[{"data":1,"prerenderedAt":68},["ShallowReactive",2],{"blog-pcb-signal-ringing-termination-en":3},{"title":4,"description":5,"date":6,"lastUpdated":6,"category":7,"image":8,"imageAlt":9,"readingTime":10,"wordCount":11,"timeRequired":12,"htmlContent":13,"faq":14,"tags":31,"slug":37,"sourceLocale":38,"availableLocales":39,"jsonld":44},"PCB Signal Ringing: Find the Cause, Then Pick the Termination","Investigate PCB signal ringing using propagation delay, circuit models and probe checks, then compare termination options for the actual interface.","2026-11-07","design","/assets/blogs/pcb-signal-ringing-termination/ringing-termination-original.webp","Conceptual point-to-point PCB signal path and illustrative waveforms with and without ringing",12,2228,"PT12M","\u003Cp>A clock edge that overshoots and rings at a receiver can reflect a mismatch along the interconnect, resonance involving package and load parasitics, power or reference disturbances, measurement loading, or a combination of these effects. The observed peak must be assessed against the actual receiver&#39;s limits and measurement uncertainty.\u003C/p>\n\u003Cp>Compare the propagation delay with the edge rate, inspect the topology and return path, and test proposed changes at the receiver. A termination is a design choice to verify against signal quality, timing, power and the interface specification.\u003C/p>\n\u003Ch2>1. Investigating Possible Causes\u003C/h2>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Possible contribution\u003C/th>\n\u003Cth>What to investigate\u003C/th>\n\u003Cth>Useful comparison\u003C/th>\n\u003C/tr>\n\u003C/thead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Interconnect reflections\u003C/td>\n\u003Ctd>Delay, characteristic impedance, source, receiver and branches\u003C/td>\n\u003Ctd>Compare receiver waveforms with a modeled or measured topology change\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Local resonance\u003C/td>\n\u003Ctd>Package and interconnect inductance, load and probe capacitance, damping\u003C/td>\n\u003Ctd>Compare a circuit model and a controlled change in loading or resistance\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Shared supply or reference disturbance\u003C/td>\n\u003Ctd>Simultaneous switching, return path and supply behavior\u003C/td>\n\u003Ctd>Compare switching and quiet nets using appropriate reference and probing\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Measurement system\u003C/td>\n\u003Ctd>Probe bandwidth, capacitance, grounding and loading\u003C/td>\n\u003Ctd>Compare safe, low-inductance probe connections or another suitable probe\u003C/td>\n\u003C/tr>\n\u003C/tbody>\u003C/table>\n\u003Cp>For an ideal series RLC model, 5 nH and 5 pF give an undamped natural frequency near 1.01 GHz and a critical total series resistance near 63 Ω, including the source resistance. These values are not predictions for an arbitrary board: actual damping, geometry and distributed effects change the response. A trace-length change can change both propagation delay and lumped parasitics, so its effect alone does not uniquely identify the cause. \u003Ca href=\"https://www.ti.com/lit/an/snla043/snla043.pdf\">TI&#39;s transmission-line discussion\u003C/a> describes how reflections can overlap.\u003C/p>\n\u003Ch2>2. When Does a Trace Need Termination?\u003C/h2>\n\u003Cp>One screening convention compares one-way delay with one-sixth of the signal rise time. It helps identify when a transmission-line model deserves attention; it does not guarantee distinct reflections, acceptable behavior below the threshold, or the need for an added resistor. \u003Ca href=\"https://ww1.microchip.com/downloads/en/Appnotes/doc6349.pdf\">Microchip&#39;s AT91 note\u003C/a> discusses the assumptions behind selected ratios and notes that capacitive loading can still cause overshoot on short traces.