<aside> ⚠️
Design recommendations, not released product specifications. This guide targets a 24V nominal, high-power-capable Mote power path. The proposed coupled-inductor configurations below have not been qualified at 50W or 100W. The Jetpack is the documented and tested 100W precedent; its archived design uses discrete, not coupled, inductors. Every source, cable, connector, intervening node, protection element and payload converter must support the chosen current. Never route high power through an unmodified standard node or you risk frying the standard node.
This page was created with the help of AI to synthesize and distill several engineering research and development projects.
</aside>
Start with the 50W coupled-inductor design; use the archived Jetpack as the 100W baseline. For a coupled 100W design, preserve useful differential inductance by using multiple compact components instead of choosing a single tiny-inductance part solely for its current rating.
These targets refer to power at the local Mote bus/power path, not a guarantee of delivered payload power over an arbitrary cable. At 24V, 50W means 2.083A and 100W means 4.167A before node consumption, conversion losses and cable losses. A payload requiring that output power needs additional input-current margin, likely on the order of ~15%.
| Change | 50W variant — proposed | 100W variant — proposed |
|---|---|---|
| Preferred coupled magnetics | MSD1514-473MED, one two-winding component per BM port; two per two-port Mote. 47µH per winding; about 188µH differential inductance with the correct series-aiding orientation. | Five MSD1514-103MED components cascaded per BM port; ten per two-port Mote. Each stage has 10µH per winding and about 40µH differential inductance; five stages give about 200µH. This avoids parallel-current-sharing problems but costs board area. |
| Coupled alternatives | MSD1514-683MED, 68µH per winding / approximately 272µH differential: potentially better droop, but only 2.2A simultaneous-winding thermal reference current, so little margin at 50W/24V. MSD1278H-473MED is smaller, but its 2.10A thermal reference and series-aiding saturation margin are too close for a comfortable 50W recommendation. Treat both as characterization options. | Four MSD1514-473MED per port, arranged as two parallel branches, each containing two coupled stages in series: approximately 188µH effective differential inductance. Each branch carries half the bus current. This is a compact-package alternative, not a drop-in or proven design; requires current-sharing, fault and parasitic-resonance validation. |
| Documented discrete baseline | The Jetpack magnetics can also be used at lower power, subject to circuit qualification. | Four Pulse PA4344.104NLT discrete inductors, 100µH each, as in the archived Jetpack. The archived BOM describes 5.3A / 118mΩ parts. Reproduce the actual schematic and port wiring; do not infer coupled-inductor behavior from this discrete design. |
| Small local decoupling | Preserve the buck input requirement: at least 4.7µF effective low-ESR capacitance, plus local high-frequency bypassing. Use DC-bias-adjusted values, not nameplate capacitance. | Same requirement; low-power logic rails do not automatically require twice the capacitance just because through-power doubles. |
| Damping starting point | Reproduce the Mote v1.1 baseline: two EEEFT1H470AP 47µF/50V electrolytics, each in series with 7.5Ω, across VBUS–GND; 10µF/50V and 100nF/100V direct bypass in the source schematic. Retune for new magnetics. Prefer additional voltage margin where practicable. | Same proven topology as a starting point, not the same guaranteed resistor values. Re-measure the resulting PDN and tune damping for the cascaded or series/parallel magnetics, wiring, input protection and payload capacitance. |
| Extra transient storage | Provision 470µF effective, nominally ≥50V and preferably 63V, low-ESR bulk capacitance near a separately soft-started payload branch. This is an engineering starting point for a full 50W step lasting roughly 100µs with ≤0.5V capacitive sag at 24V, not a measured Sofar prescription. | Provision about 940–1000µF effective in the controlled payload branch for the corresponding full 100W, 100µs step. The Jetpack archive instead uses 20 × KRM55WR71H336MH01L, 33µF/50V MLCCs = 660µF nominal on VBUS; their effective capacitance under bias must be checked. |
| Protection and layout | Re-rate fuses, load switches, copper, shunts, connectors and wiring for at least the qualified 2.083A-plus-margin operating current. Add controlled inrush, coordinated overvoltage clamping and reverse-polarity handling. | Re-rate the complete path for 4.167A-plus-margin. Do not retain an original 2A fuse or low-current payload switch as the high-power path. Bidirectional powering needs protection that works from either port. |
| Release gate | Thermal soak, biased differential impedance/return loss, droop, hot plug and simultaneous data/load testing at the final minimum local voltage. | Same, plus cascade/parasitic evaluation or branch-sharing/fault tests. Compare against the discrete Jetpack baseline. |
Compactness: MSD1514 components are approximately 15.5 × 15.5 × 14.2mm; MSD1278H is approximately 12.3 × 12.3 × 8.05mm. These are component dimensions, not a finished-board envelope. The 100W coupled options intentionally trade component count and board area for current capability without collapsing nominal inductance. Reserve clearance and cooling; closely packed or potted parts do not inherit the datasheet thermal test conditions.
Do not treat the 50W/100W labels as valid across the entire existing 16–32V bus range. At 16V the currents are 3.125A and 6.25A, respectively. The preferred selections above are 24V nominal prototype recommendations. Set a validated minimum local voltage and current ceiling; reduce power or shut down below that envelope. Do not raise the bus to 40V/48V on an unmodified Mote.
Use the manufacturer's winding diagram. Place one winding in the positive DC coupling leg and the other in the return leg so that the differential data signal sees the intended high mutual inductance. A normal common-mode-choke orientation that cancels differential flux can leave only leakage inductance and severely load the data path. Verify the orientation experimentally before applying high power.
For an ideal tightly coupled pair with self-inductance L and mutual inductance M:
$$ L_{diff}=2L+2M\approx4L $$
This is the series-aiding differential-loop value, not the value of one winding, and not a reason to short or parallel the two windings inside the ordinary two-wire PoDL circuit. With the opposite incorrect polarity the loop inductance is approximately 2L−2M, potentially orders of magnitude smaller.
Coilcraft's cited Isat is the sum of the currents in both flux-aiding windings. In the intended differential series-aiding configuration, each winding carries the full branch DC current; conservatively compare 2 × branch current against the quoted Isat. Coilcraft's both-windings Irms is the equal current in each winding simultaneously at a stated temperature rise. Do not use its one-winding thermal rating as the bus rating. Distributor labels such as “6.8A” may describe one-winding thermal performance or a saturation condition, not the intended PoDL operating current. Be very careful with these ratings from different distributors, we only use Coilcraft here as an example.