hardware: Rev 0 fan-characterization fixture schematic (stacked on #8)

hardware/fan-characterization-fixture/BRINGUP.md

@@ -0,0 +1,165 @@
+# Rev 0 bring-up: minimum safe first power
+
+Record every result in the JumpJet characterization evidence record, with the
+instrument, its setting, and a photo or capture. Do the steps in order: each
+one removes an unknown that the next depends on.
+
+**Starting state**
+
+- JP1, JP2, JP3, JP4 **OPEN**. JP5 fitted.
+- F1 holds a labelled 0 Ω link or dummy fuse, not a rated fuse. The bench
+  PSU current limit is the first protection.
+- Controller not connected to J1. Fan not connected to J3.
+
+**Operating rule (tach back-power)**
+
+- The controller's 3V3 must be powered **before JP3 or JP4 is closed**.
+- Keep 3V3 on while JP3 or JP4 is closed and fan power is on. Before
+  removing 3V3, turn fan power off or open JP3 and JP4.
+- Fan power with JP3 and JP4 both open is allowed without 3V3: the raw tach
+  lead then reaches only TP8 and has no path to the 3V3 rail.
+- See README "Tach protection and back-power".
+
+## 1. Identify the specimen
+
+1. Photograph the label: part number `9GA0424P3J001`, lot, and date code.
+2. Photograph each lead and record its colour.
+3. Record any connector or housing on the leads.
+
+## 2. Verify the leads before any 24 V
+
+1. With the fan unpowered, measure lead-to-lead resistance and diode-mode
+   readings. Record all of them.
+2. Establish each lead's function by photo, continuity, and the manufacturer
+   drawing (9GA0424P3J001 Rev C). The catalog colour convention is **not**
+   sufficient on its own.
+3. Do not land a lead on J3 until its function is established.
+4. `FAN_PWM_RAW` must never reach `FAN_24V`: Q1 is not rated for that fault.
+
+## 3. Prepare the bench supply
+
+1. Set 24.0 V and a conservative current limit based on the 0.27 A rating.
+   Record the limit. It is a bench setting, not a derived startup
+   specification.
+2. Check the output with no load.
+3. Record whether the PSU output floats or is earth-bonded.
+4. Connect J2 and confirm polarity at TP12 against TP14 **before** landing the
+   fan. The fixture has no reverse-polarity element.
+
+## 4. Fan power only, signal lines released
+
+1. Land only `FAN_24V` and `FAN_GND` on J3. Keep the control and sensor leads
+   insulated and off J3.
+2. Power up. Record the supply current: the inrush peak, if a current probe
+   is available, and the steady state.
+3. The drawing specifies that an open control lead gives the same speed as
+   100 % PWM. Record whether the specimen behaves that way.
+
+## 5. FAN_PWM_RAW with no active drive
+
+1. Land the control lead on J3.3. Keep **JP2 OPEN** for the whole step.
+2. Measure the open-circuit voltage first. Scope TP6 against TP7 and record
+   the voltage and any waveform.
+3. Continue only if both are true:
+   - the lead's identity was already established in step 2, and
+   - the measured open voltage is ≤ 5.25 V, as the manufacturer drawing
+     specifies.
+
+   If the open voltage is above 5.25 V, **STOP**. Do not connect the
+   resistor. Treat the lead identity, or the assumption about the fan's
+   control interface, as wrong until it has been investigated.
+4. Temporarily connect a known resistor from TP6 (`FAN_PWM_RAW`) to TP7
+   (GND). This is a bench test part, not a fixture component:
+   - 100 Ω
+   - 1 %
+   - ≥ 0.5 W
+5. Measure the voltage across the resistor with a DMM on its **voltage**
+   range. Do not put the DMM's current input in this circuit.
+6. Calculate the current: I = V / 100 Ω.
+   - At the drawing's 2 mA maximum source current, expect about 0.200 V.
+   - That is below the control input's documented ≤ 0.4 V LOW threshold, so
+     the resistor holds the input LOW.
+7. Record:
+   - resistor value (measured, if possible)
+   - measured voltage
+   - calculated current
+   - resulting fan behaviour (drawing: 0 % = stopped)
+8. Remove the resistor.
