Luke McCoy

Thinking differently about electricity and power consumption.

[Luke McCoy @cloud_thrifty]

I almost replaced a perfectly good battery cell today.

New 314 Ah pack, one day on the bank. Under load, cell 6 read 278 mV below all fifteen of its neighbors — the exact signature of a cell on its way out.

The cell was fine. A bolt was loose.

The tell is in the shape: the sag was perfectly proportional to current, 9.4 mΩ of resistance, and it vanished the instant the load stopped. A dying cell sags progressively and takes minutes to recover. A bad joint quits the moment the current does.

The BMS never saw it. It measures its own sense leads, not the busbar bolted next to them — it rated cell 6 at 61 mΩ against a 59 mΩ median. Right cell, useless magnitude.

One re-torque. That same cell now runs 4 mV high at 39 A.

Before you replace a cell, put a real load on the pack and check whether the sag tracks the current. Cells get blamed for what joints do.

[Luke McCoy @cloud_thrifty]

Geothermal, measured. 10 tons cooling 6,500 sq ft on a 100% off-grid house — no grid, no gas.

8 kW for an hour pulls the house down 3°F against a 25° outdoor delta.

Pictured: one of the two compressors and the loop reservoir. Horizontal ground loop, $22,210 in equipment, set by me and one contractor.

If you were building off-grid, would you put that $22k in the ground or in batteries? I went with the ground and would again.

[Luke McCoy @cloud_thrifty]

@MuelverRealty No critical-loads panel — the whole 6,500 sq ft house is behind the inverters. The real decision was investing in 10 tons of geothermal. 8 kW for an hour drops the house 3° against a 25° outdoor delta. Efficient cooling is cheaper than the battery bank bad cooling demands.

[Luke McCoy @cloud_thrifty]

64 panels staked to the ground: $7,040
Three EG4 inverters: $7,500
A battery bank I balance one cell at a time: $15,000

Three years without a power bill: priceless

[Luke McCoy @cloud_thrifty]

Worth doing the arithmetic. 5,600 Wh over 24h = a ~233W average draw. In the ~6 usable sun-hours the panels have two jobs — carry that 233W load AND refill the 5,600 Wh bank (≈930W) — so ~1,150W of real output. A 400W panel nets only ~80% real-world (~320W), so straight break-even is ~3 panels. I'd run 4 (1,600W): that 4th panel is your buffer — it covers cloudy days, panel aging, and refilling from a deeper draw, so you're not betting the bank on a perfect-sun day. Your 100W puts back about a tenth of the day, exactly like you said. Same math scaled up runs our off-grid place — the array stretches the battery, it never refills it on demand.

[Luke McCoy @cloud_thrifty]

Load calc is the right instinct — but a "5kVA" rating is only real if the battery can source it. At 48V that's ~125A continuous, so a single 100Ah pack (100A BMS) cuts out before you ever hit 5kW. And most budget 5kVA units are high-frequency — no transformer to buffer inrush, ~1.5-2x surge vs the 3-5x a low-frequency (transformer) inverter gives, which is what AC and well-pump motors need to start. So meter the real loads, then size the battery to the inverter — ours all ran over nameplate (well pump: 0.42 kWh/80 gal vs 0.11 calculated).

[Luke McCoy @cloud_thrifty]

Soft start's the right call — it'll pull the compressor inrush down to about 1.5x its running amps instead of the 4–5x a bare start pulls. Clamp an amp meter on the unit and you can watch the surge drop in real time. One caution: don't over-limit it — the compressor still needs a real slug of current to break away, and starving it of the inrush it's designed for is how you buy dramatically faster wear. Our whole off-grid setup runs on the same rule: size for the surge, not the average. Even the well pump got specced around startup, not its 1,800 W run.

[Luke McCoy @cloud_thrifty]

It's bitter melon (Momordica charantia). Thrives in July heat, climbs anything it can reach, and the fruit is exactly as bitter as advertised. Nobody's bragging rights this round — the garden stays undefeated.

[Luke McCoy @cloud_thrifty]

Garden game. This is growing in our off-grid kitchen garden right now.

