Is ANF Working Better in V3 ?

  • 1
  • Question
  • Updated 6 months ago
I've gotta feeling I'm in for it by posting this but I've just got to ask if other folks are seeing the following.

Since purchasing my 6600 several months ago, I've gone from thinking the ANF worked OK to being pretty unhappy with it.  I've found it intermittent at best and sometimes totally ineffective.  I read the manual about it and have paid attention to the "amount" setting but finally concluded it's pretty useless and rarely even bother with it.  I've read several others on the forum who feel about the same way.

I continued to ignore it after upgrading to V3 because Flex didn't list it as an item they'd worked on.  But long story short, I've just started noticing it actually seems to work pretty well now?  Is anyone else seeing that?  Is it possible Flex actually fixed it but hasn't especially advertised it for some reason?  Since it was intermittent in the first place I could just be getting "lucky" I guess but I've been trying it all day and it seems to work now ...

Photo of Russ Ravella

Russ Ravella

  • 50 Posts
  • 23 Reply Likes
  • curious

Posted 6 months ago

  • 1
Photo of Jim Gilliam

Jim Gilliam

  • 944 Posts
  • 218 Reply Likes

The ANF has always worked for me when no one is speaking. As soon as someone speaks I can hear the offending  carrier modulate up and down. I found it very annoying and stopped using it. I haven't tried it on version 3.
Photo of Russ Ravella

Russ Ravella

  • 50 Posts
  • 23 Reply Likes
Hi Jim,

Thanks for the response.  Yeah, that was one of the things I didn't like about it as well.  Also, it seemed to replace some of the offending carrier with a weird distorted mix of the in-band signals and the carrier it was trying to notch out sometimes.  I'm noticing very little of that all of a sudden.  Like I said, maybe I'm just getting lucky hitting cases it always handled better than others (whatever those would be...).
Photo of N8AUM

N8AUM

  • 118 Posts
  • 12 Reply Likes
Good question Russ, currently running V3 but I will enable it than reload 249 and see if it responds in the same way. Will let you know if I notice any difference between the 2 versions. Guess I wont get much sleep tonight lol 
Photo of K5ROX

K5ROX

  • 128 Posts
  • 30 Reply Likes
ANF on every radio I have works great.....except the flex.
Photo of Scott N8UMW

Scott N8UMW

  • 155 Posts
  • 47 Reply Likes
Yes sir. ANF and NB need much work.
Photo of Neal Pollack, N6YFM

Neal Pollack, N6YFM

  • 255 Posts
  • 127 Reply Likes
Guys:   Can you define a little more specifically what you mean, rather than
"works great except the Flex"?

If engineering in ANY company was going to act on the complaint, it helps to know
a little more specifically and verbosely what the issue is?   Can you describe what the Flex does with ANF that the other rigs do not?

Without more detail, your comments are without use.  It's like calling Microsoft and
yelling at the support person "I GOT AN ERROR MESSAGE".   They ask, what did
it specifically say, and the response is usually something like  "I DON'T KNOW, BUT
IT DON'T WORK AND I'M PISSED".   Um, how would you take action on THAT software bug?  :-)  :-)

Cheers,

Neal
Photo of N5LB - Lionel B

N5LB - Lionel B

  • 224 Posts
  • 63 Reply Likes
Agreed Neal.  
Photo of Mark  K1LSB

Mark K1LSB

  • 83 Posts
  • 27 Reply Likes
Neal,

Here's a screenshot of HDSDR with a half dozen manual notches I've defined (there aren't any actual birdies in the received signal, I've only defined the notches to illustrate the way the notches operate on the received audio):

https://drive.google.com/open?id=1nyaac_2Bp1MbsdpN3f2rSmS2KbdOfpf3

Notice how each of the narrowly defined notched lines are completely removed from the audio stream while nothing else in the audio band slice is affected at all. A casual listener would hardly notice any degradation in the audio stream at all, because the removed lines are each so narrow that the human ear can hardly discern the loss of any audio "substance" from any individual notch.  Even the presence of multiple notches would only become audibly apparent if they were all clumped closely together.

