VANILLA HAS A PULSE: 720 HOURS INSIDE MY CAMPERVAN’S ELECTRICAL BRAIN

Thirty days. Seven hundred and twenty consecutive hourly records. One 200 W solar panel. One AGM leisure battery. A Victron SmartShunt quietly counting every amp that crossed the battery terminals, and a Victron SmartSolar controller recording what the roof managed to pull out of a September sky.

I exported the lot.

What came back was not a boring spreadsheet. It was Vanilla’s electrical heartbeat.

Across the exact 720-hour window from 12:00 on 2 September to 11:00 on 2 October 2026, the data records sunshine arriving, fridges sipping power through the night, charging systems waking up, heavy loads appearing without warning, state of charge rising and falling, and one extraordinary episode where an almost-full battery was dragged down to 55 per cent before the whole system hauled itself back onto its feet the following morning.

This is probably the deepest look I have ever taken at how Vanilla actually uses electricity. Not what the brochure says. Not what a battery calculator predicts.

This is what really happened.

The first number is 8.268 kWh. That is how much solar energy Vanilla harvested during the 30-day period.

The SmartShunt recorded 519.05 Ah flowing into the leisure battery and 501.80 Ah flowing back out. Despite all the nights, the fridge, lighting, charging and general campervan life, the final balance across the month was +17.25 Ah.

Average state of charge was 91.5 per cent. Vanilla finished the analysis window at 100 per cent SOC.

That alone tells me something important. The electrical system was not spending September slowly bleeding itself dry. Taken across the entire month, the system repeatedly recovered.

But averages are sneaky little creatures. They can make a riot look like a tea party.

Buried inside that reassuring monthly average is a low point of 55 per cent SOC and a logged voltage of just 10.58 V.

More on that later, because 28 and 29 September deserve their own chapter.

The headline dashboard looks like this:

Solar harvested: 8,268 Wh / 8.268 kWh Battery charge recorded: 519.05 Ah Battery discharge recorded: 501.80 Ah Net battery balance: +17.25 Ah Average SOC: 91.5% Lowest SOC: 55% Highest SOC: 100% Lowest logged voltage: 10.58 V Highest logged voltage: 14.39 V Best solar hour: 102 Wh Strongest charge hour: +16.49 Ah Strongest discharge hour: −7.65 Ah

That 8.268 kWh works out at roughly 276 Wh per day averaged across the whole period.

This was September, not midsummer. Daylight was shortening every day, the van was not permanently parked at a laboratory-perfect angle, cloud existed, shadows existed, and yet the roof still quietly found more than eight kilowatt-hours.

The reason I can see all this so clearly is Victron.

The SmartShunt sits in the battery circuit and measures what actually crosses the battery boundary. Charge goes in, load comes out. It does not care whether I think the battery “should” be full.

It counts.

The SmartSolar MPPT does a different job. It watches the photovoltaic side of the system and continually hunts for the voltage and current combination that extracts the most useful power from the panel.

Put SmartShunt and SmartSolar history together and suddenly the campervan stops being a collection of electrical boxes.

It becomes a measurable energy system.

And once you start seeing it like that, you can become dangerously interested in numbers.

Solar production was not remotely flat.

Of course it wasn’t. British September weather does not do flat unless we are discussing cloud bases.

The standout day was 26 September with 569 Wh harvested. That is comfortably the best solar day in the entire dataset.

The best individual hour occurred at 10:00 that morning, when the MPPT logged 102 Wh.

The following hour produced another 100 Wh.

That distinction matters.

The export records energy inside hourly buckets. So 102 Wh during one hour means the panel averaged roughly 102 W across that entire hour. It was not simply a one-second flash of 102 W that happened to appear on a screen.

Sustaining that sort of average for the hour is much more meaningful.

The fascinating twist is that the best solar day was not the best battery day.

On 26 September the battery still ended the day about 16.61 Ah down.

That is one of my favourite findings in the entire investigation because it destroys a very easy assumption:

Lots of solar does not automatically mean the battery is gaining energy.

If consumption rises faster than solar generation, you can have a glorious solar day and still finish electrically poorer than you started.

Sunshine is income.

Loads are expenditure.

The battery balance is the bank statement.

Breaking the month into calendar weeks makes Vanilla’s behaviour even clearer.

The partial week beginning 31 August harvested 1,177 Wh and finished +1.91 Ah.

The week of 7 to 13 September harvested 2,138 Wh. The battery received 105.03 Ah and delivered 105.13 Ah.

Net result: −0.10 Ah.

That is almost absurdly close.

Over seven days of real campervan life, the electrical system effectively broke even to within one tenth of an amp-hour.

The week of 14 to 20 September harvested 2,041 Wh and finished +6.27 Ah.

The week of 21 to 27 September harvested 1,755 Wh and finished +5.12 Ah.

