Revision Impact Analyzer.Vianora · PCB pilot

VIANORA · PCB REVISION IMPACT ANALYZER

Rev A to Rev B: every change that matters, before you order.

Upload both versions. Get one plain report of every changed part, connection, cost and firmware reference — ranked by what will stop the board working, not by which file changed.

Founding pilot · 10 board pairs · run with you by an engineer · no card

2.9respins on the average PCB project — and a missed change is the cheapest way to earn oneLifecycle Insights, cited in Design World (2018)
$44,000average cost of one respinLifecycle Insights, cited in Design World (2018)
8.5 daysaverage schedule impact per respinLifecycle Insights, cited in Design World (2018)
~3 in 4buyers name component obsolescence a supply-chain threat — up almost 7 points year on yearElectronic Design / Source Today reader survey

THE QUIET CHANGES

The changes that cost a board are the boring ones.

Nobody misses a new connector. What gets missed is a value, a pin, a DNP flag or a line of firmware — changes small enough that no one thinks to tell anyone.

  • A value swap that moves a net

    100 nF to 1 µF is one line in the BOM. On a charge pump's flying cap or a regulator's feed-forward network it is a different circuit. A BOM compare can tell you the value changed; it cannot tell you what that did.

    Caught by: net-level comparison plus the reference design's own constraints.

  • A pin swap on a connector

    Two pins exchanged on a header. The netlist changed, the board still passes DRC, and the harness built for Rev A is now wired backwards — found at the system test, in the enclosure, after assembly.

    Caught by: pin-level net membership, not wire order.

  • DNP flips

    The part that was fitted is now not fitted, or the reverse. The assembly drawing says one thing, the BOM another and the schematic a third — and the pick-and-place file decides.

    Caught by: reconciling schematic, BOM and assembly variant in one diff.

  • Same net, different copper

    A net rerouted across a plane split, or past a 0Ω link. The netlist reads identical and the return path is gone. Electrical review of the schematic finds nothing, because the schematic did not change.

    Caught by: copper-level comparison, layer by layer, against the previous revision.

  • Substitution under pressure

    The part is out of stock, an "equivalent" goes in, and its pinout, timing or tolerance is not the same. Now the design document, the BOM and the board disagree — and only two of them are wrong.

    Caught by: part-level change with a pinout and parameter comparison.

  • The code that still points at Rev A

    Pin map, device tree, HAL defines, calibration constants, the test fixture's pinout. The hardware moved and the firmware that talks to it did not. This is the failure nobody diffs, because it lives in a different repository.

    Caught by: the report naming the code lines that depend on a changed pin or net.

WHAT YOU DO TODAY

Four ways to compare two boards, and what each one misses.

Board diffing is not an unsolved problem — it is a set of partial answers, each of which is excellent at one axis and silent on the rest.

  • Visual overlay in a PDF or Gerber viewer

    CATCHESMoved copper, a shifted footprint, a deleted silk note.

    MISSESA value, a pin assignment, a net, a cost delta, a line of firmware.

  • A BOM comparison tool

    CATCHESAdded, removed and re-valued parts; MPN changes; DNP flags — if the two exports are clean.

    MISSESConnectivity, copper, code, and any change that did not alter a BOM line.

  • Text diff of the EDA files

    CATCHESText and geometry changes in KiCad-style files, inside one toolchain, for a team already on git.

    MISSESThe meaning. "Coordinate moved" is a diff; "the CAN pair is reversed" is a decision. And nothing crosses into cost or firmware.

  • The engineer who remembers

    CATCHESEverything they were told, and everything they personally changed.

    MISSESThe change you are looking for — by definition, the one nobody remembered to mention.

We are not claiming nobody can diff a board. The gap is the join: part, connection, copper, cost and code in one report, written so that someone who did not make the change can act on it.

HOW IT WORKS

Two revisions in, one ranked report.

  1. 01

    Both revisions, as exports

    Netlist, BOM and Gerbers for Rev A and Rev B — from KiCad, Altium, OrCAD/Allegro or Eagle/Fusion. Cross-tool comparisons (an Altium design ported to KiCad) are the case where this earns the most.

  2. 02

    Normalised into one model

    Parts keyed by reference designator and MPN. Nets keyed by pin membership, not by name. Copper keyed by layer. That is what makes "the net was renamed" and "the net was actually reconnected" two different findings.

  3. 03

    Classified, then ranked

    Electrical, assembly, mechanical, cost and interface. Each row gets a severity based on what it breaks — a reversed CAN pair outranks a moved silkscreen, always.

  4. 04

    The report, and the record

    A plain-language table with before, after, why it matters and what to do. Signed off once and kept next to the design, so the next revision starts from a known state.

WHAT YOU GET

The report, ranked by what breaks.

Every row answers three questions: what changed, what it breaks, and what to do before the order goes out. Cost and supply rows are in the same table, because they gate the same order.

ItemRev ARev BClassRiskWhat it breaks / what to do
J2 pin 3 / 4CAN_H / CAN_LCAN_L / CAN_HInterfaceCriticalThe Rev A harness is now reversed. Re-pin the cable, update the assembly note, and flag every unit already built.
fw/board.h:22#define CAN_TX_PIN 4unchangedCodeCriticalHardware moved the pin; firmware still drives the old one. Update the define (and the device tree, if there is one) in the same commit as the revision.
R1410 kΩ 1%100 kΩ 1%ElectricalHighI²C pull-up is now ten times weaker: rise time fails at 400 kHz. Restore 10 kΩ or drop the bus to 100 kHz and document why.
C314.7 µF fittedDNPAssemblyHighFeed-forward cap removed from the LDO loop; stability margin drops. Confirm the DNP is intended before the assembly order goes out.
U7TPS62130 · activeTPS62130 · NRNDSupplyMediumNot end-of-life yet, but 8-week lead time and +$0.42 at 500 units. Qualify an alternative now, not at the next build.
NET_3V3_AUXrouted north of the splitcrosses the plane split under U9CopperMediumSame netlist, different return path. Check the ADC's reference noise floor on the bench before release.
BOM total @100$41.18$43.07CostLow+4.6% per unit, driven by U7 and two passives. Update the quote and re-check the margin assumption.

