Receiving a new batch of network transformer samples? Here's the complete inspection procedure — instruments, settings, pass/fail criteria, and automation ideas.
Equipment Checklist
Required:
[x] LCR meter — must support 100kHz test frequency
Budget option: Tonghui TH2816B (~$300)
Mid-range: Hioki IM3536 (~$800)
High-end: Keysight E4980A (~$3500)

[x] 4-wire milliohm meter (or LCR meter with DCR mode)

Recommended:
[~] Hipot tester — for isolation voltage verification
[~] VNA — for insertion loss / return loss (NanoVNA sufficient for 100BASE-TX)

Optional:
[~] Calipers — for dimensional / footprint verification
[~] Test PCB — transformer footprint + 100Ω termination + SMA ports
OCL Measurement (Most Critical Parameter)
python# Measurement parameters — DO NOT deviate from these values
OCL_TEST_CONFIG = {
"frequency_Hz": 100_000, # 100 kHz
"signal_V_rms": 0.1, # 0.1 V RMS — matches actual signal level
"mode": "L", # Inductance
"secondary": "open", # Secondary winding pins: floating / not connected
}

Pass / fail criteria (design to Min, not Typ)

OCL_PASS_CRITERIA = {
"10/100BASE-TX": 350e-6, # 350 µH (IEEE 802.3 Clause 40)
"1000BASE-T": 1000e-6, # 1000 µH (IEEE 802.3 Clause 25/28)
}

Common supplier trap:

Some datasheets specify OCL at 1kHz or 1V RMS → inflated reading

Always confirm: "What is your OCL test frequency and voltage?"

DCR Measurement
Winding Pins Typical Range Flag If
Primary TX+ → TX− 1–5 Ω > datasheet max
Secondary RX+ → RX− 1–5 Ω > datasheet max
CT split TX+ → CT_TX ~0.5–2.5 Ω > 60% of full DCR

Use 4-wire (Kelvin) measurement to eliminate probe contact resistance.
High DCR + high OCL = extra turns → inflated inductance, worse HF performance.
Isolation Voltage (Hipot) Test
Test voltage: 1500V AC (standard IEEE 802.3)
2500V AC (industrial / IEC 61000-4 environments)
Duration: 60 seconds
Pass criterion: No breakdown; leakage < 1 mA
Connections:
HV terminal → all primary pins shorted together
GND terminal → all secondary pins shorted together

⚠️ Safety: proper hipot enclosure + PPE + trained operator required
⚠️ Never probe hipot with standard DMM leads
Visual Inspection Checklist
bash# Per-unit checks (100% for samples ≤20 pcs; AQL II sampling for production)
grep -E "PASS|FAIL" <<'CHECKLIST'
[ ] Part number on component matches purchase order
[ ] Pin count matches datasheet
[ ] Pin pitch within ±0.05mm of datasheet (verify with calipers)
[ ] No cracked ferrite, no damaged leads/pads
[ ] Pin 1 marker clearly visible and consistent across lot
[ ] Date code / lot code recorded for traceability
CHECKLIST
Functional Test
Test PCB setup:
PHY chip (e.g., LAN8720A, DP83848) → Transformer → RJ45
Bob Smith: 75Ω + 1000pF to chassis GND on each center tap

Pass criteria:
✓ Link establishes at 100Mbps (or negotiated speed)
✓ Zero packet loss on 5-minute ping test

✓ Link stable on 50m cable (stress test for return loss)
✓ Link stable at +70°C ambient (soak test for OCL at temperature)

If functional test fails but OCL/DCR pass:
→ Check Bob Smith termination ground (chassis GND, not signal GND)
→ Check center tap pin connections in schematic
→ Verify footprint pin-out against this supplier's datasheet
Incoming Inspection Decision Matrix
Application OCL DCR Hipot Functional
────────────────────────────────────────────────────────
Eng. sample ALL ALL — 1 unit
Consumer (AQL) AQL AQL — Sampling
Industrial AQL AQL ALL ALL
PoE design AQL ALL AQL ALL
Supplier Data to Request
Before placing an order, ask:

Lot-specific OCL test data (what frequency, what voltage?)
Isolation voltage production test report (what duration?)
Datasheet with Min / Typ / Max columns clearly labeled

Voohu Technology (www.voohuele.com) provides lot test data with OCL at 100kHz/0.1V RMS. MOQ 50pcs, DHL 3–5 days.