Author Topic: GPS Year Error (1024 wk rollover) BLUE Systems with GlobalTop /Mediatek MT3333  (Read 4370 times)

0 Members and 1 Guest are viewing this topic.

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter

Developers aware of issue, considering several options, in addition to, or instead of, updating.See
https://forum.blitzortung.org/showthread.php?tid=3648 in the INTERNAL forum for owners.
Mike
« Last Edit: November 02, 2022, 07:53:21 PM by Cutty Sark Sailor »

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2325
Would not think it is a standard 1024 GPS Week rollover issue (as such) but obviously sounds like some form of GPS timing issue and most likely very specific to this hardware?

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
« Last Edit: November 02, 2022, 11:51:44 PM by Cutty Sark Sailor »

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
A beta system firmware available... system will ignore year error, first step to correcting on server... should get system back up, signals nvald since RTC time is still -24 wks.  Install so devs can check is data is being sent.  Stations likely will show 'idle'... GUI may be erratic, as the script that drives GUI has to load from server, as in FW 9.2, and not onboard in firmware, as in <FW 9.1
https://forum.blitzortung.org/showthread.php?tid=3648&pid=24682#pid24682
« Last Edit: November 03, 2022, 05:55:48 PM by Cutty Sark Sailor »

Offline 92merc

  • BismarckWeather.net
  • Forecaster
  • *****
  • Posts: 1398
  • BismarckWeather.net
    • BismarckWeather.net
Is this just a system blue problem then?  I didn't see any mention of RED systems.
https://www.BismarckWeather.net
Davis VP2, Cumulus, WeatherDisplay, Blitzortung, Saratoga Scripts, NOAA Stream via PI

Offline vreihen

  • El Niño chaser
  • Forecaster
  • *****
  • Posts: 1216
  • K2BIG
...and also if someone's System Blue station is still reporting as usual today, does this mean that they do not have the troubled GPS hardware and there's nothing to worry about?????
WU Gold Stars for everyone! :lol:

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
Issue seems currently limited only to GT g3 / Mediatek MT3333 ... with <3.2 GPS firmware.  Issue shouldn't affect later Blues with newer module.  Now, I'm guessing, since I'm not that experienced with GPS, and way behind on technology, but:
Just thinking with fingers here... #-o :roll: :twisted: :-$
Personally, I think those lower version GT/MT had a bad 'birth date', or whatever it's called,  as this lot seems to be March, 2003... whatever they certainly had the vulnerability not to roll over, running exactly 1024 wks. 

Reds Haven't apparently done 1024 wks, from their 'birth' of the older GT module but, if they're going to be affected, I'd suggest anytime between now and before about 2029, assuming they were 'birthed' in 2010 or erlier, and assuming they have original GT firmware... .
 Another thought, Since WE built, and initialized / activatedthe REDS ourselves, it may be that that would start when we first powered our reds up... hmmm... maybe could check that.  So, if I activated my Red 689 in 2013(?) that might mean I could expect issues about 2032 or something...The BLUES were 'factory' built, initialized in 'batches', and perhaps somewhere along  certain modules were either 'glitched' from the git-go, or initialized with error ... just thinking out loud about why older reds aren't whackin out    while but oldest blues are.

Keeping in mind that the 1024 weeks for a device, has nothing to do with the GPS epoch time, but the count begins when the 'device' determined it was 'born'... or whatever the technical stuff really is...  Some devices have a command that can reset the 'RTC' (real time clock), (which is different from the 'GPS' satellite time signals, I understand) , but these don't seem to have been published anywhere that I could find.

So... the Blitz FW 9.3 just published for both Reds and Blues, effectively tells the controller to 'ignore' the RTC from the GPS module. It actually has little effect on computation anyway, but the controller doesn't like the bad time..... The computation server uses the computing server's RTC, against the 'receiving server's RTC, ... using the station's known, 'DNA' (sieve), firmware, antenna type, location, filter settings, operating mode,etc.
The controller, at least yesterday, was showing the 'bad date/time' on sigs page, and 'bad date/time' on Status Page, since, so far, the new firmware hasn't told the controller to fix it, just igno the GPS error for the 'watchdog'... I think. ??? Doesn't affect sigs sent, just allows them to be sent, instead of suppressed.

