>
>>Regards,
>>Aaron
>>
>>---
>>
>>The Ten-Tec RX320 and its brother, the RX320D, are shortwave receivers
>>that are fully controlled by serial commands sent to its RS232 port.
>>There is no front panel on these radios--all radio tuning and control
>>information is provided by a program running on a host computer.
>>Furthermore, the final stage of intermediate frequency (IF) processing
>>is handled by a DSP, which demodulates the radio signals according to
>>the user's desired mode and outputs them to a speaker via an internal
>>codec chip.  (In the RX320D, this 12-kHz intermediate frequency is made
>>available on a rear-panel connector for advanced applications requiring
>>more processing power than the internal DSP can handle.)
>>
>>The heart of the RX320 radio is its Analog Devices ADSP-2101 DSP chip,
>>whose job is to handle both command decoding and signal demodulation.
>>Signal demodulation is performed by the AD1847 codec chip, which
>>digitizes the 12 kHz IF signal for processing and also processes the
>>decoded audio stream from the DSP for output to a speaker or
>>headphones.  A Motorola MC145140 PLL generates the internal frequencies
>>for various parts of the receiver circuitry.
>>
>>The ADSP-2101 firmware is stored on a DIP (dual inline package)
>>socketed 27512 64 KB EPROM, making the prospect of replacement easy and
>>practical.  Early analysis shows that only the first 8K of the EPROM
>>has data.
>>
>>The open320 project will progress in two stages:
>>1. Since Ten-Tec does not provide the firmware source code, the
>>original firmware will be reverse engineered such that the operation of
>>major subsystems (RS232 reading and writing and codec and PLL control)
>>can be mapped and studied.
>>2. Once operation details of these subsystems are known, replacements
>>can be coded.  As with Linux, it is anticipated that users will be able
>>to configure firmware with the modules that they desired (for example,
>>AM, SSB and FM demodulation, the latter of which is currently not
>>supported in the RX320 firmware).  This will allow the greatest
>>flexibility and provide a framework for future extensibility.
>>
>>There is already a version of GCC that can compile to ADSP-21xx object
>>code.  Part of the project will involve getting up to speed on these
>>tools and making cross-platform building easy and painless.
>>
>>
>>
>>__________________________________
>>Do you Yahoo!?
>>SBC Yahoo! DSL - Now only $29.95 per month!
>>http://sbc.yahoo.com
>>
>>------------------------ Yahoo! Groups Sponsor ---------------------~-->
>>Buy Coral Calcium for Greater Health - $23.95
>>http://www.challengerone.com/t/l.asp?cid=2805&lp=calcium2.asp
>>http://us.click.yahoo.com/MmkSQC/NTVGAA/ySSFAA/ELTolB/TM
>>---------------------------------------------------------------------~->
>>
>>
>>
>>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>Frank Gentges KØBRA
><fgentges at mindspring.com>
>LF web site at <http://www.amrad.org/projects/lf>
>
>
>_______________________________________________
>lf mailing list
>lf at amrad.org
>http://www.amrad.org/mailman/listinfo/lf
>
>