+
+This is a **controlled low-state current measurement** at the measured
+resistor voltage. It is not an exact zero-volt short-circuit measurement.
+The drawing's "≤ 2 mA at 0 V" figure is the limit it is compared against.
+
+Close JP2 later, in step 7, only if the open voltage was ≤ 5.25 V and the
+calculated current was ≤ 2 mA.
+
+## 6. Raw tach, before any pull-up or GPIO
+
+1. Power the controller. Connect J1.3 to controller 3V3 and J1.4 to
+   controller GND, so the fixture's 3V3 rail is live before JP3 closes
+   (operating rule). Leave J1.1 (`MCU_PWM`) and J1.2 (`TACH_GPIO`) unwired
+   and JP1 open until step 7, item 5.
+2. Land the sensor lead on J3.4. Keep JP3 and JP4 open.
+3. Scope TP8 with the fan running. The drawing's sensor specification
+   9D0001H202 says open collector: expect TP8 to float or sit near 0 V with
+   no pull-up.
+4. **If TP8 is actively driven above 3.3 V, stop** and revise the tach stage.
+5. Close **JP3** only. Record at TP8:
+   - high level
+   - low level, which must be ≤ ~0.6 V (the documented VCE(sat) max is
+     0.8 V at 5 mA)
+   - edge shape, glitches, and ringing
+6. Add no filtering.
+
+## 7. Validate the open-drain PWM stage
+
+1. Keep the controller's 3V3 on from step 6 (operating rule), but leave its
+   PWM output unconnected: J1.1 (`MCU_PWM`) stays unwired and JP1 open until
+   item 5. With JP2 open, drive TP3 (`PWM_IN`) from a signal generator:
+   - 0–3.3 V square wave
+   - 25 kHz, the drawing's specified control frequency
+2. Scope TP4 and TP5. Then close JP2 and scope TP6. Record:
+   - the low level (≤ 0.4 V required, about 4 mV expected)
+   - the edges
+   - the duty cycle against the generator
+3. Remove the generator.
+4. Flash the fan-characterization image for the N16R8 with the Rev 0 bench
+   overlay (README "Controller wiring overlay"):
+   - `GATE_GPIO=6`, `SINK_LEVEL=1`, `TACH_GPIO=7`
+   - PPR 0, internal tach pull-up off
+5. Wire J1.1 and J1.2 to the controller (J1.3 and J1.4 are already wired from
+   step 6) and close JP1.
+6. Confirm the boot log shows the line released. Confirm TP4 stays at 0 V
+   through reset, flashing, and a controller power cycle. Open JP3 before
+   the power cycle and close it again once 3V3 is back (operating rule).
+7. Run `FAN_PWM_HOLD` at 0 %, a mid duty, and 100 %. For each, compare the
+   TP4/TP6 captures with the reported `applied_sink_duty_pct`.
+
+## 8. Connect tach to GPIO7
+
+1. Continue only if step 6 showed an open-collector output with levels
+   within 0–3.3 V, and the controller's 3V3 is on.
+2. Close JP4.
+3. Check that TP9 (`TACH_GPIO`) follows TP8 within the R4/R5 divider:
+   - high ≈ 3.09 V
+   - low ≈ V(TP8)
+4. Run a hold. Compare `tach_edges` and `edge_hz` with a scope or
+   logic-analyzer count over the same window.
+5. Add a glitch filter only if the captures show a need.
+
+## 9. PPR stays unknown
+
+The manufacturer documentation suggests 2 pulses per revolution. DragonBench
+still keeps `CONFIG_DB_FAN_TACH_PPR=0` until independent evidence exists:
+preferably an optical tachometer correlated with tach frequency at several
+speeds on this specimen.
+
+Record that evidence and reference it in `CONFIG_DB_FAN_TACH_PPR_EVIDENCE`
+when PPR is set.
+
+## After bring-up
+
+- Choose F1 only after startup and locked-rotor current are measured.
+  Locked-rotor is a separate, deliberate test with its own plan.
+- Nothing measured on this fixture becomes a JumpJet threshold, frequency,
+  PPR, or safety value without its own review.