Adom Custom Electronic Load
Public Made by Adomby adom
Two-layer digitally controlled electronic-load prototype. Compact R11 is under active electrical, cooling and probe-access review; not a manufacturing release.
2311424
Three compact cooling concepts — 3D and thermal trade study
Project · Earlier 2D comparison
The taller 40 mm fin stack is the leading thermal candidate in this assumption-based study. The 50 mm cooler is the thinnest, and the Pi 5 cooler has the best documented 5 V PWM/tach interface. None has an established installed power rating for this load.
All three models use a 72 × 72 mm PCB packing target, compared with the current 72 × 104 mm PCB. This reduces PCB area by 31%, if the circuit can be placed and routed. A 5 mm carrier allowance on every edge gives an 82 × 82 mm planning envelope, versus the existing assembly's 151 × 120 mm. This is a 63% reduction in rectangular planning area, not a demonstrated scaffold occupancy reduction. Leads, intake/exhaust and probe travel may require more space.
These are mechanical mockups, with Q1, an isolation-stack reservation, a cooler, four large pins, scaffold contact locations and proposed probe/USB reservations. They are not populated or routed versions of the complete 107-reference board. No existing board file or manufacturing release was changed. The cooling source geometries are original approximations bounded by supplier dimensions; they are not manufacturer digital twins.
Open the interactive three-option 3D comparison — one shared Adom viewer, with a selector and a hide-cooler control to inspect Q1.
1. Preferred 50 mm low-profile cooler
Nominal cooler body 50 × 50 × 12 mm, based on the seller's dimensional photo. Concept assembly reaches 23.9 mm above the scaffold contact plane, excluding a future fan guard, cable and clamp details. The supplied fan is two-wire; a suitable tach fan remains necessary and may increase height. Snapshot price about $19–24 before its replacement fan and mounting hardware.
2. 40 mm taller fin stack
Seller specifies a 40 × 38 × 35 mm sink, or approximately 40 × 40 × 45 mm including the 4010 fan. The mockup uses a conservative 40 mm square sink footprint. Concept height is 56.9 mm above the scaffold contact plane. The 4020 option increases this by 10 mm without shrinking plan area. Listing snapshots: $8.67 with 4010, $9.14 with 4020; tach pinout remains unverified. The assumed fin count, thickness and base thickness are not supplier measurements.
3. Raspberry Pi 5 blower architecture
The manufacturer drawing gives approximate 63.5 × 42.5 × 13.7 mm dimensions. Rotating its long axis along the PCB allows it to fit the proposed cooling bay, but mounting ears, cold-plate offsets, support bosses and pinch access need an exact review. Concept height is 25.6 mm above the scaffold contact plane. The official specification specifies 5 V, PWM control and tachometer, with production through at least January 2036. No current purchase quote was obtained. Its published intended use is Raspberry Pi 5; using it on Q1 is an engineering adaptation requiring a new insulated interface and clamp. Our visual blower and fins are simplified, not an exact reproduction of its underside contact geometry.
Large pins and scaffold alignment
Each model has large-pin centers at (4,4), (68,4), (4,68), (68,68) mm relative to PCB edges. Thus the pin rectangle is 64 × 64 mm, exactly two 32 mm grid intervals in each direction. Placing the board origin at scaffold (12 + 32m, 12 + 32n) makes the pins land at 16 + 32k mm, matching the base scaffold contact grid. Contacts under the board are drawn gray; pin bodies are gold.
Pin envelope uses Large Short dimensions: 3.6 mm body, 4.8 mm flange and 1.85 mm tail. The source skill's 5.6 mm shoulder depth plus 0.05 mm contact lip places the molecule board top 5.65 mm above scaffold top, leaving air under the 1.6 mm PCB. Pin geometry is simplified, not a new production pin design. Grid alignment is verified by arithmetic; retention strength, pincer access and full probe swept-volume clearance are not verified. The narrow teal lane is a reservation, and its seven dots are illustrative, not a reduction of required test coverage.
