noise blanker nb

  • 3
  • Problem
  • Updated 4 years ago
  • Acknowledged
Noise blanker is still so so performance
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes

Posted 4 years ago

  • 3
Photo of WW1SS - Steve

WW1SS - Steve

  • 794 Posts
  • 268 Reply Likes
Yes it is. I wish they would take care of these BUGS. Seems more like they are more interested in bringing new stuff to market and less about the glitches.Other manufacturers will implement a fix and bring out an update in a week or two or three. It seems here you have to wait 6 months for a new version to arrive.Even then the bug might have not been gotten to.
(Edited)
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
What companies bring out fixes each week? Flex has been bringing out big updates close to each 3 months.
Photo of Burt Fisher

Burt Fisher

  • 1241 Posts
  • 480 Reply Likes
What companies have a need to bring out fixes each week?
Photo of WW1SS - Steve

WW1SS - Steve

  • 794 Posts
  • 268 Reply Likes
Not many but when there is an issue it is generally fixed fairly sooner than later. So I guess we have another 3 months to wait to see if the NB gets better. Bet you it won't as I have been waiting since 1.4 when I had a 6300.
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes
Noise blanker is still so so performance.

Over the weekend I had a very bad problem with a high level of ignition type qrm at s9+ on 80m (normally s5 noise level). I tried to track it down but it was so strong i failed to find it.

The noise blanker on the F6300 (v1.6 SmartSDR) did at best a mediocre job of taking out the noise, I still could not work a station at s9 due the ignition noise. I guess I must have gotten used to its poor performance, but do remember it was clearly better on my old F5k with PowerSDR. This was very disappointing, especially so considering that over a year ago FRS said th'd be releasing "a world class noise mitigation with v1.5".

Due to being unable to operate 80m during the weekend, I thought I'd try some other options and tests. I recently acquired a 32 year old Yaesu FT101zd mk3 from an SK. I powered it up and tuned 80m using same antenna etc. 

I then hit the NB button and seriously could not believe the ignition noise had completely vanished (s metre dropped by 3 s points), not a little bit lower but just gone! I was so surprised that I'd assumed the QRM source had stopped, but then turned NB off again and there it was very loud and clear as before.

Over the weekend with lots of playing around, I came to realize that the not-so-magic NB button can be much better on the Flex 6k. If it works so well with 32 year old analogue kit, surely there is much more improvement to come hopefully? 

Sorry FRS need to try harder on the NB, especially given the QRM rich environment we now find ourselves in.

73 de Steve G1XOW
Photo of Jim Gilliam

Jim Gilliam

  • 937 Posts
  • 218 Reply Likes
That is a very interesting observation. I have an Icom 7800 that uses DSP and I found, also, has a virtually worthless noise blanker compared to my old Kenwood 940. The 940's noise blanker runs circles around that of the 7800 and Flex 6500. Perhaps it is easier to implement noise blanking with hardware than software, or there is much work to be done of the algorithms.
Photo of Burch - K4QXX

Burch - K4QXX

  • 531 Posts
  • 130 Reply Likes
I agree that the NB on the Flex-5000 is much better than on the 6000 series.  However, the addition of the WNB has helped a little on the noise at my location.  The regular NB seems to work better now since it was added back in the software after WNB was added.  If the regular NB would take out the noise on the Panadapter like the WNB does, that would be a huge improvement.
Photo of David

David

  • 304 Posts
  • 51 Reply Likes
Are you using just the NB or have you tried the WNB, NR, ANF, and ACG in addition and combination of them? When you compared two radios were you also listening to a transmitting station or just comparing noise levels?

I agree the NB does not seem to be as effective since the WNB was added. The WNB is difficult to judge given the way it works a simple on/off doesn't always demonstrate how effective it is. The indicator does flash on and off showing it is doing something based of noise across the band. The ANF for me is a negative on the overall sound quality while might reduce noise it impacts the sound quality in an unpleasant way. The NR and ACG seem to be the most effective on reducing 'noise'. This is just my experience and I am sure a learning curving of how and when to use each. It would be nice to have a magic button that would make the noisier bands quieter and improve the intended signal quality.

