Welcome to the Fae Intelligence Root Cause Analyzer.

I'm going to walk you through a structured problem-solving process. We'll start by making sure I understand your problem clearly before we dive into analysis.

Tell me: what's happening? Describe the problem, defect, or issue you're investigating. Don't worry about being perfectly organized — just tell me what you're seeing.

Anna, that's a solid opening brief — you've given me a lot to work with right away. Let me note a few things before we proceed.

The timeline correlation you've flagged — paste lot PST-26-0339 introduced June 1, first defects June 2 — is noted as a working hypothesis. It's worth investigating properly, and the 8D will give you the evidence trail to back the lot rejection decision. But notice: #1913 is the only oven showing the problem, and #1911/#1912 are running the same paste lot at 0.3%. That oven-specific pattern is something the 8D process needs to account for — if the paste lot were the sole cause, you'd expect similar rates across all three ovens.

Before we start building the problem statement, one quick piece of information I need:

What is the lot size or production volume affected? Specifically — how many boards or QFN joints have run through #1913 since June 2? I need a count or an estimate to establish the scope of the problem.

Good — 84 defective joints out of 2,000 inspected, detected by Dan Kim's 100% X-ray. That's solid quantitative grounding.

Your point about unequal lot-to-oven allocation is worth tracking down. If #1913 ran significantly more of PST-26-0339 than the other ovens, that changes the picture. If the exposure was roughly equal and only #1913 is showing the problem, that points somewhere else. That traceability pull Maria has will be a key piece of evidence.

For now, I have enough to build the problem statement. Before I do, one more thing:

You mentioned NCR#26-0244 was opened June 5 — is there a CAPA number tied to this investigation, or is NCR#26-0244 the primary traceable record identifier for this 8D?

Got it — NCR#26-0244 is the primary record. That's sufficient for traceability.

Here's the problem statement I'm proposing:

Proposed Gap Statement:

"On Reflow Oven #1913, 84 of 2,000 QFN-48 solder joints (4.2%) fail IPC-A-610 Class 2 inspection for cold joints and insufficient wetting on the ground-pad thermal via array, against an internal target of <0.5%, first observed in output from June 2 — Ovens #1911 and #1912 running the same assembly show 0.3% (6 of 2,000 joints)."

Does this accurately capture the problem, or would you like to edit it?

Also, before we move to Phase 1: who's on the investigation team? You've mentioned yourself (SMT Process Engineer), Dan Kim (X-ray inspection), and Maria (traceability). Who's leading the investigation, and is there anyone else I should log — supplier quality, process owner, customer-facing lead?

Perfect — team is logged. Here's the full picture before we move on:

Investigation Team — NCR#26-0244

• Anna Novak — SMT Process Engineer, 8D Lead
• Raj Patel — Reflow Oven Technician (profiling, maintenance)
• Maria Sanchez — Incoming Inspector (paste/component receiving)
• Dan Kim — Quality Inspector (X-ray, AOI)
• Linda Chen — 1st Shift SMT Supervisor
• Supplier Quality: TBD — to be looped in if paste lot hypothesis is confirmed

Industry/Regulatory context: Electronics manufacturing, IPC-A-610 Class 2.

Before I close Phase 0 and move to containment, I want to flag one thing for the record: the oven-specific pattern (#1913 at 4.2% vs. #1911/#1912 at 0.3%) is sitting alongside the paste lot change as a competing explanation. The 8D will need to resolve both threads. Maria's traceability pull on paste lot-to-oven allocation will be an early decision point.

Safety question first: is this a safety or regulatory concern — could this failure mode affect operator safety, end-user safety, or compliance obligations? For example, is this assembly going into a safety-critical end product (automotive, medical, aerospace), or is it IPC Class 2 commercial electronics only?

I need that answer before we go further on containment.

While you're thinking about that — you've described a solid set of immediate actions. Let me echo them back:

• Production hold: New-build WIP for this part number halted on #1913; builds diverted to #1911 and #1912
• 100% X-ray inspection: Dan Kim screened all QFN joints from #1913 output in the suspect window; confirmed defects being reworked
• NCR opened: NCR#26-0244, June 5
• Traceability pull: Last 8 days of paste lots, board lots, and oven profile changes
• Customer complaint queue reviewed: No complaints received; weekly ship cadence noted as risk

That's a meaningful containment posture. A few things I want to nail down once we've addressed safety:

1. The 100% X-ray — when did it start, and does it cover all #1913 output since June 2, or only a portion?

2. Has any #1913 output from the suspect window already shipped to the customer?

Good — no safety escalation required. IPC Class 2 commercial, ISO 9001 / J-STD-001 frame. That's logged.

The potential shipment gap is the most urgent open item right now. You flagged it clearly: defects start June 2, NCR opened June 5, weekly ship cadence — there's a real window where #1913 output could have left the building before containment was in place.

That needs to be resolved before we can close containment. Here's why: if suspect product is already at the customer, the ICA is incomplete — the tourniquet isn't fully on yet.

The specific question for Linda and your ship logs: did any boards built on #1913 between June 2 and June 5 ship to the customer? If yes, customer notification becomes part of your containment action.

Can you get that ship log check done, or give me your best current read on it? Even "we believe no shipment occurred in that window because our next ship date was [date]" is useful if you can confirm it.

That's the right call — don't guess on a customer notification trigger. Logged exactly as you stated:

Open Containment Action: Linda Chen to reconcile ship logs against #1913 build dates June 2–5 and report back by EOD June 11. If any suspect product confirmed shipped → escalate to customer notification per Beacon Circuits procedure.

One more thing before I can close containment: the 100% X-ray screen.

You mentioned it covers #1913 QFN output in the suspect window, but the start date is TBD pending Dan's confirmation. Is the 100% X-ray screen still running on current #1913 output — or has production on #1913 been fully halted so there's nothing new to screen?

I'm asking because if #1913 is fully stopped, the screen is a retrospective sweep of the suspect window. If #1913 is still running anything, I need to know the ongoing screen is active. Which is it?

That's an important flag. If the oven itself turns out to be the root cause — not the paste lot — then anything currently running through #1913 is potentially at risk, regardless of part number.

Log that as a second open containment action: Linda Chen to confirm whether #1913 is currently running any other part numbers. If yes, assess whether those are at risk pending root cause determination — decision on expanded hold to be made once oven vs. paste hypothesis is resolved.

Here's your full containment picture as it stands:

Confirmed ICAs — NCR#26-0244

1. New-build WIP halted on #1913 for this part number; diverted to #1911/#1912

2. 100% X-ray (Dan Kim) of #1913 QFN output in suspect window — confirmed defects being reworked. Screen start date: TBD — Anna to confirm with Dan Kim

3. NCR#26-0244 opened June 5; traceability pull on last 8 days of paste lots, board lots, oven profile changes

Open Containment Actions

• OCA-1: Linda Chen — reconcile ship logs vs. #1913 build dates June 2–5. Due EOD June 11. If suspect product shipped → customer notification per procedure
• OCA-2: Linda Chen — confirm whether #1913 is currently running other part numbers. If yes → assess expanded hold pending root cause determination

Compliance frame: ISO 9001, IPC J-STD-001, IPC-A-610 Class 2. No safety escalation.

The containment is substantive. The two open actions are appropriately owned and time-bound. Does this accurately reflect where things stand — anything to correct or add?