Name it. Hint: you've probably walked past it in an Asian grocery, and almost nobody grows it in an American backyard.

Answer posted tomorrow. No prizes — just bragging rights.

[Luke McCoy @cloud_thrifty]

Hottest-load day of the summer, measured. 6,500 sq ft, 100% off-grid, AC cooling 83% of the afternoon.

The house used 77 kWh — its biggest day of the year. The ground array brought in 118. The battery bank was full by 4:30 PM and carried the evening.

Full disclosure: the diesel generator added 6 kWh at dawn, finishing what Friday's storms started. That's the whole diesel story for the week.

People say a house this size can't run off-grid. The meter disagrees.

[Luke McCoy @cloud_thrifty]

This is Tiki. Hatched in March — the first animal ever born on this 90-acre farm.

The farm runs 100% off-grid: solar and batteries for power, a well for water, septic for the rest. The only utility bill that reaches this house is internet.

She's named after our favorite dinner: chicken tikka masala. She doesn't know that, and we've agreed to keep it that way.

[Luke McCoy @cloud_thrifty]

Active, and not for the reason people usually give. Moving charge instead of burning it works both ends — the high cell drops AND a low cell fills, so every balancing amp counts double. Passive is fine until in-service drift outruns what the resistor can burn off; past that point the pack never self-balances. Same array, same day, both our packs:

[Luke McCoy @cloud_thrifty]

@rosegoldreeterd The unwritten chapter of every operator's manual: "undersized is fine if you're brave and wrong." Our 259D has taught us the same physics on the farm — the load chart is a suggestion right up until it's a headline. Glad you, the mower, and the trailer all survived.

[Luke McCoy @cloud_thrifty]

@BrewNeuron @JoeTransue @Point_of_Gold @Object_Zero_ Before buying the big one, meter your critical circuits for a week. We're 100% off-grid and every load we measured came in well above spec-sheet math — our well pump bills 0.42 kWh per 80 gallons against 0.11 calculated. Size to measured, not nameplate.

[Luke McCoy @cloud_thrifty]

@rospigge60559 That quiet hum is the whole point. One tip from our farm: meter the fridge and freezer for a week. Summer compressor duty cycle usually lands 2-3x above the label math, and that's the number that decides how many cloudy days your bank actually covers.

[Luke McCoy @cloud_thrifty]

Treating the water heater as part of the HVAC loop is the right instinct. Every big load in a house is a battery of sorts — tank temperature, house temperature. Timing them to production is storage you don't have to buy. We're 100% off-grid, and the biggest gains came from moving loads, not adding panels.

[Luke McCoy @cloud_thrifty]

Sound warning, and the same misunderstanding exists on the load side. People size systems from nameplate specs and textbook math. Our deep-well pump draws 4x what the lift physics says — 0.42 kWh per 80 gallons, measured at the inverters. Measure the real loads before buying hardware.

[Luke McCoy @cloud_thrifty]

Deep-well water has a precise price. 80 gallons, 450 ft of lift, 1,300 ft of pipe: 0.42 kWh, measured at the inverters — 15.4 minutes at 1,800 W.

An EV burns that in a mile and a half.

Physics says the lift is 0.11 kWh. The meter says 4x. If you're designing an off-grid system, that gap is the difference between a battery bank that carries you through a cloudy week and one that doesn't.

Design to measured loads. Every calculated number in your plan is a debt the generator collects.

[Luke McCoy @cloud_thrifty]

@dreamsofcode_io Are you still taking questions on the go nvim setup? I am unable to get past this.
Failed to run `config` for nvim-lspconfig

/home/luke/.config/nvim/lua/custom/configs/lspconfig.lua:12: attempt to index field 'root' (a nil value)

# stacktrace:
- nvim/lua/custom/configs/lspconfig.lua:12
- nvim/lua/custom/plugins.lua:14 _in_ **config**
- nvim/lua/core/init.lua:122

[Luke McCoy @cloud_thrifty]

@lexfridman This is a very subjective statement to support the value of an objective discipline. What is an anti-science sentiment?