Heat analysis: useful estimates, not wattage ratings
All cases retain 40°C ambient, 55°C case ceiling, the existing 15°C junction-rise screen, Rjc = 0.42 K/W, Rcs = 0.673 K/W and 10% power guard. The interface Rcs is retained as a conservative comparison assumption; the new simplified spacer stack has not been qualified to that value. Sensor lag, tolerance and normal-operation margin still need additional allowance.
| Concept | Conditional load-power ceiling at assumed h = 50 to 150 W/m²K | Broader geometry/contact sensitivity | Still-air sensitivity h = 5 to 10 |
|---|---|---|---|
| 50 mm low-profile | 3.1–5.4 W | 2.3–6.7 W | 0.47–0.90 W |
| 40 mm taller fins | 5.9–7.2 W | 4.4–9.0 W | 1.73–2.85 W |
| Pi 5 architecture | 3.9–6.0 W | 2.8–7.5 W | 0.66–1.23 W |
These intervals are scenario results, not confidence intervals or guaranteed capacities. The real cooler can fall outside them. No measured or CFD-derived fan operating point establishes any of these h values. Equal h makes a controlled geometry comparison but does not mean the three actual fans produce equal cooling. The broad intervals overlap, so the analysis cannot establish a definitive product winner.
With the same 82 × 82 mm planning envelope (67.24 cm²), the narrower scenario implies approximately 0.047–0.080 W/cm², 0.088–0.107 W/cm², and 0.058–0.090 W/cm², respectively. This ranks conditional heat rejection per reserved plan area, not verified whole-board power density. The taller design buys surface area with height; it could consume more vertical scaffold space.
Heat travelling along the fins
These plots show the analytical fin-root-to-tip temperature profile at 5 W Q1 heat. Junction and interface drops occur before the fin root. They show how stronger assumed cooling changes the gradient, not a simulated fan RPM sequence.
The still-air curves at 5 W exceed the policy limits. These are steady-state failure screens, not predictions of safe time after fan stoppage. We have not solved transient thermal capacity, sensor delay or shutdown latency. Retain load shutdown on fan failure; do not rate any concept for 5 W fan-off operation from these calculations. Large predicted temperature rises also make constant-property, constant-film approximations progressively less accurate.
Reproducible method and assumptions
This is an analytical straight-fin efficiency model, not CFD, a new Adom Fields board solve or supplier characterization. It uses m = sqrt(2h/(k*t)), eta = tanh(m*H)/(m*H) and Rsa = 1/[h*(eta*A_fin + A_base)], with k = 200 W/mK and an insulated fin-tip approximation. Junction temperature is Ta + Pq1*(Rsa + Rspread + Rcs + Rjc). For the comparison, Pq1 = 1.1 * Pload is a conservative bound assigning all DUT power plus guard to Q1; D1 and shunt losses remain separately relevant on the real board.
Nominal assumed fin geometry: 50 mm concept has 12 fins, 10 mm tall, 1 mm thick, 2 mm base, 50% effective fin-length coverage for the inset fan; 40 mm concept has 12 fins, 32 mm tall, 1 mm thick, 3 mm base, full coverage; Pi concept has 10 fins, 11.7 mm tall, 1 mm thick, 2 mm base, 60% coverage for the blower. The cavity visual meshes and these effective coverage factors are independent idealizations. Unknown exact dimensions prevent an accurate mesh-derived thermal result.
Nominal spreading resistance is 0.5 K/W, assumed, not solved from Q1's small contact. The broad sensitivity sweeps 0.2–1.0 K/W spreading and 70–130% effective fin coverage, together with h = 50, 100 and 150. Base underside convection, radiation, recirculation, fan motor heating, obstruction, nonlinear buoyancy, junction spatial variation, duct pressure drop and thermal coupling into nearby parts are omitted. Uniform convection and a uniform fin-root temperature simplify a nonuniform real flow. There is no mesh convergence claim because this is an analytical model.
Calculation inputs and full results · Calculation/model source · GLB export source · Model dimensions and group audit
Why 15–25 W is not established by a bigger fan
Under the existing junction-rise screen, required sink-plus-spreading resistance is 15/(1.1*Pload) - 0.673 - 0.42:
| Desired DUT load | Maximum sink + spreading resistance, before further margin |
|---|---|
| 5 W | 1.63 K/W |
| 10 W | 0.27 K/W |
| 15 W | Negative: impossible with these retained interface/temperature assumptions |
| 25 W | Negative: impossible with these retained interface/temperature assumptions |
Even a hypothetical perfect sink would hit the retained junction-rise screen near 12.5 W load, before additional margin. Higher power therefore needs a deliberately revised thermal architecture and protection policy, not an unreviewed firmware limit increase. Alternatives include a lower-resistance insulated contact, several independently controlled MOSFETs, or a larger heatpipe cooler. No temperature gate was weakened in this study.
