DSP-10 Project

The DSP-10 is an amateur-radio, software-defined 2-meter transceiver that can be built at home. It operates on SSB, FM and CW modes. Features are tailored to operation with weak signals on VHF, UHF and Microwave frequencies.

A three-part article by Bob Larkin, W7PUA, published in QST magazine for September, October and November of 1999 describes the project. Copies of the article can be downloaded from the ARRL Web site. The transceiver hardware consists of a single PC board for the RF parts and an Analog Devices EZ-KIT Lite demo board for the I-F and audio processing. This is all controlled by any PC through a serial port. All programs are available as 'Free Software.'

W7LHL's DSP-10 with a 1296 transverter

On the right is W7LHL's DSP-10 setup. He uses a laptop to control the transceiver. All DSP-10 hardware is inside the die-cast aluminum box. As shown here, Ernie is using a 1296 MHz transverter to receive the NU7Z beacon.

The transceiver is very useful on 2-meters as a standard radio and has a number of interesting features. But, this was not the motivating force behind the project. There is a great untapped potential for weak-signal QSO's if the proper processing is used. Version 2 of the software is a start towards achieving this goal. It makes a number of tools available to the experimenter interested in this area.

The transceiver is low power, about 20 milliwatts, but there is provision for adding an 8-Watt amplifier (see below). However, the low power can be ideal for use with transverters, since most of these use attenuators to throw away the higher power anyway.



TAPR Part Kits

Part kits for the DSP-10 RF board have been available in the past. This has been of great value to those putting the project together. As of April 2002 there is a possibility of the kits again being sold. Let TAPR know if you are interested in getting a kit. Details are available at TAPR web site. Our thanks to Steve, N7HPR, and TAPR for taking on this project!


Single Yagi, 150 Watt 2-Meter EME QSO

W7SLB and W7PUA demonstrated a QSO on 2-meter EME, using the PUA43 mode of the DSP-10. Single Yagis and transmitter powers of 150 Watts or less were used on both ends of the contact. Details are available on the weak signals page and the linked text.


QRP 5-Watt 1296 EME QSO

W7SZ's DSP-10 screen receiving W7LHL-Click to Enlarge

On 25 February 2001 Ernie, W7LHL, and Larry, W7SZ, were successful with a PUA43 mode QSO on 1296 MHz EME (Moon bounce). They were using a QRP power level of 5-Watts and backyard TVRO dishes of 10 and 12 feet.

Copy was Q5 as can be seen on the screen at the left (click on it for a bigger version). They were using a 28 character setting on PUA43 (see below for more information on the PUA43 mode). The extra characters were being used for redundancy in portions of the message. The message would begin to appear in a few minutes, but Q5 copy was taking roughly 15 to 30 minutes per message. They exchanged call signs, grid squares, and confirmations both ways. Copy at W7LHL was not quite as good as for the other direction, but both stations received a complete message in the 28 characters. This was their first attempt at using the 5-Watt level for a QSO. As was reported earlier they had been successful at higher power during January.

None of those involved are aware of any previous 5 Watt 1296 MHz EME QSO's although this may well have been done using larger antennas. Any information on this would be appreciated.

The PUA43 mode described below is ideal for this type of work. The limits of signal strength have yet to be fully explored. Both Ernie and Larry reported that they were kept busy with antenna pointing and playing screen "anagrams," so were not sure what their patience level might be if the power kept dropping. The nature of the mode is to trade-off time for increased sensitivity.


10 GHz PUA-43 Mode EME QSO

On 11 November 2001 Ernie, W7LHL, and Larry, W7SZ, had moved their experimentation to 10 GHz and were successful in using sub-audible signals for an EME (Moonbounce) 2-way contact. Details and screen shots are on the weak signals page. This contact is particularly significant in demonstrating the ability to use the mode at higher frequencies, where broadening of the signal spectrum is a factor.