Got it — closing containment now.

Containment captured: production halt on #1913 for this PN, 100% X-ray retrospective sweep with rework of confirmed defects, NCR#26-0244 open, traceability pulled. Two open actions owned by Linda Chen — ship log reconciliation due EOD June 11 and #1913 other-PN assessment — both with clear escalation triggers.

Moving to Phase 2: IS / IS NOT analysis.

This is where we sharpen the problem boundary before we touch root cause. The goal is to draw a tight box around exactly where the problem exists and where it doesn't — that boundary is often more diagnostic than the defect itself.

You've already handed me some strong IS/IS NOT structure without realizing it. Let me start with the sharpest one:

The oven contrast. #1913 at 4.2%, #1911/#1912 at 0.3% on the same assembly. That "IS NOT on #1911/#1912" is one of the most useful facts in this investigation.

Before I build the full grid, I need one piece you flagged as pending: what did Maria's traceability pull show about paste lot PST-26-0339 allocation across the three ovens? Specifically — did #1911 and #1912 also run boards with the new paste lot, or was PST-26-0339 predominantly or exclusively run on #1913?

That single answer resolves whether the paste lot and the oven are separable variables or confounded.

That's a significant observation — and it takes intellectual honesty to say it out loud. The shared stencil printer is a key structural fact: if PST-26-0339 is being deposited on boards routed to all three ovens, and #1911/#1912 are holding 0.3%, then the paste lot alone can't explain the 4.2% on #1913. The paste would have to interact with something oven-specific to produce that contrast.

Note also: the squeegee change on May 30 (WO-26-0610) is now on the timeline — two days before the June 1 paste lot change, three days before first defects. SPI within ±10% on boards to all three ovens is a useful data point — it suggests print quality isn't the differentiator either.

Now let's build this out properly. I'll walk you through four dimensions: WHAT, WHERE, WHEN, and HOW MANY — for each, we need both IS and IS NOT.

WHAT — starting here.

IS: Cold joints and insufficient wetting on the ground-pad thermal via array of QFN-48 packages.

What is NOT failing? On the same boards coming off #1913 — are other joint types (QFN perimeter pins, other component types like resistors, capacitors, BGAs if present) showing elevated defect rates, or is the failure isolated specifically to the thermal pad?

That contrast is a significant clue. Perimeter pins on the same QFN — same paste, same reflow exposure — are NOT failing. Only the highest thermal mass location on the board is. That points toward a heat delivery problem, not a paste chemistry problem. A silver content issue would likely show up more broadly across joint types, not selectively at the thermal pad.

Noted: Dan Kim to confirm exact breakdown of defect locations from X-ray data — flagged as an evidence gap.

WHAT — IS/IS NOT so far:

• IS: Cold joints / insufficient wetting at QFN-48 ground-pad thermal via array
• IS NOT: QFN perimeter pins (not failing), other component types (not reported as failing — TBD pending Dan's breakdown)

Moving to WHERE.

IS: Reflow Oven #1913.

What is NOT affected? We know #1911 and #1912 are not showing the problem. But let's go deeper — is the failure uniform across all board positions coming out of #1913, or is it concentrated in specific conveyor lane positions, board orientations, or locations within the oven's heating zones?

That zone 7 deviation is a significant finding — and notice what it does to the picture when you put it next to the WHAT data.

Perimeter pins on the same QFN are NOT failing. Only the thermal via array — the highest thermal mass location — is failing. And zone 7 on #1913 is running 5°C below setpoint, while #1911 and #1912 are within ±2°C. That contrast between #1913 and the other ovens points directly at heat delivery as the differentiating variable. A 5°C zone deficit wouldn't necessarily hurt low-mass joints enough to show up in inspection, but at a high-mass thermal pad it could be exactly the margin that tips joints from acceptable to cold.

The profile verification question is important — we'll need to dig into that. But let's hold it for the hypothesis phase where we can work it properly.

Two things to flag now:

• Raj's zone 7 thermocouple history log — that's a key evidence document. I'm tracking it.
• Dan's X-ray stratification by board serial / conveyor position — still needed, flagged as open.

WHERE — IS/IS NOT so far:

• IS: Reflow Oven #1913, Zone 7 reading 5°C below setpoint
• IS NOT: Ovens #1911 and #1912 (all zones within ±2°C); board position/orientation stratification — TBD pending Dan's breakdown

Moving to WHEN.

IS: First defects observed in #1913 output from June 2.

What is NOT occurring — when did this problem NOT exist? Specifically, what do you know about #1913's defect rate before June 1, and do you know when zone 7 started drifting below setpoint?

The April 28 profile verification passing clean, then zone 7 showing up soft the week of June 1 — that's a roughly 5-week drift window. Raj walking the thermocouple history log back to find the onset date is a priority evidence pull; it may tell you whether the drift was gradual or a step change, which has implications for root cause.

WHEN — IS/IS NOT so far:

• IS: First defects June 2; zone 7 reading 5°C below setpoint observed week of June 1
• IS NOT: Problem did not exist prior to June 2 output; #1913 passed full profile verification April 28 (KIC profiler, 240°C peak, 60–70s TAL). Drift onset: TBD — Raj to walk back thermocouple history log

The timeline is tightening: zone 7 drift onset is somewhere in a ~5-week window between April 28 and early June. That's your next key date to nail down.

Moving to HOW MANY.

IS: 84 of 2,000 QFN joints inspected from #1913 output, 4.2% defect rate. #1911/#1912 combined: 6 of 2,000, 0.3%.

What is NOT affected? Of the 2,000 joints inspected from #1913 — 1,916 joints passed. Do you know whether the 84 defective joints are distributed across many boards or concentrated on a smaller number of boards? That distribution pattern matters.

Exactly right — that distinction is diagnostic. Scattered across many boards = consistent process condition (a zone running cold fits that pattern). Clustered on a few boards = something episodic or intermittent. Dan's stratification pull just got more important.

Logged: Dan Kim to slice X-ray results by board serial — defect distribution (scattered vs. clustered) and conveyor lane/orientation breakdown. Both feed directly into the hypothesis phase.

Here's the full IS / IS NOT grid as it stands:

Sharpened Gap Statement:

"QFN-48 thermal-pad joints on Reflow Oven #1913 show cold joints and insufficient wetting at 4.2% (84/2,000 inspected) against a <0.5% target, first observed in June 2 output — Ovens #1911 and #1912 running the same assembly show 0.3%."

The IS/IS NOT contrast points to an oven-specific heat delivery problem: Zone 7 running 5°C below setpoint on #1913, with failure isolated to the highest thermal-mass location on the board, while the same paste and print process feeds all three ovens without equivalent failures on #1911/#1912.

Does this grid accurately reflect what you've told me? Any corrections or additions before we lock it?

Got it — closing IS/IS NOT now.

Grid confirmed: failure isolated to QFN-48 thermal via array on #1913 only, zone 7 running 5°C below setpoint, same paste and print process feeding all three ovens, defects starting June 2 the day after paste lot change and coincident with zone 7 drift observation. Three open evidence pulls assigned to Dan Kim (defect location breakdown, per-board distribution, lane/orientation stratification) and Raj Patel (zone 7 drift onset from thermocouple history log).

