Random Shutdowns of Kiwis [don't use the previously recommended Jameco 1A power supply]

edited March 2020 in Problems Now Fixed
I have two KiwiSDRs. For some unknown reason they both randomly shutdown together. They have separate power supplies. There is no pattern to the shutdown that I can find.

Does anyone else have this issue?

Jim Gay
wa6our

Comments

  • jksjks
    edited February 2020
    Please tell us a little more about what "shutdown" means. Do you think Linux is halting or just the Kiwi server stops or freezes? Do the 4 LEDs go dark or does the pattern freeze? (i.e. the LEDs are still on). It would be interesting to know if you can still ping the Beagle's ip address when in this state.

    Very important: what is the state of the single "power" LED to the right of the Ethernet RJ45 connector after shutdown? (opposite side of the RJ45 from the 4 other LEDs) Is it still on or is it dark? This tells us if it is a software halt or a shutdown due to the Beagle power controller shutting down.

    When you say shutting down "together" do you mean you've observed them shutting down exactly at the same time or you've happened to catch them after-the-fact both in a shutdown condition? (i.e. they could of shutdown at separate times in the past but you didn't notice this event). It is an important clue if you actually observed them shutting down simultaneously.

    I'll look at the logs and see if anything looks suspicious.
  • jksjks
    edited February 2020
    Okay, some of these questions are answered by looking at the Kiwis and their logs:
    :8073
    Feb 17 12:41:09 kiwisdr kiwid: 1d:00:31:22.977 .... 0     6000.00 kHz  am z6  "xxx.xxx.xxx.xxx" Hampstead, United Kingdom (LEAVING after 0:45:25)
    Feb 17 17:16:29 kiwisdr kiwid: 00:00:00.580   KiwiSDR v1.380 --------------------------------------------------------------------
    
    :8074
    Feb 17 09:36:04 kiwisdr kiwid: 6d:14:26:13.977 01234567          UPDATE: check scheduled
    Feb 17 17:16:28 kiwisdr kiwid: 00:00:00.154   KiwiSDR v1.378 --------------------------------------------------------------------
    (update to v1.380 occurred on reboot)
    
    They were both obviously restarted at the same time, but when they stopped can't be determined by the last log messages before restart (:8074 is a non-public, 8-channel WSPR extension Kiwi). It isn't a Kiwi server crash/panic because there are no messages to that effect, and no "core" file in /tmp. But the server can fail for other reasons (e.g. infinite loop, memory leak causing lock-up etc). The filesystems had plenty of free space and there appeared to be no damage.

    It's more important than ever to know the state of the power LED the next time it happens.
  • edited February 2020
    Today when I woke in the morning they were both down. There were no lights on. I had to disconnect and reconnect the power plug to restart them. in the latest case I restarted them when I found I couldn't reach them from my laptop. What is strange is that they both go down every time, never only one. It would be nice if the log stayed around instead of being created new each boot. Might be able to gather an idea of caused them to shut off. It looks like there was an update last night. I wonder if it could be equated with updates taking place?
  • The log is preserved as the usual Linux syslog in the /var/log/syslog* files. These are most easily viewed using the Linux commands:
    cdp
    msl        (only the most recent log entries)
    ml         (all log entries)
    
    What is not preserved is the content of the "log" tab on the admin page. That only contains the messages from the current server run. You can give the msl and ml commands in the admin "console" tab in addition to the Linux shell.

    If the power LED is not lit then this is almost certainly a power quality issue. What kind of power supplies are these? It's winter there, right? What's your opinion about how these supplies will hold up with brief line voltage disruptions you might encounter in winter (i.e. the ones where the lights flicker very briefly).

    Just look at the debacle we went through yesterday with power quality. Ethernet failures due to bad power. Something we've never seen before..
  • edited February 2020
    It might be useful to review this thread of information.
    http://forum.kiwisdr.com/discussion/comment/8733
    Ron
    KA7U
  • I use Jameco 5 volt 1 amp power bricks that use a transformer. I have many other devices on the same power outlet and know of them go down.
    Here is what the messages file shows:

    Feb 17 11:55:01 kiwisdr kiwid: 23:45:14.816 .... UPDATE: version 1.380 is current
    Feb 17 11:55:49 kiwisdr kiwid: 23:46:03.313 01.. 0 880.00 kHz am z0 "80.229.148.164" Hampstead, United Kingdom (ARRIVED)
    Feb 17 12:37:12 kiwisdr kiwid: 1d:00:27:26.701 01.. 1 22.71 kHz cw z13 "24.20.199.5" Sunnyside, Oregon, USA (ARRIVED)
    Feb 17 12:38:05 kiwisdr kiwid: 1d:00:28:19.440 01.. 1 22.71 kHz cw z13 "24.20.199.5" Sunnyside, Oregon, USA (LEAVING after 0:00:58)
    Feb 17 12:41:09 kiwisdr kiwid: 1d:00:31:22.977 .... 0 6000.00 kHz am z6 "80.229.148.164" Hampstead, United Kingdom (LEAVING after 0:45:25)
    Feb 17 13:59:51 kiwisdr rsyslogd: [origin software="rsyslogd" swVersion="8.4.2" x-pid="437" x-info="http://www.rsyslog.com"] start
    Feb 17 13:59:51 kiwisdr kernel: [ 0.000000] Booting Linux on physical CPU 0x0
    Feb 17 13:59:51 kiwisdr kernel: [ 0.000000] Initializing cgroup subsys cpuset
    Feb 17 13:59:51 kiwisdr kernel: [ 0.000000] Initializing cgroup subsys cpu
    Feb 17 13:59:51 kiwisdr kernel: [ 0.000000] Initializing cgroup subsys cpuacct
  • http://kiwisdr.com/quickstart/

    The Kiwi plus BBG draws over 1A and peaks approaching 1.5A

    1A is simply not enough, it might run while idle, but will fail under load or momentary glitch as mentioned.
    Stu
  • jksjks
    edited February 2020
    We stopped recommending the 1A Jameco supply some time ago (more than a year, possibly 2) because 1A is not enough to hold up under peak conditions (e.g. the moon and stars align just right, the current draw is too much and the input voltage sags below the Beagle's PMIC low-DC-voltage trip point). This would account for the randomness and the apparent total shutdown (power LED off).

    It's too bad because at $12 or whatever it was that Jameco linear supply was a steal. Too bad they didn't have a 2A version. Those kind of linear DC plug-packs are now technically illegal to import into the USA (and elsewhere probably) due to stupid efficiency requirements. The industry is forced to only offer SMPSs and our hobby consequently suffers. Instead you have to spend $60 on an open-frame linear or roll the dice on eBay.
  • Ahhh. That could explain things. A possible power surge that took the Kiwis down but none of the other devices. I will work to find another supply.

    Thanks for the information.
Sign In or Register to comment.