NEW VERSION 2 SOFTWARE - "Weak-Signal" Release

Version 2.0 is now available for download. This is a major upgrade of both the PC control program as well as the DSP i-f and af processing. To fully see the changes you should look at the upgrade notes which are also part of the full zipped download. But, in a nutshell the changes are:

  • Six additional audio filters for SSB and CW.
  • One audio filter that can be custom tailored from the screen.
  • 13 point audio equalizer from the configuration file.
  • New LMS auto-notch.
  • Greatly improved LMS de-noise operation.
  • Two selectable I-F bandwidths.
  • Two new spectrum analysis bandwidths down to 2.2 Hz.
  • Automatic EME Doppler correction of receiver frequency. The accuracy is a fraction of a Hz at 2-meters and within a few Hz at 1296 MHz
  • Fast Push-to-talk without delay (some minor hardware changes needed).
  • S-meter that is accurate from -140 to -25 dBm.
  • Highly accurate Sun-noise measurements.

In addition there are now four new weak-signal modes fully integrated between the hardware and software:

EME-2 Mode - Automated transmit/receive timing and Doppler correction for self-echoes on EME. Two trace spectral display provides any amount of power integration of signal and noise powers allowing echoes to be seen 30 dB below the audible levels.


LHL-7 - Seven-tone FSK mode for communications that does not require precision frequency control. Automated transmit and receive of signals too weak for audible copy, but still visible on the waterfall display. This mode is EME compatible and can be Doppler corrected. W7SLB EME Reception - Click to enlarge

PUA-43 Mode uses forty-three tone FSK with adaptable "very long-term integration." This mode provides automated transmit and receive for communication by terrestrial or EME paths, allowing an extreme ability to trade-off of data rate against signal strength. "Message Estimation" is used to always show the most-likely message along with a second alternative. The character color is indicative of the confidence level for each symbol.

To the right is a sample of EME reception of the PUA43 mode. Note that no signal is visible on the spectral waterfall, yet copy is 100%. Click on the photo to make it full-sized.

LTI Mode - This stands for Long-Term Integration (the same as averaging). Although called a "mode" in the operation of the DSP-10, this is really a measurement tool. Transmission is essentially the same as CW. Reception uses the same weak-signal noise averaging as EME-2 and PUA43 to extract spectral traces on extremely-weak signals, limited only by one's patience! Overnight tests show -185 dBm signals with well defined traces (that's a little over 0.0001 microvolts!). The display allows one to actually measure these signal levels to within a dB or two. This has been a useful and fun tool for exploring propagation paths. Like the other modes, it can be operated with EME Doppler corrections.

Hardware requirements: Full utilization of these weak signal modes requires a frequency stability of about 1 Hz at the operating frequency (it may not be 2-meters because of external transverters). The programs work with any PC with DOS, a serial port and VGA graphics. No math coprocessor is required, but it speeds some operations quite a bit. These modes require the RF hardware of the DSP-10. Some limited receiving functions are available with the Audio Processor option described below.

Availability: These programs for the PC and DSP are available from this site for no charge.

What can be done with the DSP-10 using these programs? Basically these are a set of tools that are being made available to anyone that wants to do serious weak-signal experiments. The package allows QSO's with stations that are much too weak for conventional modes. This opens many interesting possibilities for contacts. Here is a sampling of some early work:

  • The first PUA43 EME contact between W7SZ and W7LHL was on Jan 9, 2001 using 1296 MHz. Both stations were running about 40 Watts and using TVRO dishes of 10 and 12 feet. The interesting thing is that copy was truly Q5, in that the computer screen displayed 100% copy of calls, grid squares and a short message! These two stations have since followed up with a 5-Watt QSO reported above!
  • W7LHL is receiving 10 GHz EME-2 echoes with 1.2 Watts. Click for details including later results at higher power.
  • Last summer, the PUA43 code was used successfully by W7LHL and W7PUA over a 665 km Mountain-top to Mountain-top path using the 10GHz microwave band.
  • W7PUA has received his own 2-meter EME-2 Moon echoes using QRP 5-Watt power and a 4x12 element Yagi antenna.

Installation instructions for Ver 2.0 are in the README20.TXT that is part of the download. Updated 9 Feb 01


DOCUMENTATION ITEMS Several new documentation items have become available:

You can now get the Three part QST Article in PDF from the ARRL.

