7.1 Reverse Currents, Bypass Diodes and PID
Key Takeaways
- Handbook v1.3 Learning Outcome 3.1.4(b) states that modules are designed to withstand only small reverse currents, so series or string overcurrent protection might be required when parallel strings can back-feed a weak string.
- Reverse current is the feed from healthy parallel strings into a shaded, shorted or otherwise faulted string; a later ratings chapter applies 1.35 × IMOD_MAX_OCPR < (NS − 1) × ISC_MAX, and this section is the reason that test exists.
- Bypass diodes sit in the module junction box, typically one per cell-string substring, so a shaded substring is taken out of the current path and the rest of the module and string can still produce.
- Potential induced degradation reduces power output under high system voltage, humidity and temperature; some PV constructions are more prone, and the loss may be recoverable or permanent.
- A single shaded module in a ten-module string is a hot-spot problem if its diodes have failed open; adding parallel neighbour strings turns the same fault into a reverse-current problem as well.
Quick Answer: Handbook v1.3 Learning Outcome 3.1.4(b) and (c) requires four linked ideas. Modules are designed to withstand small reverse currents. Series or string overcurrent protection might be required when parallel strings can back-feed a weak string. Bypass diodes mitigate damage to and performance of panels caused by shading by taking a starved substring out of the current path. Potential induced degradation (PID) reduces power output, and some types of PV are more prone to that degradation. This OpenExamPrep section is independent study material for those 2922-34 points; it is not a City & Guilds publication.
Why a parallel string can be driven backwards
A single series string has one current, set by light on the cells and by the inverter's maximum power point tracker. Nothing behind that string is pushing extra current into it, because there is no parallel neighbour.
The moment you parallel two or more strings onto a bus — at a combiner box, at paralleled inverter input terminals, or inside a central sub-array — those strings share a voltage. Each healthy illuminated string is a current source feeding the inverter. If one string is shaded, shorted, or otherwise unable to produce, it can stop looking like a generator and start looking like a load. The remaining (NS − 1) strings can then force current backwards through the weak string. That is reverse current.
The current is still a photovoltaic current, not a kiloampere mains fault. The safety chapter already taught that short-circuit current (Isc) is only a little above operating current (Imp). Reverse current from two neighbouring strings is still measured in tens of amperes. That is enough to overheat cells, ribbons and junction boxes in the victim string. It is not enough to make an ordinary AC miniature circuit-breaker behave as if a transformer had bolted. Do not wait for a consumer-unit device to clear a reverse-fed string.
Modules withstand only a small reverse current — that is why string OCPD exists
Handbook 3.1.4(b) is two linked sentences. First: modules are designed to withstand small reverse currents. The datasheet name for that withstand is commonly the maximum reverse current, maximum series fuse rating, or IMOD_MAX_OCPR (module maximum overcurrent protection rating). Typical crystalline products sit in a band such as 15 A, 20 A or 25 A. Treat the number on that datasheet as the design value, not a handbook pass mark.
Second: series/string overcurrent protection might be required. Might is the exam word. A one-string array has NS = 1, so there are no parallel neighbours. Reverse feed from other strings is zero. You still isolate and you still rate cables, but you do not fit string fuses just to satisfy a parallel-feed rule that does not apply.
A two-string domestic roof often still sits inside the module's reverse-current withstand: one neighbour can push about one Isc into the victim, and Isc is commonly below a 20 A fuse rating. Three or more strings in parallel is where reverse current grows as (NS − 1) × Isc and the module may not be designed to take it.
A later chapter (Learning Outcome 4.2) teaches the numerical test used in UK small-system design: string overcurrent protection is required where 1.35 × IMOD_MAX_OCPR is less than (NS − 1) × ISC_MAX. Learn the why here; do the arithmetic there. Do not turn this section into a second ratings chapter.
- (NS − 1) × ISC_MAX is the current the remaining healthy strings can push into the victim, taking each remaining string at its maximum short-circuit current. The later chapter applies the UK current multiplier to obtain ISC_MAX.
- IMOD_MAX_OCPR is how much reverse current the module is designed to survive.
- 1.35 is a coordination factor. It leaves margin so that a gPV fuse or a listed DC breaker actually operates before the module's reverse-current withstand is exceeded, rather than hoping the cells will survive at the exact rated ampere.
If the inequality is true, available reverse current (checked with that 1.35 factor) outgrows the module. Then you shall fit string overcurrent protective devices (OCPD) — typically gPV fuses to BS EN 60269-6 or a DC breaker listed for the job. If the inequality is false, the module's built-in reverse-current withstand is accepted as enough and string fuses are not required by that test. That is why the handbook says might be required, not always.
Do not fit AC MCBs as string fuses. Do not invent IMOD_MAX_OCPR from how many strings you can see on the roof. A microinverter architecture is not a parallel-string reverse-current problem in the 3.1.4(b) sense: each module's DC is converted locally, so there is no parallel DC bus of strings sharing a victim.
Bypass diodes: shading damage versus shading loss
Handbook 3.1.4(c) first half: bypass diodes mitigate damage to and performance of panels caused by shading. Section 5.3 already showed that a series cell that is dark becomes reverse-biased and can hot-spot. The diode is the local relief valve.
Where they live. Almost every crystalline module puts the diodes in the rear junction box. They are not in the inverter, not in the connector, and not a sticker on the glass.
How many. Typically one diode per cell-string substring. A conventional 60-cell full-cell laminate is often three substrings of twenty cells and three diodes. 72-cell modules often follow the same three-substring idea with longer groups. Half-cut products may use three or six diodes depending on the internal parallel layout. Count that product's junction-box diagram. There is no handbook pass mark that says always three.
