Elecraft

K3 SSB Contour Filter (reduce total bandwidth)

Classic

List

Threaded

9 messages Options

Tony Fegan VE3QF

K3 SSB Contour Filter (reduce total bandwidth)

K3 SSB Contour Filter

Recently I was reminiscing about "Narrow Band Voice Modulation" which
was developed in the late 1970s by Tom Lott VE2AGF and R.W.Harris,
WB6CZX. An important part of this system relies on the fact that between
the low frequency vowel sounds and the high frequency consonants is an
almost void of several hundred Hertz. Lloyd Butler VK5BR developed a
hardware filter but I was looking for a computer filter program to
investigate and experiment with this.

After hunting around on the internet, I suddenly realized that my K3
with KRX3 was all I needed. I locked the VFOs and set the filter on the
main receiver to 350-750 Hz and the sub-receiver to 1400-2400 Hz.
Listening to both receivers is equivalent to a a single receiver with a
bandwidth of 2050 Hz and a notch from 750 Hz to 1400 Hz. This gives a
total bandwidth of 1400 Hz with acceptable audio quality. On some SSB
signals it is possible to reduce the total bandwidth down to 900 Hz.
Reduced bandwidth means better S/N ratio.

I had one signal (USB) with QRM from 1 kHz below. Almost no copy with
normal tight filter but 100% copy with 700 Hz to 1500 Hz notched out.
So, it looks promising but needs some more testing.

Setting filters on CW is relatively simple. On SSB there are so many
variables as you have to consider the transmitted signal, QRM and also
the preferences of the receive operator. It is almost essential to have
preset settings that can be recalled to suit the transmitted voice
characteristics and various levels of noise and QRM.

I am not an audio engineer or a programmer but it should be possible to
develop a firmware (SSB contour) filter with a wide adjustable notch. I
think it would be a worthwhile addition to the K3.

Give the receiver/sub-receiver method a try and see if it helps you
combat noise and QRM while contesting and chasing DX. If you like it, we
will have to talk nicely to Lyle!

73
Tony Fegan VE3QF
K3 #137

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

David Cutter

Re: K3 SSB Contour Filter (reduce total bandwidth)

Didn't they use something like this in the space programs?

David
G3UNA

> K3 SSB Contour Filter
>
> Recently I was reminiscing about "Narrow Band Voice Modulation" which
> was developed in the late 1970s by Tom Lott VE2AGF and R.W.Harris,
> WB6CZX. An important part of this system relies on the fact that between
> the low frequency vowel sounds and the high frequency consonants is an
> almost void of several hundred Hertz. Lloyd Butler VK5BR developed a
> hardware filter but I was looking for a computer filter program to
> investigate and experiment with this.
>
> After hunting around on the internet, I suddenly realized that my K3
> with KRX3 was all I needed. I locked the VFOs and set the filter on the
> main receiver to 350-750 Hz and the sub-receiver to 1400-2400 Hz.
> Listening to both receivers is equivalent to a a single receiver with a
> bandwidth of 2050 Hz and a notch from 750 Hz to 1400 Hz. This gives a
> total bandwidth of 1400 Hz with acceptable audio quality. On some SSB
> signals it is possible to reduce the total bandwidth down to 900 Hz.
> Reduced bandwidth means better S/N ratio.
>
> I had one signal (USB) with QRM from 1 kHz below. Almost no copy with
> normal tight filter but 100% copy with 700 Hz to 1500 Hz notched out.
> So, it looks promising but needs some more testing.
>
> Setting filters on CW is relatively simple. On SSB there are so many
> variables as you have to consider the transmitted signal, QRM and also
> the preferences of the receive operator. It is almost essential to have
> preset settings that can be recalled to suit the transmitted voice
> characteristics and various levels of noise and QRM.
>
> I am not an audio engineer or a programmer but it should be possible to
> develop a firmware (SSB contour) filter with a wide adjustable notch. I
> think it would be a worthwhile addition to the K3.
>
> Give the receiver/sub-receiver method a try and see if it helps you
> combat noise and QRM while contesting and chasing DX. If you like it, we
> will have to talk nicely to Lyle!
>
> 73
> Tony Fegan VE3QF
> K3 #137
>
>

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

Tony Fegan VE3QF

RE: K3 SSB Contour Filter (reduce total bandwidth)

In reply to this post by Tony Fegan VE3QF

Hi Gary,
Thanks for your interest. I am quite aware that the gurus at Elecraft
are extremely busy and need evidence that firmware changes are
beneficial. That is why I asked that those with the KRX3 should try my
method and report their findings.
This method is only good for evaluation but you don't want to tie up
the KRX3 with it.
The brain seems quite happy with a large chunk of bandwidth missing. We
are not talking Hi-Fi but intelligibility. Try it and make your own
decision.
The K3 has a great CW receiver but SSB improvements seem to be on the
back burner.