\u003C/p>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth align=\"right\">Signal rise time (10–90%)\u003C/th>\n\u003Cth align=\"right\">Critical length, FR-4 microstrip (~5.9 ps/mm)\u003C/th>\n\u003Cth align=\"right\">Critical length, FR-4 stripline (~7.0 ps/mm)\u003C/th>\n\u003C/tr>\n\u003C/thead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd align=\"right\">1 ns\u003C/td>\n\u003Ctd align=\"right\">~28 mm\u003C/td>\n\u003Ctd align=\"right\">~24 mm\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">500 ps\u003C/td>\n\u003Ctd align=\"right\">~14 mm\u003C/td>\n\u003Ctd align=\"right\">~12 mm\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">300 ps\u003C/td>\n\u003Ctd align=\"right\">~8 mm\u003C/td>\n\u003Ctd align=\"right\">~7 mm\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">100 ps\u003C/td>\n\u003Ctd align=\"right\">~3 mm\u003C/td>\n\u003Ctd align=\"right\">~2 mm\u003C/td>\n\u003C/tr>\n\u003C/tbody>\u003C/table>\n\u003Cp>Use the edge rate under the relevant driver, load and operating conditions, not clock frequency alone. For example, an assumed 300 ps edge gives a screening length near 8 mm at 5.9 ps/mm. Simulate the selected IBIS models and process, voltage and temperature corners, and check the receiver requirements; an IBIS model does not supply a single universal real edge rate.\u003C/p>\n\u003Cp>Delay per millimeter depends on the effective dielectric constant (\u003Cspan class=\"math-inline\">t\u003Csub>pd\u003C/sub> = √ε\u003Csub>eff\u003C/sub>/c\u003C/span>). The values above assume typical FR-4. Low-Dk laminates make signals faster and critical lengths slightly longer. For impedance and delay calculation, see our \u003Ca href=\"/en/blog/characteristic-impedance-pcb/\">characteristic impedance guide\u003C/a>.\u003C/p>\n\u003Ch2>3. What an Unterminated Line Does: Worked Bounce Example\u003C/h2>\n\u003Cp>Take a 3.3 V CMOS driver on a 50 Ω trace, driving a CMOS receiver that is effectively an open circuit. Two reflection coefficients decide everything:\u003C/p>\n\u003Cp>\u003Ccode>rho = (Z - Z0) / (Z + Z0)\u003C/code>\u003C/p>\n\u003Cp>For this illustrative lossless line, assume a linear resistive source, an ideal step and an ideal open load, without clamps, stubs or parasitic capacitance. The load reflection coefficient is +1; the source coefficient depends on \u003Ccode>Rs\u003C/code>. The first wave launched is \u003Ccode>V * Z0 / (Rs + Z0)\u003C/code>. Each far-end arrival adds the incident wave multiplied by \u003Ccode>(1 + rho_L)\u003C/code>; the wave returning to the source reflects again according to \u003Ccode>rho_S\u003C/code>.\u003C/p>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth align=\"right\">Source impedance \u003Cspan class=\"math-inline\">R\u003Csub>S\u003C/sub>\u003C/span>\u003C/th>\n\u003Cth align=\"right\">\u003Cspan class=\"math-inline\">ρ\u003Csub>S\u003C/sub>\u003C/span>\u003C/th>\n\u003Cth>Receiver voltage after arrivals 1, 3, 5, 7, 9 (V)\u003C/th>\n\u003Cth>Behavior\u003C/th>\n\u003C/tr>\n\u003C/thead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd align=\"right\">15 Ω (strong driver)\u003C/td>\n\u003Ctd align=\"right\">−0.54\u003C/td>\n\u003Ctd>5.08, 2.34, 3.82, 3.02, 3.45\u003C/td>\n\u003Ctd>54% overshoot, then ringing that crosses 3.3 V repeatedly\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">30 Ω\u003C/td>\n\u003Ctd align=\"right\">−0.25\u003C/td>\n\u003Ctd>4.13, 3.09, 3.35, 3.29, 3.30\u003C/td>\n\u003Ctd>25% overshoot, settles in about three round trips\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">50 Ω (matched)\u003C/td>\n\u003Ctd align=\"right\">0\u003C/td>\n\u003Ctd>3.30, 3.30, …\u003C/td>\n\u003Ctd>Clean single step at the receiver\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd align=\"right\">65 Ω\u003C/td>\n\u003Ctd align=\"right\">+0.13\u003C/td>\n\u003Ctd>2.87, 3.24, 3.29, 3.30\u003C/td>\n\u003Ctd>No overshoot; staircase approach to 3.3 V\u003C/td>\n\u003C/tr>\n\u003C/tbody>\u003C/table>\n\u003Cp>The 5.08 V peak and 2.34 V subsequent level are calculated values for this ideal model, not measured results or proof of damage or a false edge. Compare a real waveform against the specific receiver&#39;s absolute maximum voltage, clamp-current, duration and switching requirements, as well as the measurement uncertainty.\u003C/p>\n\u003Cp>Driver behavior varies with process, supply, temperature and drive setting and can differ between edges. Use the device&#39;s available models and data at relevant corners; a single linear resistance is an approximation.