The final partial week beginning 28 September harvested 1,157 Wh and finished +4.05 Ah.

Read only those weekly net numbers and the whole month looks calm and civilised.

It was not.

Average all 30 days together by time of day and Vanilla develops a remarkably clear rhythm.

Between midnight and early morning, solar is obviously zero and the battery steadily gives energy away.

Midnight averages around −0.91 Ah per hour. At 03:00 the average is around −0.87 Ah. At 06:00 it is still around −0.86 Ah.

The first meaningful change arrives around 08:00, when the average hour turns positive at about +0.81 Ah.

At 09:00 the average climbs to +1.71 Ah.

Then the roof really wakes up.

10:00 is the strongest average solar hour of the month at about 44.1 Wh, with the battery averaging +2.81 Ah.

11:00 follows at about 42.8 Wh and +2.73 Ah.

By late afternoon the curve begins to soften. At 17:00 the average battery flow is still slightly positive.

By 18:00 it has crossed back into negative territory.

At 21:00 the average drain reaches roughly −1.40 Ah, the deepest normal point of the evening profile.

From 19:00 through to 06:59, the system averages approximately −0.93 Ah per hour overall.

This is Vanilla breathing.

Evening: exhale. Night: slow discharge. Dawn: bottom of the curve. Morning: solar starts to bite. Late morning: recovery. Afternoon: consolidation.

Then the cycle begins again.

Once you see the hourly heatmap, the pattern becomes almost embarrassingly obvious. Bands of overnight discharge sit across the graph while the centre of each day fills with solar and positive battery flow.

It looks less like a spreadsheet and more like an ECG for a campervan.

At midday on 28 September, Vanilla was essentially full.

The SmartShunt showed 99.5 per cent SOC.

One hour later it was 95 per cent.

At 14:00 the shunt logged −7.65 Ah for the hour and SOC had fallen to 87 per cent.

At 15:00 another −6.72 Ah disappeared.

SOC: 79.5 per cent.

At 16:00 the battery lost another −6.30 Ah.

SOC: 72.5 per cent.

Logged voltage: 11.49 V.

At 17:00 SOC had fallen to 66.5 per cent.

At 18:00 it was 63 per cent and voltage was down to 10.88 V.

By 23:00 the SmartShunt showed 57.5 per cent.

The decline continued into the early hours of 29 September until SOC reached 55 per cent.

At 06:00 the logged battery voltage hit the lowest figure in the entire 30-day dataset:

10.58 V.

In one day, 28 September recorded 46.19 Ah leaving the battery and only 6.57 Ah going back in.

Daily net result:

−39.62 Ah.

That is not normal background campervan consumption.

That is an event.

At 07:00 on 29 September something changes.

Battery flow turns positive at +3.83 Ah.

At 08:00 the SmartShunt records +15.30 Ah and SOC jumps to 65 per cent.

At 09:00 another +11.20 Ah arrives.

SOC climbs to 78 per cent and voltage rises to 13.61 V.

At 10:00 another +7.05 Ah goes in.

SOC reaches 86.5 per cent and voltage is 14.05 V.

Across 29 September, the battery took 47.02 Ah and delivered just 10.93 Ah.

Net result:

+36.09 Ah.

The day before had removed 39.62 Ah.

The following day put 36.09 Ah back.

That pair of days is the electrical equivalent of falling down the stairs and then marching straight back up them.

What caused the heavy load on 28 September cannot be proven from an hourly export alone.

The strongest hour, −7.65 Ah at around 12 V, corresponds to roughly 92 W of average battery-side power across the full hour.

That is exactly the kind of clue that makes me look at deliberate DC charging loads, fridge demand, lighting, heater auxiliaries and anything else attached to the leisure system at the time.

The important point is not to invent a culprit.

The important point is that the data exposed the event clearly enough that I now know exactly when to investigate.

The month contained plenty of perfectly healthy-looking voltage behaviour, including charging voltages above 14 V.

But the lowest logged figure, 10.58 V, deserves respect.

Voltage under load is not the same thing as resting voltage. Cable resistance, a large temporary load, battery chemistry, temperature and the precise measurement point can all influence what appears in a log.

So one low-voltage record should not automatically be read as a laboratory measurement of true resting state of charge.

Even with that caveat, a sustained run around the low elevens and high tens on an AGM leisure battery is not something I want to make a habit of.

The data shows voltage below 11.5 V for 15 consecutive hourly records, from the afternoon of 28 September into the morning of the 29th.

That makes the event considerably more interesting than one rogue sample.

This is precisely why a proper battery monitor changes the conversation.

Without the SmartShunt I might simply have remembered that “the battery got a bit low.”

With the export, I can watch the descent hour by hour.

There is a wonderfully misleading field in the solar export.

