PV string commissioning that holds up at turnover
Commissioning a PV array is repetitive work with little margin for error. Every source circuit gets its polarity checked, its open-circuit voltage compared with what it should be, its current checked against irradiance, and its insulation tested before it's closed into a combiner or inverter. Do it for 400 strings on a hot afternoon with clouds moving through and the paperwork is where things fall apart. The free workbook above keeps it in four connected tabs: String Map → String Test → Inverter Startup → Racking Torque Log.
1. The string map comes first
Before anyone lifts a meter, map every source circuit: string ID, inverter, MPPT input, combiner and input number (or "Direct" for string inverters), modules per string, module model and wattage, block/row/table, and home-run cable ID. The String Test tab looks up modules per string from this map by string ID, so the expected voltage uses the actual count on the drawings, not an assumed one. The summary totals strings, modules, and kWdc and shows the min/max modules per string. Mixed string lengths on one MPPT are a design question to settle before energizing.
2. Expected Voc: correct for temperature or chase ghosts
Module Voc is rated at STC: 25 °C cell temperature. Real module temperatures on a sunny day run 40–60 °C, and voltage drops as temperature rises. The correction uses the datasheet coefficient:
- Expected Voc = modules × module Voc(STC) × (1 + β/100 × (Tmodule − 25))
- % deviation = (measured − expected) ÷ expected, checked against your tolerance
With β = −0.27 %/°C and a 45 °C module, a 26-module string of 49.6 V modules should read about 1,220 V, not the 1,290 V nameplate sum. Measure module temperature on the back of a module in the string's row. The template lets any row override the global temperature if conditions shift through the day.
What the deviations tell you: a string low by roughly one module's Voc (≈ 4%) usually has a missing jumper, a shorted module, or a failed bypass diode. Strings reading high are often a count error on the map. A string that sits far from its neighbors in the same conditions is worth an IV curve before it's energized.
3. Isc and irradiance
Short-circuit current tracks irradiance almost one-for-one, so the template corrects expected Isc by G/1000 and the small positive Isc temperature coefficient. Record plane-of-array irradiance from a reference cell at the moment you read Isc. When a cloud edge passes, record the new irradiance on that row rather than failing the string. If many strings read low together, suspect irradiance or soiling first.
4. Polarity, insulation resistance, and IV curves
Polarity gets checked before any string is landed in a combiner or inverter. A reversed string forces an overall FAIL. Insulation resistance is recorded positive-to-ground and negative-to-ground at the test voltage your spec or IEC 62446-1 sets for the system voltage; the minimum is an input. Record the IV-curve file name so each string's curve can be pulled up later. Steps in the curve point to shading or bypass diodes, and a low fill factor points to series resistance.
5. The cold-weather voltage check
The String Test tab also runs the design-side check the AHJ will ask about: maximum string Voc at the site's record low temperature, using the manufacturer temperature-coefficient method in NEC 690.7(A) (verify against the edition your AHJ adopted), compared with the lowest maximum-voltage rating among the inverter, combiners, disconnects, and cable. Commissioning is the last chance to catch a string built one module too long. See our PV voltage temperature correction reference for the method and the NEC table alternative.
6. Inverter startup and racking torque
The Inverter Startup tab walks pre-energization terminations and labeling, DC and AC verification, startup and functional checks (grid profile per the interconnection agreement, anti-islanding, rapid shutdown where required), communications, and documentation. The status reads COMPLETE only with no fails and no blanks. The inverter manufacturer's startup procedure governs where it differs.
The Racking Torque Log records specified torque from the racking or tracker manual, the verified reading, torque stripe, and the wrench used with its calibration due date. It flags UNDER and OVER against your acceptance band. A cluster of UNDER readings on one crew or one wrench means expanding the sample, not re-torquing one bolt.
From spreadsheet to commissioning record
A workbook works for a rooftop or a single inverter block. On a utility-scale site the hard part is keeping hundreds of strings, dozens of inverters, and several crews on one current record. Field PM's QA/QC module puts checklists, hold points, test records, and photos on the crew's phone, and QA/QC workflows sequence sign-offs so an inverter can't be marked ready before its strings are. Start a free 30-day trial. No credit card required.
Related free tools and templates
- PV voltage temperature correction: NEC 690.7 method explained
- Insulation resistance test report: megger records for feeders and equipment
- Insulation resistance testing (IEEE 43) explained
- Voltage drop calculator: check DC home runs and AC feeders
- Ground resistance test report
- Browse all free construction templates