Circuit Review


The complete 24 VAC, 100 W constant-resistance load schematic is generated and independently reviewed. It contains 67 components, AD633 conductance control, OPA197 current regulation, 50 mΩ shunt, hardware overcurrent/thermal shutdown, test points, and isolated interfaces; ERC and legibility checks are clean.
The defective IRFP250N library part was replaced with the linear-rated IXTH75N10L2, correctly mapped Gate–Drain–Source.
Open bench-validation items
  • Verify current-loop stability and tune compensation.
  • Select a real Kelvin 50 mΩ / 5 W shunt MPN.
  • Validate MOSFET cooling and DC SOA experimentally.
  • Confirm default-off latch behavior during power sequencing.
Project files:
Suggested next steps
  1. Review the schematic and scaling.
  2. Select the physical shunt and heatsink arrangement.
  3. Simulate or bench-test loop compensation.
  4. Proceed to PCB constraints and placement.
there is a lot of errors in airwires
I’ll inspect the current airwire report and layout state, determine whether the issue is missing schematic connectivity or simply an unrouted/unplaced PCB, then fix the appropriate root cause.
The 127 airwires are PCB routing errors, not schematic errors. The schematic connectivity is valid, but all 67 footprints are scattered outside the default 160 × 100 mm board and no copper has been routed yet.
Proposed PCB setup:
  • Board: 100 × 100 mm rectangular
  • Stackup: 4 layers, standard FR-4
  • Copper: preferably 2 oz outer layers for the load-current path
  • Top side: all components
  • Edges: AC input and fuse left; command/power/control connectors opposite edge
  • Power section: bridge, Q1, shunt and heavy terminals grouped near an edge
  • Q1: placed at the board edge for external heatsink mounting
  • Analog section: AD633, OPA197 and sensing network separated from bridge/gate-current paths
  • Routing: wide high-current loop first, Kelvin shunt traces separately, then analog/control signals
  • Thermal note: Q1 dissipates externally through the heatsink; the PCB must not be expected to absorb 100 W
Confirm this 100 × 100 mm, 4-layer, top-side-only arrangement with Q1 at the edge, and I’ll place and route the PCB to eliminate the airwires.

Waiting for input


Independent Schematic Review
Review date: 2026-09-11
Scope: Current schematic only; no PCB layout or cosmetic rearrangement. All 67 components, 28 nets, active-device pins, passive values, connector assignments, and requested protection/control paths were inspected independently.
Overall result
The logical schematic topology remains coherent. The previously identified IRFP250N physical pin-number defect and unsuitable linear-mode SOA evidence have now been resolved by replacing Q1 with the Flux-library IXTH75N10L2 and explicitly reconnecting its numbered terminals.
Requested-item status