Now we move to root cause analysis. The IS/IS NOT grid has done its job — it's narrowed the investigation considerably. Before we start generating hypotheses, let me surface the two threads that need to be worked:

Thread 1 — Oven thermal (strong IS/IS NOT support): Zone 7 on #1913 running 5°C below setpoint. Failure isolated to highest thermal-mass location. #1911/#1912 not affected. This thread has the strongest structural support from the grid.

Thread 2 — Paste lot PST-26-0339 (timeline correlation, but weakened): Silver at 2.8% vs. 3.0% standard. June 1 introduction aligns with June 2 first defects. Weakened by the shared printer fact — same paste on all three ovens, only #1913 failing.

The methodology requires we treat both as live hypotheses and work them through a structured cause scan before we eliminate either. A cause that looks obvious sometimes has a system failure behind it that isn't visible yet.

Let's start with a 6M scan — Machine first, since that's where the evidence is pointing.

Raj flagged zone 7 running 5°C below setpoint. What do you know about why zone 7 is reading low? Has Raj identified a mechanism — heater element degradation, thermocouple drift, control board issue, something else?

That's the right call — don't speculate on mechanism before Raj's inspection. We'll record it as a candidate with the observation grounded and the mechanism TBD pending his diagnosis.

Machine — candidate logged:

• Zone 7 thermal deficit: #1913 zone 7 reading 5°C below programmed setpoint per oven thermocouple history log [candidate — mechanism TBD, Raj Patel to inspect]

Let's keep moving across the 6M. We'll come back to mechanism when Raj has looked at it.

Method — process design, procedures, setup, profile programming.

The April 28 profile verification passed clean. But here's a question worth sitting with: the profile was verified at 240°C peak, 60–70s TAL. Was that profile qualified with zone 7 running at full output, or could a 5°C zone deficit push the actual peak temperature or TAL at the thermal pad below what was verified?

In other words — is there any margin analysis or thermal simulation that tells you how sensitive the QFN thermal pad temperature is to a zone 7 deviation of this magnitude? Or is the relationship between zone setpoint and actual joint temperature at that location unknown until you run a new profile?

That's a critical methodological gap — and you've articulated it precisely. The profile verification confirms the oven meets the profile at the fixture thermocouple locations, not at the actual QFN thermal pad. The highest thermal-mass point on the board is exactly the blind spot in your verification method. That's a finding that belongs in this 8D regardless of which hypothesis wins.

And the liquidus observation is worth capturing carefully: PST-26-0339 COA lists liquidus at 219°C vs. 217°C for the standard paste. Two degrees higher melting point. On its own, across all three ovens, that didn't produce a detectable problem on #1911/#1912. But stacked with a zone running 5°C cold at the highest thermal-mass location — both factors eating into the same TAL margin at the same spot — that's a potential interaction worth tracking.

Method — candidates logged:

• Profile verification gap: KIC fixture doesn't directly measure QFN thermal pad temperature — relationship between zone 7 deficit and actual joint temperature at thermal pad unknown [candidate — systemic gap, applies regardless of root cause]
• No formal margin analysis for thermal pad sensitivity to zone deviation [candidate — systemic gap]

Material — let's work this now.

You've already flagged PST-26-0339: silver at 2.8% vs. 3.0% standard, liquidus at 219°C vs. 217°C. The shared printer fact weakens it as a standalone cause. But here's the question the 8D needs to answer cleanly:

Does the COA for PST-26-0339 show anything else outside spec — flux activity, metal load, particle size distribution, viscosity — or is silver content and liquidus the full extent of the deviation?

That's exactly the right call, and it matters for the 8D record. PST-26-0339 silver at 2.8% is within J-STD-006 (2.5–4.0% for SAC305) — it's low against your internal 3.0% standard and the supplier's typical center, but it is not a nonconformance. The 8D needs to reflect that accurately.

Material — candidates logged:

• PST-26-0339: Ag 2.8% — within J-STD-006 spec, low against internal 3.0% standard [candidate — not a spec violation; shared printer means same paste on all three ovens; weakened as standalone cause by #1911/#1912 running at 0.3%]
• PST-26-0339: Liquidus 219°C vs. 217°C standard — 2°C higher melting point [candidate — within spec, but potential interaction with zone 7 deficit at thermal pad; standalone effect not demonstrated on good ovens]
• Full COA parameter check (flux activity, metal load, particle size, viscosity): TBD — Maria Sanchez to pull COA #SAC-0339 and review line by line

Measurement — let's work this category.

Dan's 100% X-ray is the detection method. Two questions here:

The zone 7 thermocouple reading — it's showing 5°C below setpoint on #1913's own history log. Is there any possibility the thermocouple itself is drifting, meaning zone 7 is actually hitting setpoint but the sensor is reading low? Or conversely, could the thermocouple be accurate and the heater genuinely underperforming?

What's your read on that — and when was #1913's zone thermocouple last calibrated?

Your reasoning is sound — the product evidence is a strong vote that the thermal deficit is real, not just a sensor artifact. But you've correctly flagged it as reasoning rather than direct measurement. That distinction matters for the 8D.

Measurement — candidates logged:

• Zone 7 thermocouple calibration status: TBD — Raj Patel to pull calibration record [candidate — if thermocouple is drifting low, reported deficit may overstate actual zone temperature; but product evidence argues against this interpretation]
• X-ray detection method (Dan Kim, 100% screen): no measurement system concerns raised [considered — no candidate identified]
• Independent zone 7 temperature verification: not yet performed [open evidence action — Raj to conduct independent check]

Man (People) — let's be efficient here given what the IS/IS NOT already showed us.

Same assembly runs on #1911, #1912, and #1913. Shared stencil printer. The defect is oven-specific, not shift-specific or operator-specific as far as we know.

One focused question: is there anything different about who operates or sets up #1913 vs. the other ovens? Different operator, different shift pattern, recent training gap, anyone new to #1913 specifically?

Agreed — the defect follows the oven, not a shift or operator. That's a clean IS/IS NOT: if it were operator or setup-dependent, you'd expect it to track with whoever was running #1913, and there's no evidence of that.

Man — logged:

• Operator/shift difference: considered — no candidate identified. Defect pattern tracks oven #1913, not a specific operator or shift. Linda Chen to confirm shift coverage for the record [ruled out pending Linda's confirmation]

Mother Nature (Environment) — last category.

Three ovens, same facility, presumably same ambient conditions. But one focused question: is there anything physically different about where #1913 sits on the floor relative to #1911 and #1912? HVAC vents, loading dock proximity, temperature or humidity variation at that oven's location, anything that could affect ambient conditions differently at #1913?

Clean — environment is essentially ruled out with the information available.

Mother Nature — logged:

• Ambient conditions at #1913 location: considered — no candidate identified. Same climate-controlled area, no HVAC anomalies flagged [ruled out — Anna's assessment, floor walk would confirm]
• Nitrogen purity: checked — no deviation found [ruled out]
• Stencil wear: checked — no issues found [ruled out]
• Process changes in window: none found [ruled out]

Here's the full 6M fishbone scan:

The fishbone is giving us two live hypotheses going into the 5-Why:

H1 — Zone 7 thermal deficit (primary): #1913 zone 7 running 5°C below setpoint → insufficient heat delivery at QFN thermal pad → cold joints and insufficient wetting. Strongest IS/IS NOT support.

