Averaging not as good

  • 4
  • Question
  • Updated 5 years ago
I know there was some work done on this but averaging does not seem to work nearly as well as the last version.
Photo of Greg

Greg

  • 656 Posts
  • 63 Reply Likes

Posted 5 years ago

  • 4
Photo of Michael - N5TGL

Michael - N5TGL

  • 308 Posts
  • 101 Reply Likes
Yep, that was one of the very first things I posted about on the 1.1 release:

https://community.flexradio.com/flexradio/topics/new_averaging_method_doesnt_feel_like_an_improvement

I really don't like it.  Reminds me of a really old peaking spectrum analyzer.  Some folks may like it better, that's fine..but make the averaging method switch selectable or give us access to the attack/decay settings.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1848 Posts
  • 670 Reply Likes
Averaging method and settings per Panadapter
I think it would be really nice to control the averaging parameters on a "per panadapter" basis.  That way we could do a side by side comparison of the averaging method/settings to decide which we like best.  You might decide to use both methods (old and new) or have different settings for a given method  for different modes, bands, or conditions. 

The new method (V1.1) seems better in some cases with weak signals and the old method (V1.0.24) seems better in general use.   But I'm not even sure of that since it's not possible to do a real time comparison.  

Providing options is more work but probably the only way to make everyone happy.  The questions are:
- how much time will it take
- is it worth the effort (i.e. how many PLUSes did the idea get?)
- where does it fall on the priorities list. 

Greg, Michael and others -- What do you think? Is it worth the effort?

Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com







  


Photo of Michael - N5TGL

Michael - N5TGL

  • 308 Posts
  • 101 Reply Likes
Well, it sounds to me like the change was very simple...just an adjustment of the attack/decay settings for the averaging function.  Ideally, I'd like to see those settings exposed to the end user so they can tailor the response to what they would like to see.

However, I'd be happy with a switch selection to go back to the 1.0.24 and the 1.1 method if I had no other choice.


Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9186 Posts
  • 3541 Reply Likes
This is an idea reply not really an answer to the original problem topic Please reference the new topic here: Averaging method and settings per Panadapter
Photo of Paul Christensen, W9AC

Paul Christensen, W9AC, Elmer

  • 323 Posts
  • 138 Reply Likes
One test is to look at peak signals while sliding the AVG bar from left to right.  Under v.1.1, peaks remain constant.  Under v.1.0.24, peak readings change by more than 5 dB.  While the reduction in the peak attack time may seem more visually pleasing to some, it comes with inaccuracy.  Of course, it would be nice to have the option of switching between the old and new response times to satisfy everyone.

Paul, W9AC



      

Photo of Michael - N5TGL

Michael - N5TGL

  • 308 Posts
  • 101 Reply Likes
Well, if you're averaging, that does come at the expense of peaks...that's kinda what averaging does.  :)  The averaging action of 1.0.24 was quite strong, I rarely ran it more than 11.  However, the 1.1 version requires quite a bit more, I run it at 95, and it still doesn't give me what I'm looking for.  The peaks don't matter to me as much as being able to see that there's something there.  Can't wait for panafall!

What was done in 1.0.24 was very similar to PSDR, which was what I was really after, and now no longer have...
Photo of Paul Christensen, W9AC

Paul Christensen, W9AC, Elmer

  • 323 Posts
  • 138 Reply Likes
An example where the current display is superior is when Morse CW signal strength is measured against an adjacent steady CW carrier. Under 1.0.24, equal real signal peaks would show the active Morse CW several dB below the carrier value as the AVG slider was turned up.

Not to beat a dead animal here, but other SDR programs (SpectraVue for example) do a good job of retaining the peak value under averaging while keeping the display motion very fluid. Hopefully, a choice will be offered in the future.
Photo of Michael - N5TGL

Michael - N5TGL

  • 308 Posts
  • 101 Reply Likes
Paul, yep, that absolutely makes sense, but since I'm not a CW op (yet) that has less benefit to me.  I can see how you would find it useful.
Photo of Greg

Greg

  • 656 Posts
  • 63 Reply Likes
I don't even think its a toss up.  The previous version was just better IMO.  Was there a need for this change for some other improvement?

Greg

Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 4189 Posts
  • 1334 Reply Likes
I guess it depends upon which parameter you want to average.  Are you wanting to average the total signal?  The attack times?  The decay times?  Just smooth out the total display?  I see about four different parameters that can be adjusted.  Attack time, hold, decay time, and time frame to average, perhaps more.... Frankly, I like the new display, once I got adjusted to it.  But I would support porting these parameters to some sort of "properties" tab, along with adjustments for many other functions, such as NB, NR, ANF, APF, etc.

Eventually, I am sure that they will do this.  The first task is to get these functions functional.  Then add the bells and whistles and tweak-ability.
Photo of Ned K1NJ

Ned K1NJ

  • 314 Posts
  • 81 Reply Likes
Yes, the signal does get averaged down depending on settings, but so does the noise.
 I'm after S+N/N more than accuracy.  This new technique is a good piece of work, for
sure, but I prefer the previous iteration.  Cultural noise impulses really hit the panadapter
pretty hard too.  Distant lightning might make things pretty jumpy.

Ned,  K1NJ
 
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1848 Posts
  • 670 Reply Likes
Note: since this posting was originally a QUESTION, Tim created a separate IDEA posting to add an option to pick either the NEW or OLD method and to adjust the parameters

If you would like to see this added, go to this link below and add you PLUS VOTE and comments.

https://community.flexradio.com/flexradio/topics/averaging_method_and_settings_per_panadapter

Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com