Steve Bible, N7HPR, has been working hard on the documentation project and has an excellent Assembly Manual that should be of great help with the board assembly. A couple of cautions from Steve: "Please let people know that it is draft and a work in progress. Check back often to download the latest version before/while building. And to please send in comments and suggestions to improve it." Many thanks to Steve for all his efforts.

And Beb, W7SLB has posted photos at his DSP-10 Picture Gallery. You will notice some interesting variations in the way Beb put the transceiver into the box.

Beb also has a collection of DSP-10 user pictures and would greatly appreciate more photos. Anything will do, but it is especially fun to see the people! Photos of the assembly process can be good, too.


EMAIL REFLECTOR FOR THE DSP-10

Thanks to Beb, W7SLB and Al, K3TKJ, we have an email reflector to distribute information on the DSP-10 transceiver project. The reflector is on mailman.qth.net and will cover all aspects of this project. You can send an email to the reflector, after you are a subscriber. This will then be sent to all of the other subscribers. This will give a way for all those interested in the project to share their questions, answers and experiences, quickly and informally.

To subscribe, unsubscribe, or to read the archives, go to the qth.net Web Site. Select "DSP-10" as the mailing list.


8-WATT "BRICKETTE" AMPLIFIER

This 2-meter amplifier, specifically for the DSP-10, is now ready to be built. Full details are in the June 2000 QST magazine article. A PC board package is available from Mashell Electric (Beb, W7SLB). This was the amplifier used by W7PUA for the 5-Watt EME-2 self-echo test mentioned above. Who is going to be the first to do a Brickette-to-Brickette EME contact? No part kits are presently available, but maybe someone would like to take on this project? (Updated 9 Feb 01)


LATEST INFORMATION FOR BUILDERS

COMMAND SUMMARY SHEET Steve Bible, N7HPR, has prepared a great summary sheet for the DSP-10 keyboard commands. You can download a PDF of it here. This does not yet have Ver 2 commands, but the ones on it are mostly correct for the new software.

CORRECTIONS:
1 - The connections on the EZ-Kit to P3, Pin 31 and P3, Pin 32 are backwards. A corrected portion of QST Fig 11 shows the change. It also reflects on QST Fig 10, but no wiring is changed. (16 Nov 99)
2 - R49 in the schematic and the part information should be 47K (not 1.8K.) (16 Nov 99)
3 - R21 should be 10K (not 100K); this causes the gain to be 20 dB too high. (1 Dec 99)
4 - R108 should be 4.7K Ohms (not 470). Some combinations of U107 and U104 have been found to have the data line at U104 switching too early. These are cascaded shift registers and the timing is quite close. The R108 resistor change slows down the data line going to U104, making the timing less critical. (1 Dec 99)
5 - C71 must be NPO temperature stable type. See Parts Information. (5 Nov 00)

IMPROVEMENTS:
1 - R16 can be changed from 10K to 6.8K. This centers the output Voltage on U10A better. I am unaware that the old value had actually caused any clipping on U10A output, though. (1 Dec 99)
2 - U15 Mixer Grounding can cause major degradation in sensitivity. It is very important that this is dealt with! All the information is on U15 modification page. The modification is simple but the improvement can be big. (21 Dec 01)

IMPORTANT The silk-screen legend is backwards for the 78L05 regulators, U13 and U103. The flat sides of both TO-92 packages should be towards the top of the board. There is no problem in installing the parts, of course. The sketch of the package included with the board is correct. Thanks to KD7TS for catching this. (Added 26 Oct 99)

Note 8 of the Assembly Notes says that all chip caps can be 1206 size. This is NOT true for the 0.01 uF and 470 pF values. These should be 0805 size. Also the 0.01 uF part number in the 4.1 addendum to the Assembly Notes is for the larger size. See the Part Information page for more information and supplier part numbers.