Overall I find the Flex to be able to cut noise and retain voice quality great. It would be nice to have audio DSP similar to GAP Hear-It or BHI products in addition to all the RF DSP.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
The WNB works for me, my radio at night on 80M is not workable without it, I think they did a wonderful job on it.
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Bill they are talking about NB not whole band noise filter.
Photo of Jim Gilliam

Jim Gilliam

  • 937 Posts
  • 218 Reply Likes
Even more interesting is that is seems to work wonderfully for some types of noise but not all types of noise.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
That is true and we all know that. WNB is only for wide band noise. And it is amazing on that noise. I hope soon they will tweak the NB in coming releases.
Photo of Jim Gilliam

Jim Gilliam

  • 937 Posts
  • 218 Reply Likes
Not being an expert on noise, what is the difference between wide band noise and non-wide band noise?
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
It was explained in another post,, I don't know if I could find it.
Photo of Jim Gilliam

Jim Gilliam

  • 937 Posts
  • 218 Reply Likes
I am somewhat familiar with Fourier transforms and theoretically any function periodic or otherwise that is not purely sinusoidal contains an infinite number of frequencies of varying amplitudes and phases. So it would seem all noise would be wide band.  
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
See if this helps.
4 months ago
Photo of Tim - W4TME

 Tim - W4TME, Customer Experience Manager

  • 4199 Posts
  •  
  •  1299 Reply Likes
WNB works differently from NR so you really can't compare the two.  NR is a narrow bandwidth process that once converged, doesn't usually require re-convergence.  

With WNB, since the bandwidth it operates on is very wide, signals in other parts of the band away from your signal of interest may adversely affect the aggravate signal levels requiring frequent re-training.

In our alpha testing, we received feedback that there was no visual indication that WNB was temporarily disabled while it re-trained  after a significant change in signal level and the WNB indicator.  It seemed that WNB was "broken" (not blanking) when it was actually re-training.  If you want to propose a change, make it an idea post and allow public feedback on the merit of it.

Second, while WNB does operate at the SCU level, it can be applied on a per panadapter basis.  If WNB is not applied to a particular panadapter, we decided there was no need to display the blanking state of WNB.  And all slices within the same panadapter do show the same WNB enabled/disabled state.
(Edited)
Photo of David

David

  • 304 Posts
  • 51 Reply Likes
Here is the description. In short "targeted at correlated impulse noise" looking at common noise across the band not just what you are tuned to.

https://community.flexradio.com/flexradio/topics/smartsdr-v1-5-0-is-now-available
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
I seem to have missed the distinction between scu-based WNB and slice-based WNB. How does one select between the two?
(Edited)
Photo of David

David

  • 304 Posts
  • 51 Reply Likes
I think it is per panadapter not slice. So if you have a slice on another panadapter you can open the DCP and toggle the WNB for that slice on that panadapter and the slice on another panadapter will remain unchanged. If you do that to slice in the same panadapter all the slice in that panadapter are changed.
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
Oops, I mistyped.  I actually meant SCU-based vs. Panadapter-based... (I don't like typing on the ipad much for this reason.... by the time i manipulate the softkeys I forget what I was trying to say!)  

Correct...The slice-based Noise Blanker is the NB.  
But I had not heard anything about being able to activate WNB for either the whole SCU or only on a specific panadapter.  Have I missed something here?  

Ken - NM9P
Photo of David

David

  • 304 Posts
  • 51 Reply Likes
You can turn it on/off globally/SCU using the left side list ANT and the WNB was added there.
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
I wondered why it was added there, too.  I guess I need to read the appropriate manual page!   I noticed that when I turned it on and off that the indicator also toggled on the slice flag.  But I guess I didn't watch it when I toggled the one in the DSP pull down in the slice flag.  I have more playing to do tonight....
(Edited)
Photo of Gerald - K5SDR

Gerald - K5SDR, Employee

  • 830 Posts
  • 1514 Reply Likes
NB works ONLY on the respective slice within a 24 kHz bandwidth.  WNB works at the SCU level only and applies to all slices that are enabled on that SCU as well as the panadapter itself.  NB has no affect on the panadapter since it works at the audio level.  WNB works at the RF sampling rate and thus can improve the panadapter on correlated noise.  Both may be used concurrently if needed.   NR is designed for uncorrelated noise.
Photo of pa0bie

pa0bie

  • 152 Posts
  • 20 Reply Likes

I NEVER use the receivers in my 6700.  To much noise and no effective reducer and blanker. I have to recieve through the 6700 and "TEAMSPEAK" from a friend or use my 7800. Is there a Transmit-only 6000 coming out soon ?

Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
I don't mean to be piling on but NB has never worked for me and wbn had only worked on two occasions. But when it worked it worked well.
Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3762 Posts
  • 1621 Reply Likes
Legacy NB vs DSP NB

Before the invention of DSP, Noise BLANKING - which means removing impulse noise like ignition noise was accomplished by having a hardware circuit stop the receiver when it detected noise above a set threshold Hence Legacy NB receivers sound quieter because they are hearing nothing at all. This works pretty well for AM and SSB where your brain can integrate the cut off parts of the information but is much less effective for digital modes like CW and especially RTTY and all the modern digital modes where the loss of information can destroy the message.

In the early 1980, DSP chips came into use in Ham Radio DSP attempts to do with mathematics what the older NB process used to do by removing impulse noise without losing the entire signal. DSP Based NB represented a significant improvement in Impulse noise mitigation because it did not remove all the information from the signal which meant that it worked much better with digital modes.

NR vs NB
A lot of hams on this forum are confused as to the difference between NB and Noise Reduction (NR).
Noise Blanking means removing impulse noise only.
NOISE REDUCTION Means removing pass band signal noise in addition to impulse noise

DSP also gives you the ability to analyse receive pass band noise and using mathematics remove much of the noise while still retaining the information. NR is very effective in removing noise so you can hear weak signals that would normally be obscured by the noise

WNB. Wideband Noise Blanking is a new feature of the Flex Radio 6000 series. It looks at the entire SCU receiver to find correlated noise signatures. When it identifies correlated noise it inverts the noise and sums it against the correlated noise to very effectively remove it without the massive information loss of NB. But it only works for correlated noise where it can identify a pattern or signature

So while that old FT101 NB (I used to own one) worked really well on ignition noise and made everything quieter by several SUnits it really was not hearing that well because it had blanked out the weaker signals
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Nice explanation Howard...thanks. So, by extension then, is it fair to say, as more noise signatures are added WNB will work more consistently? Another related question, where the WNB works on the entire SCU, isn't that looking for noise signatures consistent across 14MHz? If a preselector is enabled (6500/6700) would the entire SCU look dead except for the selected pass band(s)? Or is the WNB in front of the preselector(s)?
(Edited)
Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3762 Posts
  • 1621 Reply Likes
I don't know that algorithm of WNB but I suspect that it dynamically recognizes correlation. Looking at the hardware architecture all processing is after the pre selector
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
But if it compares then against known signatures there would be a finite number of cases. Don't anyone (FRS) confuse what I am asking with wanting secret sauce). Gerald just mentioned, as others have, the arching power line transformer adjacent to the Austin office. I don't have one of those in my neighborhood. Is that why I see different results than, say, Lee did in his Youtube vids?

Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3762 Posts
  • 1621 Reply Likes
I believe that algorithm dynamically identifies correlated noise signatures. Your noise may not be as correlated

Correlation is a mathematical function.
(Edited)
Photo of Gerald - K5SDR

Gerald - K5SDR, Employee

  • 830 Posts
  • 1514 Reply Likes
Unfortunately the desire for noise mitigation is like that of finding a single drug that will cure all forms of cancer.  Such a drug doesn't exist.  Just like there are many types and causes of cancer, there are a myriad of noise sources, types and signatures.  Some can be mitigated and some simply cannot.  Some sources we call noise are actually QRM that if mitigated would remove the desired signal as well.  This is much like the chemo therapy where the cure is worse than the disease.  That is why we allow you to turn the WNB setting so high that it starts to affect the desired signals.  Band conditions and the noise type/source can vary widely and so might the setting.

I have watched WNB drop the noise floor on 40m at our club station by 20 dB and make a dead band usable.  It works pure magic on that arching power line across the road from our office.  On the other hand, I have seen some types of noise at home that it will not reduce.  NB or NR may help depending on the type of noise.  None are a single bullet cure for all types "cancer."  

We will continue to do ongoing research into innovative noise mitigation techniques over time.  These are science projects.  I can guarantee that no innovation will remove all types of noise in a single algorithm until the scientists find a single cure for cancer.  
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Didn't Michael Renni bring just such a cure with him in '52 but some idiot had to go and shoot him? Probably someone from big Pharma.
Photo of Rick Hadley - W0FG

Rick Hadley - W0FG

  • 600 Posts
  • 130 Reply Likes
Klaatu barada nikto ...
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
If somebody shot and killed me I'd be less than inclined to give him the cure for cancer also.
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes
Gerald,

Understand your comments, and I understand it may be a certain type/mode of noise that might not match what they NB is trying to do, but you are missing the originally point I made.

