FMEA[VisualPro Tech Brief] What You Miss When an ECR's FMEA Impact Review Stops at the Changed Part
Summary
A design change that swaps a single part still reaches failure chains in other FMEA elements. In a demonstration transmission control unit DFMEA, changing one spot in the power supply produced eight impacted rows, and one of them was left with no retest linked to it.
A familiar scene in change review meetings
When an engineering change request (ECR) comes in, the review meeting usually revolves around the part being changed. The team updates that part's FMEA rows and agrees to rerun the tests for that part. The minutes say "FMEA updated, 2 retests."
An FMEA is not a parts list. It is a web of failure chains. A failure cause in the power supply leads through an MCU reset to a failure mode on the software side, and the noise from a switching regulator reaches the ADC and CAN on the same board. If you look only at the changed part's rows, the rows beyond those links never make it into the meeting.
The standards already require an impact assessment
Change impact assessment is not a new requirement. The standards readers already follow require it. This article is about checking whether that requirement actually reached every FMEA row and every test.
| Standard | What it requires | What tends to be missed in practice |
|---|---|---|
| IATF 16949 design and development changes (8.3.6) | Identify changes, assess and record their impact, apply them after approval | The impact assessment stopping at the changed part |
| AIAG-VDA FMEA Handbook | Treat a design change as a trigger to revisit the FMEA and reassess risk | Revisiting chains in other elements and new failure causes |
| ISO 26262-8 change management | Impact analysis of the change and identification of affected work products | Confirming that every affected row has a defined verification method |
| Customer change notification (PPAP, etc.) | Notification and approval of the change and its verification results | Checking that no row lacks verification before notifying |
Three ways the impact spreads
The demonstration used a transmission control unit (TCU) DFMEA built by VWAY. The change is as follows.
- ECR-2026-031: Replace the LDO regulator in the TCU hardware's power management block with a synchronous buck converter.
- Reason for change: The FMEA risk analysis still showed thermal shutdown due to insufficient LDO heat dissipation (AP M).
Following the failure network from this change, and adding EMC interference that is not in the network by engineering judgment, produced eight impacted rows. They fall into three groups.
A test condition issue overlaps with this. The existing high-temperature load test used an 85°C ambient condition. If the part was changed because of heat, whether this condition still covers the changed specification is also part of the impact review.
Method 1. Tag impacted rows with the change number in VisualPro
In the optimization view of the VisualPro DFMEA, we added an action row for each impacted row and prefixed the prevention and detection actions with [ECR-2026-031]. The two new failure causes were added to the power management block and linked to existing failure modes.
With this in place, entering the change number in the optimization view's search box collects the impacted rows element by element. O, D, and AP before and after the action can be compared in the same view. Below is the TCU software side with the same search. Three rows come up even though only the power supply was changed.
Method 2. Link impacted rows to retests in Recoble
FMEA tools manage rows and actions, but the review progress and retest plan for a change usually end up scattered across meeting minutes and emails. For the demonstration, we built two things in Recoble.
- ECR note: Holds the change overview, the impact review method, and the retest items (TST-01 to 05) as headings. They are high-temperature load, supply variation and cranking (ISO 16750-2), EMC emissions (CISPR 25), EMC immunity (ISO 11452), and ADC accuracy under ripple conditions.
- Impacted FMEA row board: Makes one card per impacted row, with the impact path (changed element itself, power supply sharing, EMC interference, new failure cause), the FMEA element, and the AP before action as fields.
Next, we added a "verifies" link from each retest heading to the cards that test confirms. One test can confirm several rows (EMC emissions → new EMI and CAN), and one row can be confirmed by several tests (CAN → emissions and immunity).
Method 3. Find missing verification with the trace table's "missing" column
Recoble's trace table starts from the board cards and places the linked retests in the next column. The figure at the top of this article is that view. Next to the 9 links, there is 1 missing. The new failure cause row, "output voltage undershoot due to insufficient control loop phase margin," does not yet have any retest linked to it.
This row was added to the FMEA, and its current detection control in VisualPro even says "bench load step measurement." But it was left out of the ECR's retest plan. If the approval conditions include "every impacted row has a retest linked to verify it," you have grounds to hold approval until the missing count in this table reaches zero.
What the demonstration showed, and its limits
| Item | Result |
|---|---|
| Impacted rows | 8 (changed element itself 3 · other elements 3 · new failure causes 2) |
| Existing failure causes of the power management block | 3; the other 5 are causes in other elements or newly added causes |
| Retests | 5, with 9 links |
| Missing verification | 1 row (new failure cause) |
This data was created for the demonstration, and the impacted rows were found with the failure network and engineering judgment. Physical interference such as EMC, which has no links in the FMEA network, cannot be found automatically by a tool. Putting those rows in scope was a human decision. What the tools did was gather the found rows again by change number and expose the rows with no link to a test.
What to try on your next ECR
- Starting from the changed element's failure causes, step up one level at a time toward failure modes and failure effects, and add higher-level rows in other elements to the impact list.
- Add a separate step that asks about failure mechanisms specific to the new part or technology (switching noise, loop stability, degradation of new materials, and so on).
- Tag every impacted row with the change number so you can gather them later with one search.
- Check whether existing test conditions cover the changed specification (temperature, voltage, frequency).
- Before approval, confirm that every impacted row has at least one test linked to verify it.
Closing
FAQ
Q1Can't an FMEA tool alone handle change impact analysis?
Finding impacted rows and recording actions and reassessments is done in the FMEA tool. In the demonstration, too, the impacted rows were gathered with a VisualPro search. However, the change's retest plan, progress, and approval conditions sit outside the FMEA, so it is easier to cross-check them when they are kept in a separate space linked to the FMEA rows.
Q2Isn't this too much work for every change?
It depends on the size of the change. For a minor change with only one or two impacted rows, a single note is enough. This method pays off for changes whose impact spreads across several elements.
Q3How do you find new failure causes?
Tools do not generate them automatically. The practical approach is a separate step that asks "in what ways can this part fail that the previous one could not?", using the new part's datasheet and application notes, internal failure history, and a cross-functional review.
Q4How does this relate to customer change notification?
The results of the impact review are the basis for deciding whether to notify and what to submit as verification results. The notification criteria themselves follow the customer's requirements.
Q5Can't you do this in Excel?
You can. Make two tables, one for rows and one for tests, and add a mapping column. But every time a row or test changes, the mapping has to be updated by hand, and finding missing links is also manual work. When they are connected by links, gaps show up directly in the table.
- IATF 16949:2016, 8.3.6 Design and development changes
- AIAG & VDA, FMEA Handbook, 1st ed., 2019
- ISO 26262-8:2018, Clause 8 Change management
- ISO 16750-2 (electrical loads), CISPR 25 (emission limits for the protection of on-board receivers), ISO 11452 series (component electromagnetic immunity)
Recoble is an AI collaboration workspace made by VWAY, and VisualPro is VWAY's functional safety and reliability analysis tool. The DFMEA and the design change (ECR-2026-031) in this article are data created by VWAY for the demonstration, and the impacted rows and retests are examples for this demonstration. The standard clauses are summarized at the level of each clause's title. Text inside the screenshots was translated from the Korean originals for this article, so labels may differ from the products' English interface.