Notch filter to remove CWID from repeaters

We have 4 repeaters on our Allstar Network.

On one of them we can't turn off the CWID (we have a short callsign and changed to 35wpm), but Nets still find it annoying (and for good reason).

That repeater's path is RF <-> Analog Radio <-> Switch <-> DMR radio <-> TG <-> Allstar/DVSwitch <-> Allstar <-> To whomever

The idea is at the Allstar/DVSwitch to remove the CWID.

Anyone do this?

Claude suggests:
"Notch filter at the CWID pitch. A repeater ID is a fixed tone, usually 600–1000 Hz. A narrow notch (Q around 30+) removes it almost completely and barely touches speech. Nothing in the chain does this natively, so it needs a small shim in the audio path.

  1. USRP shim with a tone detector. This is more work but gets rid of all the CW. Run a small Python script between analog_bridge and the bridge node (626794). It reads the USRP UDP packets (a 32-byte header plus 160 int16 samples at 8 kHz) and runs a Goertzel detector at the CW pitch. When it sees the tone keyed on and off, it zeroes those frames and forwards everything else untouched. It adds about 20 ms of latency. Change analog_bridge’s USRP txPort/rxPort and Asterisk’s usrp.conf so the shim sits in the middle. It’s roughly 40 lines.

  2. Drop by DMR source ID. If the x-band’s DMR side always sends the ID under a fixed radio ID, filtering on that would be cleaner. I don’t know of a source-ID filter in analog_bridge, and I’d have to confirm one exists in your build. It would also drop that repeater’s voice traffic, so it’s only useful if the ID is the only thing you want from it."

    Thanks

A notch filter would only mask the problem. You'd still get a carrier and perhaps some ringing around the frequencies when it ID's. You're better off making sure no repeater in your link ever ID's to a path received by ASL. Controllers with gated PL, if you must have RF in the path, or using a dedicated linking radio frequency or a pair of them if you want to maintain full duplex throughout your link, where there is no hangtimes or other telemetry.

Related: I don't believe the rxnotch directive in app_rpt actually works in ASL3. At least it doesn't seem to.

···

On Nov 16, 2026, at 08:30, Wayne Spivak [email protected] wrote:

Someone replied to a topic you are Watching.

| KC2NJV
November 16 |

  • | - |

We have 4 repeaters on our Allstar Network.

On one of them we can't turn off the CWID (we have a short callsign and changed to 35wpm), but Nets still find it annoying (and for good reason).

That repeater's path is RF <-> Analog Radio <-> Switch <-> DMR radio <-> TG <-> Allstar/DVSwitch <-> Allstar <-> To whomever

The idea is at the Allstar/DVSwitch to remove the CWID.

Anyone do this?

Claude suggests:
"Notch filter at the CWID pitch. A repeater ID is a fixed tone, usually 600–1000 Hz. A narrow notch (Q around 30+) removes it almost completely and barely touches speech. Nothing in the chain does this natively, so it needs a small shim in the audio path.

  1. USRP shim with a tone detector. This is more work but gets rid of all the CW. Run a small Python script between analog_bridge and the bridge node (626794). It reads the USRP UDP packets (a 32-byte header plus 160 int16 samples at 8 kHz) and runs a Goertzel detector at the CW pitch. When it sees the tone keyed on and off, it zeroes those frames and forwards everything else untouched. It adds about 20 ms of latency. Change analog_bridge’s USRP txPort/rxPort and Asterisk’s usrp.conf so the shim sits in the middle. It’s roughly 40 lines.

  2. Drop by DMR source ID. If the x-band’s DMR side always sends the ID under a fixed radio ID, filtering on that would be cleaner. I don’t know of a source-ID filter in analog_bridge, and I’d have to confirm one exists in your build. It would also drop that repeater’s voice traffic, so it’s only useful if the ID is the only thing you want from it."

    Thanks


Visit Topic or reply to this email to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, click here.

Spoke w/Repeater owner and he's going to try to turn it off.

That would solve the issue.

That path is very confusing. There's no Allstar node directly on the repeater..?

No, but we have turned off the CWID.

We are now working on turning off tail (the software says 0-9, but the drop down is 1-9…. Told the repeater owner to try entering a “0” and not use drop down)…

···

From: Robert Gaynor [email protected]
Sent: Monday, November 16, 2026 2:43 PM
To: [email protected]
Subject: AllStarLink Community Forum: Notch filter to remove CWID from repeaters

Someone replied to your post.





|

KK6RQ
November 16

|

  • | - |

That path is very confusing. There's no Allstar node directly on the repeater..?


Visit Topic or reply to this email to respond.

To unsubscribe from these emails, click here.

I was just going to say that seems like a bit of a kludge as to the way it's set up. If you're just picking the audio off the air, whatever's in that audio is going to go over the network. Best practice would be to attach an Allstar node to the repeater and link to it that way, with ASL as the repeater controller. That way, you'd be able to tweak the CWID however you want. You also wouldn't have to worry about repeater hang (or telemetry/bloops & bleeps) going over the network.

Thank you for your response.

Best practice is not always the feasible practice.

Wayne

···

From: Robert Gaynor [email protected]
Sent: Monday, November 16, 2026 3:51 PM
To: [email protected]
Subject: AllStarLink Community Forum: Notch filter to remove CWID from repeaters

Someone replied to your post.





|

KK6RQ
November 16

|

  • | - |

I was just going to say that seems like a bit of a kludge as to the way it's set up. If you're just picking the audio off the air, whatever's in that audio is going to go over the network. Best practice would be to attach an Allstar node to the repeater and link to it that way, with ASL as the repeater controller. That way, you'd be able to tweak the CWID however you want. You also wouldn't have to worry about repeater hang (or telemetry/bloops & bleeps) going over the network.


Visit Topic or reply to this email to respond.

To unsubscribe from these emails, click here.

The primary issue here is not the ID itself, but the keying of links when there is no one accessing the repeater (ie IDs and hang time). This creates havoc for connected systems. There are a few solutions:

  1. (best practice). AllStar is located at the repeater, either on a link port, or being the repeater controller itself.
  2. Dedicated duplex RF link between the repeater and AllStar. Depending on jurisdiction, the link may need to ID, but this can be managed a few ways to ensure that the AllStar node doesn't get keyed, unless someone is actually accessing the repeater.
  3. Using CTCSS, activated by user input to key the AllStar node. Encode a CTCSS tone only when there is a valid input signal. The AllStar node decodes the tone and sees no hang time. Users accessing the repeater can turn their CTCSS decoders off to hear the hang time and CWID, and the repeater remains compliant with ID regulations.

Any of these solutions will do the job.

If none of these are feasible, I'd seriously reconsider whether it's worth linking the repeater at all, it's likely to be too much trouble for the network.