Experimental changes to code optimisation

DO4CZ
Posts: 63
Joined: Sat Nov 16, 2019 8:55 am
Location: near Rostock, Germany

Re: Experimental changes to code optimisation

Post by DO4CZ » Sat Jan 18, 2020 3:04 pm

After several hours no problems with 1x GD 77 as hotspot and 1x GD77 in normal operation.
Thank you very much...

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

Re: Experimental changes to code optimisation

Post by kt4lh » Sat Jan 18, 2020 4:21 pm

Listening on a net right now with my MD380 running MD380Tools and this build on my GD77, everything sounds exactly the same between the two radios, even the delay. Does this build have the 1 frame delay in it, if so, it's even more subtle than I expected, or maybe the MD380 already has such a delay in it and I'd never noticed before.

I'll keep testing, but so no problems listening on a net, basic run through menus to set up, quick scan of DMR and FM.

Actually, interesting. I just noticed FM Scanning that a nearby repeater was busy but scanning didn't stop on it, but about 2s later I noticed it sitting on that repeater channel, even though it had been unkeyed for several seconds prior and had still been scanning when it did. Said another way; it felt like it was delayed in recognizing the repeater being active. But listening for a few more minutes, I'm not seeing this happen. Maybe just something for others to also watch out for, see if they see that happen?

EDIT: I have not seen it again since, so maybe just an error in the keyboard/chair interface. ;)
Last edited by kt4lh on Sat Jan 18, 2020 6:21 pm, edited 1 time in total.

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

Re: Experimental changes to code optimisation

Post by W1AEX » Sat Jan 18, 2020 5:35 pm

I loaded the Jan 18 firmware [0d961d6] into both radios this morning and have been running them for about 3 hours with a mix of scanning of both analog and DMR repeaters and this version has behaved fine. Good reports from other stations when I have transmitted and I have not heard a single dropout while listening to the local DMR repeaters. A few dropouts noted while listening through the Zumspot, but I think that might have been a Brandmeister issue.

Looks like a very nice release Roger!

73 and thank you, Rob W1AEX

HI8NLO
Posts: 18
Joined: Tue Nov 19, 2019 11:55 pm

Re: Experimental changes to code optimisation

Post by HI8NLO » Sat Jan 18, 2020 7:37 pm

This version solved the dropout, hickup and and BER issues I was experiencing with the Jan 16th version. I hadn't noticed before but the way the information is presented with TA showing in hotspot mode is very good and clean almost making unnecessary having a NEXTION display for a RPi.

Regarding stability I've been testing the whole day on the PAPA System and SOCAL TGs with no inconveniences.

n1id
Posts: 17
Joined: Sat Nov 30, 2019 12:30 pm

Re: Experimental changes to code optimisation

Post by n1id » Sat Jan 18, 2020 8:33 pm

Everything working fine here too!

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

Re: Experimental changes to code optimisation

Post by VK3KYY » Sat Jan 18, 2020 8:33 pm

Thanks.

I did get one report via email of crashing after 5min of Tx, but I won’t know if this means 5 mins without releasing the PTT.

So it sounds like I can copy this version, so it becomes the Latest.

BTW.

I think the small problems people noticed with scanning and the ID display, were already there in the last few versions, but as these glitches appear randomly, they can get incorrectly attributed to a specific release

I think both the scanning and the ID display still need some fine tuning

User avatar
DG1YFX
Posts: 25
Joined: Fri Dec 13, 2019 11:16 pm
Location: Wiesbaden, DE
Contact:

Re: Experimental changes to code optimisation

Post by DG1YFX » Sat Jan 18, 2020 10:31 pm

Well, the audio has improved a lot. Before they were slight glitches on Tx every second analog to the counter on the display. And on Rx sometimes dropouts. I tested both very thoroughly because at the same time I am testing my own little hblink master with echo tests. On BM, one might get auto-banned.

73s, Jens

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

Re: Experimental changes to code optimisation

Post by VK3KYY » Sat Jan 18, 2020 10:45 pm

DG1YFX wrote:
Sat Jan 18, 2020 10:31 pm
Well, the audio has improved a lot. Before they were slight glitches on Tx every second analog to the counter on the display. And on Rx sometimes dropouts. I tested both very thoroughly because at the same time I am testing my own little hblink master with echo tests. On BM, one might get auto-banned.

73s, Jens
I think a lot of the changes to audio are probably due to the change to buffer the DMR audio by 1 frame. But I have also made some other improvements to reduce CPU processing of the DMR audio, but removing one of the buffer copy actions, and also the whole sound processing source file is now running as optimised code.

F6GVE
Posts: 84
Joined: Sat Nov 16, 2019 8:52 am

Re: Experimental changes to code optimisation

Post by F6GVE » Sat Jan 18, 2020 11:08 pm

Apart from what I said before, and what was probably the same with the precedent release, I found nothing different. The reliability, the DMR/FM behaviour are the same with my 2 GD77 that I used as habitually all that day long.
I often transmit for a long time (TOT 3 mn but I immediately key again) and I found no special bug or sync problem.
The scanning is as good as habitually and I intensively use it.
I had no crash.
However I didnt used the hotspot function today.

So for me this optimisation is completely transparent.
Once again, congratulations for the nice job.

The only problem I have for a long time is this sort of desync/dropout/bug that I have when receiving DMR. It is pretty rare with one GD77 but relatively frequent with the second one (it is the same with calibration ON or OFF).
Also when scanning, if the scanner stops on a DMR RX, most of time there is a sort of hesitation (dont know how to describe) before I can clearly ear. I suspect that these two observations are linked.
I tried to adjust the "reference oscillator tuning" with no success. In fact I can almot supress the dropout (it lasts about 2 to 3 seconds every time it happens) if I set to be really off frequency. But in this case the DMR sensitivity is bad of course.
I could suspect some I Q calibration issue? But I had no way to be sure nor to do anything.

This firmware became fantastic. I have no other DMR transceiver and I dont want another one...

Luc

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

Re: Experimental changes to code optimisation

Post by VK3KYY » Sat Jan 18, 2020 11:51 pm

Thanks

I've now made this version the 'latest' official release

Post Reply