I was having trouble with my WNB on 20 Meters, ....

  • 5
  • Question
  • Updated 2 years ago
  • Answered
I was getting a lot of pumping of the WNB on 20 meters and figured that there must be some strong station with very sharp transitions (distortion) on the signal.  Guess what I found?  Radio Havana at 15.370 MHz running 60 KHz wide and very distorted.

Here is a link to my recording.  With such distortion and sharp edges (or crossover, or whatever others call it) it is no wonder that the WNB was catching impulses and pumping.

https://youtu.be/0XHk8Q9WDSk

This is a ridiculous signal....
But it made me wonder.  Is there a way to define the WNB so that it samples a more limited bandwidth of signals in order to do its work so that I can exclude such nasty out of band signals from the WNB's calculations?

Ken - NM9P
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3969 Posts
  • 1225 Reply Likes

Posted 2 years ago

  • 5
Photo of G8ZPX

G8ZPX

  • 196 Posts
  • 110 Reply Likes
Ken,

I have also seen this on the upper end of 40m band around 7.160 - 7.200. In Europe we often see some very strong international broadcast stations above 7.2. This renders the WNB almost unusable on 40m, and to an extent 80m SSB DX segment as well. So yes please, a more definable "passband" for the WNB should be very very effective.

Also, I feel that as it could be configured to just process say the current panadapor visible range then we might actually see more dynamic noise reduction because it can be better focused on the noise we are currently experiencing, not the much wider bandwidth we don't really care about.

Looking forward to the WNB focus slider...

73 de Steve G1XOW
(Edited)
Photo of Alex - DH2ID

Alex - DH2ID, Elmer

  • 891 Posts
  • 166 Reply Likes
@Ken, @ Steve I like the idea of a WNB focus slider or bandwidth sliders showing
the focus and bandwitdth numbers. That should not be difficult to implement.
73, Alex - DH2ID
Photo of K1UO - Larry

K1UO - Larry

  • 773 Posts
  • 126 Reply Likes

This is a "me too" moment... 


Great idea to allow adjustment to the Band range of interest somehow.

Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3466 Reply Likes
The answer is no, there is not a control that will allow you to define the WNB bandwidth.  Making the panadapter bandwidth smaller will effectively reduce the bandwidth WNB uses, but I do not think you can get it narrow enough.  You can try it and see if that makes a difference.

In addition, I have added these comments into a program change request (technical a feature request) as item #3807 in our bug tracker
Photo of Ted, NX6C

Ted, NX6C

  • 76 Posts
  • 26 Reply Likes
Tim,
   This is very helpful.  did I miss this fact in the documents?
Ted
NX6C
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3466 Reply Likes
I am not certain we described that particular behavior.  It really only becomes relevant if the panadapter is really wide (as in MHz)
Photo of Steven G1XOW

Steven G1XOW

  • 282 Posts
  • 119 Reply Likes
Tim

Next time I see it on 40m I will make a video for you.

Normally I have an on-screen pan of about 100kHz. Typically the QRM source is around 150-300 kHz HF of what I am looking at (in the 41m am broadcast band).

The last one I remember was a far-east station, radio china or similar. It was about s9+20 and pretty wide on my half-wave vertical. Using the WNB was thus impossible on 7.160MHz.

I will try a tacking filter on it next time and see it that helps. Not sure if the WNB sampling is pre or post TNF? 
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3969 Posts
  • 1225 Reply Likes
Hey, that's an idea I hadn't considered before! 

Would a wide TNF filter take this station out of consideration for the WNB?  I have never even tested to see how wide a TNF can go.

My guess is that the WNB takes its samples from the ADC before the tap that drives the notching or demodulation routines.  But it would be an interesting, but probably very complicated, set of calculations to have "exclusion zones" from the WNB......
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3466 Reply Likes
The WNB processing is before TNF.
Photo of HCampbell  WB4IVF

HCampbell WB4IVF

  • 170 Posts
  • 38 Reply Likes

 Ken, thanks a heap for tracking down the cause of the pumping.  I’ve noticed similar behavior many times but never took the time to investigate further.  If WNB can be improved in such cases by giving the user the option of narrowing the range that WNB searches (maybe even down to the slice receiver range?) then I’m all for it.  Great idea! 

Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3466 Reply Likes
The problem is the bandwidth sampled has to be wide enough to gather an appropriate amount of data to properly correlate the noise and effectively remove it from the spectrum.  We have optimized that in the current implementation and make the bandwidth smaller will make WNB less effective.
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3969 Posts
  • 1225 Reply Likes
That makes sense, Tim.  More information means better analysis of correlation.

It stimulates another thought, though.  (My mind drifting off again into half-informed, off the wall SDR theory......)

Knowing that you need a certain bandwidth to analyse in order to assure good performance, would it be possible to have some control as to from WHERE that bandwidth is sampled?  For example....if my antenna response is not even and some of the samples are being gathered from "dead" sections of the antennas receive response, could the "sample band" be shifted a little to take advantage of a more representative sample of the actual noise being received by the station?  

OR, if there is a nasty signal like the Radio Havana signal I posted, could the sample bandwidth be shifted AWAY from that end of the panadapter to include more of the bandwidth on the lower end and less on the high side so that the offending signal and its sharp peaks is not included into the sample to distort the algorithm?

Think of this sort of like a giant "IF Shift" knob on my old Icom rig that didn't change the actual bandwith of the filter, but shifted the location of the filter bandwidth up or down to emphasize either the higher end or the lower end of a signal.   If the sample is, for example, 500 KHz wide with a 200 KHz wide panadapter, instead of having an extra 150 KHz evenly spaced on each side of the panadapter, it could be shifted to sample 50 on one side and 250 on the other, or vice verse, (or 0 and 300) as needed to give a more effective sample to the WNB.  

Just my brain going into overdrive, thinking up more possibilities....(I probably need to visit Austin and sit down with the software engineers over a couple of Red Bull's and get this stuff out of my system.)  (;>p)

Ken - NM9P
Photo of Steven G1XOW

Steven G1XOW

  • 282 Posts
  • 119 Reply Likes
In an attempt to provide some valuable feedback regarding the WNB problem to Flex software coders, I've just made a short video to demonstrate the problem. It appears to be confined to the bands that also have a big AM broadcast segment at one end of the ham band (especially, 20, 40 and 80m in my case). It looks like the WNB is sampling at least 2-3MHz wide which will always mean that some non-ham band sigs will get included. This seems excessively wide sampling to correlate against just a few kHz that we are actually interested in. Could SSDR limit WNBs sample scope to the ham band limits, or have some definable slider upper/lower limits?  This may result in WNB being less effective when set for narrower sampling, but better that than it being entirely usable for some.

Personally I don't find much need for WNB at my quiet country location but others may need this facility to work without the overload pumping action when AM broadcasters are around.

https://youtu.be/r9Sq9ftBv-M

73 de Steve G1XOW
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3969 Posts
  • 1225 Reply Likes
Steve, I have never had any overload problems before. This signal was about S9+40 to 50. Although I don't remember turning off the preamp, I did shift antennas. The vertical was about 20 dB down, and the signal was still very wide and distorted. Even with NO antenna, it was the worst sounding, widest, AM signal I have ever heard, except for one CBer who was splattering about 1.5 MHz each side of center! I wish I had a recording of that one!

Anyway, the Radio Havana had extremely sharp modulation products and the distortion was of a type I don't think I have ever heard before. It was just strange.

If I hear it again, I will compare preamp and attenuators settings.
Photo of mikeatthebeach .

mikeatthebeach .

  • 247 Posts
  • 38 Reply Likes
Got Two Tracking Filter Boards from SV1AFN plus his preamp board via Ebay for my ICOM IC-7300 HF thru 6 meters Rig to cut out out of band Rx interference

Using an Arduino Switch Band Decoder IP Switch Unit from RemoteQTH.com 

This 9-Band switchable filter ( uses 2-in series BPF filters with a mid-section preamp ) from SV1AFN " HF Bands Preselector Boards" with a DX Engineering RTR1A
switch box or radio's with only one Antenna port & no way to add preamps/filtrs.

