Try Luc’s suggestion of connecting to the unit via Victron Connect and then see what happens.
Do you mean I should use Victron Connect to sync the RTC? OK, I’ve just done that, so we’ll see the result tomorrow.
Hi @CZvacko, this unit has indeed fallen back on determining the end of a day based on irradiance instead of using the RTC. As Duncan said, the Cerbo has no control over the SolarSense and can only listen for the BLE advertisements that the SolarSense broadcasts.
Connect to the SolarSense with VictronConnect to resynchronize the RTC so that the yield graph resets at 00:00.
Hi all, I’ll continue investigating this issue (on working days though 
). Unfortunately I haven’t been able to reproduce it yet. It seems that the fault only happens on some units, and not often. So capturing the behaviour is going to be tricky.
If you see any pattern emerge from your own units (rain, snow, heat, cold, only on mondays, etc.) please let me know so I can narrow the search. Thank you all for your help thus far!
My problem began on July 22. I went through all the data in the database and found two anomalies on that day:
-
The battery voltage reading looks as if it’s stuck between 21:37 and 4:50 (the curve shouldn’t be flat; the data points are there).
-
The TxPowerLevel value dropped briefly. In another thread, you mention: (at low battery this is reduced). It happened only that day, never before.
I checked the camera recordings from that night; around 21:37, when the voltage reading stuck, weather conditions and temperature were normal. Later, around 23:20, it started to rain. The weather that day appears to be unrelated to this event.
Observation on Solar Sense. This relates to 26 July 2026. The cumulative yield for one of the Solar Sense units reset to zero twice during the day. All 3 Solar Sense units recorded the Time Since Last Sun through the night and reset to zero at or around sunrise.
The units are recoding “time” correctly for the cumulative field “TimeSinceLastSun” which tends to indicate that the RTC on the units is functioning correctly. The “CumulativeYield” should also be using the RTC to trigger the rest to Zero at Midnight. It is however impossible to prove that all three units reset to zero at midnight because one unit was already at zero.
It would appear that the RTC is operating correctly and that there is another issue that is causing the Cumulative Yield to reset. This observation is limited only to the time frame of yesterday’s data because previous data does not support this observation.
CumYield26_27.pdf (208.3 KB)
TimeSince26_27.pdf (249.0 KB)
Hi @CZvacko, was it only the battery voltage that was stuck during this period, or did the other values (timeSinceLastSun, cellTemperature) seem stuck as well?
At that time, only the battery voltage stuck. I just checked the other days and there was a flat curve every night. But always it’s at a lower level; this time it was at a high level.
BTW, yesterday, after the RTC sync, the daily yield was correctly reset at 0:00.
FYI: To save power, the SolarSense goes into increasingly longer and deeper power-down modes as night (darkness) goes on. The flatness you are seeing is probably a reduction in measurement jitter as there are fewer measurements.
Yesterday Solar Sense Unit (Magenta) went through two rests to zero, previously posted. Just now Solar Sense (Blue) did a partial rest but not to zero. See PDF attached. SS Blue is the other unit that has displayed erratic behaviour ie. no discernible pattern although less frequently than SS Magenta.
When I look at the voltage there is a lot of fluctuations in the charging voltage for both SS Blue and SS Magenta but very little voltage fluctuation in SS Yellow. Around the time of SS Blue partial reset there are far larger voltage fluctuations when compared to SS Magenta. But tongue in cheek - it looks like SS Yellow has been prescribed Diazepam. Why the large differences in charging voltage especially when compared to yellow? This is a rhetorical question.
Volt-CumYield 27-7.pdf (455.7 KB)
Just seeking some information. In the MQTT data for the Solar Sense is an output field “UnspecifiedRemnant”. I only noticed this filed today and have not previously looked at this field.
What does this information relate to? It’s state appears to be either 0 or 1
Of the 3 Solar Sense units installed, 1 unit has held data consistently and not shown any errors. This unit is the only one that has published data in the “UnspecifiedRemnant” and its results are either 1 or 0. The 2 units that have shown errors are not reporting in this field today.
Hi @McKillop UnspecifiedRemnant indicates which version of the instantReadout record definition the SolarSense uses. All SolarSense versions should have this fixed at “1”.
In the future we may remove fields like TimeSinceLastSun and BatteryVoltage and put something else there. An UnspecifiedRemnant at “0” then signals to the decoder that it should use a different bitmapping.
I only added the Unspecified Remnant as a logged item just past midday. The results are attached. All three units oscillate between 0 and 1.
I would like to see Battery Voltage remain if at all possible. It indicates issues with batteries with far more advance notice than a “Low Battery” alarm. UnspecifiedRemnant.pdf (1.4 MB)
For Instance
BatteryVoltage.pdf (280.6 KB)
Solar Sense 750 Data Analysis
The data sheet for the units states that the battery is an “Internal battery Rechargeable 40 mAh / 3,6 V - Operating temperature range -40 to +85 °C”
The printed Circuit board of the unit (See attached pdf) states “Rechargeable battery LIR2450 only”
The battery that is installed in the unit is an LSC2440. There is no data sheet that I can find for this battery but the manufacturer website Supply LSC2440 3.6V 40mAh Rechargeable Super Lithium-Ion Capacitor Battery Wholesale Factory - YICHANG POWER GLORY TECHNOLOGY CO., LTD states that the battery is a “LSC2440 3.6V 40mAh Rechargeable Super Lithium-Ion Capacitor Battery” Temperature range “-40 - +60C” Life Expectancy >= 100,000 cycles.