Now here's a screenshot of the AutoNotch feature of HDSDR acting on an actual birdie in the audio stream:

https://drive.google.com/open?id=1VYacDEEO62Q4VYr0UVuPFRu4W7VV8oxQ

Midway through the audio waterfall window I turned on the AutoNotch. Again, notice how only the offending birdie is automatically detected and removed, with absolutely no adverse effect seen (or heard) anywhere else in the audio band slice. The listener would be completely unaware of the presence of a birdie if he weren't looking at the waterfall, and would be equally unaware of the on/off status of the AutoNotch if he didn't notice the color of the ANotch button. That's what I meant by my use of the term "transparent" in my earlier post.

I've used HDSDR for years before acquiring my Flex 6300, and the notch filter in HDSDR is one of the most impressive features I've ever seen in any SDR software. It is 100% effective at filtering out birdies while having absolutely no effect on any other aspect of the audio stream. If there aren't any birdies in the audio, the AutoNotch can be turned on and off with absolutely no deleterious effect on the audio stream - it only acts on (completely removes) any birdies found in the audio bandwidth. It's so un-intrusive I always just left it on. NOTE: The AutoNotch in HDSDR is not adjustable, it's either simply on or off.

Now contrast that with the behavior of the auto notch filter (ANF) in Flex SSDR. Whenever it is engaged, even at the least intrusive slider setting (when the adjustment slider is at the far left), the audio is badly distorted regardless of whether there are any birdies in the audio stream or not.  And the ANF is far less than 100% effective at removing any actual birdies that are found in the audio stream, if there is any other activity (such as voice) also in the stream.

In short, the ANF function seriously degrades the audio in SSDR any time it's engaged. Do this simple experiment yourself: tune in to a strong ragchew QSO on any band and then engage the ANF function, and you'll certainly see what I'm talking about. Many other users have posted the same observations in these fora.

I can't recall ever having seen so basic and glaring a flaw in any software go so long without having even been acknowledged by the software author, let alone having been fixed. For Flex to remain so profoundly mute on this very pointed topic, with it having existed as a user-noted problem for so long, is simply beyond my ability to fathom. Flex devotes large amounts of resources to adding new features into the latest $200 upgrade while ignoring glaring defects in previous releases.
Photo of Neal Pollack, N6YFM

Neal Pollack, N6YFM

  • 255 Posts
  • 127 Reply Likes
OK, good report.   That is a proper and fine example of a detailed problem report.
Thanks!
Photo of Neal Pollack, N6YFM

Neal Pollack, N6YFM

  • 255 Posts
  • 127 Reply Likes
Also, since Flex has a very small engineering group, it might be more effective for
them to license the Auto-Notch software code from HDSDR rather than spend a year inventing a new module on their own.  But they will need to decide on that.

In the mean-time, it sounds like defining a manual notch, parking it at the edge of the
band, and sliding it into place when needed, might be the best work-around I have read about in this thread.

Cheers,

Neal
Photo of SteveM

SteveM

  • 273 Posts
  • 44 Reply Likes
Yes, that's been the recommended work-around going on 5 years now - forget automatic, do it manually. Previous to v2.4.9, people were using this work-around and then began complaining that, over time, TNFs would be scattered all about the spectrum and then were too time consuming to remove. So Flex decided to add a "Remove All TNFs" function to v2.4.9. I guess that was one solution, but I would have thought fixing ANF is the better solution.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3918 Posts
  • 954 Reply Likes
I am not a Flex employee


None of the HPSDR code can work with SSDR.
(Edited)
Photo of Mark  K1LSB

Mark K1LSB

  • 83 Posts
  • 27 Reply Likes
And how do you know that?

Once again, you come across as being privy to information that only an insider could possibly know.

1) To make the statement you made, you'd need to know (as a minimum) what language each program (HDSDR and SSDR) was/is written in.

2) Furthermore, you'd need to be enough of an expert at coding digital audio processing routines to know for a certainty that the algorithms used by each program are 100% incompatible insofar as being able to share any snippet of code logic between the two.

And I'm betting you're pretty much clueless on both counts.  If you weren't you'd be flaunting your credentials in every post you make.

So I ask you, how can you sit there and make such absolute declarations as the one in your post above?

Jeez!..and you wonder why I summarily dismiss so much of what you post!

EDIT:

Bill,

I just realized that you said "None of the HPSDR code can work with SSDR", which is correct.

However, HPSDR has never been previously mentioned on this page, though HDSDR has been mentioned nearly a dozen times.

