As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
A step is a bypass diode in conduction, which means a sub string of cells has dropped out and the rest of the module is carrying current past it. A rounded knee is series resistance somewhere between the cell metallisation and the test lead, taking a bite out of the fill factor while leaving both ends of the curve where they were. A low plateau is a current shortfall, and because short circuit current tracks irradiance almost linearly, that usually means less light reaching the cells than you assumed rather than a defect in the module. The three are separable by shape alone, which is the whole reason a sweep is worth taking.
The three features of a curve that carry diagnostic weight
An I-V sweep walks a module or a string from short circuit, where voltage is zero and current is at maximum, round to open circuit, where current is zero and voltage is at maximum. In between, a healthy trace holds a nearly flat current plateau, turns through a knee and drops to the voltage axis. Three departures from that shape account for most of what a field sweep ever shows: a step in the plateau, a knee that has gone soft, and a plateau sitting lower than the light says it should.
Each departure moves a different part of the trace, and that is what makes shape diagnostic. A fault that limits current lowers the plateau while barely touching the voltage end. A fault that adds resistance in series with the cells leaves both ends close to where they were and removes power from the middle. A sub string taken out of circuit removes a block of voltage at a current the rest of the module can still supply, so the trace drops in one visible stair. Reading them apart is mostly a matter of naming which axis lost something before naming the cause.
That is the argument for an instrument that records the whole locus rather than its two ends. The HT Instruments Solar I-Ve states curve tracing of modules and strings up to 1500 V and 40 A, measurement of open circuit voltage and short circuit current, front side and rear side irradiance for bifacial work, and an internal database of module performance to compare a trace against. The curve tracer roundup covers the rest of that instrument class.
A step in the curve points at a bypass diode in conduction
A crystalline module is built as two or three sub strings of cells, each with a bypass diode across it. Cells in series all carry the same current, so the weakest illuminated cell sets the current for its entire sub string. When one cell is shaded, soiled or cracked badly enough that it can no longer pass the current the rest of the string wants, the voltage across that cell reverses and the bypass diode turns on, carrying the current around the whole group. The module keeps working, minus the voltage that group used to contribute.
On the sweep this appears as a stair. Walking down from short circuit, the trace holds full current until the operating voltage rises far enough that the weak group can no longer be carried, at which point current drops to what that group can supply and a lower plateau resumes. The width of the voltage block that vanished tells you how much of the module left the circuit: one of three sub strings in a sixty cell module is roughly a third of its voltage.
Two steps in a string sweep almost always mean two separate modules each with a diode conducting, not one module with two faults. A step that moves between sweeps as cloud crosses the array is shading. A step that sits at the same voltage every time in still, even light is physical: a bird dropping, a cracked cell, a delaminated patch or a diode that has failed short and is now bypassing a perfectly good group.
A rounded knee points at series resistance somewhere in the path
Series resistance is everything between the cell junction and the instrument terminals: cell metallisation, solder bonds, ribbon interconnects, junction box terminals, connector contacts and the cable run. It does not change open circuit voltage, because no current flows at open circuit and so nothing is dropped across it. It barely changes short circuit current either. What it does is bend the region where current and voltage are both large at once, which is precisely the region that makes power.
Fill factor is the quantity that captures the damage. It is maximum power divided by the product of open circuit voltage and short circuit current, so it measures how square the curve is. Take a module whose sweep ends at 45 V and 10 A. The rectangle those ends define is 450 W. A healthy fill factor near 0.78 puts maximum power at 351 W. Let series resistance drag the fill factor to 0.68 and maximum power falls to 306 W, a loss of nearly 13 percent, with both end points still reading exactly as they did before.
So a rounded knee is worth chasing with a resistance measurement rather than another sweep. A four wire measurement sends current down one pair of leads and senses voltage on another, so lead resistance drops out of the result and a joint of a few milliohms becomes readable. The GOWENIC four wire milliohmmeter states the four wire method, six ranges and an anti interference design; its measurement range and resolution are not stated in the listing bullets, which is the first thing to confirm before trusting it on a connector.
A low plateau points at current, and current points at light
Short circuit current is close to linear in irradiance. Halve the light and the current roughly halves. Open circuit voltage behaves very differently, depending on the logarithm of the current, so the same halving costs only a few percent of voltage. Cell temperature is the dominant influence on the voltage end instead, in the region of 0.3 percent per kelvin for crystalline silicon. A plateau that sits low while the voltage end is close to expectation is therefore a light problem until proved otherwise.
Light problems are rarely the sky. Uniform soiling, a dirty band along the lower edge of the glass, pollen, an array tilt that is not what the design assumed, an irradiance sensor lying flat on a roof instead of in the plane of the modules, or a sweep taken while thin cloud drifted across all produce an honest low plateau on a module in perfect health. Measuring irradiance in the plane of the array at the instant of the sweep removes most of that argument, which is why tracers carry a reference cell or irradiance meter input.
