Filter Skirt Issue 2.3.7.10 with WSJT-X

  • 1
  • Question
  • Updated 3 months ago
  • Answered
Running a 6400 with the new 2.3.7.10 software. In WSJT-X I am seeing distortion at the end of my filter skirts in the WSJT waterfall. I do NOT see the same thing in the Flex waterfall. I did not see this issue in WSJT with previous SmartSDR version?  My DAX settings are the same as the previous version of SmartSDR. Is anyone else seeing this behavior? It is hindering reception on the filter ends. Filter is set to auto or third position in digital mode. See screenshots below taken at 50.312 then 50.313 a few minutes later. It is not band conditions.

 TIA! Dan KG0AQ 73


Photo of Dan KG0AQ

Dan KG0AQ

  • 28 Posts
  • 1 Reply Like

Posted 3 months ago

  • 1
Photo of Paul W5PF

Paul W5PF

  • 47 Posts
  • 4 Reply Likes
I believe this is a function of the WSJT-X Wide Graph. If you set your filter for 5K the light blue on the laft and the yellow on the right will go away and your background will be uniform.

73,
Paul W5PF
Photo of John - AI4FR

John - AI4FR

  • 54 Posts
  • 16 Reply Likes
I see this occasionally with my 6700. My fix is to reload the profile and all is well again. Happened with the older version as well.
Photo of Dan KG0AQ

Dan KG0AQ

  • 28 Posts
  • 1 Reply Like
Yes, I can get rid of the right side trash/hash by widening the filter. But the little hash on the left is the real problem. Especially with KH1/KH7Z transmitting at 303hz. Any ideas on widing the left side? Can't seem to get past zero with DIGU.  Thanks for tip!
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3467 Reply Likes
Go into the filter setup (Settings->Radio Setup->Filters) and set your DIGx filters for lowest latency, not auto.

AAnds make sure the slice RX filter is equal to or wider than the WSJT-X waterfall (check the filter skirt settings, not just the size)
Photo of John - AI4FR

John - AI4FR

  • 54 Posts
  • 16 Reply Likes
Excluding 60m, go lower in frequency with the VFO.
Photo of John - AI4FR

John - AI4FR

  • 54 Posts
  • 16 Reply Likes
Oh, BTW the waterfall on WSJT-X has no bearing on decoding signals.
Photo of Doug Hall

Doug Hall

  • 187 Posts
  • 54 Reply Likes
It absolutely does. The bins/pixel setting controls the range of frequencies you will decode. You won't decode any signals higher than the highest frequency you can see in the waterfall. Don't believe it? Set bins/pixel to 1 and try it.
Photo of Jay / NO5J

Jay / NO5J

  • 1514 Posts
  • 222 Reply Likes
Doug

Interesting

Agreed, signals not seen in the wide graph don't decode. 

Somewhere I got the impression that the 40m FT8 portion began at 7.074 MHz and extended 2K above that. A setting of bins/pixel to 1,  Start 0Hz, limits the widegraph to a maximum of 1.45K, I can't widen the widegraph any wider than that. So bins/pixel 1, excludes decoding signals from 1.45K and above.

In order to decode the entire 2K FT8 portion bin/pixel 1, Start 0 Hz, just won't work for me.

Setting bin/pixel 2, Start 0 Hz, with the widegraph widened out to it's maximum gives me signals from 7.074 MHz to 7.077 MHz which is plenty wide, but the widegraph window is wider than I would normally use it.  But since I often see FT8 being worked above the FT8 portion in the JT-65 portion I wont be missing those outliers.

My point here is that I had been reducing the widegraph width to only display about 2K total, so I was missing any outliers above 7.076 MHz, I didn't know that the widegraph displays width, excluded decodes.

Now I do.

Now I think ...

Bin/pixel 4 Start 0 Hz, for me, is about optimum, for decoding FT8.

I use JTAlertX, and it's window size isn't adjustable in width, I then adjust both the WSJT-X, and widegraph windows to it's width which takes up approximately half the width of my desktop display. which gives me a widegraph display from 7.074 MHz to 7.0768 MHz, which combined with a 0 Hz to 3000 Hz DIGU filter width, will allow me to decode some of the outliers, and also move the filter edge hash out of the widegraph window.