What they do. When a substring is shaded — a chimney bar, a leaf, a soiled patch, a bird — that substring's photocurrent collapses. The other substrings and the other modules in the string still want to push current. The diode across the starved substring forward-biases (about 0.7 V for a silicon diode, a typical drop, not a handbook constant) and bypasses those cells. The shaded substring is taken out of the current path. The remaining substrings in that module, and the other modules in the string, keep producing. Yield falls by about that substring, not by the whole string.
What they do not do. They do not add extra current. They do not make shade free. They do not replace layout, later MCS shade evaluation, or module-level electronics when a roof is chopped to pieces. They prevent the worst electrical and thermal damage and they preserve the rest of the generator.
Failed diode, open. The substring cannot be bypassed. A hard shadow on those cells limits string current and reverse-biases the dark cells. Hot-spot risk is now real: cracked cells, browned encapsulant, melted backsheet, cracked glass, and in the worst case a fire initiation on the roof.
Failed diode, short. The substring is always bypassed. Module voltage is permanently low by about that substring. The string may fall out of the inverter MPPT window. Thermography shows a cold substring; the box may still look tidy.
PID: leakage that steals watts
Handbook 3.1.4(c) second half: potential induced degradation reduces power output; some types of PV are more prone to degradation.
PID is not soiling and not a blown diode. It is leakage current driven by system voltage between the cells and the earthed frame (or other earthed metalwork), through glass, encapsulant and edges. High voltage, humidity (moisture in the encapsulant or on the glass), and high temperature accelerate it. A UK maritime winter is damp; a still summer roof is hot; a long string on a transformerless inverter can sit at several hundred volts relative to earth. Those three stack.
Polarity, qualitative only. On many p-type crystalline modules, PID is worse when the cells sit at negative potential relative to the grounded frame — historically a negative-ground array, or a transformerless inverter that holds the array negative. Positive grounding was used on some older plants as a mitigation. A floating array (neither pole solidly earthed, common on transformerless domestic inverters) does not magically delete PID: the inverter's switching and leakage paths still set a voltage to earth. Name polarity, negative-ground, and floating as factors. Do not invent a percentage loss as a 2922 pass mark.
Recoverable versus permanent. Some PID is a polarization effect. Power falls on the I–V curve (lower fill factor and voltage) but can recover in the dark, or with a manufacturer PID-recovery overnight bias. Other PID is electrochemical: sodium-ion movement, corrosion, junction damage. That share is permanent. Say often recoverable versus permanent, not that every PID event comes back on Saturday.
Which products. The handbook's phrase is enough for the paper: some types are more prone. Industry observation, not a handbook table: older p-type cells (including some PERC), some thin-film constructions, and modules with moisture-hungry encapsulant or non-PID-resistant glass have shown more PID; many n-type, heterojunction, and PID-tested laminates are more resistant. Quote the datasheet PID test (IEC TS 62804 family) on a real specification; do not rank brands from memory in the exam.
Symptom versus cause versus exam action
| Symptom on site or on the paper | Likely cause in this section | Exam / site action |
|---|---|---|
| Victim string hot, cables warm, one string reverse-fed while neighbours produce | Reverse current from (NS − 1) parallel strings into a shaded, shorted or mismatched string | Explain why string OCPD might be required; apply the 1.35 × IMOD_MAX_OCPR test in the later ratings chapter; do not use an AC MCB as a string fuse |
| One substring cold or one module underperforming under a chimney bar; rest of string still carrying current | Bypass diode conducting around the shaded substring | Treat as designed behaviour; fix the shade if yield matters; the diodes have done their job |
| Local brown cell, cracked glass, melted backsheet on a shaded module; string current collapsed | Bypass diode open (or missing); reverse-biased cells hot-spotting | Module is damaged; isolate; replace; never keep loading a hot-spot |
| Whole string voltage about a third low with no shade | Bypass diode shorted | Replace the module or junction box per the manufacturer; check the MPPT window |
| Slow loss of power, worse after damp heat, no single shaded cell | PID under high voltage, humidity and temperature | Name PID; note polarity / negative-ground / floating; some types more prone; recovery may or may not be possible |
| Single-string array, no parallel neighbours | Reverse-feed from other strings is zero | String fuses not required by the parallel-feed test; still rate and isolate the DC circuit |
Scenario: one shaded module in a ten-module string
A south roof has ten 400 W-class modules in one series string into a garage inverter. A chimney casts a hard bar across one module at midday. The other nine are in full sun.
Diodes healthy. The affected substring (or substrings) inside the shaded module is bypassed. String current continues. String voltage drops by the bypassed substring voltage, not by nine modules. The inverter still sees a generator. Yield is reduced; the roof should not be on fire. That is 3.1.4(c) working as intended.
Diodes in that module failed open. The shaded cells cannot be bypassed. They limit the whole string current. The nine bright modules try to drive operating current through reverse-biased dark cells. Those cells dissipate power as heat. Hot-spot risk is now on a domestic roof, on a ten-module string, from one shadow and one failed box. Isolation, replacement, and a method statement belong next, not it is only one panel.
If that ten-module string were later paralleled with two more identical strings at a combiner (NS = 3), a faulted or heavily shaded string could also be reverse-fed by the other two. That is the 3.1.4(b) reverse-current problem, and it is why the later chapter may demand string OCPD. One shaded module and a failed diode is a hot-spot story. Parallel neighbours are a reverse-current story. The paper can ask both in one stem. Keep them in separate rows of the table.
Why might series or string overcurrent protection be required on a 2922-34 array even though modules are designed to withstand small reverse currents?
A chimney shades one substring of one module in a ten-module series string. The bypass diodes are healthy. What should a 2922-34 candidate expect?
Which statement matches handbook v1.3 on potential induced degradation?