Table


ItemStatusIndependent finding
Bridge/fuse/MOV power pathCompleted — passJ1.1 → F1 → BR1 AC, with RV1 across the fused AC pair; J1.2 returns directly to the other bridge AC terminal. The MOV is downstream of the fuse. BR1+ feeds VRECT and BR1− feeds AGND. F1 is a 6.3 A time-lag device; RV1 is the 30 VAC S20K30 variant.
No VRECT reservoir capacitorCompleted — passVRECT connects only to BR1+, Q1 drain, R1 and TP1. C1 is 1 nF on VRECT_SCALED, not VRECT. No reservoir/bulk capacitor is present across VRECT–AGND.
AD633 equation and command scalingCompleted — passAD633 uses W = (X1−X2)(Y1−Y2)/10 V + Z. X1=VRECT_SCALED, X2=AGND, Y1=G_CMD_SCALED, Y2=AGND, Z=AGND. Divider ratios are 10 kΩ/(46.4 kΩ+46.4 kΩ+10 kΩ)=0.097276 and 9.31 kΩ/(24.9 kΩ+9.31 kΩ)=0.272143. At 3.3 V command, G_CMD_SCALED=0.8981 V; with 50 mΩ shunt this commands 5.723 Ω, or 100.64 W at 24 VAC nominal. D1 clamps the post-1 kΩ command node with cathode to DAC_CLAMPED and anode to AGND.
OPA197 feedback polarityCompleted — passOPA197 DBV pins are OUT=1, V−=2, IN+=3, IN−=4, V+=5. IREF is on IN+, Kelvin shunt voltage is on IN−, and increasing output increases MOSFET current/shunt voltage, giving negative feedback.
Compensation topologyCompleted — topology pass; stability unresolvedR10=1 kΩ in series with C3=1 nF from LOOP_OUT to VSHUNT forms a series-RC high-frequency feedback path. R7=1 kΩ plus R8=100 Ω isolates/damps the MOSFET gate. This is electrically connected as intended, but the values are explicitly initial/placeholder values and were not proven stable with the actual MOSFET, wiring inductance and shunt layout. OPA197 literature supports output isolation for capacitive loading, but not this complete power-loop compensation without analysis.
Q1 D/S/G and gate clampCompleted — passQ1 is now IXTH75N10L2. Manufacturer pinout is 1=Gate, 2=Drain, 3=Source, tab=Drain. Flux terminal/pad UIDs and pad numbers were checked before replacement, then Q1 was explicitly wired as pin 1→GATE, pin 2→VRECT, pin 3→VSHUNT. D2 remains cathode at GATE/anode at VSHUNT and R9 remains the gate-source pull-down.
Kelvin shunt sensingCompleted — schematic pass; PCB realization unresolvedRSH1=50 mΩ/5 W lies from VSHUNT to AGND. U2 IN− and the OC sense branch originate at VSHUNT. Logically correct, but a two-terminal generic resistor cannot guarantee true Kelvin copper; separate force/sense routing must be enforced in PCB layout.
TL431 biasCompleted — passU5 DBZ mapping K=1, REF=2, A=3 is correct. K and REF are tied, A is at AGND, and R11=6.2 kΩ from +15 V supplies about 2.017 mA. This exceeds the listed maximum minimum-regulation current of 0.7 mA, although that extracted numeric row was not machine-verified.
6.5 A LM393 thresholdCompleted — passR12=66.5 kΩ and R13=10 kΩ divide 2.495 V to 0.32614 V. Across 50 mΩ this is 6.523 A. R14/C6 provide a 10 µs input filter; C5 filters the threshold.
LM393 polarity and pull-upCompleted — passComparator A pin 2 is inverting and pin 3 is non-inverting. OC_THRESHOLD is on pin 2 and VSHUNT_OC on pin 3, so overcurrent releases the open-collector output. R15=10 kΩ pulls OC_TRIP to +15 V, which forward-drives the fault-OR diode into FAULT_SET. Unused comparator B inputs are defined (AGND and REF_2V495), and its output is marked no-connect.
CD4013 set/reset and unused halfCompleted — passSET and RESET are asynchronous active-high. FAULT_SET drives SET1; RESET_ENABLE drives RESET1; CLOCK1 and D1 are grounded. Q1 is FAULT_LATCH and ~Q1 is ENABLED. All four unused-half inputs SET2/D2/RESET2/CLOCK2 are grounded, as required for unused CMOS inputs. Unused Q2/~Q2 outputs are no-connect.
Startup default-off pulseCompleted — topology pass; startup margin unresolvedC7=100 nF from +15 V to FAULT_SET and R17=100 kΩ to AGND create an approximately 10 ms active-high SET pulse on rail rise. The topology defaults the latch to fault/off, but operation across arbitrary +15 V ramp rates, brownouts and rail sequencing is not guaranteed without transient validation or a supervisor.
Thermostat fail-safeCompleted — passJ4.2=AGND; J4.1=THERM_FAULT. A normally closed thermostat holds THERM_FAULT low. Opening the contact lets R16=10 kΩ pull it to +15 V and D4 drives FAULT_SET high. This is fail-safe for open wire/contact, but not for a shorted thermostat/wire.
Q2 clamp polarityCompleted — passMMBT3904 pinout is B=1, E=2, C=3. E is at VSHUNT, C at GATE, and FAULT_LATCH drives B through 4.7 kΩ with 100 kΩ B-E pull-down. A latched fault therefore clamps gate to source.
Connector assignmentsCompleted — passJ1: 1=fused AC leg, 2=AC return. J2: 1=+15 V, 2=AGND, 3=−15 V. J3: 1=0–3.3 V command, 2=AGND. J4: 1=thermostat sense, 2=AGND. J5: 1=FAULT_LATCH, 2=ENABLED, 3=AGND. J6: 1=+15 V, 2=RESET_ENABLE through an external momentary contact.
DecouplingCompleted — passAD633 and OPA197 each have 100 nF from +15 V to AGND and from AGND to −15 V. LM393 and CD4013 each have 100 nF +15 V decoupling. ±15 V rail entry has 22 µF bulk per rail. TL431/protection area has local 100 nF rail bypassing.
Fresh ERC and legibility checksCompleted — passPost-replacement review at 2026-09-11T06:25:18.285Z returned no WARNING, ERROR, UNCERTAIN, OPEN or USER findings for the selected Electrical Rule Checks and Schematic Legibility Checks. No violations required remediation.
Mutations and blocked corrections
  • Electrical/connectivity fixes: Q1 was replaced and its three terminals were explicitly reconnected by physical pin number.
  • Q1 part correction: completed with IXTH75N10L2; IRFP250N was rejected because its reviewed datasheet lacks guaranteed DC linear-mode SOA evidence. The replacement is explicitly designed for linear operation and provides forward-bias SOA through DC.
  • PCB work/cosmetic rearrangement: unattempted by instruction.
Unresolved engineering risks
  1. MOSFET SOA and thermal validation: IXTH75N10L2 provides DC FBSOA data, but the actual near-100 W operating trajectory, case temperature, heatsink, airflow, and transient line waveform still require measured validation.
  2. Control-loop stability: R10/C3 and gate resistances are placeholders. No full-loop AC/transient model was available to prove phase margin over rectified-line voltage, MOSFET capacitance, heatsink temperature and wiring parasitics.
  3. Startup/brownout: The passive 10 ms SET pulse may fail with slow or non-monotonic auxiliary-supply ramps, simultaneous RESET assertion, or brief brownouts. A reset supervisor or tested Schmitt-trigger POR remains preferable.
  4. Kelvin implementation: The schematic cannot enforce separate force/sense copper on the two-terminal generic shunt.
  5. MOV margin: A 30 VAC MOV is appropriate for nominal isolated 24 VAC, but transformer regulation/high-line conditions must be checked so steady-state secondary voltage does not approach or exceed its continuous rating.
  6. Generic/placeholder parts: RSH1 is generic and its pulse/temperature coefficient/Kelvin construction are unspecified; C15/C16 are generic bulk capacitors; thermostat type, trip temperature and contact rating are external/unspecified.
  7. Command/source sequencing: The design assumes an externally isolated command source referenced to AGND and an independent isolated ±15 V supply. Behavior when command is present before the auxiliary rails was not transient-tested.
  • Overall result

  • Requested-item status

  • Mutations and blocked corrections

  • Unresolved engineering risks