The Core Decision: Series vs. Parallel Battery Topologies
When designing custom power banks, the fundamental choice between wiring batteries in series and parallel circuits dictates your pack's voltage, capacity, and physical architecture. The rule of thumb is absolute: series wiring increases voltage; parallel wiring increases capacity (amp-hours) and current delivery.
To understand why you must combine both in most practical builds, look at the physics of power delivery. Power (Watts) equals Voltage times Current. If you need to run a 600W inverter from a 12V nominal pack, you are pulling 50A of continuous current. That requires massive 4 AWG wire and generates severe I²R (heat) losses. By wiring cells in series to create a 48V nominal pack, that same 600W load only draws 12.5A, allowing you to use thinner 12 AWG wire and vastly improving efficiency.
However, a single series string of 18650 cells only gives you the capacity of one cell (usually 2.5Ah to 3.5Ah). To achieve usable runtimes, we wire cells in parallel before wiring them in series, creating a topology like 3S2P (3 series groups, 2 parallel cells per group). This yields a 11.1V nominal pack with double the amp-hour rating and double the maximum continuous discharge current.
Failure Mode Contrast: What Breaks at the Extremes?
The most critical difference between series and parallel configurations is how they handle catastrophic cell failure. Understanding these extremes is non-negotiable before you spot-weld a single busbar.
Series Circuit Extremes
- Open Circuit (Broken Link): If a series interconnect breaks or a single cell fails open, the entire pack dies. Voltage at the main terminals drops to 0V because the circuit path is broken. It is a nuisance failure, but generally safe.
- Short Circuit (Internal Cell Short): If one cell internally shorts, it effectively becomes a piece of wire. The pack remains functional, but the total voltage drops by one cell's nominal voltage (e.g., 3.6V). The remaining cells now push the same current through a lower total voltage, accelerating their degradation and risking over-discharge if the BMS low-voltage cutoff isn't calibrated for the missing cell.
Parallel Circuit Extremes
- Open Circuit (Disconnected Cell): If one cell in a parallel group disconnects, the pack still functions. However, the capacity of that specific parallel group drops, and the remaining cell must now carry 100% of the load current. If the load exceeds the single cell's maximum discharge rating, it will overheat and vent.
- Short Circuit (Internal Cell Short): This is the critical hazard. If one cell in a parallel group shorts internally, its voltage drops to near 0V. The other healthy cells in that parallel group will immediately dump their maximum short-circuit current (often 20A to 30A per cell) into the dead cell in a desperate attempt to equalize voltage. This causes rapid thermal runaway. This is why high-reliability parallel packs use individual cell fuses or rely on precisely sized nickel strips that act as sacrificial fuses to isolate a shorted cell before the parallel neighbors can ignite it.
Behavior Matrix: How Circuit Parameters Shift
Use this reference table to predict how your pack will behave when you alter the topology or when a cell degrades. The values below assume a baseline of standard 3.6V Li-ion NMC cells (like the Samsung 30Q).
| Circuit Change / Event | Effect on Pack Voltage | Effect on Pack Capacity (Ah) | Effect on Max Discharge Current |
|---|---|---|---|
| Add 1 cell in Series | Increases by 3.6V (nominal) | No change | No change (limited by weakest cell) |
| Add 1 cell in Parallel | No change | Increases by cell Ah rating | Increases by cell's max C-rating |
| One cell drops 0.5V (imbalance) | Pack voltage drops slightly under load | Usable capacity drops (BMS cuts off early) | No immediate change, but heat increases |
| One parallel cell fails open | No change | Drops by that cell's Ah rating | Drops by that cell's max current rating |
Design Walkthrough: Building a 12V 6Ah 3S2P Li-ion Bank
Let's move from theory to the bench. We are designing a 12V nominal (11.1V actual) 6Ah battery pack to run a portable 12V Wi-Fi router and LED lighting system during grid outages.
Component Selection:
- Cells: 6x Samsung INR18650-30Q (3.6V nominal, 3000mAh, 15A max continuous discharge). Total raw capacity: 6Ah. Cost: ~$35.
- BMS: Daly 3S 15A Smart BMS with UART. The 15A rating covers our 100W maximum load (8.3A at 12V) with a safe margin. Cost: ~$18.
- Interconnects: 0.15mm thick, 8mm wide pure nickel strips. At this thickness, the strip will carry 10A continuously without overheating, but will melt and isolate a cell if a short-circuit event pushes current past 40A.
Topology Mapping (3S2P):
- PG1 (Parallel Group 1): Wire Cell 1 and Cell 2 in parallel. The shared negative terminals form the Pack B- node. The shared positive terminals form the PG1+ node.
- PG2 (Parallel Group 2): Wire Cell 3 and Cell 4 in parallel. Shared negatives form PG2-. Shared positives form PG2+.
