A reference design is a valuable starting point, but a working demonstration board is not the same object as a production product. The enclosure, power source, cable set, operating environment and manufacturing process can all change the circuit’s behavior. Copying the schematic while changing those conditions transfers less evidence than the design team may expect.
This reference design adaptation checklist helps an engineering team decide which evidence can be reused, which assumptions have changed and which tests must be repeated. It supports a controlled PCB design review, rather than treating a vendor example as an automatic approval for a different application. The useful output is a change-and-verification record tied to the actual product revision.
Identify what the reference design proves and what your product changes
Freeze the reference baseline before editing
Save the original schematic, BOM, layout, stackup, firmware, test report and relevant device documentation with their revision identifiers. Record the source and retrieval date. A folder containing a schematic from one revision and firmware from another is not a controlled baseline. Preserve the original files separately from the adaptation so later reviews can see exactly what changed.
Read the demonstrated operating conditions. Identify input voltage, load range, temperature, cooling, cables, firmware settings and the measurement method behind each performance claim. A result measured at room temperature with a short cable does not establish performance at the edge of a product’s environmental range. Separate measured results from typical simulations and from features mentioned without supporting test data.
For example, Analog Devices’ describes a specific evaluated system and states that application suitability still requires testing. That boundary is useful across reference-design work: the source provides engineering evidence within its documented scope, while the product team owns verification of its adapted implementation.
Write the product requirements in testable terms
Replace broad goals such as 鈥渋ndustrial quality鈥� with measurable requirements. Define supply range, transient conditions, operating and storage environment, interfaces, startup behavior, load steps, measurement accuracy, mechanical envelope and service expectations. Identify requirements that are mandatory for product release and those that remain development goals.
Include abnormal but foreseeable states. What happens when an external cable is connected before power, a sensor is unplugged, the supply ramps slowly or firmware stops responding? A demonstration board may rely on a bench supply, host computer or operator to avoid these states. A deployed product needs a defined response without that laboratory supervision.
Map every requirement to a source of evidence. The source might be an existing reference test, a device specification, a calculation, a simulation or a new product test. Mark evidence as reusable only when its assumptions still apply. This simple distinction prevents a copied report from becoming a substitute for verification of the changed hardware.
Classify changes by the physical behavior they can affect
Create a table with the original feature, proposed change, reason, affected behavior and verification action. A smaller inductor can alter saturation margin and thermal behavior; a different capacitor can alter effective capacitance, ESR and control-loop response; a connector substitution can change current capacity and cable shielding. 鈥淓quivalent value鈥� is not enough to establish equivalence in the application.
Layout changes deserve explicit entries. Shortening the board, moving a connector or changing the ground arrangement can affect return currents, coupling and heat flow even when the netlist is unchanged. Layer count, dielectric spacing and copper distribution also matter. Use the multilayer board construction review to define a buildable stackup rather than copying only the total board thickness.
Some changes interact. Moving a switching regulator closer to a sensor while changing its operating frequency is more than two independent edits. Identify coupled risks and plan a test that can reveal them. A review that approves each substitution separately can miss the behavior of the combined product.
Preserve critical geometry while understanding its purpose
Reference layouts often embody constraints that are not obvious from the schematic. Identify high-current switching loops, sensitive feedback nodes, clock return paths, differential routing, thermal spreading and antenna clearances. Read the device-specific layout guidance before deciding which geometry can be moved. If a feature’s purpose is unclear, treat that uncertainty as a review item.
Do not preserve geometry blindly when the surrounding construction changes. The same trace width on a different stackup may have a different impedance, while the same copper shape under an enclosure may have different thermal performance. Copy the engineering requirement, then confirm that the new construction satisfies it. Keep calculations and assumptions with the design record.
A useful design review asks what evidence would reveal a bad adaptation. For a regulator that might be startup and load-step measurements; for a precision input it might be noise and drift under realistic digital activity. Choose tests around failure modes rather than collecting attractive oscilloscope screenshots that do not address the changed conditions.
