Solar Radiation Sensor Installation, Cleaning, Calibration and Abnormal Data Checks

Quick Answer: Reliable solar radiation sensor data starts with the correct measurement plane, an unobstructed view and a clean optical surface. When readings become abnormal, check raw output, wiring, scaling and timestamps before changing calibration. For a thermopile pyranometer, a small nighttime offset can occur; a persistent reading of hundreds of W/m2 in darkness calls for investigation, not an assumed solar signal.

Install for the radiation quantity you intend to measure

A horizontally mounted pyranometer measures global horizontal irradiance, or GHI. A sensor parallel to a photovoltaic array measures irradiance in that plane, commonly called plane-of-array irradiance. These are different measurements. Two functioning instruments can disagree because one is horizontal and the other follows a tilted array. Record the orientation with the data instead of trying to remove that difference by adjusting gain.

Choose a position with a clear view of the sky and direct sunlight through the day. Inspect the likely shadow paths of masts, antenna brackets, fences and nearby buildings. A location that is clear at noon may be shaded every morning. Repeat the exposure review when structures change or vegetation grows; a single installation photograph cannot establish year-round freedom from shading.

Level, orientation and cable restraint

For horizontal measurement, adjust the leveling feet until the bubble is centered, tighten the fixings and check the bubble again. Tightening can move the instrument on an uneven mounting plate. For an intentionally tilted measurement, set the sensing plane to the required tilt and azimuth; do not level it horizontally and still label the result as array-plane irradiance.

The NiuBoL NBL-W-HPRS and NBL-W-HPRS-A installation arrangements call for the cable terminal to face north. Keep this mechanical orientation distinct from the measurement plane: it is not a reason to tilt the receiving surface toward north. Route the cable below the sensing surface, secure it against movement and leave enough strain relief that wind or maintenance work cannot pull the sensor out of alignment.

Check the instrument version before testing its output

A passive thermopile, an amplified voltage output, a current transmitter and an RS485 instrument require different tests. Do not apply a generic 24 V supply to passive signal wires. Equally, a powered transmitter with its supply disconnected cannot be assessed as though it were a passive millivolt sensor.

Response times quoted at 95% and 99% are not directly interchangeable. Nonlinearity is one error component, not a complete accuracy statement. For measurement uncertainty, use the instrument's calibration information and include installation, cleanliness and acquisition errors. Specify cable length, jacket material, power consumption and environmental protection for the selected assembly rather than borrowing these values from another output version.

Clean the dome without creating a new measurement problem

Dust, salt deposits, pollen and dried water marks alter the light reaching the thermopile. Inspect the outer dome and bubble level together. Establish the inspection frequency from the actual fouling rate: a dusty construction site or coastal location may need attention much sooner than a clean, accessible reference site. A fixed monthly interval is not a reliable rule for every installation.

1. Record the time, visible contamination and current reading before cleaning. Flag the maintenance interval in the data record.
2. Remove loose material gently. Use a clean, soft cloth dampened with water; for the NBL-W-HPRS-A, alcohol is also permitted for cleaning the glass. Avoid abrasive rubbing and sharp tools.
3. Do not loosen the glass dome to clean its underside. Moisture inside the optical enclosure requires service assessment, not wiping through an opened seal.
4. Inspect the cable entry, connector and bracket, then recheck the level or intended tilt. Cleaning pressure can move an inadequately secured mount.
5. Record the post-cleaning condition. Compare readings only while allowing for changing sunlight; a passing cloud can obscure the effect of removing dirt.

The dome and leveling arrangement are part of the measurement system, not cosmetic fittings. Internal condensation in an NBL-W-HPRS-A calls for return-for-service assessment. On instruments with a serviceable desiccant arrangement, use the correct service procedure for that design; do not assume every sealed pyranometer has a user-replaceable dryer.

Separate the sensor signal from the displayed value

Raw millivolt output

For a passive thermopile, irradiance equals output voltage divided by the individual sensitivity. Keep the units consistent. If the calibration coefficient is 10 uV/(W/m2), an output of 8 mV is 8000 uV, corresponding to 800 W/m2. This is a worked example, not a coefficient to enter into every instrument.

Record the serial number and its coefficient together. Replacing a probe without updating that coefficient can produce plausible-looking but systematically wrong readings. Choose a logger input that resolves millivolts adequately and accepts the expected small negative offset. A general-purpose 0-10 V input may not provide sufficient usable resolution for a raw thermopile.

