When a receiver develops discrete spurs after its buck converter starts switching, begin with a controlled power-and-spectrum experiment. A lower-loss RF laminate does not identify the noise source or repair the current path that couples into the receiver. First establish which operating state creates the spur, how it reaches the sensitive circuit, and whether a layout or power-distribution change restores the required margin.
This diagnostic workflow is intended for civilian telemetry receivers, SDR instruments and sensor electronics. Treat an FR-4 power section as a candidate baseline when it meets the electrical, thermal and fabrication requirements. Use the Rogers versus FR-4 comparison to evaluate the RF routes separately instead of treating every noise problem as a material-selection problem.
1. Capture the converter state with each receiver spectrum
Record input voltage, output voltage, load current, switching setting, operating mode and board temperature. On the receiver side, record tuning frequency, gain state, bandwidth, sample rate, RF stimulus and termination. Save analyzer resolution bandwidth, detector and averaging settings. Lock gain where the test permits, or log gain changes so a reduced output spur is not mistaken for improved input-referred performance.
The LM76002/LM76003 datasheet, Revision A illustrates why mode matters: its light-load behavior can reduce switching frequency, while forced PWM extends fixed-frequency operation toward no load, subject to on/off-time foldback limits. This is device-specific behavior. Confirm the actual waveform and the operating restrictions of your converter before changing modes or synchronization.
Capture both a correctly terminated receiver input and a controlled wanted signal. A spur that appears only with a signal present can reveal modulation or mixing that an idle spectrum misses. If testing two allowed switching frequencies, keep the remaining conditions as constant as practical and record any accompanying changes in ripple, temperature or gain.
2. Draw the switching loop on the actual copper
For a synchronous buck, trace the current paths during both switch states and identify their difference. The fast commutation loop includes the input bypass capacitor and the high-side and low-side switches. It is not simply the entire path through the inductor and load. Analog Devices AN-139, Power Supply Layout and EMI explains this hot-loop distinction and how its geometry affects emission.
Mark this loop on the layout with component pads, vias and layer transitions included. Then mark the switch-node copper, feedback route, receiver supply branch and nearby RF or clock routes. Inspect the return path as carefully as the outgoing trace. The objective is to find a specific shared impedance or coupling opportunity that can be tested.
For the LM76003 example, TI recommends close input bypassing at PVIN and PGND, a compact switch-node connection, short feedback routing and a suitable ground plane. Follow the selected device’s grounding instructions rather than copying another regulator’s pin arrangement. A nearby plane can help control fields, but AN-139 also discusses currents induced in that plane; the ground label alone does not make every location quiet.
3. Separate a rail problem from a field-coupling problem
Use the following proposed comparison matrix. Change one condition at a time and retain the original baseline. Before substituting a supply, provide a controlled isolation point, avoid backfeeding either source, and preserve the receiver’s required voltage and sequencing. Check the substituted rail at the device pins rather than assuming a bench source is automatically quiet.
| Experiment | What to keep consistent | Interpretation to test next |
|---|---|---|
| Disable the buck and power the receiver from a separate suitable source | Receiver settings, local voltage and measurement connections | An improvement implicates the converter system, but does not yet distinguish rail coupling from fields |
| Power the receiver separately while the buck drives a representative dummy load | Buck operating state, input wiring and load behavior | A remaining spur suggests field, shared-ground or another conducted path still exists |
| Compare one sensitive supply branch before and after an evaluated filter | DC voltage, gain, load and converter state | A repeatable change supports that branch as a contributor; verify filter stability and transient behavior |
| Scan with a near-field probe at fixed height and orientation | Probe position reference, spectrum settings and board operating state | A local maximum guides inspection; it is not a calibrated prediction of receiver interference |
Keep substitute supply leads and instrument grounds documented: they can create a new coupling path or remove an existing one. Use probing appropriate to the node voltage and required bandwidth, with a short local return for rail measurements. A large probe-ground loop can pick up switching fields and exaggerate apparent rail noise.
4. Interpret spur movement without assuming direct harmonic pickup
Analog Devices’ RFIC power-management study describes supply-noise feedthrough and mixing products around the desired IF, using the ADRF6820 as its example. It also shows why supply domains need separate evaluation. Its measured sensitivities and recommended rail arrangements are not universal limits for a different receiver.
Consider a hypothetical receiver with a wanted IF tone at 8.0 MHz and a converter-related tone at 0.5 MHz. Candidate modulation sidebands are 7.5 and 8.5 MHz. If an allowed converter change moves that tone to 0.6 MHz, corresponding candidates become 7.4 and 8.6 MHz. This is a calculated diagnostic expectation, not a claimed measurement or proof of the coupling mechanism.
Compare observed movements against a short prediction list before testing: direct feedthrough, harmonics, sidebands around the wanted signal, and possible sampled aliases. Include the receiver’s frequency plan. If a peak does not move as predicted, investigate clocks, local oscillators, external interference and the measurement setup instead of forcing the data to match the first explanation.
A visually lower peak is insufficient evidence if resolution bandwidth, gain or noise floor changed. Define the acceptance metric in advance: for example, worst spur relative to a fixed wanted signal, or a specified receiver sensitivity result. Preserve absolute levels and settings alongside any normalized plot.
5. Test a focused layout or filtering correction
Choose a correction tied to the comparison results: a revised bypass placement, a smaller commutation loop, a moved sensitive route, or filtering of an identified supply branch. Give each board revision a single primary hypothesis and record other unavoidable differences. Repeat the same receiver test after the change, including the load or startup state that originally exposed the problem.
Evaluate filter impedance, capacitor bias effects, transient response and regulator stability with the selected components. If adding an LDO, check headroom, dissipation and rejection at the troublesome frequencies and actual load. Do not substitute a headline noise figure for a complete power-path check. Validate the assembled board across its required operating range.
A routing correction also needs an intact signal reference. Review the controlled-impedance geometry when moving an RF trace or changing its reference layer. If the apparent problem is a persistent frequency-response limit rather than a converter-correlated spur, the SDR balun and RF-interface checklist provides a separate diagnostic path.
6. Decide whether the RF paths justify a hybrid stackup
After reducing the identified coupling, assess RF insertion loss, phase consistency and dimensional tolerances against the product budget. A hybrid stackup may be justified for particular RF layers while other functions use a different dielectric system. That is a construction decision requiring fabrication review, not an automatic remedy for converter spurs.
Use the stackup planner to record layer roles and reference planes. Ask the fabricator to confirm the actual material combination, bonding process, finished thicknesses and copper construction. Revalidate the corrected noise behavior after a stack change because geometry and return paths may change with it.
Review the converter noise paths on your RF board. Send the converter schematic and layout, power tree, stackup, switching modes, receiver frequency plan, correlated spectra and acceptance targets. Include the tested changes and their operating conditions. Organize the fabrication and review requirements with the RFQ builder before requesting a quotation.