Convert the adapted circuit into a manufacturable, verifiable product
Remove demonstration conveniences deliberately
Evaluation boards often include jumpers, LEDs, oversized connectors, alternate circuits and measurement headers. Decide which are needed in production, which can become unpopulated options and which can be removed. Document the effect of each removal. A power LED that is irrelevant on the bench may dominate a battery product’s sleep current, while removing a jumper may eliminate a useful current-measurement boundary.
Retain sufficient access for bring-up, programming and production test. The final enclosure can make a convenient bench connector inaccessible. Review probe approach, cable insertion, board support and the stage at which programming occurs. Connect these decisions to the test-point access plan before the mechanical design is frozen.
Firmware is part of the adaptation. Record clock configuration, peripheral assignments, calibration, startup sequence and device configuration bits. A schematic-compatible board can fail because the firmware assumes a different oscillator, pin assignment or external pull state. Preserve a recoverable development configuration until the production image and programming process have been verified.
Build a verification matrix with conditions and pass criteria
For each requirement, name the test method, equipment, hardware revision, firmware revision, operating conditions and acceptance criterion. Include the number and identity of samples and explain the purpose of that sample set. A few development boards can expose design problems but do not automatically establish production reliability or statistical process capability.
Separate functional, electrical, thermal, mechanical and compliance-related questions. A functional demonstration does not replace an emissions or immunity assessment, and a thermal simulation does not replace all measurements. Where formal compliance is required, use the applicable current requirements and a suitable qualified process. Do not infer certification from a component’s approval or a reference board’s test result.
Use controlled comparisons when a change is uncertain. Measure the original implementation and the adaptation under comparable conditions if the original hardware is available. Keep probe placement, bandwidth settings, cables and load profiles consistent. Otherwise a difference in measurement setup can be mistaken for a difference caused by the PCB.
Work through an adaptation example
Consider a hypothetical sensor controller based on a reference board powered from a stable laboratory supply. The product adds a long input cable, a compact sealed enclosure and a pulsed radio load. The schematic retains the same regulator and nominal output voltage. Those similarities do not close the power-integrity or thermal questions.
The change table would identify input transients and cable impedance, reduced cooling and radio load steps. Verification would include startup across the specified input range, local rail behavior during the pulse, thermal measurements in the enclosure and fault recovery. The exact limits come from the product requirements and selected devices, not from this example.
If a larger capacitor is proposed after a rail dip, review more than capacitance. Check effective capacitance under bias, inrush, regulator stability, physical placement and discharge behavior. The corrective action should address a measured mechanism. Repeating the relevant tests after the change closes the loop between observation and design, rather than relying on a plausible but unverified fix.
Release manufacturing data as one coherent configuration
Prepare the fabrication outputs, assembly drawings, BOM, placement data and test requirements from the same approved revision. Define fitted variants and approved alternates explicitly. A revised schematic with an older BOM can send production back to the component that engineering intended to replace. Use file identifiers and checksums where practical to make the released configuration unambiguous.
The PCB manufacturing file package should state stackup, critical dimensions, controlled-impedance requirements when applicable and any special acceptance needs. The assembly package should distinguish do-not-fit parts, polarity-sensitive components and process restrictions. Do not assume that notes in a vendor’s reference document will automatically reach the manufacturing team.
Review component availability and approved substitutions before a pilot build. Supply-chain changes can reopen technical questions that were already tested with the original BOM. Preserve the approved configuration while evaluating alternatives, and record which evidence must be repeated for each material change. Procurement flexibility is useful when its engineering boundary is defined.
Use the pilot build to close risks, not merely produce units
Plan the prototype assembly around the unresolved questions. Identify which units will support destructive analysis, thermal characterization, firmware development and customer demonstrations. Record rework and deviations against each unit so test results can be interpreted against what was actually built.
At release, distinguish closed requirements, accepted limitations and open risks. Assign an owner and next action to every open item. A product should not inherit a reference design’s reputation as a substitute for this decision. The strongest handoff explains what was retained, what changed, why it changed and how the adapted product was checked.
When requesting PCB manufacturing support, provide the controlled outputs and the features that require special review. Discuss the adapted design and pilot-build requirements with the change list, BOM and remaining verification questions. That gives engineering and manufacturing a concrete basis for the next build.