Higher DUT power also requires checking Q1's DC safe operating area at actual Vds and case temperature, D1 loss versus current, shunt and conductor heating, contacts, protection and control-loop stability. A cooler calculation cannot establish a higher electrical load rating. In particular, D1's old 0.513 W cooling result cannot simply be carried into a higher-current design or a board with less copper. Firmware remains at the existing 5 W prototype envelope.
Recommended next candidate
Investigate the 40 mm fin stack with a documented feedback fan first for watts per plan area. Keep the preferred 50 mm design as the low-height candidate and Pi's 5 V blower as the documented-interface candidate. Resolve actual fin geometry, pressure/flow, cold-plate contact, insulating materials and clamp load before selecting one. Then place all components, preserve D1 copper, route the 72 mm target and replay every required probe/pincer motion. If those do not fit, grow the PCB in scaffold-compatible steps rather than claiming the packing target passed.
No fabrication geometry, installed thermal rating, physical test or higher-power release is claimed. Research/model date: 13 October 2026.
Process recording
37-second raw Pup comparison clip: switch cooler concepts, hide the cooler to inspect the Q1 interface, and compare overall height. This is source footage for the narrated edit; no new voiceover or captions are attached to this raw take.
# Three compact cooling concepts — 3D and thermal trade study
[Project](../../README.md) · [Earlier 2D comparison](../cooling-options/README.md)
**The taller 40 mm fin stack is the leading thermal candidate in this assumption-based study. The 50 mm cooler is the thinnest, and the Pi 5 cooler has the best documented 5 V PWM/tach interface. None has an established installed power rating for this load.**
All three models use a **72 × 72 mm PCB packing target**, compared with the current 72 × 104 mm PCB. This reduces PCB area by 31%, if the circuit can be placed and routed. A 5 mm carrier allowance on every edge gives an **82 × 82 mm planning envelope**, versus the existing assembly's 151 × 120 mm. This is a 63% reduction in rectangular planning area, not a demonstrated scaffold occupancy reduction. Leads, intake/exhaust and probe travel may require more space.
These are **mechanical mockups**, with Q1, an isolation-stack reservation, a cooler, four large pins, scaffold contact locations and proposed probe/USB reservations. They are not populated or routed versions of the complete 107-reference board. No existing board file or manufacturing release was changed. The cooling source geometries are original approximations bounded by supplier dimensions; they are not manufacturer digital twins.
**[Open the interactive three-option 3D comparison](https://wiki.adom.inc/adom/adom-custom-electronic-load/render/docs/cooling-3d/viewer.html)** — one shared Adom viewer, with a selector and a hide-cooler control to inspect Q1.
## 1. Preferred 50 mm low-profile cooler
[Download GLB](preferred50.glb)
Nominal cooler body **50 × 50 × 12 mm**, based on the [seller's dimensional photo](https://www.aliexpress.us/item/3256811747139271.html). Concept assembly reaches **23.9 mm above the scaffold contact plane**, excluding a future fan guard, cable and clamp details. The supplied fan is two-wire; a suitable tach fan remains necessary and may increase height. Snapshot price about $19–24 before its replacement fan and mounting hardware.
## 2. 40 mm taller fin stack
[Download GLB](chipset40.glb)
[Seller](https://www.aliexpress.us/item/3256809091061240.html) specifies a **40 × 38 × 35 mm sink**, or approximately **40 × 40 × 45 mm** including the 4010 fan. The mockup uses a conservative 40 mm square sink footprint. Concept height is **56.9 mm above the scaffold contact plane**. The 4020 option increases this by 10 mm without shrinking plan area. Listing snapshots: $8.67 with 4010, $9.14 with 4020; tach pinout remains unverified. The assumed fin count, thickness and base thickness are not supplier measurements.