A plateau that slopes downward instead of running flat is a different fault. The tilt is shunt conduction: a leakage path in parallel with the cells, from a crack, a damp junction box or a damaged backsheet, drawing a growing share of the current as voltage climbs. Series resistance and shunt resistance deform opposite ends of the same curve, and separating them is most of the value in looking at a trace rather than at a pair of numbers.
Reading the three shapes side by side
| What the trace does | Which quantity moved | Most likely cause | Confirming check |
|---|---|---|---|
| Single stair part way along the plateau | Voltage, in a discrete block | Bypass diode conducting around a shaded, soiled or damaged sub string | Clean or unshade and sweep again, then thermal scan for a warm diode |
| Knee rounded, both ends normal | Fill factor only | Series resistance in a connector, solder bond or cable run | Four wire resistance measurement at each joint along the path |
| Plateau flat but low | Current, in proportion | Less irradiance than assumed, uniform soiling, or a current limiting cell | Irradiance in the plane of the array at the moment of the sweep |
| Plateau sloping down toward the knee | Current, progressively with voltage | Shunt path from a crack, damp junction box or damaged backsheet | Insulation resistance test and a close visual on the junction box |
| Voltage end short by a whole module step | Open circuit voltage | A module missing, reversed or bypassed in the string | Count modules and compare against the designed string voltage |
The habit worth building is to name the axis before naming the fault, because the axis is observable and the fault is an inference. It also guards against the commonest error in reading any instrument, which is treating a number as though it carried its own conditions. The same trap appears when a gas detector reports parts per million without naming the gas it was calibrated against, and it appears here whenever a current is quoted without the irradiance that produced it.
What to do with a curve once you have read it
Record irradiance and module temperature with every sweep, because without them the trace cannot be translated to standard conditions and cannot be compared with anything. Keep the first good sweep of each string as a baseline, since the most reliable fault signal available later is the difference between a string and its own past shape. When a step appears, change the light and sweep again before touching hardware. When the knee softens, stop sweeping and start measuring resistance at joints.
One trap sits at the start of all of this. A high impedance meter placed across a disconnected string can show a voltage that is capacitively coupled rather than generated, which sends people hunting a module that is fine. A low impedance mode loads the circuit just enough to collapse that phantom reading, which is the problem low impedance multimeters exist to settle. Confirm the voltage end honestly first, then take the sweep, then read the shape.
Frequently Asked Questions
Does a step in the curve always mean a bypass diode?
Almost always, because a step requires a block of voltage to leave the circuit at a current the rest of the module can still carry, and that is what a conducting bypass diode does. What varies is the reason the diode turned on. Shade, soiling, a cracked cell, a delaminated region and a diode that has failed short all produce the same stair. The way to separate them is to change the light and sweep again: shading moves or disappears, physical damage does not.
Can a rounded knee be caused by the test leads rather than the array?
Yes, and it is a common false alarm. Series resistance does not care whether it lives in a cell interconnect or in a tired crocodile clip, so a corroded lead, a long thin extension or a connector that is not fully latched will round the knee of a perfectly healthy module. Repeat the sweep with the instrument moved to the other end of the run before condemning anything, and check that the tracer is using separate sense connections where it offers them.
How much irradiance do I need before a sweep is worth comparing to a datasheet?
Commissioning practice normally asks for at least 700 W/m2 in the plane of the array, measured at the moment of the sweep, along with stable conditions rather than broken cloud. Below that level the translation back to standard test conditions carries more uncertainty than the fault you are hunting, and the comparison stops being evidence. Low light sweeps are still useful for shape, which is what this article is about, but not for pass or fail against a nameplate.
Why does the inverter screen not show any of this?
An inverter reports the operating point it has settled on, which is one coordinate on the curve rather than the curve itself. It can tell you that a string is producing less than its neighbour, which is a useful prompt, but it cannot show a step, a slope or a soft knee because it never visits the rest of the trace. A tracer deliberately walks the whole locus in a few tens of milliseconds to recover the shape the inverter throws away.
Does a string sweep identify which module is at fault?
It narrows the search rather than finishing it. A string trace tells you how many sub strings have dropped out and roughly how much voltage went with them, which sets the number of modules to inspect. Finding the specific one normally means a thermal scan looking for the warm cell or warm diode, or disconnecting the string into sections and sweeping each section until the step follows one group.
Is a low plateau ever a module fault rather than a light problem?
It can be. Current mismatch inside the module, from cracked cells that no longer contribute their full share, will cap the current the whole series group can pass regardless of how bright it is. So will a partially shorted cell. The distinguishing move is to measure irradiance in the plane of the array at the moment of the sweep. If the light is where it should be and the current is not, the module is the suspect.