Noise blanker/reduction settings?

[G8VR]

In the G7CNF video demonstrating the Anan 100D using diversity to deal with noise, he also shows demos of the noise reduction/noise blanker. When he uses these tools, you can see the intensity in the waterfall drop away as the noise is dealt with. When I insert NB1 or 2 or indeed the NR1/2 or the SNB nothing ever changes in the waterfall. I can hear an audible difference in the speaker, but nothing in the waterfall. What I am looking to do is find optimal settings for the blanker and noise reduction and also get my waterfall to reduce in intensity when the blanker/reduction is enabled - always assuming it's actually doing something. Can anyone help with some settings to enable me to do that? I've already searched the board and can't find any answers to this.

I should add that when I beam at noise here, the intensity in my waterfall increases (more snow shall we say) and decreases when I beam off it, so my feeling is something not right in my blanker/reduction settings.

Many thanks in advance.

[w-u-2-o]

Do you have a link to that video?

[K9RX]

not in direct response to your posting but definitely in response to your heading for this thread: I played with the NB settings... and found that a THRESHOLD value of 15 is about a 1-2 db improvement for me over the default setting of 32. Consistent - audible - seen on the waterfall and on the S-meter (dbm). As an example of its abilities: w/o NB2 I can see the ghost signal of a station but he's not decoding and yet consistently, with NB2 on, that same station is -11 to -9.

I would add that in one direction on 12/10 I have noise to my SE that is about 12 db higher then say towards the NE ... and the blanker does nothing to it. So noise (of course) varies in its particulars and thus the ability for the NB/NB2 to handle it will be highly dependent on that factor.

Gary
K9RX

[w-u-2-o]

Kerry,

NB and NB2 are “wideband blankers” and operate on the IF passband (the entire bandwidth of your spectral display). So those two functions will have a visible effect on the spectral display. NR, NR2 operate on the receiver passband and the results are not visible on the spectral display. SNB operates on a bandwidth slightly wider than the receiver passband; however, for practical purposes it’s fine to just say it operates on the receiver passband, so again not visible on the spectral display.

The specific spectral and time domain characteristics of the noise or interference in question strongly affect how well any of these functions work singly or in parallel. That is why there are so many of these tools.

You can glean additional technical details on each noise mitigation method here:

https://github.com/TAPR/OpenHPSDR-wdsp/blob/master/WDSP%20Guide.pdf

With respect to the G7CNF video (I think I have the correct one linked), I'm afraid his demonstration of the effectiveness of the Diversity function is quite poor. He does not show a desired signal in the passband. When you have a desired signal and are trying to null out an interfering signal near it or on top of it, the way you adjust the phase and amplitude controls, and the effects on the spectral display, are quite different than pictured. Generally you only see interference suppression right around the desired signal, not across the entire display. And remember that this only works against coherent, interfering signals that come from a point source, not against random, atmospheric noise. You may find reading this topic to be helpful:

https://apache-labs.com/community/viewtopic.php?f=21&t=2362

73,

Scott

[G8VR]

Gentlemen
This is exactly the kind of information I was looking for. I'm going to go away and have a read of the links and make some of those setting changes.

I have a big noise source locally to me, it's on pretty much night and day and represents about 25db vs my quietest direction here in North London. I think it's coming from a house not far away, but esp during the lockdown here no one is going to entertain a visit!

I have to admit I can be slow to learn, but I very much appreciate the help - so a big thanks. I'll report back in due course.

[G8VR]

OK, I see why that video isn't exactly helpful or representative. Here's one I made today on 6m using the diversity to kill a huge local noise source. As you can see workable and then totally lost in the noise. Thanks to all who help others on this site.

https://www.dropbox.com/s/4o5jokb35qlov ... 3.mov?dl=0

[w-u-2-o]

[G3ZQH]

Hi Kerry
Great result with diversity on 6m. I have a similar problem on 6m with a 10-15dB interference from a specific beam heading. I can't achieve your results- no more than a few dBs of improvement at best using diversity. What did you use for your RX2 antenna?