73
Tony VE3QF

Gary Hvizdak wrote:

> Hi Tony,
>
>
>
> This sounds very cool and your approach most ingenious. However, it
> is likely that Wayne and/or Lyle would want sufficient evidence
> regarding the value of this sort of notch filtering, before expending
> the effort to implement it in firmware. In the mean time, your dual RX
> approach using the sub-receiver should provide you with a useable
> workaround method to accomplish such a notch.
>

>
>
> BTW, I just did a quick search online and found the following link …
>
>
> http://basic-programming.blogspot.com/2005/10/speech-recognition-formants-for-vowels.html
>
>
>
> … which implies that there is no such gap in the spectrum “for all vowel
> sounds”.
>
>
>
> 73,
>
> Gary KI4GGX

Barry N1EU

RE: K3 SSB Contour Filter (reduce total bandwidth)

Tony Fegan VE3QF wrote

I am quite aware that the gurus at Elecraft
are extremely busy and need evidence that firmware changes are
beneficial. That is why I asked that those with the KRX3 should try my
method and report their findings.
The K3 has a great CW receiver but SSB improvements seem to be on the
back burner.

Tony, I verified that intelligibility is maintained with a null in the passband center as you suggested, and it's an interesting effect. But I'm not convinced that this will bring an improvement in QRM immunity. QRM invariably encroaches from the top or bottom of the passband, not from the middle.

Two pieces of spectrum spanning 350-2400hz are used in your scheme. I suspect that this will actually be more vulnerable to QRM than using K3's conventional tight ssb filtering in the range of 1.3-1.8Khz width.

If the hypothesis is that the decreased total bandwidth of the scheme will yield improved intelligibility of very weak signals via better s/n ratio - I'd like to see some recordings made to test this because I wouldn't consider it a given.

73,
Barry N1EU

Ignacy

RE: K3 SSB Contour Filter (reduce total bandwidth)

In contests and especially 160m SSB there are several stations on top of each other. Assuming that the majority of power is within a few hundred Hz, the notch in the middle possibly can reduce interference from stations 500-1000 Hz away.
Ignacy

KK7P

Re: K3 SSB Contour Filter (reduce total bandwidth)

In reply to this post by Barry N1EU

>> I am quite aware that the gurus at Elecraft
>> are extremely busy and need evidence that firmware changes are
>> beneficial. That is why I asked that those with the KRX3 should try my
>> method and report their findings.
>> The K3 has a great CW receiver but SSB improvements seem to be on the
>> back burner.
>>
> Tony, I verified that intelligibility is maintained with a null in the
> passband center as you suggested, and it's an interesting effect. But I'm
> not convinced that this will bring an improvement in QRM immunity...

The NBVM people took this a step further. Once the spectrum was
notched, they shifted the upper frequency range down by the notch width,
resulting in a narrower overall spectrum while maintaining intelligibility.

Ignoring the bandwidth for the moment, the question becomes, "Is the
resulting voice signal acceptable under crowded band conditions?"

73,

Lyle KK7P
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

Tony Fegan VE3QF

Re: K3 SSB Contour Filter (reduce total bandwidth)