It may be possible to use this Band Switching Tracking Filter reading Band Data 
from DDUtil to feed the RemoteQTH.com Switch Band Decoder 

This might be the Answer for a Flex6300 (No-Preselct Filters) but may be way 
overkill for a Flex6500/Flex6700 that already has internal preselect filter built-in.

Will see if I can adapt this to my Flex6500 just as a test to see what happens.

Like I said - way overkill for a Flex6500/Flex6700  

Any comments welcome 

73 Mike WB6DJI
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3969 Posts
  • 1225 Reply Likes
Are those receive-only filters?  If so, on the 6500/6700 they would work nicely in the RXA/Loop or RXB/Loop and would never be in line with the transmitter.

In any case, for me this is not a regular, continuing situation and I posted it mostly because the signal from Radio Havana was so extraordinarily wide and distorted.  I have seen extremely sharp modulation kick a noise blanker before, but never this badly, and not this badly with WNB.  

I still do not think it is overload in my case, and I have NEVER seen any signals as strong as the ones that Steve - G1XOW posted, even with two 1200 Watt stations less than a mile away from my station.  I have been amazed at how crunch-proof this rig has been,

But the event brought up the interesting thought of what I will call "band shifting" the WNB sampling band, for lack of other technical terminology.  

It has been an interesting and enlightening discussion of the whole WNB subroutine.
Photo of mikeatthebeach .

mikeatthebeach .

  • 247 Posts
  • 38 Reply Likes
These are receive only BPF Filters
Photo of Steven G1XOW

Steven G1XOW

  • 282 Posts
  • 119 Reply Likes
On my front-end I already have the Heros pre-selector. https://www.herostechnology.co.uk/pages/SCR_Preselector.html

However, as the QRM source is only about 7-800kHz away, any passive analogue filtering will not be sharp enough when so close to the passband. You'd need a kind of tracking notch filter prior to the WNB sampling to do any good. I.e. strip our / notch the QRM from the sampling scope.
(Edited)
Photo of Steven G1XOW

Steven G1XOW

  • 282 Posts
  • 119 Reply Likes
Gerald,

As I said in my post, personally I don't need WNB so it's not an issue for me. The few experiments I did proved it was pointless as it was impossible to make the massive AM broadcast signals low enough relative to the signal we are interested in. Being up to 10 times louder than the miniscule DX in the noise I am trying to receive.

There is no facility for "spot-attenuation", and a sig 10 times louder that is just out of the 20m band cannot be reduced by filters to any large degree without badly affecting performance of the desired in-band signal too. I use the well respected Heros Filter (external to the F6k) which makes no difference with such close-in sigs.

I also tried adjusting the slider downward (it was set at about 40 in the video to ensure the demonstration). Once the slider drops below about 20 there is no benefit anyway. Also, I have notice the same pumping affect at the top end of 40m (7.170) even with the slider set at 0 and WNB turned on.

I feel there is a fundamental design flaw in the math/sampling scheme. The massive AM broadcast signals being sampled versus the tiny desired signals will never lead to happy math (weighted averaging) and thus the blanking will be ineffective when any big sigs are present. I suspect the same would be true in any big contest such as WPX SSB too.

To work really well, some tweaking is required (or the ability to notch-out the AM broadcast bands QRM prior to averaging), Otherwise its function will only be useful in a small number of cases (those without big AM BC QRM), and hence why some claim WNB is a gift from god, yet others think the opposite.

In a nutshell, it needs more work based around real-life environmental situations to be a really useful facility.

73 de Steve G1XOW
(Edited)
Photo of Fred GLENN

Fred GLENN

  • 54 Posts
  • 7 Reply Likes
FYI, nice band openings with huge pileups on 40m this morning gave me a chance to investigate a bit further. Definite problem with WNB selected but also see the noise pumping with the NB function activated. Nowhere near as bad, but quite noticeable. Able to copy strong stations in the noise with the narrow band NB on 5-10KHz away ... dissappears when turned off. 

Not really a problem for me now that I know to shut them off. Thanks again for the hints. Much appreciated. 

73,
Fred