Open Source WSJT Joe Taylor K1JT 12 th International EME Conference Würzburg August 25-27 2006
Open Source WSJT I.Motivation, brief history II.Current status, usage III.Some EME “politics” IV.Future developments V.Hardware demonstration
I. Motivation for WSJT Study coding and modulation schemes Study coding and modulation schemes Enable VHF/UHF QSOs at minimum S/N Enable VHF/UHF QSOs at minimum S/N Propagation modes: MS, EME, … Propagation modes: MS, EME, … Make a convenient, enjoyable, rewarding- to-use program Make a convenient, enjoyable, rewarding- to-use program Establish ultimate limits for weak-signal, real-time QSOs. Establish ultimate limits for weak-signal, real-time QSOs.
EME: Design Goals for JT65 Work at lowest possible signal levels Work at lowest possible signal levels Conform to standard operating practices Conform to standard operating practices Exchange all info required for valid QSOs Exchange all info required for valid QSOs Good QSO rate for contests, DXpeditions Good QSO rate for contests, DXpeditions Very high confidence in copied information Very high confidence in copied information Extremely low probability of false QSOs Extremely low probability of false QSOs
WSJT Milestones 2001: FSK441 for meteor scatter 2002: JT44 for EME 2002: JT6M for MS, ionoscatter on 6 m 2003: JT65 for EME, with FEC 2005: JT65 “Deep Search” decoder; Open Source release 2006: Linux and FreeBSD versions; many algorithmic improvements
II. WSJT Status: August 2006 Total users: >3000 Total users: >3000 Nearly all MS uses FSK441 Nearly all MS uses FSK441 JT6M popular on 6m JT6M popular on 6m JT65 EME users: 300 – 600 JT65 EME users: 300 – 600 JT65 EME QSOs: >40,000 JT65 EME QSOs: >40,000 30% of entries in 2005 ARRL EME contest used JT65 (On 2m, >70%) 30% of entries in 2005 ARRL EME contest used JT65 (On 2m, >70%)
JT65 EME, Band-by-Band 6 m: QSOs made almost daily 2 m: 1-yagi to 2-yagi QSOs common 70 cm: usage slowly increasing ? 23 cm: usage increasing With 2.3–3m TVRO dishes, 10 Watts is enough!
First and Last G4CCH transmissions VK7MO G4CCH IO93 OOO CW NOW?
JT65 DXpeditions 1.VK7MO: VK9XMO, VK9CMO 2.3Y0X, Peter I 3.ZL2RS: to many places 4.ON4IQ to HI3, FS, J4, … Single yagi or small, transportable array with modest power is enough on 2 m and 70 cm -- and even 23 cm, for working larger stations
Informal Survey (July 2006) 1.How many EME QSOs have you made using JT65? On what bands? Using JT65A, B, or C? 2.How many QSOs that you originally considered complete, turned out to be invalid? For what reasons? 3.Has anyone claimed to work you in a JT65 EME QSO that did not occur, or was not valid? 4.Are the fixed JT65 messages generally reliable, and the formats useful? 5.Do you make scheduled, or random QSOs? Call CQ? Answer CQs? Self-spot when calling CQ? Tailend other QSOs?
Informal Survey, cont’d 1. 1.QSOs by band (round numbers) 6m: 200 2m: 11,800 70 cm: 400 23 cm: 100 Total reported QSOs: 12,500 2. 2.Bad QSOs considered good by reporting station: 16* 3. 3.Bad QSOs considered good by other station: 12* 4. 4.Message structure, reliability: very good 5. 5.Typical mix: 50% random, 50% skeds *Bad QSOs mostly results of operator errors 50 responses
JT65: Differences from CW Structured messages Error-correcting code Synchronized transmissions Transmissions can be averaged Copy is “all or nothing” Roughly 10 dB advantage
JT65 Decoding Thresholds Message type KV (dB) KV Avg (dB) DS (dB) Short65 (dB) Arbitrary–24–28 BC + Grid/Rpt–24–28–28* Shorthand–32 Sync limit: –30 dB *Callsign must be in list
III. EME Politics On a web site: “… only two characters need to be decoded to print full EME messages…” Editorial in a respected journal: “… the JT65 mode when using the Deep Search Decoder … needs to ‘recognize’ just only 2 letters but displays a full callsign.” Another web site: “All JT65 QSOs made using the Deep-Search Routine are not complying with long established EME QSO guidelines and are not eligible [for this Top List] due to partial copy.” ✻✻✻ No Nein Non Нет… ✻✻✻ See “How Many Bits” paper near end of Proceedings ! Try the JT65 Hardware Demonstration, yourself !
IV. Future WSJT plans? Pulsar mode ? Enhanced meteor scatter mode ? Connection to Linrad / Winrad ? New soft-decision RS decoder EME Echo and Measure modes 30-bit JT65 messages Expanded waterfall: 0 – 5 kHz
30-bit Messages Bits 1. 1.CQ de 3Y0X 30 2. 2. de K2TXB 30 3. 3.K2TXB 3Y0X OOO 72 4. 4. 3Y0X K2TXB RO 72 5. 5.TXB de 0X RRR 30 6. 6. 73 de K2TXB 30 Notes: No grid locators Reports and Rs tagged with callsigns Numerical reports optional Messages 5 and 6 can be shorthands No need for a callsign list
Programmer’s Information GNU General Public License (GPL) Languages: Python for user interface Fortran for number crunching C for A/D, D/A, PTT, … Compile on Windows, Linux, … More details in Proceedings New contributors welcome ! http://developer.berlios.de/projects/wsjt firstname.lastname@example.org
V. Hardware Demonstration Generate JT65 (or CW) at any desired S/N Audio cables A B 1. 1.A makes “QSOs” with B ? 2. 2.Send test messages 3. 3.Listen to the signals 4. 4.Stress-test the decoders 5. 5.Play as you wish … Headphones
Hardware Demo (cont’d) 1. 1.Run WSJT on both computers, make “QSOs” Generated S/N in dB (2500 Hz BW) Format: #─26
Hardware Demo (cont’d) 2. Run WSJT on A, SimJT on B; “torture test” the decoders!
Hardware Demo (cont’d) 3. Do the WSJT tutorial using recorded EME signals