\u003C/p>\n\u003Ch2>4. Termination Options Compared\u003C/h2>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Scheme\u003C/th>\n\u003Cth>Where\u003C/th>\n\u003Cth>Static (DC) power\u003C/th>\n\u003Cth>Best fit\u003C/th>\n\u003Cth>Main limitation\u003C/th>\n\u003C/tr>\n\u003C/thead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>\u003Cstrong>Series (source)\u003C/strong>\u003C/td>\n\u003Ctd>Near the driver\u003C/td>\n\u003Ctd>Little termination DC current into an ideal high-impedance receiver\u003C/td>\n\u003Ctd>Point-to-point links where receiver timing permits\u003C/td>\n\u003Ctd>Intermediate loads can see partial steps; select for both edges and corners\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Parallel to ground\u003C/strong>\u003C/td>\n\u003Ctd>At the receiver\u003C/td>\n\u003Ctd>Sustained high-state current; the actual voltage depends on the driver\u003C/td>\n\u003Ctd>Interfaces designed for the resulting load\u003C/td>\n\u003Ctd>Check driver current, voltage levels and power\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Thevenin (split)\u003C/strong>\u003C/td>\n\u003Ctd>Two resistors at the receiver to supply and ground\u003C/td>\n\u003Ctd>Divider current flows even at steady state\u003C/td>\n\u003Ctd>Interfaces requiring a specified bias and equivalent termination\u003C/td>\n\u003Ctd>Check bias, driver current and power in all states\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>AC (RC)\u003C/strong>\u003C/td>\n\u003Ctd>Series resistor and capacitor as a load-end shunt branch\u003C/td>\n\u003Ctd>Ideal capacitor blocks DC after settling; switching consumes power\u003C/td>\n\u003Ctd>Some clock or data interfaces, after validation\u003C/td>\n\u003Ctd>Response and dissipation depend on components and data pattern\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>Differential across pair\u003C/strong>\u003C/td>\n\u003Ctd>Resistor between lines near the receiver\u003C/td>\n\u003Ctd>Depends on the interface and driver\u003C/td>\n\u003Ctd>Differential interfaces specifying a load\u003C/td>\n\u003Ctd>Differential termination alone does not establish common-mode behavior\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>On-die termination\u003C/strong>\u003C/td>\n\u003Ctd>Within a device\u003C/td>\n\u003Ctd>Device and configuration dependent\u003C/td>\n\u003Ctd>Interfaces that specify supported on-die termination\u003C/td>\n\u003Ctd>Verify configuration, range and operating conditions\u003C/td>\n\u003C/tr>\n\u003C/tbody>\u003C/table>\n\u003Cp>Choose according to the interface specification and complete topology. A source resistor is often a useful starting option for a point-to-point CMOS connection, but timing, source behavior and the receiver waveform determine whether it works.\u003C/p>\n\u003Ch2>5. Sizing Each Termination\u003C/h2>\n\u003Ch3>Series resistor\u003C/h3>\n\u003Cp>\u003Ccode>R_series = Z0 - Rs\u003C/code>\u003C/p>\n\u003Cp>For an ideal 50 Ω line and a linear 18 Ω source, this gives a 32 Ω starting value. Real output impedance is nonlinear and can differ for rising and falling edges; if the effective source impedance already exceeds the line impedance, a passive series resistor cannot produce a negative correction. \u003Ca href=\"https://www.ti.com/lit/an/scaa080a/scaa080a.pdf\">TI&#39;s clock-driver note\u003C/a> describes source matching for specific drivers and receiver conditions. In practice:\u003C/p>\n\u003Cul>\n\u003Cli>\u003Cstrong>Evaluate nearby standard values.\u003C/strong> Check overshoot, undershoot, settling and timing at the receiver for relevant corners; there is no universal round-down rule.\u003C/li>\n\u003Cli>\u003Cstrong>Check both edges.\u003C/strong> Pull-up and pull-down behavior can differ, as can load and operating conditions.\u003C/li>\n\u003Cli>\u003Cstrong>Place a source-matching resistor close to the driver.\u003C/strong> Include the package and short connection in the model. A resistor elsewhere can add damping but does not provide the same source match.\u003C/li>\n\u003Cli>\u003Cstrong>Consider a tuning footprint.