- PG3 (Parallel Group 3): Wire Cell 5 and Cell 6 in parallel. Shared negatives form PG3-. Shared positives form the Pack B+ node.
- Series Links: Connect PG1+ to PG2-. Connect PG2+ to PG3-.
According to Texas Instruments application notes on battery configurations, wiring parallel groups first and then linking them in series ensures that minor voltage mismatches between parallel cells are self-balanced before the series connections are made, preventing dangerous equalization sparks during assembly.
Step-by-Step Breadboard and Bench Testing Protocol
You cannot use a standard solderless breadboard for high-current battery testing—the spring contacts will melt at 2A. "Breadboarding" a battery pack means using a modular 18650 spring-terminal test jig or a temporary busbar setup to verify logic before committing to permanent spot-welding.
- Cell Matching: Charge all 6 cells to 4.20V. Let them rest for 2 hours. Measure with a 4-wire multimeter. Reject any cell that has drifted more than 0.02V from the group average. Mismatched parallel cells will fight each other, wasting energy as heat.
- Parallel Verification: Insert cells into your 18650 test holders. Wire only the parallel groups (PG1, PG2, PG3) using temporary jumper wires. Measure the voltage across each group. All three groups should read exactly 4.20V.
- Series Linking: Connect the series links (PG1+ to PG2-, etc.). Measure the total pack voltage at B- and B+. It must read 12.60V (3 x 4.20V). If it reads 8.4V or 4.2V, you have wired a parallel group backward. Stop and fix it.
- BMS Sense Wire Routing: Connect the BMS B- pad to the Pack B- node. Connect the B1 sense wire to PG1+ (which is also PG2-). Connect B2 to PG2+ (also PG3-). Connect B+ to the Pack B+ node. Never connect the main B- and P- discharge wires until the sense wires are verified.
- Logic Test: Power on the BMS. Use the Daly Smart BMS app via Bluetooth/UART to verify it sees exactly 3 cell groups, all reporting 4.20V.
- Load Test: Connect a 12V 50W halogen bulb (drawing ~4.2A) to the BMS P- and P+ pads. Measure the voltage drop at the terminals. A healthy 3S2P pack of 30Qs should sag no more than 0.3V under this load. If it sags below 11.5V, check your nickel strip welds or spring-terminal contacts for high resistance.
Decision Tree: Picking Your Exact Topology and BMS
Stop guessing your pack architecture. Use this decision matrix to lock in your topology based on your actual load requirements. This framework eliminates the "it depends" paralysis and forces a concrete engineering choice.
| System Requirement | Topology Choice | BMS Specification |
|---|---|---|
| Load < 50W, portable electronics, USB-C PD integration | 2S or 3S1P (7.4V / 11.1V) | 5A to 10A standard BMS |
| Load 50W - 150W, 12V routers, LED strips, small water pumps | 3S2P or 3S3P (11.1V nominal) | 15A to 20A Smart BMS with UART |
| Load 150W - 800W, RV lighting, CPAP machines, small inverters | 4S LiFePO4 (12.8V) or 12S Li-ion | 40A to 60A BMS with active balancing |
| Load > 1000W, home backup, large inverters, off-grid solar | 16S LiFePO4 (48V / 51.2V) | 100A+ BMS with RS485/CAN bus |
The Concrete Pick for 12V Off-Grid Essentials
If you are building a backup power node for a home office (running a 12V router, a mesh Wi-Fi node, and a 10W LED desk lamp, totaling roughly 35W continuous draw), you need roughly 8 hours of runtime. That requires 280Wh of usable energy.
Your exact build list: Use the 3S2P topology with Samsung 30Q cells (yielding 11.1V x 6Ah = 66.6Wh per string, wait, 6Ah * 11.1V = 66.6Wh. To get 280Wh, you need 3S7P or switch to LiFePO4). Let's correct the math for the concrete pick to ensure absolute technical accuracy.
To achieve 280Wh at 12V nominal, you need roughly 24Ah of capacity. Using 3.2V 100Ah LiFePO4 prismatic cells is the only logical path here. Therefore, for a 12V 24Ah (approx 300Wh) system, your concrete pick is a 4S1P LiFePO4 topology using four EVE LF100 (3.2V 100Ah) cells, managed by a Daly 4S 100A Smart BMS. This avoids the massive parallel-wiring complexity of 18650 cells, eliminates the parallel short-circuit thermal runaway risk entirely, and provides exactly 12.8V nominal at 100Ah (1280Wh), giving you over 30 hours of runtime for your 35W essential load.
For further reading on scaling these topologies for grid-tied systems, the National Renewable Energy Laboratory (NREL) provides extensive data on how series-parallel string sizing impacts long-term degradation in stationary storage arrays. Always size your BMS continuous current rating at least 25% higher than your inverter's maximum continuous draw to prevent nuisance tripping during inductive load startup surges.