Current and amplified voltage outputs

For an NBL-W-HPRS ordered with 4-20 mA corresponding to 0-2000 W/m2, calculate irradiance as (current in mA - 4) / 16 x 2000. Thus, 12 mA corresponds to 1000 W/m2. A 0-5 V version over the same span gives 1000 W/m2 at 2.5 V. Use these relationships only for those specified spans; do not apply them automatically to the HPRS-A or a customized transmitter.

Verify the receiving channel independently with a suitable signal source, after disconnecting and isolating the sensor according to the electrical procedure. Inject into the logger input, not into the thermopile. If the simulated value is already wrong on the display, investigate the acquisition configuration before sending the sensor for calibration.

RS485 Modbus RTU

For the HPRS-A digital option, the serial settings are 9600 bps, eight data bits, no parity and one stop bit, with function 03 used for reading. Match the configured device address and the applicable register definition. The base HPRS has its own interface details; do not copy an HPRS-A wiring color or driver into it without checking the output version.

Compare the returned register value with the local logger and remote platform at the same timestamp. An online device indicator only establishes that some communication is occurring. It does not prove that the displayed radiation value is newly sampled or correctly scaled. Diagnose address collisions, stale values and unit conversion separately.

Solar radiation sensor fault patterns and the next test

Do not force all nighttime samples to zero before investigating an offset. Thermopile thermal exchange can produce small negative values, and filtering them away can hide a developing change. Preserve the unmodified values for diagnosis; apply any operational data-quality rule in a separate processing step with its meaning documented.

Field comparison is useful, but it is not automatic recalibration

A side-by-side comparison can reveal a large discrepancy when both instruments have the same plane, clear exposure, clean domes and aligned timestamps. Compare stable periods across several irradiance levels and consider each instrument's response time. A fast reference and a slower sensor can differ during moving clouds without either having a steady-state gain fault.

Do not derive a new calibration coefficient from one sunny reading or from a nearby weather website. Recalibration requires an appropriate reference and a controlled comparison procedure. For HPRS-A service planning, use the recommended two-year interval as a starting point, with earlier assessment after damage, internal moisture or persistent unexplained disagreement. Project-specific quality requirements may call for a different schedule.

When requesting support from NiuBoL, send the model, serial number, output version, coefficient where applicable, mounting photographs, cleaning history and a short synchronized series of raw and displayed values. Include the logger configuration and when the problem first appeared. This makes it possible to distinguish a sensor service requirement from an input-range or installation correction.

Project Decision FAQ

Q1. Must a solar radiation sensor always be horizontal?

No. Horizontal mounting is used for GHI, while a tilted sensor measures radiation incident on its own plane. Match the plane to the required quantity and store the orientation with the record. Leveling a sensor intended to follow a tilted array changes what it measures.

Q2. Is rain enough to keep the dome clean?

No. Rain may leave mineral marks or redistribute dirt, and dried deposits can remain on the glass. Inspect the dome after relevant weather or site events and clean according to its condition. Do not use rainfall as evidence that the optical surface is clean.

Q3. Can a constant correction remove a shadow from a mast?

No. The obstruction changes with solar position and the mix of direct and diffuse radiation. Move the obstruction or sensor where practical, and flag affected periods. A constant multiplier cannot reconstruct the missing irradiance reliably.

Q4. Can I reuse the coefficient from a replacement sensor of the same model?

No. Raw thermopile sensitivity is individual to the instrument. Use the replacement's coefficient and verify the logger calculation. For an amplified or digital output, establish whether conversion occurs inside the sensor so that sensitivity is not applied a second time.

Q5. Does a small negative nighttime reading mean the sensor is defective?

Not necessarily. Thermopile zero offset can produce small negative values. Assess the size and trend, installation conditions and input configuration. A sudden or large change needs investigation; deleting all negative values is not a diagnostic test.

Q6. Should I recalibrate before checking wiring?

No. First establish that the raw signal, power where required, wiring and acquisition scaling are correct. Calibration cannot repair a loose connector, an incorrect current span or a stale platform value. Request recalibration when the measurement chain is functioning but an unexplained sensitivity discrepancy remains.

Summary

Solar radiation sensor maintenance combines optical inspection, alignment checks and electrical diagnosis. Keep the dome clean, distinguish horizontal from array-plane irradiance, and trace an abnormal reading back to its raw signal. Correct installation and acquisition errors first; use calibration to address sensitivity, not to compensate for shade, dirt or incorrect units.