In this guide
The most common solar project failure in student labs is not a bad panel — it is a good panel of the wrong size. A 20 W panel wired to a battery that needs 200 Wh a day will never keep up, and no charge controller setting fixes arithmetic. Sizing a solar setup is five calculations done in order: audit the load, find your sun hours, size the panel, size the battery, pick the charge controller. This guide works each step with real numbers and a complete worked example (a solar street light), plus the derating factors — temperature, dust, monsoon — that separate textbook answers from systems that survive an Indian summer. For the power-electronics context around batteries and regulators, read the batteries and regulators guide alongside; for build inspiration, the hybrid solar-wind charging station and the MPPT charge controller build show both ends of the complexity range.
Step 1: the load audit (Wh per day)
Everything starts from daily energy, in watt-hours. List every load, its power, and its hours per day:
Daily energy (Wh) = power (W) × hours per day (h)
| Load | Power | Hours/day | Energy |
|---|---|---|---|
| 20 W LED street light | 20 W | 10 h (dusk to dawn) | 200 Wh |
| Charge controller self-consumption | ~2 W | 24 h | ~48 Wh |
| Total | ~248 Wh/day |
Be honest about the hours: a "12-hour night" in the load table should use the actual timer setting, and intermittent loads (a pump that runs 3 hours, a sensor node awake 10% of the time) get their duty cycle, not 24 hours. For an irrigation pump node the pump's daily runtime dominates everything else — get that number from the application, not from optimism. Add 10–15% margin for loads you forgot; every real build has one.
Our running example: a 20 W LED solar street light running 10 hours nightly, total ~250 Wh/day after margin. Every subsequent step uses this number.
Step 2: peak sun hours for your location
A solar panel's wattage rating is measured at 1000 W/m² irradiance — laboratory noon sun. Your site does not get that all day. The honest unit is peak sun hours (PSH): the equivalent number of hours at full rated irradiance. A site with 5 PSH receives the same energy as 5 hours of perfect noon sun.
Indicative PSH ranges for India (annual averages; check NASA POWER or your state's solar atlas for the specific site):
| Region | Typical PSH | Design guidance |
|---|---|---|
| Rajasthan, Gujarat, western MP | 5.5–6.0 | Generous resource; still design with margin |
| Deccan plateau, most of Maharashtra/Karnataka/AP/Telangana | 5.0–5.5 | The default assumption for most student projects |
| Indo-Gangetic plain (UP, Bihar, Punjab) | 4.5–5.0 | Winter fog weeks need autonomy margin |
| East coast, Northeast | 4.0–4.5 | Cloudier; size conservatively |
| Monsoon months (June–September, most of India) | 2.5–3.5 effective | Design the battery for this, not the panel |
For student designs, use 4.5–5 PSH as the default unless you have site-specific data, and treat monsoon as a battery-autonomy problem (step 4), not a reason to triple the panel. Our example uses 5 PSH.
Step 3: panel wattage (with honest derating)
The panel must deliver the daily energy within the sun hours, after real-world losses:
Panel wattage = daily Wh / (PSH × derate factor)
The derate factor folds in everything the lab rating ignores:
- Temperature: panels lose roughly 0.35–0.45%/°C above the 25 °C cell rating temperature. An Indian rooftop cell routinely hits 60–65 °C — that is ~15% gone on a hot afternoon. (Check the panel datasheet's temperature coefficient; it is printed there.)
- Dust and soiling: 5–15% in dusty areas if the panel is not cleaned regularly. A fortnightly wipe is part of the design, not an afterthought.
- Wiring and controller losses: 3–8% depending on wire runs and controller type.
- Age: panels degrade ~0.5–0.8% per year; negligible for a student project, worth knowing.
A combined derate factor of 0.70–0.75 is the standard honest value for a rooftop student installation.
Worked example: 250 Wh/day ÷ (5 PSH × 0.72) ≈ 69 W → pick the next standard size up, a 75 W panel (or 2× 40 W in parallel). Notice the margin philosophy: rounding up to standard sizes, not down. An undersized panel fails silently over weeks as the battery walks down to empty; an oversized panel just finishes charging by noon.
