Vanilla versus the darknessThe 600 Wh
midnight mystery
Three Victron exports. 1,584 solar-hour rows. 624 shunt-hour rows. Twenty-five reconstructed nights. A 600 Wh double mystery — and an energy ledger so forensic it needs its own evidence room.
EVERY AMP-HOUR
UNDER OATH.
This is the part where the cosy campervan blog turns into an electrical crime scene. Three raw exports. Hundreds of time-stamped battery readings. Competing energy sources. Midnight spikes. A daylight-saving complication. And, crucially, no individual appliance meter. We can test the arithmetic brutally. We cannot manufacture measurements that were never taken.
66 days × 24 rows; 1,564 have solar, load and voltage values.
6–31 Oct; 611 have charge and SOC values.
Usable daily summaries; 1–5 Oct are missing in this export.
19:00–07:00 twelve-row windows, 6–30 Oct.
What the shunt does measure
Signed hourly charge in amp-hours and estimated battery state of charge. Pairing the Ah with the logged battery voltage gives an approximate watt-hour equivalent for the hour. During a wholly negative-charge window this captures net battery energy out — not a separate fridge or heater reading.
What the export cannot reveal
Exact instantaneous wattage, Webasto duty cycle, individual fridge compressor cycles, which devices charged, where the 20 and 23 October nights occurred, and whether a separate Jackery fed any devices. There is no installed inverter, despite earlier assumptions.
Watt-hours ≠ watts
601 Wh spread over twelve hours corresponds to about 50 W average. The worst measured hour was approximately 90 Wh — equivalent to 90 W averaged over that hour. None of these establish a 601 W instantaneous load.
25 October clock-change problem
The UK clocks went back on 25 October 2026. The export has only one row labelled 01:00 and no UTC offset. That means an absolute 25 October timeline cannot be fully reconstructed from these timestamp strings alone.
20 October, 21:00: −6.86 Ah × 13.13 V = approximately −90.07 WhThe Ah value is signed; negative means net battery discharge. Watt-hours calculated with one hourly voltage sample approximate, rather than independently meter, the true integrated energy.
Twelve hours.
Two campsites.
One pattern.
At Daventry and in the Peak District, Vanilla's battery recorded almost identical overnight energy use — despite different locations and conditions. That gives us an unusually useful real-world baseline.
30.32 Ah discharged across the 19:00–07:00 window, estimated using logged battery voltage. Average around 33 W.
30.98 Ah across the matching window. Average around 34 W. Only about 2.3% more energy than Daventry.
These weren't laboratory tests. They were actual nights in a campervan: your 50-litre compressor fridge operating, the Webasto diesel heater set to approximately 21°C, interior electrical equipment active, and portable devices being charged.
The important detail is that all these consumers share the battery. The SmartShunt doesn't label individual appliances, but it does reveal the total net energy leaving the leisure battery. And, on these two dates, the combined demand was strikingly repeatable.
The takeaway: Around 400 Wh per inhabited overnight period is a defensible description of those two recorded camping nights. It is not a universal daily average, and it is not proof that the heater alone consumed any particular share.
The two-night comparison.
Different roads. Different skies. Almost the same draw from the Fogstar battery. Actual readings: 400 Wh versus 409 Wh.
Follow every hour.
Choose any overnight window. The trace below is reconstructed from the hourly signed amp-hour readings multiplied by the corresponding logged battery voltage. Taller peaks mean more energy left the battery in that hour.
20 Oct mystery
Select a bar to inspect its recorded energy discharge.
The hourly figures are period-labelled export values; they are not instant readings. The chart is based on a 12-hour row window, ending before the 07:00 row.
600 Wh twice.
Two different signatures.
The headline says “600 Wh mystery”. The detective work says something far more interesting: 20–21 October and 23–24 October look different when the clock is allowed to testify. Select the night, then interrogate the 12 hourly observations.
Five hours before midnight account for 68% of the twelve-hour battery discharge.
408.3 Wh + 192.3 Wh = 600.6 WhFive rows cover 19:00–23:00; seven cover 00:00–06:00. Date and location cannot be inferred from this shape alone.
Hourly discharge fingerprint
Requires JavaScript for live trace. Facts and comparisons are printed below.68% of this 601 Wh night came before midnight.