## 3. Raspberry Pi 5 blower architecture
[Download GLB](pi5.glb)
The [manufacturer drawing](https://datasheets.raspberrypi.com/cooling/raspberry-pi-active-cooler-mechanical-drawing.pdf) gives approximate **63.5 × 42.5 × 13.7 mm** dimensions. Rotating its long axis along the PCB allows it to fit the proposed cooling bay, but mounting ears, cold-plate offsets, support bosses and pinch access need an exact review. Concept height is **25.6 mm above the scaffold contact plane**. The [official specification](https://datasheets.raspberrypi.com/cooling/raspberry-pi-active-cooler-product-brief.pdf) specifies **5 V, PWM control and tachometer**, with production through at least January 2036. No current purchase quote was obtained. Its published intended use is Raspberry Pi 5; using it on Q1 is an engineering adaptation requiring a new insulated interface and clamp. Our visual blower and fins are simplified, not an exact reproduction of its underside contact geometry.
## Large pins and scaffold alignment
Each model has large-pin centers at **(4,4), (68,4), (4,68), (68,68) mm** relative to PCB edges. Thus the pin rectangle is **64 × 64 mm**, exactly two 32 mm grid intervals in each direction. Placing the board origin at scaffold **(12 + 32m, 12 + 32n)** makes the pins land at **16 + 32k mm**, matching the [base scaffold contact grid](https://wiki.adom.inc/adom/base-scaffold-system). Contacts under the board are drawn gray; pin bodies are gold.
Pin envelope uses [Large Short dimensions](https://wiki.adom.inc/adom/machinepinlargeshort): 3.6 mm body, 4.8 mm flange and 1.85 mm tail. The source skill's 5.6 mm shoulder depth plus 0.05 mm contact lip places the molecule board top 5.65 mm above scaffold top, leaving air under the 1.6 mm PCB. Pin geometry is simplified, not a new production pin design. Grid alignment is verified by arithmetic; retention strength, pincer access and full probe swept-volume clearance are not verified. The narrow teal lane is a reservation, and its seven dots are illustrative, not a reduction of required test coverage.
## Heat analysis: useful estimates, not wattage ratings
All cases retain **40°C ambient, 55°C case ceiling, the existing 15°C junction-rise screen, Rjc = 0.42 K/W, Rcs = 0.673 K/W and 10% power guard**. The interface Rcs is retained as a conservative comparison assumption; the new simplified spacer stack has not been qualified to that value. Sensor lag, tolerance and normal-operation margin still need additional allowance.
| Concept | Conditional load-power ceiling at assumed h = 50 to 150 W/m²K | Broader geometry/contact sensitivity | Still-air sensitivity h = 5 to 10 |
|---|---:|---:|---:|
| 50 mm low-profile | **3.1–5.4 W** | 2.3–6.7 W | 0.47–0.90 W |
| 40 mm taller fins | **5.9–7.2 W** | 4.4–9.0 W | 1.73–2.85 W |
| Pi 5 architecture | **3.9–6.0 W** | 2.8–7.5 W | 0.66–1.23 W |
**These intervals are scenario results, not confidence intervals or guaranteed capacities.** The real cooler can fall outside them. No measured or CFD-derived fan operating point establishes any of these h values. Equal h makes a controlled geometry comparison but does not mean the three actual fans produce equal cooling. The broad intervals overlap, so the analysis cannot establish a definitive product winner.
With the same 82 × 82 mm planning envelope (67.24 cm²), the narrower scenario implies approximately **0.047–0.080 W/cm²**, **0.088–0.107 W/cm²**, and **0.058–0.090 W/cm²**, respectively. This ranks conditional heat rejection per reserved plan area, not verified whole-board power density. The taller design buys surface area with height; it could consume more vertical scaffold space.
### Heat travelling along the fins
These plots show the analytical fin-root-to-tip temperature profile at **5 W Q1 heat**. Junction and interface drops occur before the fin root. They show how stronger assumed cooling changes the gradient, not a simulated fan RPM sequence.
The still-air curves at 5 W exceed the policy limits. These are steady-state failure screens, not predictions of safe time after fan stoppage. We have not solved transient thermal capacity, sensor delay or shutdown latency. Retain load shutdown on fan failure; do not rate any concept for 5 W fan-off operation from these calculations. Large predicted temperature rises also make constant-property, constant-film approximations progressively less accurate.