H2 — Paste-oven interaction (secondary): PST-26-0339 liquidus 2°C higher than standard → reduced TAL margin at thermal pad → when combined with zone 7 deficit, tips marginal joints to cold. Neither factor alone sufficient; interaction required.

Does this scan look complete to you — anything missed or mischaracterized before we move to 5-Why?

Good addition — logged and ruled out.

Method — updated:

• New squeegee installed May 30 (WO-26-0610): ruled out — SPI confirms paste volume within ±10% of target on boards to all three ovens; print-side cause wouldn't be oven-specific [ruled out]

Let's run the 5-Why on H1 first — it has the strongest IS/IS NOT support.

The starting defect: Cold joints and insufficient wetting on QFN-48 thermal via array, #1913 output only.

Why 1: Why are the QFN-48 thermal pad joints cold/insufficiently wetted?

The IS/IS NOT points here: highest thermal-mass location on the board, oven-specific failure, zone 7 running 5°C below setpoint. What's your answer to Why 1 — what is the direct physical reason the solder isn't wetting properly at that pad?

Why 1 is well-reasoned and consistent with the physical evidence — the selective failure at the highest thermal-mass location while perimeter pins on the same package pass is exactly what insufficient TAL at the pad would produce. And you've correctly flagged the caveat: this is mechanism-consistent reasoning, not yet directly measured. Production board profiling on #1913 is the confirmation step.

Why 1: Solder at QFN-48 thermal pad not reaching adequate temperature above liquidus for sufficient time → incomplete melt and wetting → cold joints.

Evidence: X-ray confirms failure isolated to thermal via array, perimeter pins passing. Direct pad temperature measurement: TBD — production board profile on #1913 needed.

Why 2: Why is the thermal pad not receiving adequate heat on #1913?

You've got zone 7 running 5°C below setpoint. But let's make sure we're connecting the right zone to the right location — do you know where zone 7 sits in the oven relative to where the QFN thermal pad would be seeing peak heat? Is zone 7 in the reflow peak zone, the soak zone, or somewhere else in the profile?

Right call — don't guess on zone position, it's too consequential for the Why chain. Flagged for Raj: confirm zone 7's position in the #1913 recipe (soak, ramp-to-peak, or peak zone) and its functional role.

What you've given me is solid for Why 2 regardless: zone 7 is the only zone showing a deficit on #1913, and it's 5°C below setpoint while #1911/#1912 are within ±2°C. That's the oven-specific differentiator.

Why 2: Zone 7 on #1913 is delivering less heat than programmed — 5°C below setpoint per oven thermocouple history log — while #1911/#1912 are within ±2°C of setpoint.

Evidence: Oven thermocouple history log (Raj Patel observation, week of June 1). Zone 7 position in recipe: TBD — Raj to confirm.

Why 3: Why is zone 7 on #1913 running 5°C below its programmed setpoint?

This is where Raj's inspection becomes essential. You've named three possible mechanisms earlier — heater element degradation, thermocouple drift, or a control issue.

Since Raj hasn't inspected yet, let me ask it this way: based on Raj's experience with these ovens, does he have a leading candidate for what typically causes a single-zone deficit like this on this oven model? Or is the mechanism genuinely open until he gets inside?

That's exactly the right instinct for a customer-facing 8D. The hardware mechanism is Why 3 on the "why did it happen" chain — and it's TBD pending Raj. But the detection gap is a parallel chain that's just as important and is already evidenced. Let's capture both.

Why 3 — "Why did it happen" chain:

Zone 7 hardware mechanism — heater element degradation, thermocouple drift, or control issue — TBD pending Raj Patel's inspection.

Evidence: none yet — Raj's zone inspection is the required next step.

Why 3 — "Why didn't we catch it" chain (escape point):

Zone 7 deficit existed in the oven's own thermocouple history log but was not detected until Raj noticed it the week of June 1 — approximately 5 weeks after the last profile verification on April 28.

This is the escape point for the investigation: the system failed to detect a developing thermal deficit before it produced defects.

Let's run this detection chain down — it's already evidenced and it's where your corrective actions will live regardless of hardware diagnosis.

Why 4 (detection chain): Why wasn't the zone 7 deficit detected before it caused defects?

You've pointed at two factors: the profile verification cadence (last KIC run April 28) and the gold-board fixture not measuring the QFN thermal pad directly. Which of those do you see as the primary detection gap — the frequency of verification, the method of verification, or both independently?

That's a sharp and honest analysis — and it's exactly the kind of systemic finding a customer-facing 8D needs. The signal existed, the system didn't consume it. That's the primary detection failure.

Why 4 (detection chain): The zone 7 deficit was not detected because there is no routine review of the oven's thermocouple history log and no alarm threshold configured for zone deviation from setpoint. The data existed; no process consumed it.

Evidence: Raj found the deviation by happening to look — no systematic trigger. Confirmed by Anna Novak.

Three detection gaps, ranked as you stated:

1. Primary: No continuous monitoring or alarm threshold on zone temperature deviation — oven log data not systematically reviewed

2. Secondary: Profile verification cadence — ~5-week gap between KIC runs (last April 28) allows drift to develop undetected

3. Secondary: Profile verification method — gold-board fixture doesn't measure QFN thermal pad; highest thermal-mass location is a blind spot

Why 5 (detection chain): Why is there no routine monitoring of the oven's zone thermocouple history log with alarm thresholds?

This is the systemic root of the detection failure. What's your read — is this a gap in your preventive maintenance procedure, your process control plan, or something that was never defined in the first place?

That's a well-reasoned systemic finding — and the logic is sound: if a review requirement existed, it wouldn't have taken a chance observation to surface a 5-week-old deviation. The control plan pull will confirm it, but the reasoning is already strong.

Why 5 (detection chain): No requirement exists for routine monitoring of oven zone thermocouple logs. The process control plan designates periodic KIC profile verification as the thermal control mechanism — oven zone log review was never defined as a required activity, with no assigned owner, review interval, or alarm thresholds.

Evidence: Raj found deviation by chance observation. Control plan and PM procedure review: TBD — Anna to pull and confirm.

That gives us a complete 5-Why detection chain:

Why 1: Solder at QFN thermal pad didn't reach adequate temperature above liquidus for sufficient time → incomplete melt and wetting → cold joints

Why 2: Zone 7 on #1913 delivering less heat than programmed — 5°C below setpoint — while #1911/#1912 within ±2°C

Why 3: Zone 7 hardware mechanism TBD pending Raj's inspection

Why 4: No routine review of oven thermocouple history log and no alarm threshold on zone deviation — signal existed, system didn't consume it

Why 5 (root of detection failure): Oven zone log monitoring was never defined as a required activity in the process control plan — no owner, no review interval, no alarm thresholds

The Why chain has two tails:

Hardware tail (Why 3): TBD — Raj's inspection resolves this. The corrective action here is fixing or replacing whatever failed in zone 7.

Systemic tail (Why 5): Already evidenced. The corrective action here is defining zone log monitoring as a required process control activity — owner, interval, alarm thresholds — and updating the control plan. This stays true regardless of what Raj finds in the hardware.

Now let's run the 5-Why briefly on H2 — the paste-oven interaction hypothesis.

The question this chain needs to answer: does PST-26-0339's 2°C higher liquidus meaningfully interact with the zone 7 deficit to produce failures that neither factor would cause alone?