Every hourly record reports Solar Load as 0 Wh.

Clearly Vanilla did not consume zero energy for 30 days.

The SmartShunt alone counted 501.80 Ah leaving the battery.

The explanation is measurement architecture.

The MPPT load reading only knows about loads measured through the controller's relevant monitoring path. Vanilla’s actual habitation loads are being measured at the battery by the SmartShunt instead.

So when I want to know how much energy the van actually used, the SmartShunt is the witness I trust.

The solar controller tells me what arrived from the sky.

The SmartShunt tells me what happened at the battery.

Together they tell the story.

I originally wanted to analyse everything by week, day, hour and minute.

There is one snag.

These Victron CSV exports contain one measurement bucket per hour. They do not contain 60 separate one-minute samples inside each hour.

That means I can produce a legitimate minute-equivalent rate, but I cannot pretend to know what happened at 14:17 compared with 14:18.

The best solar hour of 102 Wh works out to an average 1.700 Wh per minute across that hour.

The strongest charge hour, +16.49 Ah, is equivalent to an average +0.2748 Ah per minute.

The strongest discharge hour, −7.65 Ah, is equivalent to −0.1275 Ah per minute.

Across the whole 720-hour dataset, solar averages about 0.191 Wh per minute when nights are included.

Looking only at hours where solar was actually producing, that rises to roughly 0.380 Wh per minute.

Those are useful normalised figures, but they are not genuine 60-second logs.

That distinction matters.

Good analysis should make the data more exciting, not make it fictional.

The daily solar chart shows a wonderfully uneven September skyline, with 26 September towering above everything at 569 Wh.

The daily net-Ah chart tells a completely different story. It exposes days where good solar still failed to keep pace with consumption, then throws the 28 September −39.62 Ah event off the bottom of the graph like somebody dropped an anchor through it.

The SOC chart is calmer until late September, when it suddenly dives towards 55 per cent.

The voltage chart follows the same drama but with even sharper teeth, collapsing towards 10.58 V before snapping upwards as charging returns on the 29th.

The hourly solar heatmap shows the sun marching across each day.

The battery-flow heatmap overlays human behaviour on top of it: evenings, nights, morning charging, larger loads and recovery periods.

My favourite chart may be cumulative amp-hours.

Over 30 days, positive and negative energy flows chase one another like two lines in a very slow race, yet finish remarkably close together.

That graph explains the entire system better than almost anything else.

Vanilla is not surviving because the battery never gets used.

She is surviving because energy repeatedly comes back.

I used to think about leisure batteries mainly in terms of one headline number:

Amp-hours.

This data makes that feel painfully simplistic.

What matters is not merely how many amp-hours sit under the passenger seat. It is the relationship between storage, charge rate, solar opportunity, driving charge, overnight base load, occasional heavy loads and how deeply the chemistry is being cycled.

A 100-ish Ah battery that is repeatedly replenished can be remarkably useful.

The same battery with poor replenishment can feel tiny.

Likewise, 200 W written on the back of a solar panel is not 200 W arriving every hour.

The panel lives in the real world, not on a specification sheet.

What I now have is something far more valuable than a theoretical calculator:

A measured baseline for my own van.

The timing of this export could hardly be better because it gives me a detailed AGM baseline before the planned Fogstar lithium upgrade.

When the lithium battery goes in, I will be able to run exactly the same analysis again and compare like with like: overnight SOC drop, voltage stability under load, usable capacity, charging acceptance, time spent near full, depth of discharge and recovery behaviour.

That is the experiment I am now genuinely excited about.

Instead of saying “lithium feels better”, I should be able to quantify exactly how much better it is inside Vanilla.

The September export is now the control sample.

If I had to compress all 720 hours into a single conclusion, it would be this:

Vanilla’s current electrical system is remarkably good at recovering what it uses, the 200 W solar installation is making a meaningful contribution even in September, but the data also proves how quickly a substantial sustained load can expose the limited usable reserve and voltage stability of the present AGM setup.

That is the bit a glossy product brochure could never tell me.

I know it because Vanilla recorded it herself.

8.268 kWh harvested from the roof.

519.05 Ah pushed into the battery.

501.80 Ah taken back out.

+17.25 Ah left on the scoreboard.

91.5 per cent average SOC.

569 Wh on the best solar day.

102 Wh in the best solar hour.

55 per cent at the deepest SOC point.

10.58 V at the lowest logged voltage.

−39.62 Ah on the hardest day.

+36.09 Ah on the comeback day.

720 hours analysed.

And at 11:00 on 2 October, the final record in the export showed Vanilla back at 100 per cent SOC, sitting at 14.34 V and still charging.

After a month of clouds, nights, loads, solar, deep discharge and recovery, she ends the story full.

That feels like a rather appropriate full stop.

Real data. Real van. No brochure maths.