JT9 - WOW! Very sensitive indeed!

Discussion in 'Working Different Modes' started by W7UUU, Aug 18, 2014.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: abrind-2
ad: L-MFJ
ad: L-Geochron
ad: Left-3
  1. AA9G

    AA9G Ham Member QRZ Page

    Yesterday in searching for a homebrew tube tester write up I came across THIS LINK and then in reading it I wondered...are you using deep search or normal? I don't even know what to think about this fellows assertions other than they make at least superficial sense to me. I have used deep search myself most of the time and now I'm just...WTF...Makes me want to erase those contacts from my log and start over using only normal. Or expedite my CW learning.
  2. W7UUU

    W7UUU Principal Moderator Lifetime Member 133 Administrator Volunteer Moderator Platinum Subscriber Life Member QRZ Page

    What a horrible website! (Too many ham radio sites like that I'm afraid!)

    But I get what they're saying - that "deep search" may be injecting data from a database and
    only decoding 2 actual characters?

    However, that is an OLD website! Even the revision page is 5 and half years ago, the original
    data is from 2006. I'd have to see some more modern (and readable!) corroboration before
    I'd change how I do things. Something about what they're suggesting doesn't sound right.

    Need to "Hit the Googs" and see if there's any more recent scrutiny of how DS works in JT9

    Thanks for posting this up

  3. AA9G

    AA9G Ham Member QRZ Page

    Ok. I look forward to hearing your results. Since I am using WSJT-X...and it is pretty new. I don't know how it might have changed...or not.
  4. N2ADV

    N2ADV XML Subscriber QRZ Page

    Thanks for the contact on 20, Dave. :)

    Were you using your handy-dandy new beam?
  5. W7UUU

    W7UUU Principal Moderator Lifetime Member 133 Administrator Volunteer Moderator Platinum Subscriber Life Member QRZ Page


    And No! 2 watts into the G5RG. Just got home from a week in NC so will
    be routing cables all afternoon and hope to have the beams (!!) on the
    air tonight and tomorrow - just a tease though, as we leave on vacation
    on Sunday for a week... ARG! But I need a vacation so that's good.

    Oh, and just cuz I can, I emailed Joe Taylor to ask about the Deep Search
    thing and is it really legit - who knows? Maybe he'll reply back!! I'm sure
    he gets 10,000 emails a day but ya never know.

  6. N2ADV

    N2ADV XML Subscriber QRZ Page

    In the current version of WSJT-X, there are 3 different levels of decode: Fast, Normal, and Deepest. Each one takes a bit more processing power but as you get progressively "deeper" you get more accurate decoding in more adverse conditions. It's under the "Decode" section in the menu at the top of the main WSJT-X window. If you get a response, I'd love to know about how the "nuts and bolts" of this work!

    In the old JT65HF program, you could tell when the program used the deepest level of decode because you would see a "K" to denote the level of decode used.
  7. W7UUU

    W7UUU Principal Moderator Lifetime Member 133 Administrator Volunteer Moderator Platinum Subscriber Life Member QRZ Page

    Understood. But if you go back and read that horrid web site, what's being asserted in many
    blogs and sites is that in "Deep Search" the software only actually decodes 2 characters and
    using a sophisticated algorithm and a database of every callsign ever reported via PSK Reporter
    is being scanned - and the "closest match" is what gets reported and "decoded". The claim is
    that it's entirely possible, if this is true, that you grant a QSL to a station you never actually worked!

    The "Normal" mode reportedly doesn't reference the callsign database

    Just scuttlebut hearsay - I've not found any hard evidence to corroborate this claim and based on
    the convoluted poorly written and confusing web sites I've found on the topic, I'm pretty suspect

  8. K4VLF

    K4VLF Ham Member QRZ Page

    I know you guys are anxious to hear the follow up to my saga ... I used a faster laptop and now all is well.
  9. K0RGR

    K0RGR Premium Subscriber QRZ Page

    If in doubt, disconnect from the Internet and see how it works! I've never noticed any advantage in being online. On HF we also don't have the predetermined exchanges used for moonbounce, we often switch things up, so it's fairly easy to see what's being decoded or not.

    As Joe Taylor explains here, the DS function does not result in false positives, it helps reduce false negatives - http://physics.princeton.edu/pulsar/K1JT/DS_Q+A_SV1BTR.htm
  10. W7UUU

    W7UUU Principal Moderator Lifetime Member 133 Administrator Volunteer Moderator Platinum Subscriber Life Member QRZ Page

    OK - here's the straight skinny from the source (no, not JT himself but another ham who was part of the initial team).
    Take these comments in that light - but I have pretty good confidence.

    Early on, the reporting network was saved as a database into the JT65 EME software - and early experiments
    were done whereby 2 characters of an incoming signal would be compared to any other signals having been heard
    and reported - so a simple search of the CALL.txt file could locate the proper callsign based on the grid square and
    previous activity as reported and stored in the database.

    That was abandoned early on - but the software still needs to see the CALL.txt file present but it's a null file (I didn't
    check this myself) - the CW crowd was most vocal in calling JT modes "cheating" because of this, and that's a big
    part of why it was dropped.

    All versions of current JT software do NOT reference a database - and "Normal" is exactly the same as "deep" - it
    was just easier to drop the function and not have to redo the interface.

    Just the messenger, and hopefully this a correct relay of the info that was sent to me. If the fellow who sent it
    reads this and either wants to comment/correct, or allow me to post the email, just let me know

    But I think I've summed it up correctly.

    Bottom line: NO CHEATING happens in any JT mode

Thread Status:
Not open for further replies.

Share This Page