SmartSDR ANF still not usable.

  • 5
  • Problem
  • Updated 3 years ago
  • Acknowledged
The Auto Notch Filter causes audio distortion and really doesn't effectively null the carrier when vice and present at times on 75 meters my primary band.   

When this this basic feature going to work as well as it did in PowerSDr on my Flex3000?

Doing a search I am seeing posts about it from 2 years ago. 
This is a basic function of radios in the Signature Series class
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes

Posted 3 years ago

  • 5
Photo of James Del Principe

James Del Principe

  • 306 Posts
  • 45 Reply Likes
Steve, can you tell us what SW rel you are running?    I am at 1.6.x and it works exceptionally well but that is just my experience. Others may disagree. I have never had a radio (lots of radios tried over the years tho) that had such a notch filter. I am amazed at how it works.     73, Jim
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
James
I am on 1.7.32.15

I hear audio distortion.
I am comparing it to the same function on PowerSDR with a Flex 3000 and a Kenwood TS 2000.
The operation is very sub par. There should be no perceptible distortion if it's working correctly.
You should notice nothing but the lack of a carrier or absence of the tunerupers.   
Photo of James Del Principe

James Del Principe

  • 306 Posts
  • 45 Reply Likes
Steve, you are correct that there should be no discernable audio distortion. I suppose, since this is my first SDR, (6500) that it is just amazing to me how it blanks out all hetrodyning carriers. I have nothing to compare it to, however.    I hope some others can wade in here and add to this. Best 73, Jim
Photo of Don Stefanik

Don Stefanik

  • 91 Posts
  • 16 Reply Likes
i have distortion plus if some one is talking the anf will still let a tuners whine come through.
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
That is what I experience too. 
Since there is no TNF on the Maestro yet ANF needs some work. 
Photo of K0UNX

K0UNX

  • 107 Posts
  • 30 Reply Likes
The ANF does a good job in taking out heterodynes for me, but it DOES introduce audio distortion.  So I run without it until someone tunes up.  My Icom doesn't distort the audio, so I leave ANF ON at all times with the Icom.  I've assigned "F1" on the MAESTRO for ANF making it very fast to toggle it.  

BTW, for a very tiny speaker, the Maestro sounds VERY NICE!

Jim
K0UNX
Photo of Philip KA4KOE

Philip KA4KOE

  • 175 Posts
  • 22 Reply Likes
I never use it as TNF is plenty good by itself.
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
Two issues with that.
TNF is not available on the Maestro.
If using SmartSDR you have to set the TNF on the carrier. 

ANF is automatic. It works perfectly on PowerSDR.
On SmartSDR it introduces distortion and the carrier can be heard under the voice at times. This is something FlexRadio can fix. It's not something we should do without because TNF works.
Photo of Philip KA4KOE

Philip KA4KOE

  • 175 Posts
  • 22 Reply Likes
Noted. But I wouldn't know that as my Maestro hasn't arrived yet.
Photo of Lee

Lee

  • 84 Posts
  • 19 Reply Likes
Me too....ANF,is not good...back to my original complaint...let's get this radio finished!
The DSP features are still not where they should be for a expensive radio.
Lee
Photo of Gary L. Robinson

Gary L. Robinson

  • 254 Posts
  • 56 Reply Likes
I think I understand what you are saying Lee and I agree 99%. I think some things need working on and I would like to see some of the features PSDR has end up in SSDR. However, with SDR is there really such a thing as a "finished" radio - I hope not hihi!

---Gary
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9188 Posts
  • 3550 Reply Likes
Official Response
We are planning on taking a look at the ANF code to see if improvements can be made.  This is on the list.
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes
Tim,

Hopefully the NB too?   These functions are not terrible, but not great either for a benchmark-setting radio.
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9188 Posts
  • 3550 Reply Likes
Yes, the NB too.
Photo of Burch - K4QXX

Burch - K4QXX

  • 534 Posts
  • 132 Reply Likes
The one thing I noticed about the NB (not the WNB) is a lot of times it works really well but I keep having to turn it off and back on.  It degrades over time.  If it could keep stable it would work pretty good for my local powerline noise.
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9188 Posts
  • 3550 Reply Likes
It degrades because the noise signature changes over time and unless you change frequencies to force it to retrain, you will have to toggle it to force a retraining on the changed noise source.
(Edited)
Photo of Burch - K4QXX