Panel voltage must suit the controller and battery. A "12 V panel" actually has a Vmp (max-power voltage) around 17–18 V and Voc (open-circuit) around 21–22 V — deliberately above battery voltage so charging works in less-than-perfect sun. With an MPPT controller you can use higher-voltage (24 V / grid-tie) panels on a 12 V battery, which is often cheaper per watt. Match Voc to the controller's maximum input voltage with margin for cold mornings (voltage rises when cold).
Step 4: battery sizing (autonomy days and depth of discharge)
The battery bridges nights and cloudy stretches. Size it from the daily energy, the system voltage, how many sunless days it must ride through, and how deeply the chemistry may be discharged:
Battery Ah = (daily Wh × autonomy days) / (system voltage × depth of discharge)
- Autonomy days: 2–3 for most Indian sites; 3–5 if the site sees long monsoon overcast or winter fog weeks. More days = more cost and weight; this is the main budget lever.
- Depth of discharge (DoD): 50% for lead-acid (regular deep-cycle/SMF — going deeper shortens life sharply), 80–90% for LiFePO4. Never size lithium-ion/LiFePO4 without a proper BMS.
Worked example, 12 V system, 2 autonomy days:
- Lead-acid: 250 × 2 / (12 × 0.5) = 83 Ah → standard 100 Ah deep-cycle battery (rounding up to the available size).
- LiFePO4: 250 × 2 / (12 × 0.8) = 52 Ah → 50–60 Ah LiFePO4 with BMS.
The chemistry trade-off is honest: lead-acid is cheaper upfront, heavier, bulkier, and dies early if routinely deep-discharged; LiFePO4 costs roughly 2–3× more, lasts many more cycles, and delivers its rated capacity without the 50% rule. For a one-semester demo, lead-acid is the economical choice; for a deployment meant to run a year, LiFePO4 earns its price. Either way, fuse the battery positive terminal — a shorted 100 Ah battery is a welding machine.
Step 5: charge controller — PWM vs MPPT, and current rating
The controller sits between panel and battery, preventing overcharge and managing the charge profile. Two technologies:
| PWM | MPPT | |
|---|---|---|
| How it works | Connects panel directly to battery, pulling panel voltage down to battery voltage | DC-DC converter that holds the panel at its max-power point and converts excess voltage into extra current |
| Efficiency gain over PWM | Baseline | ~10–30% more harvest, most when the panel is cold or Vmp >> battery voltage |
| Panel voltage | Must roughly match battery (12 V panel → 12 V battery) | Flexible — 36/60-cell or series panels on 12/24 V batteries |
| Cost | Low (a few hundred rupees for 10 A) | 3–5× PWM for the same current |
| Right when | Small systems (<150–200 W), matched panel voltage, tight budget | Larger systems, mismatched/high-voltage panels, every watt counts |
Current rating rule: controller amps ≥ panel short-circuit current (Isc) × 1.25. Our 75 W panel: Isc ≈ 4.6 A → 4.6 × 1.25 ≈ 5.75 A → a 10 A controller (the next standard size, with headroom for a second panel later). Undersized controllers either current-limit (wasting your panel) or fail.
For students who want to understand rather than just buy, building the controller is a legitimate project in itself — the MPPT buck-converter charge controller project implements the tracking algorithm on a microcontroller, and the solar battery-bank monitor adds the dashboard layer that turns raw charging into visible data.
The sizing calculator (Python)
The whole method above compresses into a script you can reuse for every project — and paste into your report's methodology section:
def size_solar_system(loads_wh_per_day, sun_hours, autonomy_days,
system_voltage=12, battery_dod=0.5, derate=0.72):
# --- panel ---
panel_w = loads_wh_per_day / (sun_hours * derate)
# --- battery ---
battery_ah = (loads_wh_per_day * autonomy_days) / (system_voltage * battery_dod)
# --- controller: needs panel Isc; estimate from wattage at ~17.5 Vmp ---
imp_est = (panel_w * 1.15) / 17.5 # next-size-up panel, conservative
isc_est = imp_est * 1.1
controller_a = isc_est * 1.25
return {
"panel_W": round(panel_w, 1),
"panel_next_standard_W": int(-(-panel_w // 10 * 10)), # round up to 10 W
"battery_Ah": round(battery_ah, 1),
"controller_A": round(controller_a, 1),
}
# street-light example: 250 Wh/day, 5 sun hours, 2 autonomy days, lead-acid
print(size_solar_system(250, 5, 2, battery_dod=0.5))
# {'panel_W': 69.4, 'panel_next_standard_W': 70, 'battery_Ah': 83.3, 'controller_A': 6.3}
# (controller estimate uses pre-round-up wattage; feed the real panel Isc from its
# datasheet once selected — here 4.6 A x 1.25 = 5.75 A, so a standard 10 A unit)
(The controller estimate assumes a nominal 12 V panel; feed in the real Isc from the panel datasheet when you have selected the panel — datasheet values always beat estimates.) Round the controller up to the next standard size (10 A here), exactly as the worked example did.