About 27.5 W as an hourly average.
46% of this 597 Wh night came before midnight.
54% after midnight; a different temporal shape.
The 8-night thermal-style discharge matrix
Each square is one hour of net discharge in Wh, from 19:00 through 06:00. Darker means quieter; brighter orange means heavier demand. This is a power-use heatmap, not a thermal camera photograph.
Times are the labels in the source exports; timezone and measurement-window semantics are not independently verified.
The evening that
changed everything.
On 20 October, the discharge didn't rise evenly through the night. It surged in the evening and then settled down around midnight. That's the clue.
Approximately two-thirds of the total overnight energy was recorded during the first five hours. The combined draw averaged about 82 W across that period.
The subsequent seven recorded hours averaged roughly 27 W, much closer to the steady middle-of-night levels seen during camping.
Total ≈601 Wh from the two periods combined. Values are rounded.
This is why describing the whole night as a heater problem would be premature. At a normal running draw, a diesel heater alone doesn't explain a sustained roughly 82 W battery demand. Device charging, additional lighting, compressor activity or another temporary load could all contribute.
There's also a second approximately 597 Wh period on 23–24 October. We do not have reliable location or appliance records for that night. The correct forensic conclusion is that the load was high — not that we know what caused it.
Don't confuse energy and power: 601 Wh is the energy used over the twelve-hour interval; it is not a continuous 601 W electrical load. The overnight average was approximately 50 W.
Every watt has
somewhere to go.
The fridge, Webasto and device chargers can all operate together. The SmartShunt sees their combined net effect — not separate appliance totals.
THE COMPLETE
MIDNIGHT LEDGER.
We haven't just shown the interesting nights. This ranking includes all 25 complete 12-row overnight windows from 6 to 30 October. Six contain positive battery-charge intervals. Their net totals can be negative; that means more energy was added than removed over the window, not that the fridge somehow generated electricity.
| Night starting | Net battery out Wh | Charging intervals | SOC start → end | Context |
|---|---|---|---|---|
| 20 Oct | +601 | None | 81.2 → 37.4% | Location unconfirmed |
| 23 Oct | +597 | None | 92.5 → 51.8% | Location unconfirmed |
| 09 Oct | +505 | None | — | Location unconfirmed |
A signed net energy equivalent is used consistently: −Σ(Ah × voltage). It is not gross appliance consumption on any night containing positive charging. Six such nights are therefore separated from direct no-charge comparison.
When the sun is down,
who is charging?
Some positive battery-charge readings occur when the solar export reports zero watt-hours. That's not a paradox: Vanilla has an alternator-fed Victron Orion and a mains charger. But the logs cannot tell which one was operating. And a positive net Ah value cannot be treated as fridge power.
Solar logged 0 Wh. Positive battery charging is established; source unknown.
Solar logged 4 Wh. A substantial non-solar contribution is implied.
Hourly solar logged just 8 Wh.
Most hourly load entries show zero; that is not whole-campervan consumption.
The energy balance
Δ battery energy ≈ solar + alternator + mains − fridge − heater − charging devices − lighting − standbyThat is a conceptual system balance. The SmartShunt sees the net battery-side result, provided all charging and loads pass through its shunt. Source-by-source attribution needs separate meters or a controlled test.
Top positive-charge observations
These are interval Ah values with a voltage-derived energy equivalent; neither the input source nor the true instantaneous charger current is proved.
| Local row | Net Ah | Approx. Wh | Solar Wh |
|---|---|---|---|
| 29 Oct 22:00 | +26.58 | +366.5 | 0 |
| 30 Oct 08:00 | +27.87 | +389.1 | 4 |
Fridge. Heat.
Charging. Repeat.
The fridge stays switched on, but its compressor cycles. The Webasto runs to maintain a 21°C setting, yet its electrical draw changes with fan speed and starts. Device charging adds bursts of demand. None is individually metered in these exports.
On four quieter comparison nights, total battery use was approximately 88–110 Wh over twelve hours — around 7–9 W on average. That is a useful system baseline, not a fridge-only reading. Standby electronics and anything else left running are included.