Burch - K4QXX

  • 534 Posts
  • 132 Reply Likes
Change frequencies to force it to retain?  I'm not understanding that part.  I'll give you an example from last night.  I was listening on 40 meters to 3 stations rag chewing last night.  I have about an S6 noise level on 40 meters right now.  One station was right above the noise level but perfectly copy-able.  The 2nd station was about S8 and the 3rd about S9.  After listening for a couple of minutes, the "buzz" sound was back.  I'd cycle the NB off and back ON and the buzz was gone.  The powerline noise never changed and was a steady S6 all night. It got to be really annoying to have to keep turning off and on the NB every minute or two.  I never had to do this with my 1000/1500/5000a radios running PowerSDR.  If there was a way to keep this stable, it would greatly improve the NB on SSDR.  I understand that if the noise increases the NB will not be as effective or change but I have never experienced this on/off type experience with any radio I have ever owned.  PowerSDR definitely didn't work this way and I really miss it's NB.  I would still have my 5000a if is wasn't for the WNB addition to SSDR.  It is not perfect but it greatly helped my noise situation at my location. 
(Edited)
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9188 Posts
  • 3550 Reply Likes
RF noise is chaotic in nature and is constantly changing; it has one of the most complex waveforms (signatures) you will experience on the bands.  Just because the magnitude of it stayed the same, doesn't mean it did not change.

I suspect that you may need to adjust the threshold settings.
Photo of Burch - K4QXX

Burch - K4QXX

  • 534 Posts
  • 132 Reply Likes
I have played with the threshold settings for hours on end but over time, the NB always degrades to the point I have to turn it off and back on.  Doesn't matter what setting or what band.  Thankfully I don't use the NB as much as I used to since the WNB has been working pretty good for me lately.
Photo of Steven Hess

Steven Hess

  • 84 Posts
  • 19 Reply Likes
Tim.
I'm pretty sure improvements can be made in ANF. After all this still just works on the legacy Flex rigs. I don't think FlexRadio has unlearned how to do ANF. Since TNF is not available on the Maestro I'd think you'd make this a priority. 
Photo of Norm - W7CK

Norm - W7CK

  • 759 Posts
  • 164 Reply Likes
Tim,

You said "We are planning on taking a look at the ANF code to see if improvements can be made.   The "if improvements can be made." implies that it may not be possible to fix this issue.   I hope that's not what your saying?

Norm - W7CK
(Edited)
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9188 Posts
  • 3550 Reply Likes
I am saying until we perform the required due diligence and investigate to determine the root cause of the issue, it is irresponsible for me or anyone else at FlexRadio to state definitively that a particular issue has a fix.  This is the Schrodinger's Cat of software development.  Until we open to box (find root cause) the cat (issue) is both alive (fixable) and dead (not fixable).
Photo of James Del Principe

James Del Principe

  • 306 Posts
  • 45 Reply Likes
Well put, Tim.   Excellent!     73, Jim
Photo of SuperFabian40 .

SuperFabian40 .

  • 3 Posts
  • 0 Reply Likes
Hi, I'm new and do not want to be repetitive or awkward in my comments, but coming from using for a while flex 1500 to 6300 and now passing me, I can mensionar what I miss him, though not much.1) the ANR, impeccable in 1500 but in 6300 still there to work on, but blocks the heterodyne when a ssb modulation, heterodyne is mounted above and in every word that annoying hiss appears.
2) NB, NB strange both 1500 and settings at will, and good walk!
3) I can not find a third party! The truth are doing an excellent job with the SmartSDR, I think iran filing by little these rough edges, thanks !!
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
Making ANF work properly ought to be on the top of the list. 
FlexRadio already knows the math involved. 
This is the big wart on my SSDR experience.
SSDR 1.X won't be finished until ANF works right.
Just workes on the 5000, 3000 and 1500. 

