RX Equalizer tailored for CW

  • 8
  • Idea
  • Updated 5 years ago
  • Under Consideration
RX Equalizer tailored for CW
I have been using the RX equalizer on CW. With the 500HZ slider set to +10db and the other frequencies set to -10DB it provides a small but beneficial amount of audio peaking.

Would it be feasible to have the RX equalizer tailored for CW with a narrower range of frequency selections? This should allow the peaking to be greater over a smaller frequency range like on some other radios. Does this seem logical and feasible? See snapshot, CLICK to enlarge.

Regards, Al / NN4ZZ

Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes

Posted 5 years ago

  • 8
Photo of Sergey, R5AU

Sergey, R5AU

  • 835 Posts
  • 109 Reply Likes
Al, what do you think regarding APF implementation in this case with flat EQ for CW - result will much better
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Sergey,
I don't know how the approach you mentioned compares (or if I understood it correctly) but it would be interesting to see what Steve/engineering thinks.

There is an older thread about adding an Audio Peaking control. My idea is to see what engineering thinks about tweaking the RX Equalizer. If they agree it is feasible, it might be easier to implement than adding a new control

On some other radios with audio DSP it works very well. A sharp boost in the audio chain around the CW frequency (e.g. 600HZ or your preferred frequency) along with a reduction on the surrounding audio frequencies does make a weak CW signal pop. Or you can leave the other frequencies flat and just boost the frequency you want. Audio DSP coupled with the narrow 50HZ or 100 HZ filter could make a big difference for weak signal CW DX.

Regards, Al / NN4ZZ
Photo of Sergey, R5AU

Sergey, R5AU

  • 835 Posts
  • 109 Reply Likes
I got it Al, agree, combination of the narrow filters with additional DSP functionality can give dedicated effect instead with regular APF functionality, BTW it should be a part of the further DSP improvements, I think we need flex team statement on this case.

P.S. Yaesu, Kenwood, Icom has own point of view on this issue.
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3970 Posts
  • 1225 Reply Likes
It seems as though we have a natural frequency parameter in the CW offset tone control. If they could program an audio peaking filter centered on the CW offset frequency they wouldn't need another control, just a button to turn it on.

Then for Digi modes, they could have it centered on the "sweet spot" that could be defined on a panel for Digi modes similar to the SSB & CW TX panels.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Ken,
A potential advantage of using the RX Equalizer for this is you can tailor to the audio width you prefer. Moving a single slider to +10DB and all others to -10DB would be narrowest. Moving a single slider to +10DB and the adjacent sliders flat or slightly up would be a little wider, etc.

And it can still be turned on/off by engaging the RX Equalizer button. Anyway just some thoughts until we hear from the experts in engineering.

Regards, Al / NN4ZZ
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3970 Posts
  • 1225 Reply Likes
Could this be handled easier by having the center frequency track the CW offset and then having a "Depth" or "Gain" control to adjust how deeply to attenuate the side frequencies, and a "Width" control to determine how far on each side of center to peak the signal? It seems easier to me to adjust two controls than 8 RX EQ bars. The gain difference could also be made greater than 20dB, too!
Photo of W5XZ - dan

W5XZ - dan

  • 564 Posts
  • 83 Reply Likes
I tried this yesterday on 160m; normal 'beacon' HL5IVL was just barely readable on 1823...seemed to help, but the path was just too crummy...QSB just took him out...

top band WAS lousy at my QTH, anyway...

look forward to playing with it some more...73
Photo of Dale KB5VE

Dale KB5VE

  • 411 Posts
  • 56 Reply Likes
I have used this method on copying cw, I used the Agc threashold to do final touch after the receive eq. If nothing else it makes the copy more pleasant to listen to.
Photo of Ned K1NJ

Ned K1NJ

  • 301 Posts
  • 75 Reply Likes
I've been doing this for a while. It works. In PowerSDR the equalizer is
recalled in definable profiles as a part of its "persistence".

When you define an equalization profile in smartsdr for CW, or any mode,
it should be recalled when using that mode or profile again. For CW
exclusively, an APF, an audio peaking filter, would be right to the point.
I like your suggestion, but I also feel that "persistence" should cover it.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Ned,
Agree on persistence of the profile by mode. The main idea for tailoring the RX Equalizer for CW is to provide a much narrower set of ranges for CW. For SSB the wider range (as it is today) is good for voice. For if to be effective for CW peaking, it should be much narrower.

Existing range between sliders is 500 HZ or more
Suggested range is 50 HZ (maybe even less would be better)

Regards, Al / NN4ZZ

Photo of Peter Driessen

Peter Driessen

  • 17 Posts
  • 1 Reply Like
Why do we need audio peaking filter (APF)? The FIR filters operating on the IQ signals before the weaver demodulator can in principle be adjusted to be as narrow as desired (at the cost of more filter taps and more delay). The APF cannot eliminate interference inside the FIR filter passband. Am I missing something?
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Peter,
A peaking filter in the audio chain can make a weak DX signal pop compared to the the noise because it also negates any unwanted audio frequencies that are also present withing the FIR filter range. At least that is the effect on other radios I have used and you can hear it even now on the 6x00.