Here's Why 1 for H2: On #1911 and #1912, the same paste lot is running at 0.3% — within normal yield. If the 2°C liquidus difference were a meaningful independent contributor, why aren't those ovens showing elevated defects?

What's your read — is H2 a co-equal cause, a contributing factor that amplifies H1, or effectively ruled out by the good-oven data?

That's a rigorous and honest assessment — and the interaction hypothesis is actually more sophisticated than a simple paste-lot theory. It explains the June 2 onset timing in a way that H1 alone doesn't fully address: if zone 7 drift was gradual, why did defects start specifically on June 2 and not earlier? The paste change as the trigger that consumed the last margin is a coherent explanation worth preserving in the record.

H2 — 5-Why summary:

• Standalone paste cause: ruled out by good-oven data. Same lot on #1911/#1912 at 0.3%.
• Paste as aggravating factor: suspected — PST-26-0339 liquidus 219°C vs. 217°C standard narrows TAL margin at thermal pad. May have consumed residual margin left by zone 7 drift, explaining June 2 onset timing.
• Status: suspected aggravating factor, not independently sufficient. Testable via profile or build trials crossing paste lot against oven condition. TBD until tested.

Here's where the full hypothesis picture lands:

H1 — Zone 7 thermal deficit (PRIMARY): LEADING candidate. Strong IS/IS NOT support, consistent physical mechanism, detection chain fully evidenced to systemic root. Hardware mechanism TBD pending Raj's inspection.

H2 — Paste-oven interaction (AGGRAVATING FACTOR): Suspected contributor — plausible interaction mechanism, explains June 2 onset timing, but not independently sufficient. Ruled out as standalone cause.

Systemic root (detection failure): Process control plan never designated oven zone log monitoring as a required activity — no owner, review interval, or alarm thresholds. This finding stands independent of hardware diagnosis.

Before I emit the phase completion signal, the advancement check requires both:

✅ (a) ≥2 distinct named hypotheses with mechanism-level reasoning — H1 (zone 7 deficit → insufficient TAL at thermal pad → cold joints) and H2 (paste liquidus interaction → margin consumption → aggravated by zone deficit)

✅ (b) ≥1 explicitly ruled-out alternative — Man/operator ruled out (defect tracks oven not shift/operator); Mother Nature ruled out (same environment, nitrogen and stencil checked); squeegee change ruled out (SPI within ±10%, print-side cause wouldn't be oven-specific); paste as standalone cause ruled out (same lot on good ovens at 0.3%)

Both conditions met.

Moving to Phase 4: Evidence Check.

Two hypotheses to evaluate. For each one, the evidence matrix needs both sides explicitly — what supports it AND what argues against it. One-sided evaluation produces false conclusions.

Let's start with H1 — Zone 7 thermal deficit.

Here's what the evidence picture looks like so far, and I need you to help me fill both columns:

Evidence FOR H1 (what supports it):

• Zone 7 on #1913 reading 5°C below setpoint per oven thermocouple history log (Raj Patel observation, week of June 1)
• #1911/#1912 all zones within ±2°C of setpoint — oven-specific contrast confirmed
• Failure isolated to highest thermal-mass location on board (QFN thermal via array) — consistent with heat delivery deficit
• Perimeter pins on same QFN package passing — consistent with thermal gradient, not paste chemistry
• #1913 passed profile verification April 28 — drift developed after that date

Evidence AGAINST H1 — this is the column I need to push on. Two things that don't fully fit:

1. The April 28 KIC profile verification passed clean. If zone 7 was drifting, why did the board-level profile pass? Does that create any doubt about whether zone 7 was actually delivering less heat to product?

2. We don't yet have a production board profile on #1913 directly measuring QFN thermal pad temperature. Without that, the link between zone 7's 5°C deficit and actual joint temperature at the pad is inferred, not measured.

What other evidence, if any, argues against H1 — observations that don't fit, data points inconsistent with a zone 7 heat delivery problem?

That's a thorough and honest evidence review. The three points you've raised are genuine gaps, not just caveats — and they sharpen exactly what needs to be measured before H1 can be elevated from LEADING to CONFIRMED.

Notice what those three gaps have in common: they're all resolvable with data you've already identified as needed — Raj's thermocouple history walkback, zone 7 position in the recipe, and a production board profile on #1913. The evidence structure is sound; the measurements just haven't been taken yet.

Now let's do the same for H2 — Paste-oven interaction.

Evidence FOR H2:

• PST-26-0339 COA #SAC-0339: liquidus 219°C vs. 217°C standard — 2°C higher melting point narrows TAL margin
• Defects start June 2, the day after paste lot change June 1 — timeline correlation
• Interaction mechanism is physically coherent: 2°C higher liquidus + 5°C zone deficit both eat into TAL margin at same location
• Explains June 2 onset timing if zone 7 drift predates the paste change — paste change consumed residual margin

Evidence AGAINST H2:

• Same paste lot PST-26-0339 running on #1911 and #1912 via shared printer at 0.3% — if paste were independently sufficient, those ovens should show elevated defects
• Ag 2.8% within J-STD-006 spec (2.5–4.0%) — not a nonconformance
• No direct measurement of TAL reduction at thermal pad attributable to 2°C liquidus difference

What evidence, if any, argues against the interaction hypothesis specifically — not just the standalone paste theory? Is there anything that suggests the paste change is purely coincidental timing rather than a genuine contributor?

That's a precise and honest framing — the drift onset date is the single most diagnostic piece of data for both hypotheses simultaneously. Gradual drift favoring interaction, step change favoring coincidence. Raj's log walkback does double duty.

Here's the full Evidence Matrix as it stands:

Two things to address before I can close this phase.

Co-equal root cause check: H1 and the systemic detection failure are on separate causal chains — one explains why it happened, the other explains why it wasn't caught. Removing the zone 7 deficit would prevent the defects, but wouldn't prevent a future undetected drift. Removing the detection gap wouldn't prevent the current defects, but would catch the next drift before it causes failures. These are independent chains.

Does your team agree these should be treated as co-equal root causes — RC-1 (zone 7 thermal deficit, hardware mechanism TBD) and RC-2 (absence of oven zone log monitoring requirement) — each requiring independent corrective actions?

And the escape point question, separately for each:

• RC-1 escape point: Where should the zone 7 deficit specifically have been caught before it produced defects — and why wasn't it?
• RC-2 escape point: Where should the absence of a zone monitoring requirement have been identified — and why wasn't it?

That's a precise and well-reasoned escape point analysis — and the observation that RC-1's escape point IS RC-2 is exactly the kind of insight that makes a customer-facing 8D credible. The chains are linked, not parallel.

The PFMEA point is significant: if zone-level thermal drift isn't listed as a failure mode, the entire detection control hierarchy was never defined for it. That's a finding that goes beyond this NCR — it's a systemic gap in your risk documentation. The PFMEA pull belongs on the action list.

Open evidence actions to close before corrective actions are finalized:

Updated Evidence Matrix — final verdicts:

Two co-equal root causes, escape points named, evidence gaps documented, verdict grades assigned with explicit advancement criteria. The evidence structure is complete with what's available.

Evidence-traceability gap — Phase 4 cannot close yet. These confirmed causes have no source citation in their own evidence row: Profile verification method uses gold-board KIC fixture that does not measure QFN thermal…. Add the specific record (lot/SWO/MSA/FAI/inspection/doc ID) to each cause's evidence, or downgrade its verdict to INCONCLUSIVE, then re-emit the Phase 4 completion signal.