Making TNF work on my Maestro would be second on things that needs to be on the top of the list. 
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
Even on my 1500 the ANF was more convenient than TNF, but not nearly as effective as a quickly initiated manual TNF filter.  I hardly ever used ANF on the 1500.  I don't use it much on the 6500 now, either.

The advantage is that I can set up multiple TNF's very quickly and deal with multiple noise sources (on some bands, I have a couple of frequencies where my cordless phone, my router, and my neighbor's TV all converge in the same SSB bandwidth) 

I wonder if the TNF could have an option that would auto-detect and set a temporary TNF with the press of a button, and then either slowly track for drift, or stay put until the button is pushed again...perhaps with a long-press to delete it?  Or the standard right-click and edit?

As it is, I think the current ANF has too short a time constant and is trying to track the interfering signal too closely and gets confused by other legitimate signals in the passband.  But then I haven't much of a clue as to the actual algorithms used in either ANF or TNF.  But if they could match the fidelity of the TNF in the ANF then it would be heavenly!

Ken - NM9P
Photo of SuperFabian40 .

SuperFabian40 .

  • 3 Posts
  • 0 Reply Likes
Good reviewed the gift flex hear it and your will be done !!
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
Ken 
 the problem two fold at this point. TNF requires action on your part. You need to use the mouse and set the TNF.  And neither setting a TNF or a decent working ANF is implemented on the Maestro.  I'm thinking fixing the ANF and getting it working right will be easier to implement than setting TNFs on the Maestro. Fix ANF and a lot of people will be very happy. 
Photo of WW1SS - Steve

WW1SS - Steve

  • 797 Posts
  • 268 Reply Likes
I've been saying this for a long long time. It just seems to always get put aside for the next big bang.
Photo of Steve Gw0gei

Steve Gw0gei

  • 193 Posts
  • 50 Reply Likes
I rarely use anf but have tried using it this last week on tuner causing qrm to an 80m net. Using 1.9.7 i had the same issue of ineffective operation on the signal plus always audio distortion on the wanted ssb signal - tried all levels of anf and even at low level same distortion problem.
Smooth upgrade to 1.9.9.170 last night but i soon confirmed on 40m ssb that same ineffectice anf problem and audio distortion present on the wanted ssb signal. Not a deal breaker for me as i use tnf on persistent carriers but anf does need fixing in a future release.
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
It didn't get any better because they didn't address it in this release. They were targeting other specific issues as described in the release notes.

They will address ANF effectiveness eventually, after they deal with a host of other issues that are more pressing. In the mean time, the TNF is extremely effective and does not color the audio at all, except for the loss of SSB audio caused by the notch. And you can have more than one active in an audio passband at the same time.
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
Ken
TNF does not work on the Maestro. So TNF is not a replacement for a functional ANF in that situation. I bought my Maestro so I wouldn't have to boot a computer. Essentially what you are saying is use a computer.  I am sure you can see why some of us are not satisfied with that as an answer. 
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
Understood.  Not having a Maestro, I forgot that they don't have TNF working on Maestro yet.

I don't know which will be easier for them to implement on Maestro - improving the ANF, or implementing some sort of rudimentary tunable TNF (perhaps using the VFO B knob to adjust the notch frequency?).  I hope that they soon will get both working for all clients.  
Photo of Steven Hess

Steven Hess

  • 85 Posts
  • 19 Reply Likes
I  likewise don't know how they'll implement TNF on the Maestro.  You have two settings you need to accomplish. Location and width. That said getting ANF working right might not be as simple. I think I quit forgetting the processor in my Flex 6300 is not a powerful as it is on my 6 core Tower I use with my Flex 3000 and Flex 1500.  That was half the radio.

 I don't even have my 3000 connected any more. Just the Flex 1500. 
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
I don't think it is a matter of processing speed - the FPGA, I understand may be clocked at a slower speed, but is designed for the task.  But there are LOTS of things involved in getting one function to work correctly without messing up something else...  Something I have a much greater appreciation for since joining the Alpha Team and seeing the "sausage" made!  These folks - programmers and alpha testers alike, put an amazing amount of work into the simplest of updates!  Much of it is sorting out elusive cockroaches in the system unleashed by what some call "A Simple Matter Of Programming!"