This is an improvement.

Thanks!

#FlexRadio IRC chat

   73, Jay - NO5J

Photo of Al K0VM

Al K0VM, Elmer

  • 581 Posts
  • 94 Reply Likes
DO you see it if you turn off the wide graph 'flatten' function.. ?
AL, K0VM
Photo of John - AI4FR

John - AI4FR

  • 54 Posts
  • 16 Reply Likes
Turn down the RX audio in DAX. BTW, that level meter on the left hand side of WSJT-x doesn't do much for decodes either. There is no perfect spot such as a level of 25 is better than a level of 30. Keep it in the green. If you have concerns about the waterfall and reception/decodes, turn the waterfall off and watch what happens. I'd bet WSJT-x still decodes just fine with no waterfall. Good luck with it.
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3467 Reply Likes
Adjust your audio levels.
Photo of Jay / NO5J

Jay / NO5J

  • 1514 Posts
  • 222 Reply Likes
Dan
If you need help with WSJT-X adjustments, I'm fairly experienced with it, and would be available in the #FlexRadio IRC chat channel. It might be quicker, if we discuss it in realtime. Anyone else interested is welcome too. 

#FlexRadio IRC chat

   73, Jay - NO5J

Photo of Al K0VM

Al K0VM, Elmer

  • 581 Posts
  • 94 Reply Likes
With Flatten off, you will need to adjust the gain/zero sliders to the right of the flatten button in wide graph..  The down side is that when the ssdr noise floor/agc-t etc. change, the sliders will need to be adjusted again.. .. But it will show the 'real' levels in the wide graph, not what flatten chooses to make them..  I normally prefer to use flattens despite the fringes that flatten creates..
AL, K0VM
Photo of Dan KG0AQ

Dan KG0AQ

  • 28 Posts
  • 1 Reply Like
I had DAX receive at 32. Now with flatten off I'm at 13. Very touchy without flatten though. I get a black & blue background at 13 and at 14 I get a blue background with  hash. Sigs on WSJT meter are around 25 to 30. I might give you a call later IRC Jay. Thanks,

See new screenshot without flatten and DAX at 13

Photo of Paul W5PF

Paul W5PF

  • 47 Posts
  • 4 Reply Likes
Dan,

That little bit of hash goes away for me when I use the 5K filter. I don't think there is really any signal there that would interfere with the KH1 below 500 hz. You might pose the question on the wsjtgroup reflector on yahoo groups. Lot's of help there.

Paul
Photo of Jim  KJ7S

Jim KJ7S

  • 130 Posts
  • 22 Reply Likes
My experience has been that with the hash at either end of the visible spread, has been on the laptop but not the main desktop, so perhaps something in the machine is generating or letting in the hash.
Photo of Dan KG0AQ

Dan KG0AQ

  • 28 Posts
  • 1 Reply Like
Thanks to all that chimed in with suggestions. I have now changed a few settings and all is good.
Check out the cleaner waterfall and traffic on 6.

1. Widened filter to 5K
2. Cut my DAX gain down from 32 to 18
3. I went back to flatten because without it the gain requires periodic adjustment

73 Dan KG0AQ and hope to catch you on FT8 sometime. This is truly a knowledge base!

OBTW  I received my return shipping label for the PEN and Preamp RMDR enhancement
Thanks Flex,

Photo of John - AI4FR

John - AI4FR

  • 54 Posts
  • 16 Reply Likes
Glad you got it going. I use flatten as well, with 3khz filter but cut in from 0 100hz as there is never any signals down there. So in short, it is a 2.9Khz filter. Instructions state to start at 200hz which is probably better than my 100hz starting point.
W5PF's advice above is excellent. The WSJT reflector has covered the waterfall issue several times in the past discussing what it does and even having the designers chime in with clarification. As Paul stated, lots of good info there.

This conversation is no longer open for comments or replies.