Flaws with new versions (scan/recognising burst)

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

Re: Flaws with new versions (scan/recognising burst)

Post by VK3KYY » Fri Sep 02, 2022 11:35 pm

Firstly, all the developers do this in their spare time, if they have any at all, and the firmware is totally free of charge.
As its not a commerical project, it has no guaranteed support or even any guarantee anyone will reply to any posts on the forum.

The GD77 etc, are a cheap radio. Originally sub $100 USD, the hardware is not as good as in radios from Anytone, Yaesu, Kenwood, Icom etc etc.

The hardware is primarily designed for use by people like security guards, not amataur radio.

Originally we attempted to push the DMR hardware beyond its design limits, and this did yeild some good results, however the hardware did not always correctly identify the ID of the incoming signal, expecially when there were transmissions on both timeslots.

People posted multiple bug reports of this miss-identification, and the firmware had to be changed a lot to go back to using the DMR hardware as it was intended to be used by the manufacturer

The downside of this was that it now takes a long time to lock on to a DMR signal to 100% identify its ID and the TG etc being transmitted

Hence with short burst transmission, which you notice are not being detected when scanning, and probably not detected if the radio is set to the frequency of the transmission, the signal will not be detected.


Basically, the hardware could be used so it detects quickly but may have the wrong ID and TG etc, or it can be detected slowly and correctly identifys the signal.

The consunsus on the forum, for normal operation, and even the majority of scanning, that better detection accuracy is what people wanted.

Personally I was not worried about occasional miss-idenification, and I would prefer faster detection..

But my personal opinion is not that important, its what the majority of users want that is important, and most people are happy with the detection accuracy versus scan sensitivity compromise

kt4lh
Posts: 272
Joined: Sun Jan 12, 2020 4:27 am

Re: Flaws with new versions (scan/recognising burst)

Post by kt4lh » Sun Sep 04, 2022 3:53 pm

VK3KYY wrote:
Fri Sep 02, 2022 11:35 pm
People posted multiple bug reports of this miss-identification, and the firmware had to be changed a lot to go back to using the DMR hardware as it was intended to be used by the manufacturer

The downside of this was that it now takes a long time to lock on to a DMR signal to 100% identify its ID and the TG etc being transmitted

Hence with short burst transmission, which you notice are not being detected when scanning, and probably not detected if the radio is set to the frequency of the transmission, the signal will not be detected.
Not specifically related, but I find this interesting... My Anytone 878 will not pick up any Tier 3 stuff, every so often just a slight bit of garbled voice. The GD77 picks it up just fine. Of course, it's tier 3 so it's bits and pieces sometimes, but it works pretty okay. I have a commercial DMR system in town I listen to every now and then when I'm bored.

Just interesting that OpenGD77 is slow in this regard. :)

S58SMS
Posts: 153
Joined: Mon Dec 20, 2021 8:00 am

Re: Flaws with new versions (scan/recognising burst)

Post by S58SMS » Mon Sep 05, 2022 7:37 am

I will prepare a video, and post a link later today.

S58SMS
Posts: 153
Joined: Mon Dec 20, 2021 8:00 am

Re: Flaws with new versions (scan/recognising burst)

Post by S58SMS » Tue Sep 06, 2022 5:34 am

https://youtu.be/18CO8_GOePY sorry for bad quality and not perfect english, but i think everything can be seen:
- data burst interruptions
- wierd noises
- rx errors

I have to mention, that on the older FW, there is NO problem with correct name and other data, showed on a screen. Bad CRC CHECKSUM also works better on older FW: it shows id, group etc. the voice is always as it should be. New FW doesn't show id, groups etc. voice doesn't come always from the speaker immediately + the same is also sometimes on classic ham tier 2 repeaters!


youtu.be/18CO8_GOePY

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

Re: Flaws with new versions (scan/recognising burst)

Post by VK3KYY » Tue Sep 06, 2022 6:23 am

S58SMS wrote:
Tue Sep 06, 2022 5:34 am
https://youtu.be/18CO8_GOePY sorry for bad quality and not perfect english, but i think everything can be seen:
- data burst interruptions
- wierd noises
- rx errors

I have to mention, that on the older FW, there is NO problem with correct name and other data, showed on a screen. Bad CRC CHECKSUM also works better on older FW: it shows id, group etc. the voice is always as it should be. New FW doesn't show id, groups etc. voice doesn't come always from the speaker immediately + the same is also sometimes on classic ham tier 2 repeaters!
I see in your video you have CC scanning enabled.
You did not seem to mention this in your posts.