The comparison radio was a 32 year old Yaesu FT-101zd (of the valve/transistor hybrid era)!

At best the NB function of the 101zd would use a handful of very old-school analogue components and it worked superbly on the noise I saw last weekend. The same cannot be said for the NB on the F6k which did something, but nothing like the same level as the 101zd. I love my SDR and was disappointed to see it being beaten hands-down by a boat anchor.

The 101zd made a QSO perfectly possible and a pleasure to use. I'd would not have been aware of the noise if NB was already turned on. With all noise reduction options tried on the f6k (and in combinations) it only got as far a "tolerable annoyance" for a QSO.

I know the WNB works well on some noise types, but surely it's not asking too much for the Flex NB to be at least as good as a 101ZD is it?

Next time the noise happens I will video and youtube the results with like-for-like comparison of functions/bandwidth/antenna etc.

73 de Steve G1XOW
Photo of Jim Gilliam

Jim Gilliam

  • 933 Posts
  • 218 Reply Likes
Yes, you point has been well taken. Lots of fancy words but no noise blanking.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3871 Posts
  • 943 Reply Likes
No, it is not a good point at all, Steven missed the point, As Gerald said: We will continue to do ongoing research into innovative noise mitigation techniques over time.  These are science projects:
Noise mitigation came out on V1.5 and is still in development. The WNB was a huge step, under the conditions it is made for it is simply astounding. more work will be done on the others as well. That's were things seem to sit for now, some don't like it.
Photo of Jim Gilliam

Jim Gilliam

  • 933 Posts
  • 218 Reply Likes

Maybe someone should rent a Model T Ford and idle it outside the Flex labs! :)


Jim, K6QE

Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Bill, if FRS isn't already they should be paying you for your efforts as PR spokesperson.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3871 Posts
  • 943 Reply Likes
Funny Walt, Just trying to remind us that the radio is still and will likely always be developing. that is what we signed up for.
Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3761 Posts
  • 1621 Reply Likes
@Steve G1XOW

You seem to have missed my point above about Legacy vs DSP Noise Blanking that explains why that FT-101 seems to do such a good job of Noise Blanking

So....
"Before the invention of DSP, Noise BLANKING - (which means removing impulse noise like ignition noise) was accomplished by having a hardware circuit stop the receiver when it detected noise above a set threshold

Hence Legacy NB receivers sound quieter because they are hearing nothing at all. This works pretty well for AM and SSB where your brain can integrate the cut off parts of the information but is much less effective for digital modes like CW and especially RTTY and all the modern digital modes where the loss of information can destroy the message.

Modern DSP based technology tries to correct this serious deficiency of Legacy NB by using mathematics to recover as much signal as possible while still trying to remove the impulse peaks... Hence DSP circuits do not drop signal levels as dramatically as Legacy NB
Photo of Jim Gilliam

Jim Gilliam

  • 933 Posts
  • 218 Reply Likes
Maybe the could use mathematics to emulate legacy NB's for us who just want to listen.
(Edited)
Photo of Ned K1NJ

Ned K1NJ

  • 316 Posts
  • 82 Reply Likes
      The one in the "legacy" Flex 5000 was really good.  I go back to it when
 I have to.  It's sometimes worth 12 dB. or so.  (Yes, I know it means re-coding
 the algorithm, not just copying a code segment.)

       Ned,  K1NJ
Photo of Burch - K4QXX

Burch - K4QXX

  • 530 Posts
  • 129 Reply Likes
I agree.  That is one reason that I have not sold my 5000.  NB works great.
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes
As a digital comms engineer for over 32 years I am pretty sure I have not missed the point. I worked on below the noise ciphers on VLF for the military when most hams thought CW was advanced!

My metric is results based. I measure result and achievement by the output of any process/circuit/design etc, and also the acceptance testing of the user-base, not the technical and often incidental methods that may or may not have been used to get there.

In short, if users find it lacking compared to legacy methods then somewhere somehow we have missed the goal by a large margin.

In this example the 101zd allowed an unusable band to be used in the clear for perfect QSOs, the F6k failed the same like-for-like test. That is the outcome that matters to most users (aka buyers and recommenders). 