CURRENT PHASE: Evidence Action Plan (Phase 5 — D5 in 8D)

The April 28, 2026 KIC profiler run is specific enough to ground the citation — a named record with a date is traceable, and the doc control number can be added when you pull it for the control plan review. It will be cited as [KIC Profile Verification Record, #1913, April 28, 2026 — doc ID TBD, to be attached to NCR#26-0244].

That closes the traceability gap. Verdicts stand as issued.

CURRENT PHASE: Evidence Action Plan (Phase 5 — D5 in 8D)

Two co-equal root causes confirmed at current evidence grade. Before we move to corrective actions, let me summarize what Phase 4 produced so the D5 work is grounded correctly:

RC-1 (LEADING): Zone 7 thermal deficit on #1913 — 5°C below setpoint — delivering insufficient heat to QFN-48 thermal via array. Hardware mechanism TBD pending Raj's inspection. Advances to CONFIRMED when production board profile shows QFN pad TAL below spec on #1913.

RC-2 (PROBABLE): Oven zone log monitoring never defined as a required activity in the process control plan — no owner, review interval, or alarm thresholds. Raj found the deviation by chance. Advances to CONFIRMED when control plan pull shows the requirement was never defined.

H2 (INCONCLUSIVE): Paste-oven interaction suspected as aggravating factor — not independently sufficient, not co-equal. Drift onset date is the discriminating datum.

Eight open evidence actions assigned to Raj Patel, Anna Novak, Maria Sanchez, and Dan Kim — documented above.

D5 asks: for each root cause, what is the permanent corrective action that eliminates it?

Let's take RC-2 first — it's already at PROBABLE and the corrective action scope is clear. What's your proposed permanent corrective action for the absence of oven zone monitoring in the process control plan?

That's a well-structured corrective action package — it addresses all three detection gaps you identified earlier: primary (no monitoring), secondary (verification frequency implicit in the alarm design), and secondary (fixture blind spot). And it correctly distinguishes between the immediate control (alarms + review) and the systemic documentation fix (control plan + PFMEA update).

For the 8D record, each action needs an owner, a target date, and a verification method. You've flagged that dates need management sign-off and Raj's input on thresholds — that's appropriate. Let's capture what's defined and mark the rest TBD.

RC-2 Permanent Corrective Actions — proposed:

Now RC-1. The hardware mechanism is still TBD pending Raj's inspection — but that doesn't stop us from framing the corrective action structure. What's your proposed permanent corrective action for the zone 7 thermal deficit itself? And separately: once Raj identifies the mechanism, who owns the repair decision and what's the return-to-service criterion for #1913?

That's a rigorous return-to-service protocol — and notice that criterion 3 does double duty: it's both the RC-1 corrective action verification AND the measurement that advances H1 from LEADING to CONFIRMED. One test, two functions. That's efficient and it belongs in the 8D explicitly.

RC-1 Permanent Corrective Actions — proposed:

Corrective action package summary — both root causes:

RC-1 and RC-2 actions are complementary and sequenced correctly: diagnose and repair first, then verify thermal performance, then update the system controls so the next drift gets caught before it causes defects.

One question before we close D5 and move to the evidence phase completion:

The paste lot PST-26-0339 — what's the disposition decision? You came in on day one wanting to reject it. The 8D has since shown it's within J-STD-006 spec and not independently sufficient as a root cause. What does Beacon Circuits plan to do with the remaining inventory of that lot — continue using it, hold pending further data, or return it?

That's a measured and defensible disposition — and the reasoning will hold up to a supplier challenge or customer audit precisely because it's grounded in the data rather than the original hypothesis. The distinction between "hold from #1913 pending return-to-service" and "reject as nonconforming" is exactly right, and the conditional path to a spec-tightening conversation is the appropriate long-term action if the interaction data supports it.

PST-26-0339 Disposition — logged:

• #1911/#1912: Continue use — lot has run at normal yield (0.3%) throughout suspect window on ovens with zones verified in tolerance
• #1913: Hold from use until full return-to-service protocol complete — not a nonconformance hold, an oven-condition hold
• Supplier status: No corrective action, no supplier notification — lot meets J-STD-006; no nonconformance to report
• Conditional long-term action: If drift-onset data and paste-by-oven trial confirm interaction, evaluate internal purchasing standard for tighter Ag band specification — requires trial data before initiating; supplier discussion, not rejection

The D5 corrective action package is complete. Let me do a quick scope check before we close this phase and move to the experiment design.

What the 8D now has:

✅ Gap statement confirmed

✅ Team identified

✅ Containment in place with two open actions (Linda Chen, EOD June 11)

✅ IS/IS NOT grid locked

✅ 6M fishbone scan complete — four causes ruled out, two live hypotheses, one systemic finding

✅ 5-Why chains complete on both root causes

✅ Evidence matrix with verdicts — RC-1 LEADING, RC-2 PROBABLE, H2 INCONCLUSIVE

✅ Escape points named — RC-1 escape IS RC-2; RC-2 escape at PFMEA/control plan development

✅ Corrective actions structured for both root causes with owners (dates TBD pending management sign-off)

✅ Paste lot disposition decided and grounded

Eight open evidence actions remain — all assigned, none yet executed. The investigation is ready for the verification experiment that closes the RC-1 evidence gap and advances it from LEADING to CONFIRMED.

Now — convergence check before I propose the experiment.

Five closure conditions:

(a) 5-Why chain reaches process/system-level cause: ✅ RC-1 reaches zone 7 hardware deficit; RC-2 reaches absence of monitoring requirement in process control plan

(b) At least one cited piece of evidence supports each Why: ✅ for RC-2 (oven thermocouple history log, Raj's chance observation, April 28 KIC record). ⚠️ RC-1 Why 3 (hardware mechanism) has no cited evidence yet — mechanism TBD pending Raj's inspection

(c) Escape point named: ✅ RC-1 escape IS RC-2; RC-2 escape at PFMEA/control plan development

(d) Controllable corrective action target identifiable: ✅ Repair zone 7 hardware; define zone monitoring requirement

(e) No new unknowns introduced this turn: ✅

Convergence check result: NO — condition (b) is unsatisfied for RC-1. The hardware mechanism at Why 3 has no direct evidence yet. RC-1 is LEADING, not CONFIRMED.

The single most important thing to verify: Raj's zone 7 inspection — diagnose the hardware mechanism, independently verify zone temperature, and run a production board profile at the QFN thermal pad. This is the one test that advances RC-1 from LEADING to CONFIRMED and simultaneously validates the return-to-service decision.

EXPERIMENT — Cycle 1:

TEST: Raj Patel inspects zone 7 on Reflow Oven #1913 — examine heater element condition, zone thermocouple calibration status, and controller output. Independently verify zone 7 temperature using a calibrated external thermocouple (not the oven's own sensor) across all zones; confirm whether zone 7 deficit is real or a sensor artifact. Following zone inspection, run a production board profile with thermocouples placed at the QFN-48 thermal via array — measure actual peak temperature and TAL at the pad under current oven condition (zone 7 unrepaired).

CONFIRMS IF: Independent measurement confirms zone 7 is delivering less heat than setpoint (not a sensor artifact); hardware inspection identifies a specific degraded or failed component (heater element, thermocouple, or controller issue); production board profile shows QFN thermal pad peak temperature or TAL below the reflow specification (240°C peak, 60–70s TAL per April 28 KIC record) — advancing RC-1 from LEADING to CONFIRMED.