This function we originally called CC Filter: Disable, but that was not how this function actually works. So we recently renamed it to CC:Scan, as this correctly describes how this function works.

The DMR chip does not have the ability to disable the CC selection, so instead we scan for the CC of the signal by constantly changing the CC of the DMR chip, and waiting for a lock of the CC because it matches the CC that we applied to the DMR chip.

It takes many seconds to find the CC of the signal, and while the CC is not correct, there is the possibility of signals being miss-interpreted by the DMR chip, hence why you hear strange noises.

The CC Scan function is intended find the CC of a signal that you already found the frequency. It was not intended to be used at the same time as scanning.

I will discuss this problem with G4EML who wrote the CC scan etc, but I think probably we may need to disabled the CC Scan function when channel or VFO scanning is enabled.


Also.

I see that you are using 2 different types of radio for your tests. The RF hardware in the MD9600 is totally different to the DMN1801, but they do use the same DMR chip.
In fact the only things which are the same in both radios are the DMR chip and the display controller, everything else is different. So its very likely the results of scanning on these 2 radios will be a lot different.

S58SMS
Posts: 153
Joined: Mon Dec 20, 2021 8:00 am

Re: Flaws with new versions (scan/recognising burst)

Post by S58SMS » Tue Sep 06, 2022 6:40 am

Hello,

thank you for your quick response. I have all the CC of the frequencies, so i will turn this OFF and will report later today if there is any change. Regarding difference of the radios - when i've owned gd77 + baofeng 1801, i have also tested one with latest FW and one with july 2021 version and the results were the same as in the video i think (rt90/1801).

User avatar
F1RMB
Posts: 2518
Joined: Sat Nov 16, 2019 5:42 am
Location: Grenoble, France

Re: Flaws with new versions (scan/recognising burst)

Post by F1RMB » Tue Sep 06, 2022 10:24 am

VK3KYY wrote:
Tue Sep 06, 2022 6:23 am

I will discuss this problem with G4EML who wrote the CC scan etc, but I think probably we may need to disabled the CC Scan function when channel or VFO scanning is enabled.
Or maybe users will start to read the manuals, understand what they are doing... Well, being Ham in short.
CC scan while scanning is a feature, with possible (and known) side effects.

Now, if we have to disable features because we have to deal with noobs, I guess the firmware has changed its audience.


Cheers.
---
Daniel

PS: sorry for being a bit harsh.

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

Re: Flaws with new versions (scan/recognising burst)

Post by VK3KYY » Tue Sep 06, 2022 11:25 am

F1RMB wrote:
Tue Sep 06, 2022 10:24 am

Now, if we have to disable features because we have to deal with noobs, I guess the firmware has changed its audience.

Cheers.
---
Daniel

PS: sorry for being a bit harsh.
Good point.

We can't make the firmware operation foolproof.

S58SMS
Posts: 153
Joined: Mon Dec 20, 2021 8:00 am

Re: Flaws with new versions (scan/recognising burst)

Post by S58SMS » Tue Sep 06, 2022 11:31 am

Hello, no need to be rude, we are just explaining experiences, what is wrong with that? - I tried CC scan to be turned off - It makes no difference in scan speed + scan interruption still persist sometimes, as in the video. Wierd sound is gone now. If somebody has the latest FW on GD77 or BF 1801, he can compare the speed also with my video, but I had gd77 few months ago and made comparisation with bf1801 (also different fw) and the scan was also slower.

Also I have found out, that when i recieve a voice call (scan or without scan), it cuts off start of the voice call for a 2-3 seconds sometimes.

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

Re: Flaws with new versions (scan/recognising burst)

Post by G4EML » Tue Sep 06, 2022 5:09 pm

What you are describing is common to all recent releases. It is not a bug and will be similar on all radios.

The DMR signals are being checked more thoroughly than they used to be. For most people this results in more reliable DMR scanning although it will be slower because it takes time to receive and check the data. Your use case with many short burst data signals is not typical and may not benefit from these changes.

We are currently looking at ways of speeding up the scan rate without loosing the benefits of more thorough checking. This may or may not be possible.

Your observation regarding the occasional loss of a few words at the beginning of a transmission is most likely caused by having the Timeslot filter disabled. The radio then has to switch alternately between the two timeslots looking for a valid signal, again this takes time.

If you want to capture the speech as quickly as possible then it is best to make sure the channel is configured for the correct timeslot and TS filtering is turned on.

Post Reply