A photovoltaic cell measured under a light field that is 6 % brighter in the center than at the edges returns an efficiency number that depends on where the cell sat, not on what the cell does. For a lab certifying module performance, that spatial variation is a measurement error built into the light source. IEC 60904-9 Edition 3 puts a hard limit on it: a Class A solar simulator must hold spatial non-uniformity of irradiance below 2 % across the test area, evaluated on a grid of at least 25 zones with each detector no larger than 1/5 of the smallest test-area dimension.
The design problem is to build a source, usually an LED array today, whose irradiance varies by less than 2 % across a defined plane at a defined working distance, while holding the beam within the angular range the test method allows. TracePro addresses this by modeling the LED array as real sources, tracing the flux through the collimating and homogenizing optics with Monte Carlo ray tracing, and computing the irradiance map on the test plane so the spatial non-uniformity is a number checked against the Class A limit. This article covers solar simulator design for uniformity in TracePro, from source layout to the 25-zone uniformity calculation.
IEC 60904-9 rates a simulator on three separate axes, and a solar simulator design has to treat them independently because they are governed by different parts of the hardware. Spatial non-uniformity is the axis the optical layout controls. The standard defines it from the maximum and minimum irradiance measured across the test plane, computed as the difference over the sum of those extremes. Class A requires that figure below 2 %, Class B below 5 %, Class C below 10 %. The measurement uses a grid with a detector area no larger than 1/5 of the smallest test-area dimension, which for a square test area means a 5 by 5 grid of 25 zones at minimum. The design target is therefore not a smooth-looking beam but a measured spread across those zones that stays under the 2 % line.
Spectral match and temporal stability are the other two axes. Spectral match compares the source output against the reference AM1.5G spectrum in defined wavelength bands, and it is set by the choice and mix of LED wavelengths rather than by the imaging optics. Temporal instability is a drive-electronics and thermal problem. Because the optical geometry drives spatial uniformity and has little effect on the other two, the TracePro model focuses on the spatial axis, and the simulator earns its spatial-uniformity class from the irradiance map the trace produces.
A solar simulator design that reaches Class A starts with an accurate model of the LED array, because the array layout is the first thing that sets uniformity. TracePro represents each LED as a source with its real position, radiant flux, and angular distribution, so the combined field on the test plane reflects the actual array rather than an averaged sheet of light.
Modern LED simulators use several wavelengths of LED interleaved across the array to build the AM1.5G spectrum, and each wavelength group can carry its own flux and angular profile. Importing measured ray files for each LED type reproduces the true emission, while a surface source with an assigned angular distribution serves when ray files are not available. The array pitch and the arrangement of the different-wavelength emitters both affect uniformity, so laying them out at their real board coordinates lets the trace show whether the raw array, before any homogenizing optics, is close or far from the target. A bare array almost never meets 2 % on its own. The near field directly above the board shows the discrete LED footprints, and the field falls off toward the array edges where fewer emitters contribute. The value of modeling the raw array first is that it quantifies how much correction the downstream optics have to supply, which sizes the collimator and the homogenizer for the job.
The test plane usually needs 1 sun, near 1000 watts per square meter integrated over the AM1.5G band. Setting each LED to its operating flux lets the trace report irradiance in real units at the test plane, so the model confirms both the uniformity and the absolute level in a single run. If the level falls short, the trace shows whether the shortfall is array output or optical loss, which points to the fix.
A solar simulator does not only need uniform irradiance. It needs the light arriving at the test plane within a bounded angular range, because a module tested under highly divergent light behaves differently from one under the near-collimated sun. Collimation and uniformity pull against each other, and a solar simulator design has to balance them in the same optical train.
TracePro traces the collimating optics, whether a lens, a reflector, or a combination, against the real array so the angular distribution at the test plane is a measured output. Tightening collimation narrows the cone of angles but tends to sharpen the LED footprints, which hurts uniformity. Relaxing it smooths uniformity but widens the angular spread. Tracing a family of collimator designs shows where the two requirements can both be met, and it does so with the real source divergence rather than an assumption that every LED emits identically. The working distance is part of this balance. The uniformity and the angular spread both change with distance from the last optic, so the model evaluates the test plane at the intended working distance, not at an arbitrary plane. A design that passes at 200 millimeters may fail at 150, and the trace makes that sensitivity visible before the mechanical layout is fixed.
Getting from a raw array at 8 or 10 % non-uniformity down under 2 % is the job of the homogenizing optics, and this is where a solar simulator design is won or lost. TracePro traces each homogenizing approach and reports the resulting non-uniformity, so the choice rests on the simulated number. A diffuser is the most direct tool. Applying a bidirectional scattering distribution function to a diffuser surface spreads each ray into a controlled angular cone, blending the discrete LED footprints into a continuous field. A stronger diffuser mixes better but widens the angular spread and costs some flux, so it trades against the collimation requirement. A microlens or lenslet array is the second approach, subdividing and overlapping the beam so that many small images of the array superimpose on the test plane, which evens the field while preserving more directionality than a heavy diffuser. A mixing chamber with diffusely reflecting walls is a third option for compact systems. Because each approach affects uniformity, angular spread, and throughput differently, the model compares them on all three at once. A design that reaches 1.5 % non-uniformity, holds the angular spread within the test method limit, and delivers 1 sun at the working distance is ready to prototype as a Class A spatial-uniformity simulator.
The uniformity class is decided on the test plane, and TracePro computes it the way the standard does. An irradiance map on the test plane, divided into the 25-zone grid the method requires, gives the maximum and minimum zone irradiance, and the non-uniformity follows directly from those two values. The result is the same figure a certification lab would measure with a mapped detector, produced from the trace before any hardware exists.
Reading the map also localizes the residual. A center-bright pattern points to an array that is denser or a collimator that concentrates on axis, and it calls for a stronger edge contribution or a flatter collimator. A corner-dark pattern points to array edge roll-off that a larger array or a wider homogenizer would fix. Because the map is computed at the real working distance with the real optical train, the diagnosis is specific to the design under test.
Uniformity depends on how large a test area the simulator has to cover. A design that holds 2 % across a 156 millimeter cell may exceed it across a full 2 meter module. Tracing the irradiance map at the largest intended test area sizes the array and optics for the real product, and it shows the tradeoff between test-area size and achievable class, so the specification is set on data rather than on hope.
Class A solar simulator design is a uniformity problem bounded by an angular requirement, and it is decided on a 25-zone irradiance map at the test plane. TracePro brings the whole train into one trace: real LED array sources, the collimating optics that set the angular spread, and the diffuser or lenslet homogenizer that drives the last few % of uniformity. Computing spatial non-uniformity the way IEC 60904-9 defines it turns the design from a build-and-map cycle into a set of optical choices verified against the 2 % line before the first array is populated.
Request a TracePro trial and model your solar simulator to a Class A uniformity target, or contact Lambda Research to request a demo focused on irradiance uniformity for PV test systems.