So your statement, even though probably correct, is nonetheless irrelevant to the discussion.

I must concede I missed the nuance that you weren't talking about HDSDR, while you had a dozen opportunities to catch (and still missed) the fact that we were never talking about HPSDR. So I don't feel so bad.

Regards,

Mark K1LSB
(Edited)
Photo of Dan KG0AQ

Dan KG0AQ

  • 62 Posts
  • 1 Reply Like
I remember my old SDR-1000 and early software revisions of my Flex-5000 working great. I think when they went away from the AF based ANF it went to hell. Has not worked since.
Photo of Mark  K1LSB

Mark K1LSB

  • 83 Posts
  • 27 Reply Likes
What is "AF based ANF", please?
Photo of James Whiteway

James Whiteway

  • 1048 Posts
  • 299 Reply Likes
Mark, AF based means the filter is working in the audio domain, instead of the digital domain. AF filtering was common in a lot of radios before DSP came along. Some worked quite well. Others not so well.
James
WD5GWY
Photo of Mark  K1LSB

Mark K1LSB

  • 83 Posts
  • 27 Reply Likes
Thanks James.

For comparison, HDSDR and SDR Console both have very effective (and virtually transparent) ANF, and they're both free so it's not like it takes a team of professionals to get a proper job done.
Photo of James Whiteway

James Whiteway

  • 1048 Posts
  • 299 Reply Likes
Don't go there Mark !
:-)

I will say this though, writing an effective Digital Notch Filter is not trivial. But, it obviously can, and has been, done.
Noise Blankers on the other hand, are not so easy. What works great for one person's noise issues will be useless for another's. In the case of Noise Blankers, one size fits all does not apply.
  I have no doubts that FRS's developers can get a handle on the ANF issues I and others have had issues with. It's simply a matter of when, is the issue to me.
I am hoping that v2.50 comes thru with the improvements to ANF that's needed.
If so, I'll be getting v3 at that point as I like to pretend I'm a programmer and want the multi-client feature so my GUI app will run alongside SSDR at the same time.
James
WD5GWY
(Edited)
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3874 Posts
  • 943 Reply Likes
I am not a Flex employee


I would not expect any changes to the DSP in V2.5. I have not heard anything about it being worked on.
The DSP has not really been worked on since it was first introduced a long time ago. 
Imagine, even without some of those filters not working so well, the Flex is still a wonderful receiver, I wonder how that could be?
Photo of James Whiteway

James Whiteway

  • 1048 Posts
  • 299 Reply Likes
Bill, why would you think that the DSP functions will not be addressed? I would hope with all the various posts made on this forum, that FRS would see how important these features are to a lot of their user base.
 I cannot imagine with hardware that can be leveraged to the level that FRS has so far, that they cannot or will not get back to the basics for a while and address those issues.
I understand things like the Noise Blanker not being able to handle every single situation, but, it can be improved. And the ANF can be improved as well. Along with other bugs and features that don't work as well as their counterparts from other manufacturers.
  I agree that the 6000 Series radio receivers are great. But, there is a lot of room for improvement. Even the 1500 I had with PSDR had an ANF that worked great.
Other SDR radios have better ANF than the 6000 Series. To me, that is sad.
Not every ham needs remoting or multi-client. But, to capture the rest of the market, FRS needs more of the basics that work, and work equal to, or better than the competition.
   As far as the receiver being wonderful, it is simple, it has a lot to do with Direct Conversion and not having several stages the signal has to travel thru like older analog receivers do. No phase noise, and a host of other reasons.
But, interference fighting tools that work are standard on all radios now. And are the expected norm. Not something that "might" be fixed added someday.
Since you have a disclaimer stating you "are not a Flex employee" (are you an Alpha Tester?) how do you know or expect to hear that the DSP functions are not being worked on?
I have had actual Flex employees (including their head of engineering) tell me that by v2.49 things like the ANF and other issues would be looked at and  corrected. That didn't happen. Instead, v3.x came out. So, hopefully, v2.50 will be the version that finally fixes things. If not, I expect there will be a LOT more posts like those of last weekend.
James
WD5GWY



Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3874 Posts
  • 943 Reply Likes
I am not a Flex employee...

James, this has been a discussion so many times, so I will not say much more. Nothing new has been said and everything has been well documented.