\u003C/strong> Confirm the starting value and any 0 Ω option against the device and interface requirements.\u003C/li>\n\u003C/ul>\n\u003Ch3>Parallel and Thevenin\u003C/h3>\n\u003Cp>Parallel termination uses \u003Cspan class=\"math-inline\">R = Z\u003Csub>0\u003C/sub>\u003C/span> at the receiver. For Thevenin termination, pick the resistor ratio for the bias level and the parallel combination for \u003Cspan class=\"math-inline\">Z\u003Csub>0\u003C/sub>\u003C/span>:\u003C/p>\n\u003Cp>\u003Ccode>(R1 * R2) / (R1 + R2) = Z0\u003C/code> and \u003Ccode>V_bias = Vcc * R2 / (R1 + R2)\u003C/code>\u003C/p>\n\u003Cp>For \u003Cspan class=\"math-inline\">Z\u003Csub>0\u003C/sub> = 50 Ω\u003C/span> and a 1.65 V bias on 3.3 V, \u003Cspan class=\"math-inline\">R\u003Csub>1\u003C/sub> = R\u003Csub>2\u003C/sub> = 100 Ω\u003C/span>. Check that the driver can deliver the resulting current while still meeting the receiver&#39;s input high and low levels.\u003C/p>\n\u003Ch3>AC (RC) termination\u003C/h3>\n\u003Cp>An RC termination uses a series resistor and capacitor as a shunt branch at the load; it is different from a series AC-coupling capacitor in the signal path. The capacitor blocks DC after settling in an ideal circuit, but transitions draw current and dissipate power. Select the branch using the particular interface, edge, data pattern, driver and load, and verify the waveform and power over the intended conditions. \u003Ca href=\"https://www.ti.com/lit/an/scaa045a/scaa045a.pdf\">TI&#39;s CDCVF2505 note\u003C/a> illustrates this topology for that device; it is not a universal component-sizing rule.\u003C/p>\n\u003Ch2>6. Topology: Multiple Loads on One Line\u003C/h2>\n\u003Cp>In the ideal matched-source, open-load model, the initial wave is half the final source voltage and doubles at the far end. A load along the line can see a partial step before the reflection returns. Its actual waveform and threshold crossings depend on the complete topology and loading.\u003C/p>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Topology\u003C/th>\n\u003Cth>Recommended approach\u003C/th>\n\u003C/tr>\n\u003C/thead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Single load at the end\u003C/td>\n\u003Ctd>Consider source termination; check receiver waveform and timing\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Several loads clustered at the far end\u003C/td>\n\u003Ctd>Evaluate source termination and the loading and delay of each stub\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Loads spread along the line (daisy chain)\u003C/td>\n\u003Ctd>Evaluate load-end termination, stubs and each receiver&#39;s threshold and timing\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Several loads in different directions (star)\u003C/td>\n\u003Ctd>Consider a suitable fanout buffer with each branch designed separately\u003C/td>\n\u003C/tr>\n\u003Ctr>\n\u003Ctd>Branches with a resistor in each branch\u003C/td>\n\u003Ctd>Model the branch junction and driver load; compare alternatives\u003C/td>\n\u003C/tr>\n\u003C/tbody>\u003C/table>\n\u003Cp>Stubs can create additional reflections and loading. Include them in the channel model rather than assuming that an end termination corrects every discontinuity.\u003C/p>\n\u003Ch2>7. Measuring Ringing Without Creating It\u003C/h2>\n\u003Cp>The probe and its ground connection can load a circuit and contribute ringing. Probe capacitance, connection inductance, scope and probe bandwidth, reference location and the signal itself all affect the displayed waveform. \u003Ca href=\"https://download.tek.com/document/02_ABCs-of-Probes-Primer.pdf\">Tektronix&#39;s probe primer\u003C/a> discusses these effects.\u003C/p>\n\u003Cul>\n\u003Cli>\u003Cstrong>Use a suitable low-inductance connection.\u003C/strong> Where electrically safe and supported by the probe, compare a short spring ground or appropriate probe with the original setup. Do not defeat protective grounding.\u003C/li>\n\u003Cli>\u003Cstrong>Check measurement bandwidth and loading together.\u003C/strong> The familiar \u003Ccode>0.35 / rise_time\u003C/code> relation applies to particular response shapes; meeting it alone does not guarantee accurate peak measurement.