With the suggested change and using the setting in the example below , It bumps up the tone you want to hear (e.g. 600HZ) and reduces the frequencies you don't (e.g. 650 and up and 550 and down).

Even with the wide RX equalization available now you can hear the effect although it is not dramatic as it could be. .

So that is the theory, but I'd like to hear from the FRS engineers to get their feedback on the viability of this idea in the 6x00 radio architecture.

Regards, Al / NN4ZZ

Photo of Peter Driessen

Peter Driessen

  • 17 Posts
  • 1 Reply Like
The FIR filter can in principle be made just as narrow as the APF (at the cost of delay and processing cycles) and has the advantage of eliminating interference before the AGC.

Unwanted frequencies within the FIR filter passband will affect the AGC, and if they are significantly stronger than the desired signal, may cause the desired signal to be suppressed (since the AGC level will be determined by the stronger signal).
Photo of K1UO - Larry

K1UO - Larry

  • 773 Posts
  • 126 Reply Likes
It is good the DSP filtering does have steep "brick wall" skirts and is very effective eliminating qrm. and it is true the better "brick wall filtered" radios will benefit less from an APF but they will still benefit. the APF is a filter with a very narrow peak (usually about 30 Hz
at the -3 dB points) and very broad skirts. The narrow peak brings up
the signal amplitude slightly, while the broad skirts prevent noise
from being amplified and delayed in such a way as to dominate the
signal. Even in this case I find it is better to set the FIR passband around 200-250Hz. As an example, the APF function used in my Elecraft K3 is a modern DSP interpretation of the hardware APF circuit found on some legacy transceivers likethe FT1000. Some swear by its effectiveness and some don't.
The APF in earlier models of the FT1000 I had was very effective.... that was before Yaesu "improved" the radio in their later released versions. those units required a hardware Mod to get the APF back to its earlier effective performance.
Im still not sure how Flex could/wouldl implement the APF feature but I too would like to hear from the Flex Engineers concerning its viability.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
This is also being discussed on the Yahoo forum. Here are two excerpts from a recent post by Rob Sherwood / NC0B.

...Just narrowing the IF bandwidth isn't the same as a good APF for weak CW in noise...

... an APF (audio peak filter), and yes Gerald said it could easily be done back in May when he was at my QTH,..

This sounds encouraging so I'm looking forward to feedback from FRS engineering. .

Regards, Al / NN4ZZ
Photo of W5UN_Dave

W5UN_Dave

  • 315 Posts
  • 30 Reply Likes
Larry, it can be done. As I explained it to Gerald Youngblood, he responded "Oh, an inverted notch filter". I think we are making progress on selling the implementation. of an APF
Photo of K1UO - Larry

K1UO - Larry

  • 773 Posts
  • 126 Reply Likes
I hope so Dave..they got my vote to implement it!
Photo of Peter Driessen

Peter Driessen

  • 17 Posts
  • 1 Reply Like
A close-to-optimal filter for CW would be a filter matched to the shape of a dit. The bandwidth will depend on the shape of the keying envelope. A nice explanation is at http://www.w8ji.com/cw_bandwidth_desc....

For reception by computer, the matched filter can be synchronized to the timing of the bauds. An APF won't help the matched filter work any better.

For reception by ear, it may be a different story. Can the effect of an APF be replicated by a suitable IF (IQ) filter or is it essential that the APF be after the weaver demodulator?
Photo of James Nelson

James Nelson

  • 111 Posts
  • 11 Reply Likes
That makes very good sense. What a great idea .....
Photo of Sergey, R5AU

Sergey, R5AU

  • 835 Posts
  • 109 Reply Likes
After several testing i fully agree with Al idea:
unnecessay noises can be suppressed well with signal separation, APF a bit different type of the functionality what should also be implemented
but need to keep in mind:
1. CW EQ should have more sliders to cover all CW filters passbans
2. Definetely i think EQ should be a part of the internal CW menu and part of persistance - in case you turn it once, further is not necessary. You are turn you diferent passbands with appropriate EQ characteristics for each filter passband
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Eric - KE5DTO and
Steve - N5AC

What do you think about the idea, can we get this added to the "under consideration" list?

Regards, Al / NN4ZZ
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3970 Posts
  • 1225 Reply Likes
I agree, some sort of audio enhancement that reduces audio hiss at frequencies on either side of the CW audio tone would be helpful. Done well it could enhance readability of signals and reduce the "fatigue factor" as well.

