Experimental version to help DMR breakup

VK3KYY
Posts: 7478
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia

Experimental version to help DMR breakup

Post by VK3KYY » Wed Feb 19, 2020 11:04 am

Guys

Those people who had problems with the signal breaking up while mobile, could you try this version.

I started to notice breakup on one of my local repeaters, which I've not noticed before, and have tracked down a possible cause of the problem.

The radio should receive dmr data every frame (30ms), but if this does not happen for a few frames, there sort of fail-safe system which resets the DMR processor.

It looks like the fail-safe was kicking in too soon, and on weak signals its better that the DMR chip waits for the signal to re-appear.

The setting we had been using was to reset if the signal was lost for more than 2 frames (60ms), but I found that changing this to a larger value of around 6 frames causes a lot less breakup

For the developers, I changed

Code: Select all

#define END_TICK_TIMEOUT 2
to

Code: Select all

#define END_TICK_TIMEOUT 6
in fw_HR-C6000.c

I'm not sure what the optimum value is, but it looks like 2 is too low

With the type of flutter I was experiencing, which is caused by the movement of a large tree just behind my house.
The value of 6 seemed to be OK. I tried 8 but it was the same as 6.

Possibly 4 may be too low, I'd need to experiment, or potentially add this as another setting if different users require different values
Attachments

[The extension sgl has been deactivated and can no longer be displayed.]


User avatar
Ik0nwg
Posts: 242
Joined: Sat Nov 16, 2019 7:23 am
Location: JN61VG
Contact:

Re: Experimental version to help DMR breakup

Post by Ik0nwg » Wed Feb 19, 2020 11:48 am

It seems to work properly, if I notice any problems during the day I will warn you. Thanks Roger.
Sal

G4EML
Posts: 919
Joined: Sat Nov 16, 2019 10:01 am

Re: Experimental version to help DMR breakup

Post by G4EML » Wed Feb 19, 2020 11:59 am

Hi Roger,

I had a read of the ETSI spec looking for clues. It looks like the timer involved here is called T_ChSyncTo in the spec. This is defined as the channel sync timeout timer and is used to determine how long to wait before moving from the in sync state to the out of sync state in the event of loss of sync signals.

The recommended minimum value for this timer is 390ms. That corresponds to 13 lost frames. It seems a long time but it could easily happen if moving through a signal null.

Colin.

User avatar
W1AEX
Posts: 126
Joined: Sat Nov 16, 2019 9:00 pm
Location: Connecticut, USA
Contact:

Re: Experimental version to help DMR breakup

Post by W1AEX » Wed Feb 19, 2020 2:24 pm

Roger and Colin,

I loaded this version into one of my GD77 radios and will compare it to my other GD77 that has 2-16-2020 loaded. In theory this experimental version sure seems like a great idea. I wonder if it will help in a case where running through a hotspot a few packets are lost here and there?

73, Rob W1AEX

EDIT: Maybe I am jumping the gun here but after listening to lots of DMR chit-chat (through my zumspot and through several local repeaters) for the past 2 hours I have not heard a single dropout. Now and then I can recognize that something might be on the verge of dropping out but much to my surprise the radio handles it and instead of breaking into a burst of digital noise it continues on fairly smoothly. I think this adjustment might be a big step forwad. 73, Rob W1AEX

User avatar
EB3AM
Posts: 204
Joined: Fri Jan 24, 2020 1:40 pm
Location: Catalonia, not Spain
Contact:

Re: Experimental version to help DMR breakup

Post by EB3AM » Wed Feb 19, 2020 6:10 pm

downloading... and testing tomorrow morning when run about 200km

Thanks!

User avatar
EA5SW
Posts: 91
Joined: Sat Nov 16, 2019 9:45 am
Location: Valencia, Spain

Re: Experimental version to help DMR breakup

Post by EA5SW » Wed Feb 19, 2020 9:06 pm

BÏG change in the GD77 of my friend with DMR audio cuts.. now appears to work !!!

VK3KYY
Posts: 7478
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia

Re: Experimental version to help DMR breakup

Post by VK3KYY » Wed Feb 19, 2020 9:07 pm

Colin

Thanks

I found setting it to 6 seemed adequate for me, to cope with quite slow flutter caused by the movement of the tree which disturbs my signal path. This value equates to 360mS which is almost what you found the ETSI doc recommends.

I think originally Kai used a value of 10 frames for this, but I reduced it to solve some other problem ages ago.
I suspect the value of 10 was just an arbitrary number, that seemed to work ok in the Tier1 code.


I can’t remember why I reduced the value. It would have been something to do with the DMR Rx being held open for a long time after the incoming signal had ended.

But things have changed a lot in the code over the last 6 months, and using a larger value now seems to not cause any problems.

I probably need to do some more testing, as I am surprised that dropping up to 3 frames of the TS that the radio is listening to, does not cause the decoded audio to sound terrible.

We only buffer 1 full frame of audio, so there will in reality be dropouts in the audio signal, but I think the human ear and brain are adapted to cope with this and can still reconstruct the conversation, even with bits missing

VK3KYY
Posts: 7478
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia

Re: Experimental version to help DMR breakup

Post by VK3KYY » Wed Feb 19, 2020 9:11 pm

EA5SW wrote:
Wed Feb 19, 2020 9:06 pm
BÏG change in the GD77 of my friend with DMR audio cuts.. now appears to work !!!
You may need to try different values for that timeout.

I set it to 6, as this seemed to fix the QSB problem I have with one repeater.

Colin says the recommended setting may be 390ms which is a value of 13 frames, but that seems unnecessarily high, and may cause other problems

VK3KYY
Posts: 7478
Joined: Sat Nov 16, 2019 3:25 am
Location: Melbourne, Australia

Re: Experimental version to help DMR breakup

Post by VK3KYY » Wed Feb 19, 2020 9:13 pm

W1AEX wrote:
Wed Feb 19, 2020 2:24 pm
Roger and Colin,

I loaded this version into one of my GD77 radios and will compare it to my other GD77 that has 2-16-2020 loaded. In theory this experimental version sure seems like a great idea. I wonder if it will help in a case where running through a hotspot a few packets are lost here and there?

73, Rob W1AEX

EDIT: Maybe I am jumping the gun here but after listening to lots of DMR chit-chat (through my zumspot and through several local repeaters) for the past 2 hours I have not heard a single dropout. Now and then I can recognize that something might be on the verge of dropping out but much to my surprise the radio handles it and instead of breaking into a burst of digital noise it continues on fairly smoothly. I think this adjustment might be a big step forwad. 73, Rob W1AEX
Rob.

I was also surprised at how well it worked.

The value may need a bit of adjustment, and we need to test under a variety of conditions, but it’s definitely a step in the right direction

User avatar
EB3AM
Posts: 204
Joined: Fri Jan 24, 2020 1:40 pm
Location: Catalonia, not Spain
Contact:

Re: Experimental version to help DMR breakup

Post by EB3AM » Wed Feb 19, 2020 9:18 pm

https://youtu.be/yn-_Bifhux4

no words! see video :mrgreen:

Post Reply