\u003C/li>\n\u003Cli>\u003Cstrong>Measure at the receiver when practical.\u003C/strong> Document the probe location, reference and conditions; another point on the line can have a different waveform.\u003C/li>\n\u003Cli>\u003Cstrong>Compare with simulation and other measurements.\u003C/strong> Model the selected device corners and stackup, and account for the probe. TDR on a suitable coupon or board can inform the impedance model when its setup and reference plane are known; see our \u003Ca href=\"/en/blog/tdr-testing-pcb/\">TDR testing guide\u003C/a>. Agreement does not by itself prove a unique cause.\u003C/li>\n\u003C/ul>\n\u003Ch2>8. Layout Rules and Fabrication Inputs\u003C/h2>\n\u003Col>\n\u003Cli>\u003Cstrong>Plan a continuous return path.\u003C/strong> Examine reference changes, vias and width changes as part of the transition; see our \u003Ca href=\"/en/blog/routing-discontinuity-hf-pcb/\">routing discontinuity guide\u003C/a> and \u003Ca href=\"/en/blog/pcb-parasitic-capacitance/\">parasitic capacitance guide\u003C/a>.\u003C/li>\n\u003Cli>\u003Cstrong>Place load-end termination for the actual topology.\u003C/strong> Include the termination connection and remaining stubs in the model.\u003C/li>\n\u003Cli>\u003Cstrong>Investigate shared disturbances.\u003C/strong> A quiet line changing when other outputs switch may indicate shared return or supply coupling, but it is not conclusive by itself. Compare with suitable probing and controlled switching patterns.\u003C/li>\n\u003Cli>\u003Cstrong>Include impedance tolerance.\u003C/strong> A nominal 50 Ω line with ±10% tolerance spans 45–55 Ω. An ideal 50 Ω load at those line extremes gives reflection coefficients of approximately +5.3% and −4.8%, respectively; real driver, load and geometry tolerances also matter.\u003C/li>\n\u003C/ol>\n\u003Cp>For \u003Ca href=\"/en/products/high-speed-pcb/\">high-speed PCBs\u003C/a> and \u003Ca href=\"/en/products/multilayer-pcb/\">multilayer PCBs\u003C/a>, list controlled-impedance nets, targets and tolerances with your stackup and Gerber or ODB++ data in a \u003Ca href=\"/en/quote/\">quote request\u003C/a>. Agree on the achievable tolerance, trace-width changes, coupon and TDR report scope for the specific order. See our \u003Ca href=\"/en/blog/impedance-control-pcb/\">impedance control guide\u003C/a>.\u003C/p>\n\u003Cdiv data-component=\"BlogQuickQuoteInline\">\u003C/div>\n\n\u003C!-- faq:start -->\n\u003Ch2>FAQ\u003C/h2>\n\u003Ch3>How do I know if a PCB trace needs termination?\u003C/h3>\n\u003Cp>Compare one-way delay with the relevant edge rate as a screening step. A one-sixth ratio corresponds to about 28 mm for a 1 ns edge or 8 mm for a 300 ps edge at an assumed 5.9 ps/mm, but it does not determine whether termination is required. Check the interface, topology and receiver waveform.\u003C/p>\n\u003Ch3>What value series termination resistor should I use?\u003C/h3>\n\u003Cp>For a simple linear matched-source model, \u003Ccode>R_series = Z0 - Rs\u003C/code> is a starting estimate: 50 Ω minus 18 Ω gives 32 Ω. Check nearby values against both edges, relevant corners, timing and receiver limits; place a source-matching resistor near the driver.\u003C/p>\n\u003Ch3>Why does my signal still ring after adding a series resistor?\u003C/h3>\n\u003Cp>Possible contributors include source mismatch, interconnect and package parasitics, stubs, shared return paths and probe loading. Compare safe probing setups and controlled topology changes with a suitable circuit or channel model; no single comparison uniquely identifies every cause.\u003C/p>\n\u003Ch3>Is parallel termination better than series termination?\u003C/h3>\n\u003Cp>It depends on the interface, topology and available drive current. A 50 Ω resistor with an ideal 3.3 V across it draws 66 mA; a real driver&#39;s output may drop under that load. Source termination can reduce steady-state load current but must meet timing at each receiver.\u003C/p>\n\u003Ch3>Can a scope probe cause ringing?\u003C/h3>\n\u003Cp>Yes. Ground connection inductance and probe capacitance can alter the measured waveform. Compare appropriate low-inductance connections and account for bandwidth and loading while following the probe&#39;s electrical safety requirements.