I like the F6k a lot, but technical and mathematical reasons why it can't be as good as a boat anchor because it really is "better" simply don't wash.
Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3761 Posts
  • 1621 Reply Likes
Since FRS seems to be missing the mark set by Boat Anchor NB, I am sure that FRS would be happy if you could share better NB algorithms with them so that everyone could benefit from your 32 years Digital Comms Experience.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3871 Posts
  • 943 Reply Likes
As was said before Steven,,give it time. As a flex owner you signed up for a radio in progress. They may change it up just for your type of noise problem. But it may not work so well on another kind of noise. As Gerald said all the noise mitigation is working as is by desine, not a mistake or missed testing. I suppose they could make it blanket everything for every noise and make the radio almost death as is in older radios, but that is not what DSP is to do.
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Bill, you say that, work in progress, like one size fits all, for all time. Certainly, for those that ordered their radio in 2012, clearly that was a work in progress. Two and a half years later there is likely a different expectation. For those that purchased last month I don't think it's reasonable to assume they have the same expiry timer original buyers had when the software was at 0.92. From what Steve Hicks said, rel 1 is completed. What, I believe, you are hearing is some people had expectations release 1 would be further along than it got.
(Edited)
Photo of G8ZPX

G8ZPX

  • 197 Posts
  • 110 Reply Likes
Bill, more than happy to give it more time. All SDR software is a constantly moving feast and will be for some years to come. No issue with that.

I just feel that a new SDR design should be assumed to be at least as good as a boat anchor. Surely it should be a given and a minimum prerequisite before the first release of v1.0, not still being promised years down the line. Basics first, gadgets later.

I just hope that those rushing to defend a request for basic functionality improvements are not suggesting that SmartSDR cannot be as good as a boat anchor
Photo of KY6LA - Howard

KY6LA - Howard, Elmer

  • 3762 Posts
  • 1621 Reply Likes
Science will cure cancer before we get rid of all noise b
Photo of Jim Gilliam

Jim Gilliam

  • 936 Posts
  • 218 Reply Likes
Well put Howard. Correlation is a good way of mathematically determining noise is there just like we can tell when cancer is there. But the mathematical algorithms like cancer cures have a long way to go.
Photo of Phil - N6ERP

Phil - N6ERP

  • 27 Posts
  • 5 Reply Likes
SW 1.16:
My experiences:

WNB - Not Good (has not helped me yet - Adds artifacts when set over 70)
NB - Not Good (has not helped me yet)
NR - Pretty Good (Helps a lot. Still could be improved, but great start)
ANF - Less than Marginal (It is affective at cutting tones, but ruins audio quality of recv'd station)

"Software Defined Radio"
Photo of Phil - N6ERP

Phil - N6ERP

  • 27 Posts
  • 5 Reply Likes
I agree with an earlier suggestion,

Would be nice to have an Audio DSP feature, such as ClearSpeech, BHI, NIR, etc. some day...we have the processing power for it..

Still loving the Flex 6500 though.. Great system. Audio quality is much better than the heavily filtered K3 was. (Although it had a much better NB and ANF).

Thank You FR.
(Edited)
Photo of Tom

Tom

  • 41 Posts
  • 13 Reply Likes
Speaking of DSP, did you aso notice that few versions ago CW APF was much better / efficient than today or this is just my impression?

Tom
Photo of Dan -- KC4GO

Dan -- KC4GO

  • 340 Posts
  • 70 Reply Likes
I worked Winter Field Day lat week end and had an opportunity to run my 6500 around 4 small generators including my on-board Onan. The NB works for me. Listen for yourself.
https://drive.google.com/folderview?id=0B5gpcTVs7PtoZ2dDU29ldXdGSmc&usp=sharing

There are two files one on 80 and one on 40 the when the NB is on is quite clear. :)
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3888 Posts
  • 947 Reply Likes
The noise you had must be what it was setup for...
Photo of Phil - N6ERP

Phil - N6ERP

  • 27 Posts
  • 5 Reply Likes
Glad it worked for you, the dam thing is worthless on mine. Also ANF badly distorts received audio quality. This is re: SW 1.16. NR works reasonably well. The NB on K3 was much better, ANF on Icom 7600 is night and day better.
Photo of K0UNX

K0UNX

  • 107 Posts
  • 30 Reply Likes
I agree on all counts.  I have never found noise that either NB or WNB will help.  And the ANF on my Icom 7600 works MUCH better.  When I've tried to use NB or WNB, all I get is a little bit of distortion of the audio.  But I still wouldn't go back to the Icom.