"SR105" Diode When the DSP program is first started and before the PC program is run, the DSP-10 can come up with BOTH transmit and receive turned on. This sets up a loop that oscillates at full power! Very bad. My DSP-10 shows about +16 or +17 dBm out. The happens because the shift registers (U107, U108) come up uninitialized. The eventual solution will be to program an EPROM that includes the setting of the shift registers at DSP startup. Here is a work-around until you have an EPROM version: Acquire a Schottky diode and solder the cathode (bar end) to the junction of R132 and Q104-collector. Solder the anode (non-bar) end to the junction of R130 & R131 (base of Q103). It must be a low barrier voltage diode like the Schottky, but it can be either a rectifier or small-signal type. I used a SR105 from Digi-Key, but I notice that Radio Shack has 1N5817-1N5819 or 1N5820-1N5822 on special order status. W7LHL suggested that a 1N34 type of germanium could work also. This work-around prevents the 'transmit' from being enabled when 'receive' is enabled. (Added 5 Sept 01, based on 27 Jan 00 reflector posting.)

DIAGNOSTIC WEB PAGES: Three diagnostic pages that will help to find problems with the boards are now available. For general verification, the Tips and Measurements page. has various tips, lists scope measurements and shows screen pictures under specified conditions. For the Serial connections between the EZ-Kit and the main board there is a diagnostic program and other information at the Serial Diagnostics page. Various hints, a second diagnostic program and waveform photos for the Phase-locked Loops are at the PLL Diagnostics page. Also, do not forget to check the archives that go with the email reflector, listed above. They can be accessed from the qth.net Web Site. From "Select List," choose "DSP-10." Then click on "Archives." (Updated 28 May 00)

HINTS:The Hints Page has a collection of tips that may help in the assembly, test and use of your DSP-10.

DSP-10 & Frequency Reference Info: Along with building their DSP-10's, W7CQ and KO7N have been busy with the construction of great looking 10 MHz standards, using the Brooks Shera controller board. Take a look at their DSP-10 web site. This links to the info on their frequency standards as well as having additional material on the DSP-10.

To assist in the assembly of the main PC Board, there is an Assembly Part List that has a separate line for each part on the board. The value and location of each part is listed in inches and mm. This is an assembly and test aid and not for purchasing parts. Use the 'Part Buying Information' below for this. (Added 24 Oct 99.)

All of the part buying information is being moved to a single page of Part Buying Information. I will try to indicate the update status of the part page here. To that end, note that the Spread Sheet of buying information, put together by Rich, W5RXP has been updated, as of 28 Oct 99. Also more part source links and information have been listed and the 0805 chip correction, listed above, was added.


THE INSIDES

DSP-10 Hardware (1/4 size)

For those that don't have the QST articles, here is a peek inside a DSP-10. This one was carefully crafted by Ernie Manly, W7LHL, and is being used on the air. The brass covered box holds the EZ-Kit Lite DSP. The main PC board is at the bottom of the box.
Click on the image for a full size screen shot (W7SLB photo).

To aid with assembly, W7SLB took detailed pictures of his assembled PC board. Ignore the MOSFETs, Q105 and Q106, in the lower right picture, as these are not the specified package. The board has been divided into four parts. Click on any of the little pictures for the corner that you want to see:

PCB-Upper Left, Click for Bigger PCB-Upper Right, Click for Bigger
PCB-Lower Left, Click for Bigger PCB-Lower Right, Click for Bigger


HARDWARE AND SOFTWARE AVAILABILITY

As indicated above, be sure to check Part Buying Information. for updates.

PART KITS: As was noted above, information on kits is available from TAPR web site.

PC BOARDS: You can Order PC Boards from Mashell Electric (W7SLB.) The boards are double-sided with plated-through holes. They have a solder mask and silk screen legends on both sides. Included is a set of drawings. See the Web link for more details.

Assembly Notes, Rev 4.5, are available for the hardware. These will be updated from time-to-time so watch this page. Revision information at the bottom of the Assembly Notes indicates all changes. Note: Rev 4.5 has new board assembly information. (Updated 4 Dec 99.)

README FILE ONLY: If you want to read more, but not download the programs at this time, take a look at V2.0 "read me" notes along with V1.6 "read me" notes.