V2.5 as far as I know is not about noise mitigation, and Flex has never suggested that it was. Pehaps your correct in thinking it is about DSP, but I don't expect it.

One reason I see for this, is that refining the DSP in SSDR is in itslf a large project and will take most if not all the company resouces to work on. So I really don't think it will be in  update 2.5 coming up.

It is true there are things that other radios do a better job on, but bye the same tocken there are things the Flex does that other radio don't.

The real magic of the receiver in SSDR is the AGC-T something other radios don't have, a big advantage for Flex.

I say when I post that I am not a Flex employee because there have been a few on the community slaming me for acting as an emplyee,,so now I make it very clear so these people will not continue to be confused.
(Edited)
Photo of Mark  K1LSB

Mark K1LSB

  • 83 Posts
  • 27 Reply Likes
Bill, it's the way you talk about things that only a Flex insider would likely have any in-depth knowledge of, that misleads people.

For example, your statement above, "refining the DSP in SSDR is in itself a large project and will take most if not all the company resources to work on", implies that 1) you're a programmer who also happens to be intimately familiar with the SSDR source code, and 2) you know the current size of Flex's software team and and the depth (or lack thereof) of their capabilities.

In fact, you don't know any more about those things than any of the rest of us. You're nothing more than a Flex apologist, as is evidenced by your typical gushing pro-Flex statements such as the one above, "Imagine, even without some of those filters not working so well, the Flex is still a wonderful receiver, I wonder how that could be?"

It's not my intent to beat up on you, Bill, I just grow weary of seeing you so incessantly post stuff that only an unabashed (and uninformed) fanboi apologist would post.

HDSDR and SDR Console are both one-man efforts, yet they both got a very firm grip on the ANF issue years ago. Simon Brown listens to his users thru his very active groups.io page and has a "release early, release often" philosophy to his software (I often wonder how he even finds any time to program, given all of the attention he devotes to his users). If even one user finds something wrong with his software, he typically fixes it pronto. He may also be working on other things in the meantime, but he'll still post up and comment, "the problem is fixed, it will be in the next release". Contrast that to Flex's management's continued silence after how long a time of people complaining about the ANF?

And per James' strongly voiced advice, I'm done.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3874 Posts
  • 943 Reply Likes
I wish both of you would not make things so personal, if you disagree with me that's great,,does it matter? but the constant attacks?

Kevin, how much clear can I be, I don't work for Flex.
Did you hear anything about update 2.5 being about DSP? did you? Not me, is what I said. So I would not expect anything on it.

Mark, my comment is from comments Gerald made a long time ago about the man hours and testing needed to introduce the noise mitigation, It was something they worked on for a while.

example":
Gerald...
The current noise blanker does work but not as well as we would like.  As announced in the prior Flex Insider, the theme of v1.5 will be noise and interference mitigation.  We don't plan to go for just parity with other radios but to raise bar in this area just as we did with the Panafall.   We have several "science projects" on tap with a few PhDs contributing to the effort.  No promises yet as to what that will look like but it will be the main focus of v1.5.


Many many times Flex has said that the problem has been fixed in the next release, did you miss all of them?

And it matters nothing to me if either of you aprove of my comments. They are simply my opinions based on comments made from Flex in the past.

So please show more respect, if you don't like what you read,,please turn the page.
Photo of Scott N8UMW

Scott N8UMW

  • 155 Posts
  • 47 Reply Likes
The part about the noise blanker. Hilarious. V1.5 was a while ago and they still have not achieved parity with the old rigs and software, let alone raise the bar. And we're up to v3. Yay
Photo of Craig Williams

Craig Williams

  • 234 Posts
  • 72 Reply Likes
Guess I don't understand. My ANF works fine on my 6400 with V2.9.x
Photo of Volker Geith

Volker Geith

  • 72 Posts
  • 24 Reply Likes
Craig, it does not work fine, maybe fine enough for you but compared to what is possible with other rigs it simply does not the job even
Photo of Volker Geith

Volker Geith

  • 72 Posts
  • 24 Reply Likes
Craig, it does not work fine, maybe fine enough for you but compared to what is possible with other rigs it simply does not do the job even with simple single tones. I do not use it so I don't care too much.