### Reproducible method and assumptions
This is an **analytical straight-fin efficiency model**, not CFD, a new Adom Fields board solve or supplier characterization. It uses `m = sqrt(2h/(k*t))`, `eta = tanh(m*H)/(m*H)` and `Rsa = 1/[h*(eta*A_fin + A_base)]`, with k = 200 W/mK and an insulated fin-tip approximation. Junction temperature is `Ta + Pq1*(Rsa + Rspread + Rcs + Rjc)`. For the comparison, `Pq1 = 1.1 * Pload` is a conservative bound assigning all DUT power plus guard to Q1; D1 and shunt losses remain separately relevant on the real board.
Nominal assumed fin geometry: 50 mm concept has 12 fins, 10 mm tall, 1 mm thick, 2 mm base, 50% effective fin-length coverage for the inset fan; 40 mm concept has 12 fins, 32 mm tall, 1 mm thick, 3 mm base, full coverage; Pi concept has 10 fins, 11.7 mm tall, 1 mm thick, 2 mm base, 60% coverage for the blower. The cavity visual meshes and these effective coverage factors are independent idealizations. Unknown exact dimensions prevent an accurate mesh-derived thermal result.
Nominal spreading resistance is **0.5 K/W, assumed**, not solved from Q1's small contact. The broad sensitivity sweeps **0.2–1.0 K/W** spreading and **70–130%** effective fin coverage, together with h = 50, 100 and 150. Base underside convection, radiation, recirculation, fan motor heating, obstruction, nonlinear buoyancy, junction spatial variation, duct pressure drop and thermal coupling into nearby parts are omitted. Uniform convection and a uniform fin-root temperature simplify a nonuniform real flow. There is no mesh convergence claim because this is an analytical model.
[Calculation inputs and full results](../../analysis/cooling-3d/results.json) · [Calculation/model source](../../tools/compare_compact_coolers.py) · [GLB export source](../../tools/export_cooling_concepts.py) · [Model dimensions and group audit](model-audit.json)
## Why 15–25 W is not established by a bigger fan
Under the existing junction-rise screen, required sink-plus-spreading resistance is `15/(1.1*Pload) - 0.673 - 0.42`:
| Desired DUT load | Maximum sink + spreading resistance, before further margin |
|---|---:|
| 5 W | 1.63 K/W |
| 10 W | 0.27 K/W |
| 15 W | Negative: impossible with these retained interface/temperature assumptions |
| 25 W | Negative: impossible with these retained interface/temperature assumptions |
Even a hypothetical perfect sink would hit the retained junction-rise screen near **12.5 W load**, before additional margin. Higher power therefore needs a deliberately revised thermal architecture and protection policy, not an unreviewed firmware limit increase. Alternatives include a lower-resistance insulated contact, several independently controlled MOSFETs, or a larger heatpipe cooler. No temperature gate was weakened in this study.
Higher DUT power also requires checking Q1's **DC safe operating area at actual Vds and case temperature**, D1 loss versus current, shunt and conductor heating, contacts, protection and control-loop stability. A cooler calculation cannot establish a higher electrical load rating. In particular, D1's old 0.513 W cooling result cannot simply be carried into a higher-current design or a board with less copper. Firmware remains at the existing 5 W prototype envelope.
## Recommended next candidate
Investigate the **40 mm fin stack with a documented feedback fan** first for watts per plan area. Keep the preferred 50 mm design as the low-height candidate and Pi's 5 V blower as the documented-interface candidate. Resolve actual fin geometry, pressure/flow, cold-plate contact, insulating materials and clamp load before selecting one. Then place all components, preserve D1 copper, route the 72 mm target and replay every required probe/pincer motion. If those do not fit, grow the PCB in scaffold-compatible steps rather than claiming the packing target passed.
No fabrication geometry, installed thermal rating, physical test or higher-power release is claimed. Research/model date: 13 October 2026.
## Process recording
[37-second raw Pup comparison clip](../../videos/cooling-3d/three-concepts-raw.mp4): switch cooler concepts, hide the cooler to inspect the Q1 interface, and compare overall height. This is source footage for the narrated edit; no new voiceover or captions are attached to this raw take.