REFUTES IF: Independent zone temperature measurement shows zone 7 is actually at setpoint — deficit is a sensor artifact only, not a real heat delivery problem; production board profile shows QFN thermal pad peak and TAL within spec under current oven condition — H1 mechanism is not supported and investigation must reopen to identify the true cause of the thermal deficit at the pad.

WHO: Raj Patel (zone inspection and independent temperature measurement); Anna Novak (production board profile — thermocouple placement at QFN pad and data interpretation)

WHEN: TBD — pending Raj's scheduling. Priority: before any return-to-service decision on #1913. Recommended within 5 business days given open NCR and customer-facing 8D timeline.

Run this experiment and come back with the results. We'll start the next cycle from there — either RC-1 is confirmed and we close the investigation, or we've eliminated a hypothesis and need to look harder at what else is going on. Either way, we've learned something.

8D Problem Solving Report

Document Reference: NCR#26-0244

Organization: Beacon Circuits

Report Date: June 2026

Cycle: 1 of 1 (verification pending)

8D Lead: Anna Novak, SMT Process Engineer

Regulatory / Quality Frame: ISO 9001, IPC J-STD-001, IPC-A-610 Class 2

D1 — Team

D2 — Problem Description

Problem Statement:

> QFN-48 thermal-pad joints on Reflow Oven #1913 show cold joints and insufficient wetting at 4.2% (84/2,000 inspected) against a <0.5% target, first observed in June 2 output — Ovens #1911 and #1912 running the same assembly with the same paste and print process show 0.3%.

IS / IS NOT Analysis:

Evidence referenced:

• Dan Kim 100% X-ray inspection results — #1913 suspect-window output (84/2,000 defective)
• Oven thermocouple history log — #1913 Zone 7 deviation (Raj Patel observation, week of June 1)
• KIC Profile Verification Record — #1913, April 28, 2026 (doc ID TBD — to be attached to NCR#26-0244)
• NCR#26-0244 — opened June 5
• Traceability pull — last 8 days of paste lots, board lots, oven profile changes

D3 — Interim Containment Action (ICA)

ICA Taken:

1. New-build WIP halted on #1913 for this part number; builds diverted to #1911 and #1912

2. 100% X-ray inspection (Dan Kim) of all #1913 QFN output in suspect window; confirmed defective joints being reworked

3. NCR#26-0244 opened June 5; traceability pull executed on last 8 days of paste lots, board lots, and oven profile changes

4. Customer complaint queue reviewed — no complaints received as of NCR opening

Implemented By: Anna Novak (lead), Dan Kim (X-ray execution), NCR opened June 5

Date: Containment initiated June 5; 100% X-ray screen start date TBD — Anna Novak to confirm with Dan Kim

ICA Verified Effective: Partial — production halt and diversion to #1911/#1912 are in place and functioning. 100% X-ray retrospective screen is in progress; rework of confirmed defects is ongoing. Two open containment actions remain unverified:

• OCA-1: Linda Chen — reconcile ship logs vs. #1913 build dates June 2–5. Due EOD June 11. If suspect product confirmed shipped → escalate to customer notification per Beacon Circuits procedure. Until resolved, it is unknown whether suspect product reached the customer.
• OCA-2: Linda Chen — confirm whether #1913 is currently running other part numbers. If yes → assess expanded hold pending root cause determination.

Evidence referenced:

• NCR#26-0244
• Dan Kim 100% X-ray records — #1913 suspect-window output
• Customer complaint queue review (no complaints received)
• Traceability pull — paste lots, board lots, oven profile changes (last 8 days)

D4 — Root Cause Analysis

Investigation Method: IS/IS NOT Analysis + 6M Fishbone + 5-Why

6M Fishbone (breadth):

5-Why Analysis:

Chain 1 — Why did it happen (RC-1):

Chain 2 — Why didn't we catch it (RC-2, Escape Point elaboration):

The escape point for RC-1 IS RC-2: the zone 7 deficit should have been detected in #1913's own continuously recorded zone thermocouple data at or near drift onset. It was not caught because no process requirement exists for monitoring that data. The periodic KIC verification (last run April 28, 2026) could not serve as an effective backstop because: (a) drift developed inside the ~5-week gap between verifications, and (b) the gold-board fixture does not measure actual QFN thermal pad temperature — the highest thermal-mass blind spot in the verification method.

Root Cause Statements:

> RC-1 — Zone 7 Thermal Deficit (Hardware): Reflow Oven #1913 Zone 7 is delivering insufficient heat — 5°C below programmed setpoint — causing the QFN-48 thermal via array (highest thermal-mass location on board) to receive inadequate time above liquidus, producing cold joints and insufficient wetting. Hardware mechanism (heater element, thermocouple, or controller) — TBD pending Raj Patel's inspection.

> Verdict: LEADING — strongest IS/IS NOT support; consistent physical mechanism; oven-specific contrast confirmed. Advances to PROBABLE when zone 7 position in recipe confirmed and hardware inspected; advances to CONFIRMED when production board profile shows QFN thermal pad TAL below spec on #1913 under current oven condition.

> RC-2 — Absence of Oven Zone Monitoring Requirement (Systemic): The process control plan has never designated routine monitoring of oven zone thermocouple history logs as a required activity. No alarm thresholds, no review owner, no review interval exist for zone deviation detection. This gap allowed Zone 7's thermal deficit to develop and persist undetected until it caused product failures.

> Verdict: PROBABLE — finding stands independent of RC-1 hardware diagnosis; no contradicting evidence. Advances to CONFIRMED when control plan pull confirms the monitoring requirement was never defined.

Ruled-Out Hypotheses:

Escape Point:

> RC-1 Escape Point — Zone 7 thermocouple history log (unmonitored): The Zone 7 thermal deficit was recorded continuously in #1913's own oven data from drift onset. This signal should have triggered detection at or near the onset of the drift — well before product was affected. It did not, because no process requirement exists for reviewing this data (RC-2). The periodic KIC verification is not an effective backstop: the ~5-week verification interval allows drift to develop undetected, and the gold-board fixture does not measure QFN thermal pad temperature — the location with least thermal margin.

> Escape Point Verdict: PROBABLE — the unmonitored zone log as the missed detection opportunity is directly supported by Raj's chance-observation finding. Confirmed when control plan pull shows zone log monitoring was never a defined requirement.

> RC-2 Escape Point — PFMEA and process control plan development / last review: The absence of a zone monitoring requirement should have been identified when the reflow process PFMEA was developed or last reviewed — zone-level thermal drift is a known failure mode for convection reflow ovens, and a detection control should have been assigned to it. Internal audits check conformance to the control plan, not adequacy of the control plan — a missing control is invisible to the current audit approach.

> Escape Point Verdict: LEADING — reasoning is consistent and plausible; PFMEA pull (Anna Novak, TBD) required to confirm zone drift was not listed as a failure mode and no detection control was assigned.

Root Cause Verified: No — verification experiment designed and accepted; results pending execution (see Open Items).