Volker
Photo of N5LB - Lionel B

N5LB - Lionel B

  • 224 Posts
  • 64 Reply Likes
ANF seems okay to me but I mostly use CW.  When you say compared to what is possible with other dogs, what does that mean?  I really want to know because I have no reference in the shack to compare the Flex ANF.  What is it that other mfg ANF does, or does better, that sets the bar for Flex? 

As I said, it seems okay to me and maybe its because I have not used any other ANF.
Photo of Bob G   W1GLV

Bob G W1GLV

  • 815 Posts
  • 145 Reply Likes
Read the "mitigating noise" on the Flexradio website.
Photo of SteveM

SteveM

  • 273 Posts
  • 44 Reply Likes
Why? Does it state whether Flex fixed ANF in v3?
Photo of Burch - K4QXX

Burch - K4QXX

  • 538 Posts
  • 132 Reply Likes
As others have said, the ANF works fine assuming you don't use it over a moderate or strong signal.  If you do, at least on my radio, you will get the ringing from the birdie or carrier you are trying to suppress on the signal's voice peaks.  Very annoying.  If you use it on an offending noise that isn't on top of a signal, it works fine.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3972 Posts
  • 967 Reply Likes
This is why I never use the ANF. what I do is creat a TNF and park it just out side the pass band, then if a signal comes up I simply grab it and slide it over the signal ,,all done.

Till they look into this again.
Photo of Mike W9OJ

Mike W9OJ

  • 104 Posts
  • 38 Reply Likes
"If you use it on an offending noise that isn't on top of a signal, it works fine."

But that's the only reason to use it in the first place.
Photo of Mark  K1LSB

Mark K1LSB

  • 86 Posts
  • 27 Reply Likes
"This is why I never use the ANF. what I do is creat a TNF and park it just out side the pass band, then if a signal comes up I simply grab it and slide it over the signal ,,all done." -- Bill VA3WTB

Lol, an open admission from Flex's biggest fanboi that the best way to use the ANF is to avoid using it at all!
Photo of Craig Williams

Craig Williams

  • 234 Posts
  • 72 Reply Likes
Really, really, boring conversations not bring any useful information on using and enjoying Ham Radio.
Photo of Mark  K1LSB

Mark K1LSB

  • 86 Posts
  • 27 Reply Likes
Then to borrow a quote from Bill VA3WTB (wow, who ever thought I'd ever say that!?), "if you don't like what you read, please turn the page."
Photo of Craig Williams

Craig Williams

  • 234 Posts
  • 72 Reply Likes
So, I am not allowed to have an opinion?
Photo of SteveM

SteveM

  • 273 Posts
  • 44 Reply Likes
Yes you are allowed an opinion, Craig. Even a really, really boring opinion which "not bring any useful information".
(Edited)
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3972 Posts
  • 967 Reply Likes
This was a comment from the discussion.

Also, since Flex has a very small engineering group, it might be more effective for

them to license the Auto-Notch software code from HDSDR rather than spend a year inventing a new module on their own.
  My coment was:  None of the HPSDR code can work with SSDR.

My mystake, I'm sorry that I misread, I read HPSDR instead of HDSDR
I was not aware that the source code for HDSDR could work in SSDR code.
Photo of Neal Pollack, N6YFM

Neal Pollack, N6YFM

  • 255 Posts
  • 127 Reply Likes
Bill:

I have no knowledge of HPSDR or HDSDR either, other than having used HDSDR.
I was simply suggesting that sometimes it is far faster to license an algorithm
than to invent it yourself.    Since I have not looked at the source code, you are
right, perhaps the code licenses, GPL, etc, come into play, blocking use by
commercial companies.

That said, I know for a fact that the actual author of any software, open or closed
source, GPL or whatever, has the right to dual-license it as they see fit.
In my "day job" with a very large corp, we would regularly approach the holder
of a GPL licensed code module we wanted to use in a commercial product, and
ask them to also license that same code directly to us under the BSD or MIT 
license so that we could actually put the code into a commercial product.
For the appropriate compensation most were usually happy to do so.
Some, not so much.

It is really Flex's discretion to decide to try to do it all in-house, or to try to
leverage work that is already completed by making agreements.  Their choice.

Cheers,

Neal
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3972 Posts
  • 967 Reply Likes
Interesting thoughts Neal.
Using Geralds remarks, Flex is going about this in a very uncontional way, using wide band code. I don't think it is a matter of them not being able to pull this off as much as just not using the resources to get back to the project.