\u003C/p>\n\u003C!-- faq:end -->\n\u003Csection class=\"related-links\" aria-label=\"Related\">\u003Ch3>Related links\u003C/h3>\u003Cul>\u003Cli>\u003Ca href=\"/en/blog/characteristic-impedance-pcb/\">Controlled Impedance PCB Design and TDR Testing\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/blog/tdr-testing-pcb/\">TDR Testing PCB: Advanced Quality Control for Mission-Critical Electronics\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/blog/routing-discontinuity-hf-pcb/\">Routing Discontinuity in HF PCB: Managing Impedance Transitions and Reflections\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/blog/pcb-parasitic-capacitance/\">PCB Parasitic Capacitance: How Much Is Too Much, and How to Cut It\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/products/high-speed-pcb/\">high-speed PCBs\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/products/multilayer-pcb/\">multilayer PCBs\u003C/a>\u003C/li>\u003Cli>\u003Ca href=\"/en/quote/\">quote request\u003C/a>\u003C/li>\u003C/ul>\u003C/section>",[15,18,22,25,28],{"question":16,"answer":17,"answerText":17},"How do I know if a PCB trace needs termination?","Compare one-way delay with the relevant edge rate as a screening step. A one-sixth ratio corresponds to about 28 mm for a 1 ns edge or 8 mm for a 300 ps edge at an assumed 5.9 ps/mm, but it does not determine whether termination is required. Check the interface, topology and receiver waveform.",{"question":19,"answer":20,"answerText":21},"What value series termination resistor should I use?","For a simple linear matched-source model, `R_series = Z0 - Rs` is a starting estimate: 50 Ω minus 18 Ω gives 32 Ω. Check nearby values against both edges, relevant corners, timing and receiver limits; place a source-matching resistor near the driver.","For a simple linear matched-source model, R_series = Z0 - Rs is a starting estimate: 50 Ω minus 18 Ω gives 32 Ω. Check nearby values against both edges, relevant corners, timing and receiver limits; place a source-matching resistor near the driver.",{"question":23,"answer":24,"answerText":24},"Why does my signal still ring after adding a series resistor?","Possible contributors include source mismatch, interconnect and package parasitics, stubs, shared return paths and probe loading. Compare safe probing setups and controlled topology changes with a suitable circuit or channel model; no single comparison uniquely identifies every cause.",{"question":26,"answer":27,"answerText":27},"Is parallel termination better than series termination?","It depends on the interface, topology and available drive current. A 50 Ω resistor with an ideal 3.3 V across it draws 66 mA; a real driver's output may drop under that load. Source termination can reduce steady-state load current but must meet timing at each receiver.",{"question":29,"answer":30,"answerText":30},"Can a scope probe cause ringing?","Yes. Ground connection inductance and probe capacitance can alter the measured waveform. Compare appropriate low-inductance connections and account for bandwidth and loading while following the probe's electrical safety requirements.",[32,33,34,35,36],"PCB ringing","series termination resistor","parallel termination","signal reflection","overshoot","pcb-signal-ringing-termination","en",[38,40,41,42,43],"de","cn","fr","ru",{"blog":45,"breadcrumb":54},{"@context":46,"@type":47,"headline":4,"description":5,"image":8,"url":48,"datePublished":6,"dateModified":6,"timeRequired":12,"keywords":49,"articleSection":7,"author":50,"publisher":53},"https://schema.org","BlogPosting","https://hilpcb.com/en/blog/pcb-signal-ringing-termination/","PCB ringing, series termination resistor, parallel termination, signal reflection, overshoot",{"@type":51,"name":52},"Organization","HILPCB",{"@type":51,"name":52},{"@context":46,"@type":55,"itemListElement":56},"BreadcrumbList",[57,62,66],{"@type":58,"position":59,"name":60,"item":61},"ListItem",1,"Home","https://hilpcb.com/",{"@type":58,"position":63,"name":64,"item":65},2,"Blog","https://hilpcb.com/en/blog/",{"@type":58,"position":67,"name":37,"item":48},3,1795278538170]