On the two confirmed camping nights, demand averaged 33–34 W. A crude difference of around 25 W above baseline is consistent with heating and habitation, but cannot be assigned uniquely to the Webasto because device charging, lighting and fridge cycling also change.
Build a 12-hour camping night
Adjust the inputs to see how fridge, heater, USB charging and other loads might add up. This is a scenario model, not recovered appliance telemetry.
Even a perfect total match does not prove the appliance split. Plenty of different combinations could produce the same 400 Wh.
One critical detail: If devices are charged from a separate Jackery power station, that energy is not counted by the Fogstar SmartShunt — unless the Jackery itself is being recharged from the Fogstar system. The charging source changes the interpretation.
105 Ah says one thing.
SOC sometimes says another.
A forensic energy audit should cross-check two independent descriptions of a night's battery use: summed amp-hours and the reported fall in state of charge. They are related, but they don't match perfectly — and pretending otherwise would make this investigation less reliable.
The 105 Ah sanity-check table
Expected SOC fall = discharged Ah ÷ 105 Ah × 100. Disagreement can arise from timestamp alignment, SmartShunt calibration or SOC synchronisation, battery-current integration and rounding. It is not by itself proof of a defective battery.
| Night starting | Net Ah out | Implied SOC fall | Logged SOC fall | Disagreement |
|---|---|---|---|---|
| 10 Oct | 30.32 Ah | 28.9% | 28.3% | −0.6 pp |
| 19 Oct | 30.98 Ah | 29.5% | 25.7% | −3.8 pp |
| 20 Oct | 45.86 Ah | 43.7% | 43.8% | +0.1 pp |
| 23 Oct | 45.48 Ah | 43.3% | 40.7% | −2.6 pp |
pp = percentage points. This comparison assumes the SmartShunt's capacity is correctly configured to 105 Ah. SOC endpoint labels and interval Ah labels may not be exactly aligned.
30.98 Ah ÷ 105 Ah = 29.5% nominal capacity used; logged SOC fell 25.7%; difference ≈ 3.8 percentage points.That discrepancy makes the charge and SOC streams useful cross-checks, not interchangeable measurements. A proper calibration audit should precede conclusions about actual usable capacity or battery health.
200 watts on the roof.
October disagrees.
Your rated 200 W rooftop panel does not supply 200 W all day. The hourly records capture the seasonal decline as autumn sunlight weakens and days shorten. They also show why good solar alone is not guaranteed to offset a cold camping night.
Every October day
Actual daily harvest from the hourly Solar History CSV. Tap any bar to inspect it.
Across 30 days, based on 8,382 Wh recorded generation.
A drop of approximately 38% relative to September. 31 October's record is partial.
From 20–24 October, the solar system harvested just 447 Wh in five days. That's only around 89 Wh per day — roughly a quarter of a single 400 Wh camping night. This is why the alternator charger and mains charger matter so much in autumn.
Try the off-grid
survival calculator.
The Fogstar Drift is rated at 105 Ah, approximately 1,344 Wh nominal. Planning conservatively from 100% down to 20% leaves about 1,075 Wh before any losses. Move the controls to build a scenario.
Assumes a full battery, no alternator/mains charging and repeating the same solar harvest and electrical demand each day.
Solar is variable. Use 0–90 Wh as a tougher autumn-weather scenario, not a guaranteed forecast.
There is no single correct answer to how long a campervan lasts off-grid. On a quiet parked day, consumption may be low. On a cold inhabited night, demand rises. Device charging and power-station use complicate the picture further.
Most importantly, the SmartShunt's hourly data can show battery energy movement, but not a guaranteed future runtime. The battery's real usable capacity, the external chargers and unpredictable winter sunshine all influence the result.
The meter audit
gets uncomfortable.
This is where we stop admiring the graphs and start questioning the instrumentation. The daily MPPT history and the hourly solar export describe the same system, but their daily totals are not identical. Some differences are small; one is large enough to demand attention.
6–31 Oct with valid MPPT daily yield records.
For those same 26 calendar dates only.
For the same subset of dates.
15 Oct: 180 Wh daily record versus 224 Wh hourly sum.
Daily yield reconciliation
Choose a day from the 26 overlapping daily controller records. A difference is evidence of two aggregation methods disagreeing; it is not automatically lost power or a faulty panel.