DSP-10 DSP INTERFACE To assist those "talking" with the DSP unit, there is now a summary of the interface. This has all the commands and the status definitions and is here for download (20K text).

PUA43 SPEC To define the PUA43 mode, there is now a specification available for download (6K zipped).

Ver 2.0 SOURCE CODE: You can download the 'Free Software' for the PC software (Borland C) (289K zipped) and the DSP software source (82K Zipped). This second item is written in Analog Devices assembly language and can be assembled into UHF3.EXE, Rev 2.0 with the tools supplied with the EZ-Kit. (8 May 01)

Note: If you downloaded the DSP software source befor 5 Sept 01, it was missing the file HP10FIR.DAT, which is available here. Later downloads include the file. (Added 5 Sept 01.)

Ver 2.1 HEX CODE: This is the Intel Hex file for a version of the DSP program UHF3.EXE that also includes a monitor program for the EZKIT-Lite. It will run the DSP-10 at startup, but will also respond to a monitor wakeup call. Details are in the text file that comes with the Intel Hex file. The latter is ready to program an 27C512 EPROM that can then be plugged into the EZKIT-Lite, replacing the PROM that came with the board. Download the zipped file here. (Added 18 March 02.)

EZFAST Shareware: For ver 2.0 of the DSP program, you can use the loader provided by Analog Devices with the EZ-KIT Lite. Much faster is Dwight Elvey's shareware EZFAST.COM, but you must have the current Ver 1.03. It is available for download. Be sure to study the EZFAST.DOC readme file that is with it (it is a pure text document). Big mistake: Until 13 Sept 01 this was the wrong version of EZFast. Check your version by typing a DOS command 'EZFAST' and see if it signs on as 'EZFAST ver1.03x'. If not, get the download file here. Sorry for the inconvenience and thanks to KD7TS for catching it. Update 9 Feb 01 and 13 Sept 01.


SCREEN SHOT

Screen Image of DSP-10 Transceiver (1/2 size)

As a sample of the front panel, here is a shot by W7SZ of the full transceiver, using 2.0 software to receive W7PUA on 10 GHz. The path is about 100 miles, home station-to-home station, with a number of hills in the way. The signal was being scattered by water/ice or irregularities in the troposphere that produced about 45 Hertz of Doppler as well as spectral spreading. The sky was hazy with no rain in the path. At the bottom of the waterfall is a PUA43 message, producing the text at the top. The upper portion of the waterfall is at a 1200 Hz scale and the PUA43 portion is at a 4800 Hz width. Click on the image for a full size screen shot.
For further information on DSP-based weak signal detection, look at Rein Smit W6/PA0ZN's and Mike Cook, AF9Y's web pages.



AUDIO PROCESSOR

NEW VERSION! The Audio Processor configuration continues to be available in Ver2.0 of the DSP-10 software. It is now restricted to reception only. This is a good general purpose audio processor with all the display of a built-in spectrum analyzer. In addition, the input can be tuned up to at least 20 kHz and serves as a general purpose test instrument. The relative dB calibration is excellent. See the Version 2 information above for downloading information.

The audio processor requires an EZ-KIT Lite, but no other hardware outside of a PC.

IMPORTANT NOTE: Users of the Audio Processor must have the configuration variable 'hardware' set to 0. This is done by a line 'hardware 0' anywhere in the configuration file UHFA.CFG. The file can be edited with any text editor such as 'Note Pad' and the existing 'hardware 1' line should be changed. The issued file is for the full rf processor and has 'hardware 1'. The undesired symptom is the screen showing 'TRANSMIT' in the upper left and no updating of the spectral display. The next revision of the README16 file will cover this. 29 Sept 99, tnx to K3IB for pointing this out.

Screen Image of Audio Processor (1/4 size)

Here is a sample of the front panel with the audio processor (running the Ver 1.6 software).
Click on the image for a full sized screen shot.



Back to Home | Send me email

This page was last updated 28 April 02, RSL

As an aid in searching: ABCDSP10ABCD