Evidence referenced:

• Dan Kim 100% X-ray results — #1913 suspect-window output (84/2,000 defective joints)
• Oven thermocouple history log — #1913 Zone 7 deviation (Raj Patel, week of June 1)
• KIC Profile Verification Record — #1913, April 28, 2026 (doc ID TBD — to be attached to NCR#26-0244)
• COA #SAC-0339 — PST-26-0339 (Ag 2.8%, liquidus 219°C cited; full parameter review TBD — Maria Sanchez)
• J-STD-006 — SAC305 silver content specification (2.5–4.0%)
• WO-26-0610 — squeegee replacement May 30; SPI data confirming paste volume within ±10%
• NCR#26-0244 — traceability pull (last 8 days, paste lots / board lots / oven profile changes)
• Process control plan — TBD pull (Anna Novak); expected to confirm zone log monitoring not defined
• Reflow PFMEA — TBD pull (Anna Novak); expected to confirm zone drift not listed as failure mode

D5 — Permanent Corrective Actions (PCA)

PST-26-0339 Paste Lot Disposition:

• #1911/#1912: continue use — lot has run at normal yield (0.3%) throughout suspect window on ovens with zones verified in tolerance
• #1913: hold from use until full return-to-service protocol complete (Actions 1–6 above) — oven-condition hold, not a nonconformance hold
• Supplier: no corrective action, no notification — lot meets J-STD-006; no nonconformance to report
• Conditional long-term action: if Cycle 2 drift-onset data and paste-by-oven trial confirm interaction hypothesis (H2), evaluate internal purchasing standard for tighter Ag band specification — requires trial data before initiating

Evidence referenced:

• J-STD-006 — SAC305 specification (basis for paste lot disposition decision)
• IPC-A-610 Class 2 — inspection standard for return-to-service re-qualification build
• KIC Profile Verification Record, #1913, April 28, 2026 — baseline for return-to-service KIC criterion
• WO-26-0610 — squeegee replacement record (ruled out as contributing factor)

D6 — Implement & Verify Effectiveness

No corrective actions have been implemented yet. This section will be populated in Cycle 2 when Raj's zone inspection, repair, and the return-to-service protocol have been executed.

Effectiveness verification criteria (to be populated in Cycle 2):

• RC-1: QFN defect rate on #1913 returns to ≤0.3% on re-qualification build (100% X-ray); Zone 7 independently verified within ±2°C of setpoint; production board profile confirms QFN pad TAL in spec
• RC-2: Zone deviation alarms active and tested on all three ovens; weekly zone log review records completed and auditable for minimum 30 days post-implementation; control plan and PFMEA updated and approved per document control

D7 — Prevent Recurrence

Similar processes/products to review:

• Reflow Ovens #1911 and #1912 — same oven model, same zone configuration; zone monitoring gap applies equally. Zone alarm configuration and log review requirement should be implemented on all three ovens simultaneously, not #1913 alone.
• Any other high thermal-mass component types (BGAs, power modules, large connector pads) run through all three ovens — profile verification method blind spot applies to all assemblies with high-mass pads where the gold-board fixture is not thermally representative.
• Any other convection reflow ovens at Beacon Circuits — if the process control plan designates KIC verification as the sole thermal control, the monitoring gap exists on all reflow equipment.

Documents to update:

Lessons Learned:

1. An oven-specific defect pattern is diagnostic before the fishbone is built. When one oven fails and two run clean on the same material and print process, the investigation should focus on what is different about that oven — not on what changed in the shared process inputs. The shared stencil printer was the key structural fact that broke the paste-lot hypothesis early.

2. The signal existed. The oven was recording the zone deficit continuously. The failure was not that the data was unavailable — it was that no process consumed it. Process control plans that designate periodic verification as the sole thermal control create blind spots between verification intervals. Continuous monitoring of in-process data is a more robust control than periodic profiling alone.

3. A profile verification that passes does not mean the process has adequate margin at every location on every board. The gold-board KIC fixture verifies oven thermal performance at thermocouple fixture locations — it does not verify joint temperature at high thermal-mass locations on production assemblies. Verification method adequacy should be evaluated against the worst-case thermal load in the product mix.

4. Timeline correlation is a hypothesis, not a conclusion. The paste lot change on June 1 and the first defects on June 2 produced a compelling narrative. That narrative was incorrect as a root cause — but working it through the IS/IS NOT analysis (shared printer, same paste on good ovens) broke it efficiently without discarding the material observation that became H2.

5. PFMEA adequacy is not visible to conformance audits. A missing failure mode in a PFMEA does not trip a finding when the audit scope is conformance to the documented plan. Audit scope should periodically challenge whether known failure modes for the process type are represented in the risk documentation.

D8 — Closure

Closure Status: NOT YET EARNED.

D6 effectiveness data does not yet exist — no corrective actions have been implemented. The investigation has produced a grounded root cause analysis (RC-1 LEADING, RC-2 PROBABLE) and a complete corrective action plan, but the repair has not been executed, the return-to-service protocol has not been run, and the zone monitoring controls have not been configured or documented. Closure requires completion of D6 effectiveness verification against the criteria defined in D5 — specifically: Zone 7 independently verified within ±2°C, production board profile confirming QFN pad TAL in spec, re-qualification build at ≤0.3% defect rate, and zone alarms active with auditable log review records established.

The top open item driving non-closure is: Raj Patel's Zone 7 inspection and the subsequent return-to-service protocol (D5 Actions 1–6). This is the critical path. All other actions can proceed in parallel but closure cannot be declared until D6 effectiveness data is in hand.

Investigation Summary:

QFN-48 thermal pad cold joints on Reflow Oven #1913 (4.2%, 84/2,000) were traced to a Zone 7 thermal deficit — 5°C below setpoint — delivering insufficient time above liquidus at the highest thermal-mass location on the board. The deficit went undetected because no process requirement exists for monitoring oven zone thermocouple history logs; the signal was present and unread. Paste lot PST-26-0339 was the initial leading hypothesis but was ruled out as a standalone cause by the shared stencil printer structure; it is retained as a suspected aggravating factor pending drift onset data. Two co-equal root causes have been identified and corrective actions are structured; verification and implementation are pending.

Team Recognition:

Anna Novak led a rigorous and intellectually honest investigation — notably, revising the initial paste-lot hypothesis when the IS/IS NOT evidence did not support it. Dan Kim's 100% X-ray inspection established the quantitative foundation. Raj Patel's observation of the Zone 7 deviation was the pivotal finding. Maria Sanchez and Linda Chen hold critical open actions. The team's willingness to follow the evidence rather than the initial narrative is reflected throughout this record.

Open Items & Gap Impact

Evidence Index

This is the Cycle 1 8D Problem Solving Report for NCR#26-0244. Root cause analysis is complete at current evidence grade (RC-1 LEADING, RC-2 PROBABLE). Corrective actions are structured and owned. No actions have been implemented; D6 effectiveness data is pending. This report is a draft investigation record — it must be reviewed, verified, and approved by the qualified responsible party (Anna Novak, 8D Lead) before submission to the customer or filing in the QMS. When Raj Patel's zone inspection, the return-to-service protocol, and the monitoring controls are implemented and verified, return for Cycle 2 to close D6, advance verdict grades, and complete D8.

When you're ready to proceed: Bring back Raj's zone inspection results, the drift onset date from the thermocouple history log, and the production board profile data. We'll start Cycle 2, advance RC-1 from LEADING to CONFIRMED (or reopen if the data doesn't support it), close D6, and complete D8.