This means we should make darn sure our configuration for the controller is absolutely as accurate and stable as possible.. antenna type and manual mode with as few automatics as possible.  Everything computed is 'relative' to other receivers, so we want ours stable, and hopefully 'theirs'will be also.  Egon has been hinting at operators using only manual mode, as have I and some others, for several years now, and letting the system / network work as it's designed, with 'stable' stations, not with widely varying gains and band widths, etc.... also if we DON'T send noise signals, and go interference as designed, the smarter algorithms now used won't get smarter...

End of brain hurricane.

Mike
« Last Edit: November 05, 2022, 12:38:40 PM by Cutty Sark Sailor »

Offline Mattk

  • Forecaster
  • *****
  • Posts: 2325
The 1024 week rollover for a "device" as opposed to GPS (as such) is basically dependent on what the device manufacturer has engineered into the chip as part of their design, software wise. This isn't the first time manufacturers have been caught out fiddling the chipsets. 'Born' I suppose is one way of putting that and went on to live well past their expected design life. 

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
The 1024 week rollover for a "device" as opposed to GPS (as such) is basically dependent on what the device manufacturer has engineered into the chip as part of their design, software wise. This isn't the first time manufacturers have been caught out fiddling the chipsets. 'Born' I suppose is one way of putting that and went on to live well past their expected design life. 
Yes,... IF you can upload later firmware to this chip, which some have done sucessfully, (BUT I can't seem to get the AXN bin to load (yet) on my pre-release development system V19.1) the upgrade solves the issue. We know that.   The controller FW update, for both Reds and Blues is merely a work around... and does leave the 'wrong' year date on a couple of GUI status pages, including signals. 
Operators should study this thread on the BT internal board.
https://forum.blitzortung.org/showthread.php?tid=3648

Offline 92merc

  • BismarckWeather.net
  • Forecaster
  • *****
  • Posts: 1398
  • BismarckWeather.net
    • BismarckWeather.net
I was running 9.1 because the graphs on my controller page don't work on 9.2.   I see that's still the case on 9.3.  But I guess if that's the only way to fix the possible GPS issue, I'll stick with 9.3.
https://www.BismarckWeather.net
Davis VP2, Cumulus, WeatherDisplay, Blitzortung, Saratoga Scripts, NOAA Stream via PI

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
My 1439 BLUE is the first in NA, and 3rd online worldwide, a pre-release, call it 'development' system, much patched modified, and repaired.
 I had to declare war to get the GPS module FW updated ... heh ...
 Partly because the 75yr old grey cells need recharging, partly because of Network and Machine securities / etc. I strongly suspect.
 
 anyway... attached is what I had de-volved to environmentally through trial and error, and achieved success.
 Some of the tips in this garbled set of notes may help somebody,  and I have no intention of trying to streamline, or simplify them...
 at this point, “I'm grateful to just be alive!
 
 Mike

Offline miraculon

  • Sunrise Side Weather
  • Forecaster
  • *****
  • Posts: 4140
  • KE8DAF
    • Sunrise Side Weather in Rogers City MI USA
9.3 Fixed my Blue, GPS is working again. I updated the RED to 9.3 as well, but GPS looked OK there.

Greg H.



Blitzortung Stations #706 and #1682
CoCoRaHS: MI-PI-1
CWOP: CW4114 and KE8DAF-13
WU: KMIROGER7
Amateur Radio Callsign: KE8DAF

Offline Cutty Sark Sailor

  • WxElement panel
  • Forecaster
  • *****
  • Posts: 3410
    • Frankfort Weather - TwinHollies WeatherCenter
IF you've NOT updated to FW 9.4, suggest do so, both RED and BLUE systems.
Solves several issues. THIS HOWEVER does not update the GPS module firmware.
BLUEs upgraded to at least GPS FW 3.8 on the GT/Mediatek MT3333 module may see obvious improvement in PDOP/HDOP/VDOPand 1PPS accuracy with GPS/GLONASS and SBAS ... at least mine did... wasn't bad before, but all now generally <1 with mean absolute of about 8ns ... REDs running around 95ns mean absolute with P's 1-2+ generally

689
791
1439
« Last Edit: November 13, 2022, 09:45:33 AM by Cutty Sark Sailor »

Offline miraculon

  • Sunrise Side Weather
  • Forecaster
  • *****
  • Posts: 4140
  • KE8DAF
    • Sunrise Side Weather in Rogers City MI USA
Updated both to 9.4. Thanks for the notice...

Greg H.


Blitzortung Stations #706 and #1682
CoCoRaHS: MI-PI-1
CWOP: CW4114 and KE8DAF-13
WU: KMIROGER7
Amateur Radio Callsign: KE8DAF

 

anything