Wednesday, 1 February 2017

Impressive JTDX performance

I got a e-mail from Igor UA3DJY today. Followed by a short e-mail discussion. He did read my blogpost and asked me to evaluate 2 WAV files to show what the AGCc is actually doing. Besides that he explained the hint decoder to me which I will save for another post. I downloaded and installed the latest version of JTDX (17.5.1) and did some tests which were impressive, really impressive. Remembering the comment from Paul PC4T that he had better results with WSJT-X receiving the same time with both JTDX as WSJT-X. So I did some tests with WSJT-X to compare receive capability with the same WAV files.

First wav file:


Above the screenshot from WSJT-X 1.7. No more, no less. 3 decodes. I tried other input levels and repeated the pass, still I got the same 3 decodes.

Above the screenshot from JTDX 17.5.1. As JTDX has more possebilities I used them in the following order (clicked decode after switching filters): Without any filter 3 decodes but different from WSJT-X.  I switched on the AGCc and decoded 10 stations including those that were decoded with WSJT-X. 3rd pass AGCc still on and SWL on no difference and still 10 decodes. With AGCc, SWL and hint on still no difference. As you can see the strong signal will push away any other weak signal from decoding, with AGC compensation those will be decoded.

Second wav file: 

Above the screenshot from WSJT-X, look at the waterfall with many signals near to each other. As with the first wav file I tried several times with different input levels. Still the best it could do was 3 decodes.

Above the screenshot from JTDX 17.5.1. 11 decodes without any filtering, isn't it amazing?
Wait, now comes the impressive AGCc on and decoded 14 stations. AGCc and SWL on gives still 14 decodes. AGCc, SWL and hint on.....15 decodes, note M6YPZ the last decode has a star behind it as a sign it was spotted with hint decoder on.

What AGCc does is to compensate the loss of weak signals when a strong signal is received. Any receiver with AGC will reduce gain to prevent strong signal(s) will make signal distortion. Actually most modern receivers will do that. I think this is especially useful with strong ES propagation on 10m in summer.

2-2-2017 Supplement from Igor UA3DJY: 
In fact AGCc function trying to reduce noise level at beginning and end of the interval so that it is equal to the noise at the signal reception.
It is needed to provide correct operation of the correlation function being used in WSJT-X/JTDX. 
While AGC being triggered very often on the HF bands AGCc functionality lets to decode more signals.

When I tried this compensation only for JT65 signals I had got a bit better results, but later I have had to move this functionality upstairs in the source code to cover both JT65 and JT9 signal reception.

73 Igor


Compare VK7XX -16dB in WSJT-X and -15dB in JTDX, LU8HGI -20dB in WSJT-X and -22 in JTDX, IT9CCB -15dB in WSJT-X and -14 in JTDX. Overall signal reports are close, but how important is a signal report if you don't receive anything at all?

My conclusion: the AGC compensation works pretty well. But of course to show that it is working you need at least 1 very strong signal on the frequency.

I hope this is a objective test. If any reader is not shure about my evaluation capabilities let me know. If you want to test the wav files yourself I would be happy to send them to you for a second opinion so to say.


9 comments:

  1. Very interesting finding I'll have to try JTDX. I'll look to see if there is a download for Mac.

    ReplyDelete
  2. I tried the previous version of JTDX last week but kept getting an error after the first set of decodes. I'll try this new version and contact Igor if the error persists.

    73 - John AE5X

    ReplyDelete
    Replies
    1. Hello John, this error has to do with the file CALL3.TXT that is in the package. In the latest version you will get a text in the rx window that you need to place the latest CALL3.TXT in the same directory were the log is. Usually the path looks something like users\name of user\appdata\local\JTDX. Of course you can look for CALL3.TXT as it is placed in the right directory but is empty. I hope this helps. Good luck. 73, Bas

      Delete
    2. That's exactly what the error was - thanks! I'll give it a try this weekend now that I know how to fix it.

      Delete
  3. Hallo Bas, jij hebt je vergelijking gemaakt met JT65 neem ik aan? Ik namelijk alleen met JT9. Nu zag ik in de handleiding staan: "decoding efficienсy of JT9 mode is the same as in WSJT-X" Misschien verklaart dat het verschil, of juist niet ;-) Ik moet dus de vergelijking zoeken met JT65A. 73 Paul PC4T

    ReplyDelete
    Replies
    1. Hallo Paul, JT9 is nog niet geoptimaliseerd. Dat komt misschien in de toekomst. Neem in gedachten dat uiteindelijk de decoder hetzelfde is. JTDX kan daarnaast heel goed omgaan met een drukke band zoals 40m JT65. Dus signalen die over elkaar heen zitten e.d. zijn geen probleem. Voor WSJT-X dus wel. 73, Bas

      Delete
    2. Hallo Bas, dat zou dan mooi zijn want die JT65 freq zitten aardig vol. 73 Paul

      Delete

Thanks for your comment. Bedankt voor je reactie. 73, Bas