Noise Reduction needs improvement - feature request

  • 5
  • Idea
  • Updated 12 months ago
I worked the CQWW SSB contest this weekend with my FlexRadio 6300 as 2nd radio in an SO2R with my Yaesu FTdx5000. I am very impressed with the sensitivity of the 6300 as well as the selectivity - both compare very favorably to the Yaesu.

That said, I was less than thrilled with the noise reduction (DSP) features of the 6300 compared to the FTdx5000. I compared several bands, including 10 M, which had some rather severe broadband noise (I'd call it buzzing), and I couldn't get the 6300 to reduce noise anywhere near the Yaesu, even playing with the DNR, NB and WNB. The Yaesu was consistently better at making weak SSB signals readable.

There are probably several parts to this - one is that the Yaesu has four preamp/IPO settings, while the 6300 only has the option to turn the preamp on or off. Even playing around with all the different noise blanker and selectivity settings (where adjacent interference was an issue), the DNR slider didn't have nearly the effect I would have liked.

I'm hoping that there are improvements in the works, since I really like the 6300. I see some folks have complained about the automatic notch filter, but when I turned it on, it worked very well (some tuner-uppers on frequency).

Eric NC6K
Photo of Eric Gruff

Eric Gruff

  • 27 Posts
  • 10 Reply Likes

Posted 2 years ago

  • 5
Photo of Jorgen

Jorgen

  • 6 Posts
  • 2 Reply Likes
I am also i bit disappointed about the NR facility of the 6300. Here my ICOM IC-7800 functions better. About the NB facility the IC-7800 removes all the "clicks" from an electric fence surrounding a horse farm 3/4 of a mile from my home. The The 6300 NB is not so effective. Approvements are highly welcome.

 Vy 73 de OV6A Jorgen
Photo of Ross - K9COX

Ross - K9COX

  • 338 Posts
  • 104 Reply Likes
AGC levels can be effective in controlling noise. The NB and NR is another story.
Photo of pa0bie

pa0bie

  • 152 Posts
  • 20 Reply Likes
I have the same situation with my 7800 and even my 706 Icoms. The noise handling from the 6700 is completly useless from the beginning. Something should happening very urgent at FLEX or I lose my patient and dump the 6700.
Photo of G8ZPX

G8ZPX

  • 196 Posts
  • 110 Reply Likes
Eric,

In my experience the WNB is almost useless when being presented with very loud signals within its passband - such as found during big contests. In fact it can make the situation very much worse.

Even with the slider set at just 2-5% it will make a lot of additional wideband noise. I have an S9+20 machine signal/noise just above 3.8MHz so trying to use the WNB in the DX segment around 3.790 is impossible, so it's turned off most of the time nowadays.

The NB (as apposed to the WNB) is okay but not as good as say the Icom 7600v2 or the 7300 that I have in the shack. I now tend to use the 7600 exclusively on the low bands, and the F6k gets turned on when chasing DX above 20m.
Photo of Peter K1PGV

Peter K1PGV, Elmer

  • 541 Posts
  • 315 Reply Likes
The NR can use some work, I agree.

I will remind you that the AGC-T adjustment is very different from AGC on other rigs. And setting the AGC-T correctly along with the NR makes a big difference.

In my experience all things Fkex start with proper adjustment of the AGC-T.

Peter
K1PGV
Photo of Eric Gruff

Eric Gruff

  • 27 Posts
  • 10 Reply Likes
Thanks to all who replied. To Peter, K1PGV, thanks - I will try the AGC-T. My goal is to maximize my enjoyment of the 6300, not bash it, and I appreciate the willingness of the manufacturer to look into it. I'm happy to be a beta tester too.