Installation details that decide real performance
- Tilt and orientation: face true south (in India), tilted at roughly your latitude angle. A panel lying flat on a roof loses ~10–15% annually versus tilted; a panel facing the wrong way loses far more. Seasonal adjustment (steeper in winter) is a refinement, not a requirement.
- Shading: even partial shading disproportionately kills output — one shaded cell string drags its whole series section down (bypass diodes limit the damage to the shaded section, they do not eliminate it). Survey the site across the day: the shadow that is not there at 10 AM may cover the panel at 3 PM.
- Wiring: keep panel-to-controller runs short and use adequate gauge — at 12 V, a 3% voltage drop needs surprisingly thick wire over long runs. Fuse the battery positive as close to the battery as practical, and fuse each parallel panel string in multi-panel arrays.
- Ventilation behind the panel: an air gap under the panel keeps cell temperature a few degrees lower, which directly recovers some of the temperature derating.
AC loads: adding the inverter to the chain
So far everything is DC. If the load needs 230 V AC — a demonstration room, a small workshop rig — an inverter joins the chain between battery and load, and it takes its cut:
- Inverter efficiency: budget 85–90% for a decent pure-sine-wave inverter. Divide the AC energy by 0.85 to get the DC energy the battery must supply. Modified-sine inverters are cheaper and noisier — fine for resistive loads and basic tools, a poor choice for anything with a motor or sensitive electronics.
- Surge rating: motors and compressors draw 3–7× their running current at start. Size the inverter's surge rating for the largest starting load, not the running total.
- Idle consumption: inverters draw 10–30 W just being on. For a load that runs a few hours a day, switch the inverter off (or use its power-save mode) rather than leaving it idling — 20 W × 24 h is 480 Wh/day of nothing.
Second worked example: a 500 W AC demonstration load running 4 hours daily (2000 Wh AC). DC energy needed: 2000 / 0.85 ≈ 2350 Wh/day. Panel: 2350 / (5 × 0.72) ≈ 653 W → 700 W of panels (e.g. 2× 350 W in series on an MPPT controller). Battery, 24 V system, LiFePO4, 2 autonomy days: 2350 × 2 / (24 × 0.8) ≈ 245 Ah → 24 V 250 Ah bank. Inverter: 500 W running with, say, a small motor starting surge → 1.5 kVA pure-sine-wave unit. This is a serious, heavy, expensive rig — which is exactly why the load audit comes first: many "AC load" projects turn out to have DC-native alternatives (DC pumps, DC lighting) that delete the inverter, its losses and its cost in one move.
Grid-tie vs off-grid vs hybrid: which is your project?
| Architecture | How it works | Right when |
|---|---|---|
| Off-grid (standalone) | Panels → controller → battery → loads; no mains connection | Remote sites, street lights, field sensor stations, and nearly all student demo rigs |
| Grid-tie | Panels → grid-tie inverter → mains; no battery; exports surplus | Rooftop generation where the goal is the electricity bill — needs net-metering approval, not a student lab project |
| Hybrid | Both: battery backup plus grid interaction | Homes and institutions wanting backup + savings; complexity and cost beyond typical student scope |
Unless your project explicitly studies grid interaction, build off-grid. It is self-contained, testable in the lab, and every sizing step in this guide applies directly.
Series vs parallel: wiring multiple panels
- Series: voltages add, current stays the same. Two 18-Vmp panels in series give ~36 Vmp — the natural feed for an MPPT controller on a 12/24 V battery, with thinner wire for the same power. Shade on one panel, however, chokes the whole string (bypass diodes limit the loss to the shaded section).
