Date
1 - 8 of 8
New WAV file overload monitoring feature. ADC overload reporting coming soon. More 'wd...' commands and a help menu
|
Rob Robinett
WD status commands enhanced and with new wd-help command
if you 'git pull', restart WD, and then execute 'source ~/wsprdaemon/.wd_bash_aliases' you will gain access to these new features and commands. WD WAV file overload monitoring and viewing One of those is a new WD feature which logs the minimum and maximum wav file sample values found in the 16 bit samples of each 2 minute long wav file fed to the 'wsprd' decoder. A size-limited 'wav_stats.log file' contains a list of those results and the new 'wdov'' command finds all the log lines on all those log files which record 1 or more max sample value of 1.0000 and/or min sample value of -1.000. By occasionally running 'wdov' you will over time learn when/if your wav files are overloading. If users report finding many such wav overloads, then I will work to address that problem by making the AGG setting give to kwiirecorder a configurable value. WD ADC overload monitoring and viewing TDB There is a second ADC overload monitoring feature which I will next address. ADC overloads are due to excessive RF levels on the Kiwi's SMA input jack and those can only be addressed by introducing pads and/or filters in the RF feed to the Kiwi. Such overloads are becoming much more common with the much improved propagation we are experiencing, so I know that users need access to that information. Until then, you can find a running count of your Kiiw's ADC overloads on among the lines on its /status page, e.g at KFS. 'adc_ov=13543'. |
|
|
|
Rob Robinett
I fixed and refined the wd-query command which is mostly useful because it will show alll FST4W mode spots, including the -120 ones which are mixed with WSPR-2 by wsprnet.
|
|
|
|
John
I am trying to see if I have KIWI overload errors.
I did a wd -le command and received more than a page of lines. Should I be looking for the word "ERROR"? I did a couple of CR and see a new line with the same one line of info with a new time stamp. I tryed to exit the command with a ctrl C but no joy, I could not get out of the command. I had to cancel the SSH session to exit the command. I need more training. John TI4JWC |
|
|
|
Rob Robinett
Run 'wdov' to get a list of all the overload events. Since there are many, you will want to pipe its output to 'less' and/or 'grep'. It looks like your overloads happen mostly around greyline times: wsprdaemon@daemonServer:~$ wdov | grep "KIWI_1/80/" /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2202.wav: 288 286 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2204.wav: 528 240 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2248.wav: 530 2 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2254.wav: 1143 613 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2256.wav: 1315 172 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221112_2302.wav: 1909 594 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0040.wav: 2567 658 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0044.wav: 2947 380 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0046.wav: 3995 1048 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0048.wav: 4486 491 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0054.wav: 6310 1824 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0056.wav: 9220 2910 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0058.wav: 10977 1757 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0100.wav: 11827 850 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0102.wav: 13468 1641 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0104.wav: 15291 1823 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0106.wav: 16676 1385 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0108.wav: 17293 617 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_0110.wav: 17357 64 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1314.wav: 17710 353 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1316.wav: 18008 298 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1324.wav: 19252 1244 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1416.wav: 19261 9 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1418.wav: 19553 292 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1444.wav: 20303 750 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1456.wav: 20614 311 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1500.wav: 21015 401 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1552.wav: 21428 413 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1624.wav: 21994 566 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1632.wav: 21998 4 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1638.wav: 22298 300 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1644.wav: 22450 152 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1646.wav: 22866 416 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1656.wav: 23512 646 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1722.wav: 24157 645 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1724.wav: 26002 1845 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1726.wav: 27138 1136 /dev/shm/wsprdaemon/recording.d/KIWI_1/80/kiwi_ovs.log: 221113_1758.wav: 27578 305 wsprdaemon@daemonServer:~$ I am trying to see if I have KIWI overload errors. --
|
|
|
|
John
Thank you Rob. Now if I could move these numbers onto the Noise graph plots. J |
|
|
|
Rob - Is the OV_TOTAL column in the log file is the count of Kiwi A/D samples that equal A/D full scale 2^14? Since it's coming from a demodulated stream, I wouldn't think so and would be more likely to be the recorded stream has maxed out. Possibly Kiwirecorder just includes the number of overflows the kiwi code shows at the /status page and thus there is no accumulation of actual clipped samples. If the latter is the case, then I wonder how many A/D samples it takes to set the overflow flag on the kiwi?
What I'm trying to understand is how catastrophic an occasional clipping is to decoding a WSPR signal. At my station, I keep a log of the /status page of the Kiwi overflows with the following bash script and occasionally look at it to see when overflows happen: when=$(date -u +%D' '%T);echo -n $when " " >> /home/wayne/adc94_history.dat curl -s http://192.168.1.94:8073/status | grep adc_ov >> /home/wayne/adc94_history.dat
Today I accumulated over 2000 overflows from 1342Z to 1354Z (per your log file). I dropped the padding after the LNA by 3db which should reduce overflows. I am wondering what the tradeoff will be between decoding weak signals vs those lost by overflow during those times. Wayne WA2N |
|
|
|
Rob Robinett
Yes, ov_total is reported on the Kiwi's /status page and is I believe the total number of 66 Mspsp samples since Kiwi powerup/reboot where the input signal reached the ADC's full scale value. At the end of each 2 minute WSPR cycle WD compares the the current count with the count reported at the end of the previous 2 minute cycle, Our measurements of Kiwi's RF and Audio gain structure led us to choose to configure the Kiwi in a fixed -60 dB AGC so that the 16 bit audio range completely includes the 14 bit ADC range. With that AGC setting all ADC samples should output more than 2-4 lsb in the wav file. As a result, I infer that at most sites seeing more than 1-9 overloads in a cycle is cause to reduce the RF gain in the path to the Kiwi's input. So I expect that you can add even more than 3 dB of padding without impairing your system's SNR. On Mon, Nov 14, 2022 at 10:29 AM wayne roth <io@...> wrote:
--
|
|
|
|
John
I have so few unique 80m spots in a day 1 -- 15 max. I am guessing that a pad might not make any change in my count.
John |
|
|