It could be done either via an expanded RX EQ for CW/DIGI modes as Al, Sergey and others have suggested, or via a tracking audio peak filter that is automatically centered on the CW offset tone or DIGI "Sweet Spot" and has a "depth" or "gain" control and a "Width" control in order to tailor the degree of peaking and the depth of rejection of noise and hiss outside the desired passband.
Photo of Peter Driessen

Peter Driessen

  • 17 Posts
  • 1 Reply Like
Further to my previous note, what is the benefit of an APF over an IF (IQ) filter of comparable bandwidth? There should not be any audio hiss if the only input to the audio stage is the output of the IQ filter. (The audio stage presumably has at least 16 bits of dynamic range)
Photo of W5UN_Dave

W5UN_Dave

  • 315 Posts
  • 30 Reply Likes
Here is my take on how the APF should operate: it should be very narrow (Elecraft K3 has it at 5 hz); it should be adjustable independent of the radio frequency; when enabled, it should appear on the chosen signal at the existing audio pitch. It should be independent of the filter setting, but its pitch adjustment within that filter should be by fine tuning of the FlexKnob.. Whenever the APF is enabled, the FlexControl knob should also be enabled for adjusting it's pitch. When the APF is disabled, the FlexKnob should revert back to frequency adjustment.

On the other hand; is this sort of thing possible in the frequency domain rather than the audio domain?

As Mr. Youngblood remarked at the BBQ: "Oh, its the same as an inverted notch filter". except in the audio domain.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Copied from another thread......

Dave,
If the FRS team would like to utilize the RX Equalizer control, how about this idea.

Instead of specific frequencies to enhance or attenuate, set it up to be plus and minus offsets from the PITCH frequency. That way you can set it to be very narrow (down to 4 HZ) or wider (30 HZ in my example) based on your preference. The filter can still be toggled on/off easily and the width can be adjusted easily also.

Do you think this has merit?

Regards, Al / NN4ZZ

CLICK to Enlarge

Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3970 Posts
  • 1225 Reply Likes
AL,
I like this idea better than the other one with EQ steps every 50 Hz.
Tracking the CW Pitch is the way to go.

I still think that 2 simple sliders, one for width (from 250 Hz to 5 Hz) and one for "Gain" (flat to 30 dB or 40dB) would be easier to adjust. Like this:

[Break-IN] [Iambic] [Pitch] [< 650 >]

APF Width..[150 Hz].....[250 : : : : : : [|] : : : : : : : : : : : : :5]

APF Gain....[ 20 dB].....[0 : : : : : : :: : : : : : : [|] : : : : : : 30]
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Ken,
I like your idea too. One potential downside is that is doesn't support the ability to negatively impact the frequencies outside the peaking range. (i.e. up to -10db for frequencies you don't want to hear)

Another reason I suggested the same equalizer control was from a coding perspective it might be easier for FRS to implement it using the same control. So easier might mean faster to deliver and more likely to be considered.

Regards, Al / NN4ZZ
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 3970 Posts
  • 1225 Reply Likes
My assumption, perhaps poorly expressed, is that the "APF gain" control would simultaneously peak the desired frequency and cut the undesired frequencies. So if it was set for "10 dB" it would peak the center by 5 dB and drop the rest by 5 dB. A setting of "20 dB" should peak the desired signal by 10 dB and cut the signals outside the skirts by 10 dB, etc... until we get to about 30 or 40 dB total peak & cut...(or perhaps it should cut more than it peaks, perhaps a 1/3 by 2/3 split....so that raising the APF gain doesn't make the desired signal way too loud compared with the same signal without the APF?)

Perhaps "APF Sharpness" would be a better term, so as to avoid confusion...
Photo of W5XZ - dan

W5XZ - dan

  • 564 Posts
  • 83 Reply Likes
Why not just make it track the pitch control, automatically??
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Dan,
Yes that is what I meant to convey. Track the pitch setting automatically but use the sliders to set the width of the filter ( 4 HZ to 30 HZ).

Regards, Al / NN4ZZ
Photo of W5XZ - dan

W5XZ - dan

  • 564 Posts
  • 83 Reply Likes
with adjustable 'Q' ??
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Yes, the sliders would allow you to set the width and shape of the Audio Filter to be very narrow ( as little as 4 HZ) and also control the amount of peaking ( up to +10 db) as well as the attenuation (down to -10 db) for the adjacent frequencies.

Regards, Al / NN4ZZ
Photo of W5UN_Dave

W5UN_Dave

  • 315 Posts
  • 30 Reply Likes
Al, that will work
Photo of W5UN_Dave

W5UN_Dave

  • 315 Posts
  • 30 Reply Likes
See my earlier comment in this post concerning how I think the APF should be implemented.

You other users of a K3 for cw DXing, please weigh in on this.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Here is a short video clip of the RX Equalizer used on CW as it exists today. You should be able to hear the difference when it is toggled ON and OFF.

http://youtu.be/Brrd8iA7AlA

Regards, Al / NN4ZZ

Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1706 Posts
  • 580 Reply Likes
Tim - W4TME
Eric - KE5DTO
Steve - N5AC

Can we get some FRS feedback and a disposition on this idea?

Regards, Al / NN4ZZ

Automatically track the PITCH, use sliders to control the peaking width.

Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9148 Posts
  • 3466 Reply Likes
Al, thanks for your idea. There is nothing to report at this time. We take all ideas that have been voted up in the Community and consider them for inclusion once a planning period. Right now we are working very hard on achieving the roadmap deliverables for SmartSDR v1.1 so we are not speculating about new features until later.