Scope and operating assumptions
This note covers the temperature-acquisition function of battery management systems in series-string LFP traction packs for golf carts and comparable light electric vehicles — for instance, packs considered for Club Car or E-Z-GO style fleet vehicles, where the brand name marks only an application context, not chemistry, fit, or approval. The topic is how temperature is sensed, what the measurements feed into, and how a service team should read the resulting telemetry. It assumes an architecture of the kind described in the supplied technical material: an analog front end sampling cell taps and dedicated temperature channels, a microcontroller running protection firmware, switching devices in the charge and discharge paths, and a communication port such as CAN or UART for reporting. No JTM-specific thresholds, sensor part numbers, or channel assignments exist in the supplied evidence, and none are stated; everything below is a general mechanism explanation, not a product specification.
Working principle and control logic
- Sensing elements. Packs in this class commonly use NTC thermistors, whose resistance decreases predictably as temperature rises. The measurement chain is: thermistor in a resistor divider → AFE samples the divider voltage → firmware converts it to temperature using a calibrated curve, often the beta relationship 1/T = 1/T0 + (1/β)·ln(R/R0), where T and T0 are absolute temperatures (K), R is the measured thermistor resistance (Ω), R0 is the resistance at T0, and β is the device constant (K). Conversion accuracy depends on firmware holding the correct curve for the installed thermistor, good thermal contact between sensor and surface, and intact leads. Thermocouples respond faster and cover wider ranges but are less typical in this low-voltage class.
- Placement rationale. Heat generation is not uniform across a pack: current-carrying busbars, terminals, and geometrically central cells tend to run warmer than perimeter regions during high-current discharge. A multi-point layout — for illustration edge, center, and opposite-side positions per module — gives the firmware visibility of gradient as well as absolute temperature, which a single sensor cannot provide. The supplied material presents this layout as a design practice for detecting uneven heating before it becomes damage; it does not define a mandatory sensor count.
- From measurement to decision. In the general control model, temperature inputs are compared in firmware against configured boundaries, and the comparison result drives outputs: a charge-permission signal or gate state change in the charge/discharge switches, and alarm or relay outputs. Causally: measured thermistor voltage → converted temperature → boundary comparison → permission/switch state change → later re-permission once readings move back inside the allowed window. The supplied sources establish that such temperature-window enforcement exists (charging outside the permitted window must be prevented, ideally by a charger controlled through a remote on/off line from the BMS) but they do not document which sensor reading wins when several disagree, nor the release hysteresis. Any rule such as 'coldest sensor gates cold-charge inhibit' is a plausible design convention and must be confirmed from the applicable product's documentation, not assumed.
- Gradient as a diagnostic quantity. A widening spread between sensors at steady load can point to a localized resistance or airflow problem rather than a whole-pack thermal event. Whether a given BMS flags the spread itself, or the service team must derive it from logged per-channel values, is product-dependent; nothing in the supplied evidence describes an automatic gradient alarm.
Parameters and interfaces
- Channel count and mapping. Which channel watches which cell group, busbar, or terminal determines how precisely a thermal anomaly can be localized. The significance of a single channel failure depends on firmware policy; a common conservative convention is to restrict operation when thermal supervision loses a channel, but this is a design choice, not a documented behavior of any specific pack.
- Thermistor parameters (R0, β, tolerance, measurement range). These convert raw voltage to temperature. If a replacement thermistor with different characteristics is installed, every converted value shifts and all protection boundaries are effectively displaced without any threshold being edited — a configuration-level failure mode.
- Sampling and filtering. Averaging over more samples smooths noise but delays the response to genuinely fast events, such as a degrading connection heating under load. The tradeoff is inherent to the firmware design; no numeric values are supplied here.
- Telemetry content. Communication interfaces in this class are described in the source material as carrying live cell voltages, pack current, temperature, and state data to external controllers or displays. Whether per-channel temperature, min/max values, and fault flags are exposed is a property of the specific implementation; where they are, log granularity should allow temperature readings to be correlated with the current history in the same time window.
Verification and fault diagnosis
Diagnosis should rely on telemetry and recorded logs; do not open an energized pack or bypass protection to test it.
- Normal pattern. Channels track each other within a modest spread; peak readings follow load current and decay after the load ends; permission state changes coincide with temperature crossing the applicable boundary and does not chatter at the boundary.
- One channel frozen or pinned near a rail value while others move. Consistent with an open or shorted sensor lead or an AFE channel fault. Note the pack's response — whether it merely reports or also restricts operation — from its own documentation.
- All channels offset together. More likely thermistor curve mismatch or poor thermal coupling than real heating; investigate the sensing path before the cells.
- Spread growing at constant load. Compare against voltage sag and cell-voltage skew in the same log window; a thermal gradient with an electrical signature points toward a connection, whereas a purely thermal gradient suggests airflow or placement.
- Validation logic. A safe field check is to compare logged temperature rise against the charge or discharge current profile across repeated, identical duty cycles and look for a consistent upward drift. This is an indicative trend only — no quantitative pass/fail criterion is supplied by the evidence, and it should not be treated as a certified condition metric. Formal verification of thermal protection behavior belongs in qualified laboratory type-test reports.
Limitations
Surface-mounted thermistors estimate, not measure, cell internal temperature; during fast transients the interior can lead the surface, so protection acts with inherent lag. Sensor counts, placement, thresholds, hysteresis, arbitration rules, and telemetry contents are firmware and configuration items specific to each BMS; figures from another manufacturer's pack must not be transferred to any JTM battery, and no JTM product data was supplied for this note. Named vehicle platforms define the applicability boundary only — without model-specific evidence, no assumption may be made that a given vehicle uses LFP chemistry or any particular BMS behavior. Nothing here addresses certification, warranty, or hands-on servicing of live packs.