Illustrative rows for a sample four-layer board, Rev A → Rev B. Your report is generated from your files; these are examples of the shape and the severity ranking.

THE ORDER GATE

The last check before the fabrication order.

A revision is not finished when the schematic is saved; it is finished when the order is placed. So the report sits at that gate and answers four questions in one place: what changed electrically, what changed in cost at your quantity, which parts changed lifecycle status, and which parts are no longer in stock.

2–5 years

the lifecycle most components are now designed for — hundreds of thousands of parts reach end-of-life every year

Part Analytics, Component Obsolescence in 2026

  • Cost at your quantity

    Unit and extended cost deltas, not a percentage someone typed into a spreadsheet.

  • Lifecycle status

    Active → NRND → EOL, with the lead time you will actually be quoted.

  • Availability

    Which changed parts have no distributor stock, and which are single-sourced after the change.

  • Footprint drift

    The same MPN in a different package is a new footprint, a new stencil and a new assembly risk.

THE HANDOVER

A revision record a client can actually read.

If you design for other people, the change record is the deliverable that decides whether the next order is right. The report is written for the person placing the order: no CAD licence, no viewer, no tribal knowledge — a table, a severity, an action, and a date.

  • 01

    Signed off once per revision, kept with the design files.

  • 02

    Plain language, so a fab, a client or a new engineer can act on it.

  • 03

    Every row traceable to the file and line it came from.

WHAT IT IS NOT

The limits, before you sign up.

  • Not a PLM and not a git replacement — it reads your two revisions, it does not become your system of record.

  • It never modifies a design. Every comparison is read-only, and the artifacts stay yours.

  • Not a DRC and not a signal-integrity analysis. It finds what changed; it does not certify that the new version is good.

  • Not a safety or compliance record. Medical, high-voltage and safety-critical products are out of scope.

  • Changes made on the bench (a rework wire, a swapped part) have to be entered by hand — the tool can only compare what is in the files.

THE PILOT

Ten boards, run by hand, before we build anything.

We are not launching a product on this page. We are running ten real projects through the tooling with an engineer on the call, because the only thing that makes this worth building is that it survives your board.

  1. 01Send the exportsThe files we need, and nothing else. No CAD licence, no firmware, NDA on request — the exports are used for your artefact and deleted on request.
  2. 02We build it with youAn engineer runs the tooling against your design and reviews every line before it reaches you — nothing generated is sent unchecked.
  3. 03You get the artefactThe plan or the report, in your inbox, within two business days. Use it on the board in front of you.
  4. 04You decide if it is worth $29At the end of the pilot you can pre-order at $29/month or walk away. The pilot itself costs nothing.

There are no case studies on this page, no client logos and no testimonials, because there are no customers yet. The pilot exists to produce them — and to tell us whether this should be a product at all.

GET THE PILOT

Starts the paid plan today: charged on Stripe, cancel any month, price locked for the first ten teams.

Pilot: free · Paid plan: $29/month from today · We reply from info@vianora.ai. No newsletter, ever.

The pilot is free. Starting the paid plan now charges $29/month today, cancels any month, and locks the price for the first ten teams.

QUESTIONS WE ACTUALLY GET

The objections, answered straight.

BOM comparison is a good answer to a narrower question: it sees parts. It cannot see that the same part moved to a different pin, that a net was reconnected under a renamed wire, that copper now crosses a plane split, that the cost delta changes your unit economics, or that firmware still references the old mapping. We diff the netlist and the copper as well as the BOM — and we rank by what breaks, not by file.

FIGURES AND SOURCES

Where the numbers come from.

  • Lifecycle Insights — PCB design respin study

    Cited in Design World Online, "Modern Practices in PCB Design Verification" (2018): 2.9 respins per project on average, $44,000 per respin, 8.5 days of schedule impact. An average across commercial projects, quoted as published.

  • Electronic Design / Source Today — reader survey on consolidation and obsolescence

    Roughly three-quarters of respondents flagged component obsolescence as a supply-chain threat, an increase of almost seven percentage points on the previous year.

  • Part Analytics — Component Obsolescence in 2026

    Hundreds of thousands of components reach end-of-life every year, with average lifecycles now measured in roughly two to five years.

  • Existing comparison tooling — BOM compare (Sierra Circuits, RAYPCB), file diff (kidiff, AllSpice), vendor compare (Altium 365)

    Named because they are the real alternatives, and because the report has to be better than them at the join — parts, connectivity, copper, cost and code in one ranked list — rather than better at any single one.

The sample report rows are illustrative examples on a sample board, not output from a customer design. Revision Impact Analyzer is in pilot: reports are produced by an engineer using our tooling while the product is built. Figures are cited as published and may be revised by their authors.

NEXT ORDER

Know what changed before the boards are made.

Send the two revisions. We will send back the ranked report — parts, pins, copper, cost and the code that depends on them — before the order goes out.

Run my last two revisions

Founding pilot · 10 board pairs · free · reply within one business day