The LIR 2450 is Rechargeable Lithium Ion Button Battery with a nominal capacity of 100 mAh. Temperature range Charge “0 - 40C” Discharge “-20 - +60C”. Life Expectancy “>500 cycles at 0.2C mA discharge”
The 31 July 2026 approximately 0930 hrs, I replaced one of the units with a new unit, changed the VRM device instance of the new unit to that of the old unit (20) and the old unit was given a device instance number of (24). This enabled continuity of data for the Solar array that the original unit was recording. Note: when viewing the historical attachments to this thread the blue graph line up until 31/07/2026 is the original unit and onwards it is the new replacement unit. The Orange line (24) is the originally installed init previously tagged as unit (20)
The original unit was disassembled and the button battery was removed. The voltage reading was 3.77 VDC. The battery was replaced and the unit installed in close proximity tho the one unit (21) that had not indicated any data anomalies.
During the day the recorded voltage of unit (20) displayed a voltage variation within a similar band to unit (21). See graphs of both units (20) and (21) for 2 periods 28/07/26 and 01/08/26 0600 hrs - 1900 hrs.
01 August 2026 approximately 1500 hrs. I removed the battery from the other unit that was displaying data anomalies (22). When removing the battery it had magnetic properties that held it to a small blade screwdriver that movement of approximately 0.3/ms did not break the magnetic field strength. (Not a very scientific methodology but I do not have Hall effect or Gauss meters). The unit was reassembled and remained in its original location. The voltage variation of unit (22) displayed a similar tight voltage range compared to its historical range. Now units (20) (21) (22) have similar voltage ranges.
The presence of a magnetic field in the LSC2440 battery is possibly capable of causing data corruption and voltage variations that has been observed in the SolarSense units. In this installation with 3 units, one performing and two displaying data anomalies, tends to indicate that the battery performance is varying, which would explain why some units operate correctly and others do not, hence the results are not reproducible.
As of today unit (22) is still displaying data corruption even though it has been connected twice to VictronConnect as suggested by Luc. I should receive 5x LIR2450 batteries on 03 August (tomorrow) and will replace the LSC2440 capacitor battery with a LIR2450 battery on unit (22) and will see how the unit performs in terms of data stability.
At present the indicators of what may be causing data corruption is the performance of the LSC2440 battery but this is purely a very preliminary hypothesis.
Good(21)Internal.pdf (16.4 MB)
Bad(20)Renamed(24)Internal.pdf (15.3 MB)
Last24Hours02Aug.pdf (139.5 KB)
2026-08-02 0600-1900 20 and 21.pdf (160.9 KB)
2026-07-28 0600-1900 20 and 21.pdf (325.7 KB)
2026-07-28 0600-1900 (22).pdf (235.8 KB)
Last2daysVoltage 4 units.pdf (271.6 KB)
Monday 3 August 2026 1520 hrs. One battery was swapped in the SolarSense unit (22). The replacement battery is an LIR2450 and at installation had a voltage of 3.762 VDC. The unit was connected with the Victron Connect App. The unit reset accumulated yield at 0001 hrs (approx) as did the other 3 units. Voltage readings during the night were stable and recharge after sunrise has the voltage reading at a stable 3.80 VDC. No errors in any of the data fields was recorded over this period.
4 August 2026 0820 hrs. The battery in the other SolarSense unit displaying data anomalies (24). The voltage of the battery was 3.945 VDC. This is higher than the average reading of the units over the past 4 days which is a fairly stable 3.8 VDC. I anticipate that the unit will stabilise at 3.8 VDC during the course of the day. This unit (24) had stable voltage reading during the past 4 days, after the removal and reinstallation of the original LSC2440 battery.
At present there are now 2 units with LSC2440 batteries of which neither of the 2 units had displayed data anomalies. 2 units which had displayed data anomalies now have LIR2450 batteries installed. If data anomalies occur with the 2 units with replacement batteries over the next month then the issue with the units is another issue(s) that is not associated with the batteries. If no data anomalies are recorded this would indicate that the original battery LSC2440 was the most probable cause of the data anomalies.
Hi @McKillop thank you for the deep analysis and photos. Your observation of a magnetized battery is interesting, but I wouldn’t expect a static magnetic field to influence the operation of the SolarSense. Also: are you sure that the tip of your screwdriver isn’t a bit magnetized? The battery housing is slightly ferromagnetic.
The variability in battery voltage for unit (20) does look like a HW problem related to the battery (or the mounting of the battery). Lets see how it will behave from now on.
FYI: The PCB silkscreen still says only to use LIR2450 but we changed it with LSC2440 due to improved high-temperature performance. LSC2440 is rated for high-temperature operation, albeit at a reduced lifetime. LIR2450 is not rated for high-temperature operation.