MPPT 180 Wh − hourly sum 224 Wh = −44 WhPotential contributors include export rounding, differences in sample aggregation and clock/window boundaries. We cannot decide which without the original controller records and logging configuration.
All 26 days, no cherry-picking
| Oct date | Controller | Hourly sum | Δ Wh |
|---|---|---|---|
| 15 Oct | 180 | 224 | −44 |
| 30 Oct | 130 | 122 | +8 |
Every hypothesis
must balance.
Try to explain the 20 October 19:00–23:00 spike — 408.3 Wh in five hours — using a hypothetical fridge/standby baseline, average heater draw and device charging. This is a constrained accounting exercise, not automatic appliance identification.
Reconstruct the five-hour spike
These are user-adjustable assumptions. A charger delivering 90 Wh to devices may consume more at the battery due to conversion losses. Heater-start surges are averaged into the scenario if they occurred.
Wh measured-equivalent
Wh hypothetically assigned
Wh unassigned (+) or overallocated (−)
A nonzero residual means this selection doesn't account for the measured battery demand.
Critical: Even a zero residual is not proof. With four adjustable inputs and one measured total, countless explanations can balance the same equation. This is an underdetermined system.
How to actually solve it
Fridge: Watch SmartShunt live amps with heater and chargers off. Record compressor-on and compressor-off currents; repeat several cycles without changing fridge settings.
Webasto: Keep fridge conditions stable; compare shunt current before startup, during startup and after reaching 21°C. Do not disconnect heater power during shutdown.
Device charging: Connect one charger at a time for a known interval. Repeat for the iPhone, iPad, MacBook or Jackery, documenting their starting and ending charge.
Instrumentation: Verify charger paths all pass through the SmartShunt. Log direct shunt measurements and capture SOC before and after, without simultaneous driving or mains hookup.
Don't take my word
for a single watt-hour.
The article carries its own derived, time-labelled night ledger. Export the comparison values and recalculate the numbers yourself. The released CSV is a derived evidence table, not a replacement for the three original raw files.
Download forensic hourly CSV ↓Generated locally from the data embedded in this HTML; requires JavaScript and browser download permission. No external analytics or API connection.
Not yet. The measured energy is real. The cause of every individual watt-hour is still an open question. That's what makes it interesting.
The verdict.
Not a battery mystery.
A usage mystery.
Vanilla's records strongly support different usage modes rather than a constant hidden battery drain. The two confirmed camping nights are consistent. The high-demand evenings are real, but their individual causes remain unidentified.
Working, but seasonal.
The MPPT logged meaningful solar generation; October output was lower than September. Its dedicated load-output field is not a whole-van energy meter.
The battery is supplying real loads.
Camping nights around 400 Wh are consistent with an active campervan. Overnight readings alone cannot prove battery health, capacity or exact heater demand.
Measure individual suspects.
Use the SmartShunt's live current reading with the fridge cycling, heater running and device chargers plugged in separately. That is how the remaining mystery gets solved.
The big revelation: Your normal quiet-night baseline is around 7–9 W, while the confirmed camping nights averaged 33–34 W. A later evening briefly averaged roughly 82 W. Those are three different operating states — and treating them as one would lead to the wrong diagnosis.
Real journeys. Real logs. Serious curiosity.
Source exports: Solar History.csv, PV and Battery Data.csv and Solar Charge and SOC Data.csv, recorded through 31 October 2026. For each overnight period, 12 labelled hourly entries from 19:00 through 06:00 are summed, using estimated Wh = negative Ah × recorded battery voltage. Values are rounded; hourly timestamp conventions and charge metering may shift exact hour-to-hour alignment. SmartShunt net charge is not a device-level submeter, and the MPPT load output does not represent all leisure-battery appliances. All hero illustrations are fictional visualisations; images' embedded graphics must not be interpreted as actual telemetry, weather, location coordinates or campsite details. The campervan has no installed inverter. The October 25 clock-change is ambiguous because timestamps lack UTC offsets; summed hourly solar and controller daily yields show measurable differences. Decorative hero imagery contains invented telemetry, including values exceeding the real 200 W solar rating. Designed for the NIX Stories self-contained immersive format.