FWIW - I have twin daughters (fraternal), and I try to use that situation as a model when comparing the Yaesu and FlexRadio. That is, both have their own positive aspects, but both could "learn" from the other. The FlexRadio's waterfall display is fantastic and blows away the panadapter on the Yaesu, for example.
Photo of Mike - ZL1MRC

Mike - ZL1MRC

  • 61 Posts
  • 15 Reply Likes
Hello

I am also see that my icom 7600 outperforms the flex 6500 on filtering. I have followed the instructions setting up the AGC-T however weak signals do better on the icom. I am not sure what flex is doing to address this.
Photo of Eric Gruff

Eric Gruff

  • 27 Posts
  • 10 Reply Likes
Just to follow up - adjusting the AGC-T did help a bit. I did find that I reached a point where I would have liked to turn it down a bit more, but couldn't get enough audio drive into my headphones to hear low signals. Definitely an improvement - thanks to all who offered suggestions.
(Edited)
Photo of Bob G   W1GLV

Bob G W1GLV

  • 676 Posts
  • 114 Reply Likes
I've been waiting 3 years for DSP improvements. I can't figure out what is so difficult. FRS had it working fine on PowerSDR. 
Photo of KC8P

KC8P

  • 34 Posts
  • 4 Reply Likes
Another vote for NB and WNB improvement, the algorithm implemented now is very selective and sometime works sometime not at all, highly depends of the noise characteristics.
Photo of KF4HR

KF4HR

  • 544 Posts
  • 158 Reply Likes
I've had my 6700 for about a year now and I've never been able to hear (or see) any difference when I activate WNB. 

Peter is absolutely correct about the AGC-T adjustment.  It makes a huge difference in background noise.
Photo of Bill -VA3WTB

Bill -VA3WTB

  • 3002 Posts
  • 660 Reply Likes
The WNB is the most misunderstood feature of the radio it seems. People expect it to work much like any other radio. Most radio's simply blankets everything so it works on anything. But the Flex? it is made to hone in on one type of noise, it samples it and programs it self to block it out, but only if you have the right type of noise. So for many no difference can be heard if the noise is the wrong type. For me, 40 meters is not usable at night with out the WNB on it compleately clears it up.
Photo of Burch - K4QXX

Burch - K4QXX

  • 377 Posts
  • 65 Reply Likes
The WNB works pretty well for me.  As Bill stated, 40 meters (and 80 meters at my location) is almost not usable without using the WNB.  You do need to get it setup correctly and you also must adjust it when the conditions change or it can introduced "noise" in the audio and on the panadapter.  That being said, it still doesn't work as well as the NB on PowerSDR.  The NB on PowerSDR worked VERY well at my location.  That's the only thing I miss about getting rid of my Flex 5000.
Photo of Ross - K9COX

Ross - K9COX

  • 338 Posts
  • 104 Reply Likes
WNB only works on highly correlated or periodic noise such as electric fences, some types of power line noise or in my case precipitation static. Here is a demonstration of "rain noise" https://www.youtube.com/watch?v=whLREuE15SA
Photo of Kevin K4VD, Elroy

Kevin K4VD, Elroy

  • 775 Posts
  • 171 Reply Likes
Nice demo of WNB. I'm not familiar with precipitation static. Any time I see my noise jump up that far it's because someone turned on some device that I've yet to permanently disable (brute force RFI reduction cheaper than ferrite). I've not found any noise that WNF is effective at reducing, yet. This is probably a good thing for me I think. The Noise Blanker is pretty effective or at least it has an audible affect.

In general, my trace is usually down around -120 db on the spectrum at 20 meters and not much higher at 40 or 80 meters. I'm never quite sure if I'm just in a quiet area or my antennas are just not up to par.
Photo of Jay / NO5J

Jay / NO5J

  • 1577 Posts
  • 228 Reply Likes
Kevin

I suspect you really are in a quiet area, I suspect I am too. There's not all that much electricity in use within about 15 miles of my QTH. I'm in a technological desert. at least one of the nearby craftsman does own an arc welder, but thankfully it seems he only uses it for emergencies.  The regular NB helps a lot during those emergencies.
I've seen the WNB work from time to time here. Band noise at -120 dB is typical for me too, especially when the band has any activity. Late at night. or whenever the band is dead the noise floor heads lower towards -140 dB it never actually get's much lower than -130 dB though. My dummy load's noise floor is normally -128 dB for a 100 Hz wide filter on 20 M, the white noise floor trace on the panadapter makes it appear lower, best guesstimate around -133 dB. I get the -128 dB reading by hovering my mouse over the S-meter on the slice flag, but even that varies +- 2-3 dB from hover to hover. I glad I don't have much use for the WNB. It makes me feel much better about my retirement real estate decision. Thankfully signals heard are usually stronger than -120 dB, and they seem to be plentiful. My Flexradio seems to be working.

