Possible bug? very long recordings

All bug reports here please
User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Possible bug? very long recordings

#1

Unread post by G0OFE »

Here's a weird one.

I need to dig deeper with this and see if I can reproduce it, however, bear with me...

CQWW SSB. Recorded 40m SSB, and 20m SSB for the entire length of the contest, both on the same machine, both using separate Elad S2s.
20m recording was fine. However, 40m on playback exhibited symptoms akin to a loose antenna connection, even though this was not apparent while the recording was being made (I made periodic checks), and the same radio has been fine before and since. There were rapid jumps in signals and noise. The issue was not present for a period of hours, then would start again later in the recording.
I put this down to some sort of weirdness.

Fast forward a month to CQWW CW. This time on 40m I used an S2, on 20m I used an Airspy HF+. All signals fine while recording was taking place.

On playback, 40m exhibited the same behaviour as before. Perhaps on a less consistent basis than in the SSB leg.
Strangely, 20m exhibited the same behaviour this time, but to a severe extent.

And here is the strange thing.

The problem was far worse when signals were generally weaker. on 20m, at the start of the recording at 00:00 on the Saturday, the noise floor on playback was very low, below -160dbm. Band was probably dead anyway. As the band began to open, the noise floor began to jump in time with the signals. I could actually copy CW from the bursts of the noise, for example an LZ station. As signals became stronger, the recording began to behave properly, and everything eventfully became as I would expect.

It's as if there needs to be enough RF coming in to "make" the receiver work properly. This was much clearer on the CW recordings.
As the band started to close later in the day, the problem began to manifest itself more and more until it was there all the time, only to repeat the next day.

On the 40m recording, the receiver seemed to "collapse" when the contest ended and signals were no longer there, albeit a little under 2 minutes past midnight. See screenshot from the analysis I did of the file:
Screenshot-2021-12-03-150514.jpeg
So what the hell is going on here? I don't understand it.

Is it a disk write issue?
Is it caused by trying to record a 48h RF64 file? I have made much bigger RF64 files when recording Band 2, without issue.
Is it caused by recording the display range, rather than the bandwidth of the radio?

If it's a disk write problem, then why was there no error messages. If it's a receiver problem, then why did that not manifest itself while listening live?

I'm really scratching my head on this one. I need to run some experiments to see if I can reproduce the issue, and reproduce it on another machine.

Each file is less than 100GB. If it helps, I can try and drop-box them, or pull an extract using Data File Editor.

I have attached a recording. At the start, you can hear the noise pulsing with the CW signals... ES9C is apparent.. I then tune to the station itself. Then I jump to later in the day when signals are generally stronger. The radio appears to be behaving normally.


SDR Console 2021-12-03 14h 56m 15s.mp4
(43.84 MiB) Downloaded 35 times
Jim, Bournemouth IO90BR
Image

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: Possible bug? very long recordings

#2

Unread post by jdow »

What did you use to record this? I wonder wonder if the no signal environment is less than 1 bit peak to peak. If it is and signals that are more than 1 bit peak to peak appear the radio suddenly will start to work with extensive averaging effects making sub-bit sized signals quite visible and readable.

(The USC-28 DISCS SatCom modem had this effect. To make it work in the lab we had to insert a signal so that the back to back testing could proceed with the signal attenuated to the levels expected from the satellite.)

{^_^}

User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Re: Possible bug? very long recordings

#3

Unread post by G0OFE »

jdow wrote: Sat Dec 04, 2021 8:39 am What did you use to record this? I wonder wonder if the no signal environment is less than 1 bit peak to peak. If it is and signals that are more than 1 bit peak to peak appear the radio suddenly will start to work with extensive averaging effects making sub-bit sized signals quite visible and readable.


{^_^}
Elad S2 during CQWW SSB, then S2 on 40m, Airspy HF+ on 20m during the CW leg.

Could this (your explanation above) be happening during recording, when everything looked fine at the time of recording, only manifesting itself on the recording itself. THAT'S what I don't understand. There was no issue listening live.
Jim, Bournemouth IO90BR
Image

User avatar
ukdx
Posts: 41
Joined: Wed Nov 18, 2020 7:52 am

Re: Possible bug? very long recordings

#4

Unread post by ukdx »

Jim, overloading and broadcast FM are completely unrelated to this but I had a similar experience in the summer when watching and listening live looked and sounded OK but playback of the recording was a wall of noise with just the strongest signals breaking through the overload. I never did get the bottom of it and why reception was fine but playback unusable. FWIW this was on an Elad S3 using 16 bit sample recording.

Paul

User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Re: Possible bug? very long recordings

#5

Unread post by G0OFE »

Hi Paul, yes that rings a bell. I remember you mentioning that some time back. Did it only happen the once?

It does sound similar to my situation.. reception was fine, but the recording was buggered in some way.

In your case it was apparent overload, in mine it was loss of all signals, as if the feeder or antenna was intermittent. In fact it was worse than that, in that had the antenna been disconnected, the background noise still would have been higher.

Simon, in trying to diagnose this, would the logfile be of any help in future? The only issue is, once the problem is discovered, the logfile will almost certainly be lost unless we actually always save it after making a recording.
Jim, Bournemouth IO90BR
Image

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: Possible bug? very long recordings

#6

Unread post by jdow »

Jim, I am not familiar with SDRC recording tools. I know SDRSharp's server tool and SDRServer both having tools to handle narrow bandwidth links. They chop a small chunk out of the spectrum and may be able to send 8 bits of a larger size sample. Get those 8 bits wrong and you get something quite similar to what is being described here. So I figured something like this might be going on. It is a serious shot in the dark, though.

{^_^}

User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Re: Possible bug? very long recordings

#7

Unread post by G0OFE »

Thanks, Joanne.. what do you mean by "narrow bandwidth links"?
Do you mean when I am recording a section of the spectrum being sent from the SDR (in one case 110khz out of the 384khz radio bandwidth)?
Jim, Bournemouth IO90BR
Image

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: Possible bug? very long recordings

#8

Unread post by jdow »

Something like my DSL connection that only allows about 600 kbps uplink when sometimes you must trim the bits of precision sent per sample as well as the rate of those samples.

{^_^}

User avatar
G0OFE
Posts: 435
Joined: Thu Aug 06, 2020 7:33 am
Location: Sunny Sunny Bournemouth IO90BR
Contact:

Re: Possible bug? very long recordings

#9

Unread post by G0OFE »

ok, so perhaps it's not liking my recording a segment of the radio range, rather then the whole of it, and it's dropping bits of the whole lot as a result?
Jim, Bournemouth IO90BR
Image

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: Possible bug? very long recordings

#10

Unread post by jdow »

If you can generate a weak signal in the band you are considering run it up and down to see if that can affect the effect you are seeing. An off frequency tone might mitigate the problem.

{^_^}

Post Reply