Hi Lyle,

Thanks for your comments. Yes with NBVM, at the transmitter end, voice
frequencies above 1500 Hz were folded back (sideband reversal) into the
gap . The process was reversed in the receiver. As far as I remember,
Tom VE2AGF told me they even found it unnecessary to null out the middle
frequencies. I assume (shouldn't do that!) that the received processed
audio middle frequencies would contain both the high end vowel
frequencies with reverse sideband consonant frequencies. The reverse
sideband frequencies would dominate.
I don't think we even want to consider NBVM transmitting. There are
enough different modes now and a digital solution would be more
bandwidth efficient.
The contour (for want of a better name) approach does seem to help in
those situations where stations are spaced 1 kc apart. Using USB for
example, the station 1 kc lower will will have its high frequency
consonants mainly in the gap (null) of the receiving station. The middle
frequencies (low power unless heavily processed) will fall over the
vowel frequencies of the wanted signal. The station higher in frequency
will have its vowel frequencies at the high end of the gap and its
middle frequencies will be over the high consonant frequency band.
I have not had time to give it a good test. The noise reduction of the
narrower bandwidth is not significant but annoyance of the band noise in
the gap is eliminated. It is debatable whether the built in NR is better.
I think filter settings need to be set up with macros. Manual
adjustment in the heat of the battle is way too slow. I suggested before
that maybe the XFIL (or SPOT or CWT) which are not used on SSB could be
used for customizable filter settings (in sequence).

It does need a crowded band SSB contest style to give it a good test.
It will need fast switching for comparison. Set your filter macros in
the "K3 Utility" Command Tester.

73
Tony VE3QF

Lyle Johnson wrote:

>
> The NBVM people took this a step further. Once the spectrum was
> notched, they shifted the upper frequency range down by the notch width,
> resulting in a narrower overall spectrum while maintaining intelligibility.
>
> Ignoring the bandwidth for the moment, the question becomes, "Is the
> resulting voice signal acceptable under crowded band conditions?"
>
> 73,
>
> Lyle KK7P

Joe Subich, W4TV-3

Re: K3 SSB Contour Filter (reduce total bandwidth)

In reply to this post by Tony Fegan VE3QF

Tony,

> I have not had time to give it a good test. The noise
> reduction of the narrower bandwidth is not significant but
> annoyance of the band noise in the gap is eliminated. It is
> debatable whether the built in NR is better.
> I think filter settings need to be set up with macros.
> Manual adjustment in the heat of the battle is way too slow.
> I suggested before that maybe the XFIL (or SPOT or CWT) which
> are not used on SSB could be used for customizable filter
> settings (in sequence).

I played around with the RX EQ and manual notch a bit the other
day trying to simulate a SSB contour filter. The experiments
lead to a couple of ideas for Lyle to play with ....

1) "Dual PB" is not operational in SSB ... how about an overlay
filter than simply provides a broad 25 to 30 dB notch from
800Hz to 1300 Hz?

The depth and exact width might be fertile ground for some
further experimentation but it is this 800-1300 Hz area (+/-
100 Hz on either end) that lacks voiced energy in human
speech. Fundamental, second and third harmonics generally
exist below 800 Hz (up to 1000 Hz in some female voices)
while the sibilance generally begins in the range above
1500 Hz.

2) Since the 50 Hz band of the RX EQ has very little if any
effect even when the audio response is extended down to
100 Hz, eliminate that band and add a new band at 1200 Hz.
In addition, extend the range of "cut" available to -30 dB
(from -16 dB).

With the additional capabilities, the user can use RX EQ
to create his own "SSB contour" filter by dialing in as
much or little cut at 800 and 1200 Hz as is appropriate.

> Ignoring the bandwidth for the moment, the question becomes,
> "Is the resulting voice signal acceptable under crowded band
> conditions?"

Lyle, on receive it's possible to demonstrate some of the benefit
of a "voice contouring filter" simply by setting the 800 Hz RX EQ
to -16 dB and the manual notch to 1100 or 1200 Hz.

Audio engineers have made use of a 1000 - 1200 Hz notch for many
years to make band limited audio sound less muffled and make
voices more intelligible.

73,

... Joe, W4TV

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

Stephen Prior

Re: K3 SSB Contour Filter (reduce total bandwidth)

>
> Lyle, on receive it's possible to demonstrate some of the benefit
> of a "voice contouring filter" simply by setting the 800 Hz RX EQ
> to -16 dB and the manual notch to 1100 or 1200 Hz.
>
> Audio engineers have made use of a 1000 - 1200 Hz notch for many
> years to make band limited audio sound less muffled and make
> voices more intelligible.
>
> 73,
>
> ... Joe, W4TV
>
>

This is really very impressive Joe. I'm listening on 18.143 at the moment
and adding the notch and equalising as you suggest above actually turn
barely readable audio into quite comfortable copy. Someone your side of the
pond is talking about the K3, not sure of the callsign though!

73 and Happy Christmas!

Stephen G4SJP

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com