 SDRgadgets

73, Jay - NO5J

Photo of G8ZPX

G8ZPX

  • 196 Posts
  • 110 Reply Likes
Hi Ross,

Nice to see it does work for some people.

Not sure what that noise is, but I am pretty certain it is not rain induced static as such. It is way too uniform and structured 9a blanket effect) for static. Rain static is by nature very random and variable. What you may be seeing is something like a power-line or national grid circuit lighting up when it gets wet. If you zoom right in you will see the form better and more like a buzz I suspect.

Rain static (and we get lots here in the UK) often sounds just like an elevated noise floor of white noise. For example on 80m I can tell if its raining or foggy outside just by looking at my noise floor.

Either way, it is obviously doing something useful for you, so you have "the right kind of noise", for WNB. Many people don't get anything meaningful from WNB, therefore cannot be subscribing to the right kind of noise.

I think to be globally useful it needs to be much more "tweakable" as I observe that a fair number of people in fact see little distinct benefit from it.

As Paul suggest, it may be more appropriate to spend development hours on speech extraction / clean-up methods, rather than filters that might or might not work depending upon your location, geography etc.

73 de Steve G1XOW
Photo of Ross - K9COX

Ross - K9COX

  • 338 Posts
  • 104 Reply Likes
Hi Steve, what I can say is that it only happens when it rains, is antenna dependent, has occurred at different QTH's and a friend has also experienced it. Begins at a low level and them reaches a crescendo and the zap its gone. Almost like a charge buildup.
Photo of Paul

Paul

  • 481 Posts
  • 145 Reply Likes
Given the multitude of noise characteristics and the relatively well defined nature of communications quality speech, I can't help wondering if it would be worth FRS considering "speech enhancement" as opposed to (or together with) noise reduction.

I use some freely available software with my receive-only SDR. This in particular offers an algorithm based on an approach proposed by messrs Ephrahim & Malah back in 1984.

I must say it's extremely effective at extracting weak SSB from a wide variety of noise types (and not just that which is "well corrolated"). It is automatic and doesn't seem to colour the audio very much. This approach may be worthy of consideration by FRS (if it isn't already). Just a thought.

http://www-sipl.technion.ac.il/new/St...
Photo of G8ZPX

G8ZPX

  • 196 Posts
  • 110 Reply Likes
Paul,

Is this software available for PC Windows?   I would like to try it.
Photo of Paul

Paul

  • 481 Posts
  • 145 Reply Likes
Hi Steve, yes it is in fact only available for windows. It's written by Simon Brown of HRD fame, and called simply SDRconsole. It has been evolving for some years now so you may have come across it already.

Current version 3 is in "preview" form and seems to be a major re-think compared to the step from v1 to v2. It supports a long list of SDR's but currently not flex (although I believe it's possible to import IQ).

As well as effective speech enhancment/ noise mitigation, there is a server which is compatible with both LAN & WAN remote. Not bad for gratis or a voluntary donation.

Home: http://sdr-radio.com/

yahoo support:

https://groups.yahoo.com/neo/groups/s...

Hope you find it useful.
(Edited)
Photo of G8ZPX

G8ZPX

  • 196 Posts
  • 110 Reply Likes
Thanks Paul, I know Simon from his HRD fame days, so I will have a play with the app and see how it works.
Photo of AA0KM

AA0KM

  • 347 Posts
  • 60 Reply Likes
Fix this one too.