On Mar 23, 2024, at 11:50, Reginald Beardsley via groups.io <pulaskite@...> wrote:

?I've been working on the general concept of a mashup of QRSS and CCW in earnest for the last few weeks and have an algorithm to present. The algorithm implements a statistical non-linear operator which eliminates QRN from an incoming audio stream. The output is a clean keyed audio tone.

The Coherent CW concept goes back many years, but for various reasons has never become popular. My hope is this will change that.

The fundamental notion is that all Morse symbol edges sent by a station are aligned at one of several dit frame lengths. 100, 75 & 50 ms for 12, 16 & 24 wpm for audio QSOs and then 1-10 s dit frames for QRSS message transfer.

At startup on an unused frequency, the receiver listens for 10 seconds and calculates the mean value and standard deviation of the absolute amplitude to produce a statistical estimate for QRN.

Once that has been done a signal is detected by testing the mean value of a dit frame to see if it exceeds a threshold of alpha*sigma + the mean value of the QRN. If it does key a sidetone dit.

For optimal performance, the clock timing at the receiving station needs to be synchronized to the incoming signal. The initial sync is best done by using a Gold code dit stream. Gold codes have the property that they have optimal correlation properties making determining propagation delays more accurate.

Once a QSO has started, every Nth character that gets decoded gets a resync performed by FFT.

Keying the transmitter requires buffering the start of a dit or dah by as much as one frame. If the operator heard the latency while keying it would be very uncomfortable. However, only the receiving station hears the latency and it doesn't affect them. Sender hears normal sidetone and receiver hears the signal with at most a single dit frame delay added to the propagation delay.

All of the requisite supporting infrastructure is present in any of several MCU based keyer designs such as the K3NG, Morserino, etc.

I have guests coming for the eclipse, so I won't have time to bring up a development environment to compile one of the keyer codes so I can implement the changes, but from the description, it should be clear this is not a complex algorithm.

Have Fun!
Reg

On Mar 23, 2024, at 13:17, Reginald Beardsley via groups.io <pulaskite@...> wrote:

?

I've had to think about it for years to get to this point!

On Saturday, March 23, 2024 at 11:09:00 AM CDT, Dave Daniel <kc0wjn@...> wrote:

I have to think about this before rendering an opinion.

Thanks.

DaveD

> On Mar 23, 2024, at 11:50, Reginald Beardsley via groups.io <pulaskite@...> wrote:
>
> ?I've been working on the general concept of a mashup of QRSS and CCW in earnest for the last few weeks and have an algorithm to present.? The algorithm implements a statistical non-linear operator which eliminates QRN from an incoming audio stream.? The output is a clean keyed audio tone.
>
> The Coherent CW concept goes back many years, but for various reasons has never become popular.? My hope is this will change that.
>
> The fundamental notion is that all Morse symbol edges sent by a station are aligned at one of several dit frame lengths.? 100, 75 & 50 ms for 12, 16 & 24 wpm for audio QSOs and then 1-10 s dit frames for QRSS message transfer.
>
> At startup on an unused frequency, the receiver listens for 10 seconds and calculates the mean value and standard deviation of the absolute amplitude to produce a statistical estimate for QRN.
>
> Once that has been done a signal is detected by testing the mean value of a dit frame to see if it exceeds a threshold of alpha*sigma + the mean value of the QRN.? If it does key a sidetone dit.
>
> For optimal performance, the clock timing at the receiving station needs to be synchronized to the incoming signal.? The initial sync is best done by using a Gold code dit stream.? Gold codes have the property that they have optimal correlation properties making determining propagation delays? more accurate.
>
> Once? a QSO has started, every Nth character that gets decoded gets a resync performed by FFT.
>
> Keying the transmitter requires buffering the start of a dit or dah by as much as one frame.? If the operator heard the latency while keying it would be very uncomfortable.? However, only the receiving station hears the latency and it doesn't affect them.? Sender hears normal sidetone and receiver hears the signal with at most a single dit frame delay added to the propagation delay.
>
> All of the requisite supporting infrastructure is present in any of several MCU based keyer designs such as the K3NG, Morserino, etc.
>
> I have guests coming for the eclipse, so I won't have time to bring up a development environment to compile one of the keyer codes so I can implement the changes, but from the description, it should be clear this is not a complex algorithm.
>
> Have Fun!
> Reg
>
>
>
>
>

?

On 03/23/2024 5:55 PM EDT Reginald Beardsley via groups.io <pulaskite@...> wrote:


Mike,

The speed can be infinitely variable, but 12, 16 & 24 have proven comfortable speeds and appears to be the range for most people who hand key. It's also convenient frame rates 100, 75 and 50 ms. Sychronization is via a PRN with optimal correlation properties.

My apologies for having made my post a bit obtuse. I wanted to introduce the concept as briefly as possible, but didn't organize it well. This is probably explanation attempt number 20 something. It's super simple once you get the idea. Clearly I did not get the presentation right.

Please post further questions to the list after you've reread the original post a time or two as I clearly ordered the ideas poorly. I'll expand on key points based on your questions. The Gold code sequence is an interesting side topic of critical value in the synchronization process. One could also use it for selective calling by means of a Gold code generated based on call sign. Gold introduced the concept of PRN bit streams with optimal correlation and cross correlation properties. Auto correlation a narrow spike and cross correlation among the set near zero.

Have Fun!
Reg

On 03/24/2024 11:04 AM EDT Reginald Beardsley via groups.io <pulaskite@...> wrote:


Mike,

The fact that it lets you head copy very weak signals without having to listen to the hiss of the QRN seems rather valuable in something such as the QMX. The synchronization to the sender is a per QSO link initiation that accounts for propagation delay and station clock sync. Once that's been done all transmissions will be in sync.

The sender's sidetone is asynchronous, but the transmit keying is delayed by at most a single dit frame. Detection imposes another dit frame latency. But because the receiver has no reference for the start of the keying by the sender, just the emitted signal, it's transparent.

For QRSS mode you would key your message into a buffer at a comfortable rate which you could edit as one does message memories for contests. The radio would then send the message at 1-10 s dit frame rates. The receiving station would then play back the buffer contents using the side tone at the end of message or on command from the buffer.

My goal was digital mode performance without a computer other than the MCU in the QMX. Then I realized *any* keyer/decoder could implement a protocol this simple. That makes it easy for any radio to use CCW/QRSS modes. Thus this effort to lay out all the details and get it implemented in at least the K3NG and Morserino.

Have Fun!
Reg