- Parallel: currents add, voltage stays the same. Suits PWM controllers and keeps partial-shading losses local to the shaded panel. Needs a fuse per string and heavier wire.
Rule of thumb: MPPT + series for multi-panel arrays; PWM + parallel for small matched systems. Either way, keep panels in an array identical (same model and age) — mismatched panels in series are limited by the weakest, in parallel by the lowest voltage.
Earthing and lightning: do not skip this
Panel frames, mounting structures and the controller/inverter chassis get bonded to earth. In lightning-prone regions, a surge protection device (SPD) on the DC side and proper array earthing are part of a responsible installation — a direct strike will destroy unprotected electronics regardless, but induced surges from nearby strikes are the common killer and SPDs handle those. For a small student demo rig this can be as simple as solid frame earthing and keeping the installation disconnectable during storms; for anything permanent, follow the local electrical code rather than improvising.
What goes wrong in year two
Student projects are usually evaluated in months, but designing like the system must survive teaches better engineering:
- Lead-acid batteries are the consumable. Even well-treated, expect 2–4 years; routinely deep-discharged, far less. Budget the replacement on day one so it is a plan, not a surprise.
- Soiling is cumulative. A panel never cleaned in a dusty city loses output month after month. The fortnightly wipe in the maintenance checklist is the highest-ROI task in solar.
- Connections loosen. Thermal cycling works terminals loose over seasons; the quarterly torque check exists because of this.
- Loads creep. Someone plugs an extra light into the system "temporarily" and the sizing math silently breaks. Label the system's rated load at the distribution point.
Verify the panel is healthy (two multimeter tests)
Panels degrade and shippers damage them; test on arrival, in clear sun near midday:
- Open-circuit voltage (Voc): disconnect everything, measure across the panel terminals. It should read close to the datasheet Voc (typically within 5–10%, higher on cold mornings). Far below spec means damaged cells or a broken bypass diode.
- Short-circuit current (Isc): set the meter to a 10 A DC range and briefly short the panel through the meter in full sun. It should read near datasheet Isc. Do this quickly (a few seconds) and never on a large array — this test is for single small panels with a properly rated meter.
Log both numbers with the date; repeating the test yearly shows degradation honestly.
Troubleshooting table
| Symptom | Likely cause | Check |
|---|---|---|
| Battery never reaches full, even on sunny days | Panel undersized for the load; or heavy dust/shading | Re-run the sizing math with measured sun hours; clean panel; check for new shadows |
| Battery voltage collapses every night | Battery undersized or aged; DoD exceeded nightly | Measure specific gravity (lead-acid) or capacity-test; increase autonomy days in the sizing |
| Controller shows full by 10 AM but battery dies early | Battery capacity lost (sulphated/aged) — voltage rises fast because there is little capacity to fill | Load-test the battery independently of the solar setup |
| Panel Voc normal but almost no charging current | High-resistance connection, blown fuse, or controller in wrong battery-type mode | Check every connection under load; verify controller battery-type setting matches chemistry |
| Everything worked, then stopped after monsoon | Water ingress in connections/MCB box; corroded terminals | Inspect all outdoor connections; use weatherproof enclosures and cable glands |
| Inverter/load cuts out at night despite "full" battery | Voltage sag under load on an undersized or cold battery | Measure battery voltage under actual load, not open-circuit |
Maintenance checklist (the unglamorous part that makes it last)
- Clean the panel surface fortnightly in dusty areas (soft cloth, water — no abrasives).
- Check all outdoor connections quarterly for corrosion and tightness.
- Verify controller settings match the battery chemistry after any reset or replacement.
- For lead-acid: check electrolyte levels (flooded types) and equalise per manufacturer guidance; never leave discharged for weeks.
- Re-run the Voc/Isc tests yearly and log the trend.
- Trim vegetation and check for new shadows (buildings, growing trees) twice a year.
The short version
Audit the load in Wh/day, divide by (sun hours × 0.72 derate) for panel watts, size the battery from (Wh × autonomy days) / (voltage × depth of discharge), and pick a controller rated ≥ Isc × 1.25 — PWM for small matched systems, MPPT when panels are oversized in voltage or every watt matters. Round up to standard sizes at each step, verify the panel with Voc/Isc tests on arrival, and maintain it like the infrastructure it is. The Electrical branch hub collects related builds if you want to see these numbers inside real projects.