Of interest is that the noise I have is almost completely suppressed by the NB noise blanker BUT on FT8 there is considerable distortion of signals which means that it is not usuable in practice,

Here, I have an Anan 200D with 5el 6m and using a 10m doublet as the RX2 antenna. Possibly I need something different fro RX2?

73 Dave G3ZQH

[G8VR]

Hi Dave
Yes, the result was better than I could have imagined. The noise and the station were coming from the same direction so it was perfect.

The noise blanker does nothing at all to this! When I use it I too, I see the distortion you refer to when running FT-8, though as I say no loss of noise.

The noise antenna is my topband inverted l! I need to try a proper 6m antenna to see what difference that makes. As the noise is so strong we wonder if it's actually making it's way across the pcb into the supply for ADC2 and why I'm getting such a good result with such a poor sense antenna

Here, I have an Anan 8k with 2 x 6 ele OP-DES antennas for 6m at 1/2 wave spacing.

73's

Kerry G8VR

[G3ZQH]

Hi Kerry
OK, interesting. We seem to have different types of noise, though I wouldn't have thought this would affect the noise cancelling function. I was surprised that my doublet seems to RX well on 6m during Es propagation, though obviously not as good as the 6m yagi.

I tried to use a 160m Beverage as the noise antenna, but again this did not work. It was worth a try and I was encouraged by your great success. Maybe for the next noise source it will work better (there is always a new noise to deal with in my experience...).

73 Dave G3ZQH

[w-u-2-o]

Remember that for "diversity" (really coherent beam forming) to work, the noise must be from a discrete, point source so that it is coherent across both receiver channels (has a non-random, non-time varying phase relationship). Thus atmospheric noise, which exhibits a random and time-varying phase relationship between both receiver channels, cannot be eliminated by using diversity.

So if you've got something coming from a plasma TV or air conditioner that can usually be dealt with. More physically distributed noise like that from a long length of power lines can be difficult to suppress. Again, atmospheric noise is a non-starter, that's what the various NB and NR options are for.

In the work that I do we call a coherent signal like this an "interferer" or "interfering signal" in order to distinguish it from atmospheric and other random noise.

[G3ZQH]

Hi Scott
I understand what you are saying about noise- what I have is not atmospheric noise, but power line noise might be a possibility. I wish I knew more about the nature of the noise. It comes in the direction where there are many houses, a few factories, a building site and so on and would be difficult/impossible to trace. It is at least several MHz wide, 10-15dB above background and is there day and night. It has appeared only over the last 3 months.

Kerry G8VR would understand that at 90-120 degrees beam heading from UK it has been very frustrating on 6m to see others working so much good DX in Asia (VU, 9M, 4S, XV etc) which I can't RX at all because of the noise! To the west from UK my noise is low and I get excellent copy of NA on 6m.

The other observation I had was the effectiveness of the NB function in almost completely removing the unwanted noise across the whole band. This looks very impressive indeed. However, as I mentioned before, the NB function also distorts FT8 signals reducing decoding. Maybe this is inherent in the NB algorithm but possibly there might be a solution in future to avoid this specific problem?

73 Dave G3ZQH

[Tony EI7BMB]

Hi Dave, I use NB quite a bit on 6m FT8 and although it does create images of the signal i have not found to to lessen the decode efficiency. What level do you have it set at if I may ask ?

[G3ZQH]

Hi Tony
I have my NB threshold set on 15 and have not changed any of the NB Times (all set to 0.01 ms). Tried a few settings and found this to be best. I must admit that once I saw the fragmentation affect by NB on the FT8 waterfall, I assumed the decodes would be badly affected, so I didn't actually check if this was the case. I'll experiment further with this after what you have said.

I have noticed that there is no distortion of weak FT8 signals with NB and it is only the stronger signals (greater than about -5 dB) that are affected.

73 Dave G3ZQH

[Tony EI7BMB]

Yep that would be my experience too Dave