ATC700: in Online Sleep the clock is set back to the time of the retained position

Hello, I found another problem with my Teltonika ATC700 tracker (previous reports: stops sending after queued commands, setparam response lost, movement does not turn on GNSS (cannot put third link :frowning: )). While the device sleeps, it periodically sets its clock back to the timestamp of the retained position, so records get timestamps hours or days in the past.

Setup

Device: ATC700 (Hw:ATC700 Mod:21), FW 3.20.0-rc.1+7.exp.autonomous, modem FW EG915UEUABR03A01M08_01.215.01.215

Configuration: Asset movement mode Movement (12150:1), Online Sleep (102:3), records every 3600 s when stopped, Send period 0, 107:0 (but the same happens also with 107:2 and with records every 300 s).

Problem

I left the device at home near the window for almost tow days. Then I looked in Traccar logs and obtained device dump/logs using the TCT.

The device falls asleep with a GNSS fix. GPSD_PWU_CTRL then turns on GNSS every hour and gets a fix (PWU fix captured, holding). When it goes back to sleep, the GNSS source changes to Retained and the time sync uses the retained position’s timestamp instead of the new fix. Every second hour the difference is “significant” and the clock jumps back to the time the device fell asleep.

The records are still sent every hour, but their timestamps drift further into the past: 2 h 02 min after 2 hours, 4 h 04 min after 4 hours, and over 40 hours after two days. The server shows them as old positions.

Steps to reproduce

  1. Configure Online Sleep, get a GNSS fix and let the device fall asleep in a place where it still gets a fix (e.g. by a window).
  2. Leave it still for a few hours.
  3. Compare the record timestamps with the time they arrive at the server.

Records from 21–23 Oct (device asleep since 21 Oct 23:58:38 UTC):

Received (UTC) Record time (UTC) Offset
22 Oct 01:58 22 Oct 01:58:39 0
22 Oct 03:58 22 Oct 01:56:27 2 h 02 min
22 Oct 05:58 22 Oct 01:54:14 4 h 04 min
23 Oct 07:58 22 Oct 01:32:44 30 h 26 min

Logs (from the Dump)

[M:GPSD_PWU_CTRL]: PWU fix captured, holding
[M:GPSD_PWU_CTRL]: PWU wait hold elapsed, back to sleeping
[M:GPSD_PWU_CTRL]: PWU scheduled timer for 3600s (state 1)
[M:GPSD]: GNSS source changed to -> 5
[M:GPSD__TIME_SYNC]: Unsupported GNSS source for time sync
[M:GPSD__TIME_SYNC]: Time difference is significant, syncing time
[M:TIME]: Syncing to [1792627118]
[M:TIME]: Synchronizing with undefined source

1792627118 is 2026-10-21 23:58:38 UTC, the time of the fix the device fell asleep with. This sequence appears 17 times in two days, always with the same value. With 107:2 it also happened after PWU wait fix failure.

The Dump and the server-side log are available. I can send them privately.

Expected

The clock should be synced only from a fresh GNSS fix, NTP or the network, never from the retained position.

Good Day,

When the device does not have a current GNSS fix, it may retain and use the last known GNSS coordinates. However, the record timestamp should come from the device’s latest successful time synchronization rather than from the retained position itself.

Kindly set parameter 107 to 2 (After Time Sync). With this setting, the device should save and send a record only after its internal time has been synchronized, preventing records with an unsynchronized timestamp from being transmitted.

Since you mentioned that the issue may also occur with 107:2, please apply the setting again, restart the device using the cpureset command and repeat the test. If records are still transmitted with timestamps from the retained GNSS position, the complete dump and server-side logs will be required for further investigation.

Please contact your sales manager to create a Partner Portal ticket and attach the relevant logs, configuration file, device IMEI and exact test timestamps.

Thanks, and best regards,
Imran S.