Spread spectrum clocking, or SSC, deliberately varies a clock’s frequency so its spectral energy is distributed over a range instead of concentrated at fixed frequencies. It can lower measured emission peaks, but it also changes the timing seen by the receiving circuit. Use it only after confirming that the entire clock path supports the modulation, then verify both emissions and system operation.
The first decision is not “how much spread gives the biggest dB improvement?” It is “what frequency range, modulation rate and timing behavior can this system accept?” This guide explains those parameters, compares center and down spreading, and gives a calculation and release worksheet. Numerical examples are analytical teaching cases, not measurements of a manufactured PCB.

Spread spectrum clocking meaning: intentional frequency variation
A fixed clock produces spectral components around its repetition frequency and harmonics. SSC modulates that repetition frequency in a controlled way. A receiver measuring a limited frequency band may then observe less energy concentrated in one peak. The technique redistributes energy; it does not make every source of electromagnetic interference disappear.
Distinguish intentional modulation from unwanted random jitter. Both affect edge timing, but they have different descriptions and validation requirements. An interface designed to track a documented SSC profile can behave differently from a precision sampling circuit that needs a low-noise, spectrally clean clock.
It is also different from fixing the radiation path. A poor return path, an unsuitable connector transition or a cable carrying common-mode current can still radiate. Correct those design issues rather than using SSC to hide an unexplained failure at one analyzer setting.
The practical benefit is additional emission margin when the signal and system support it. The trade-off is a wider frequency excursion and a timing condition that must remain inside the receiver’s allowed behavior.
Spread spectrum clock modulation: read the four key parameters
Define nominal frequency, actual minimum and maximum frequency, modulation rate and modulation profile. A percentage is not enough because vendors can define “depth” as one-sided excursion, total span or a field interpreted differently by the selected mode.
| Parameter | Practical meaning and analytical example |
| Nominal clock frequency | 100 MHz in this example; the unmodulated period is 10 ns. This is not the modulation rate. |
| Center spread ±0.5% | 99.5–100.5 MHz, a total frequency span of 1 MHz. |
| Down spread 1% | 99–100 MHz, also a 1 MHz span, but without exceeding the nominal upper frequency in the idealized profile. |
| Down spread 0.5% | 99.5–100 MHz, only a 0.5 MHz span. It is not the same excursion as center spread ±0.5%. |
| Modulation frequency | Assume 30 kHz: one modulation cycle takes 33.3333 µs. Many 10 ns clock cycles occur during that sweep. |
| Modulation profile | A triangle, sinusoid or another supported profile determines how frequency moves with time. Use the device’s defined profile rather than assuming all SSC generators behave alike. |
Frequency and period move inversely. For the center-spread example, the period at 100.5 MHz is 9.95025 ns, while at 99.5 MHz it is 10.05025 ns. For the 99–100 MHz down-spread example, the periods range from 10 ns to 10.10101 ns. These are instantaneous-period endpoints, not RMS jitter values.
Write the frequency endpoints explicitly in a design review. That makes a factor-of-two depth misunderstanding visible before anyone programs a register.
Center spread vs down spread: compare the actual frequency limits
Center spreading moves above and below a nominal frequency. Down spreading remains at or below the nominal frequency in its specified modulation profile. The choice depends on frequency ceilings, receiver tracking and application requirements, not on which name sounds quieter.
| Mode for a nominal 100 MHz clock | Decision, cost and boundary |
| No intentional spreading | Use when the receiver or precision task does not support SSC. Timing still contains real oscillator and distribution imperfections; “SSC off” does not mean zero jitter. |
| Center ±0.5%, 99.5–100.5 MHz | Provides a 1 MHz excursion around nominal. The receiving path must tolerate the upper frequency as well as the lower one; check absolute limits and all additional tolerances. |
| Down 1%, 99–100 MHz | Uses the same idealized 1 MHz span without the center mode’s upward excursion. Its lower average frequency and longer period endpoints may affect throughput or synchronization. |
| Down 0.5%, 99.5–100 MHz | Has half the span of the center ±0.5% example. Do not claim equal spectral spreading merely because both settings contain “0.5%.” |
TI’s SPRABY0A configuration report cautions that center spread raises the highest interface frequency, while down spreading lowers average frequency. Those observations explain the basic trade-off; the report’s specific SoC settings are not universal operating limits for other devices.

Spread spectrum clocking configuration: start at the receiver
Map the clock tree from source to every receiving function. Identify where modulation is generated and what each intervening buffer, divider, multiplier or PLL does with it. A control labeled “spread spectrum” does not establish which branches are affected.
- Read the receiver/interface requirements: allowed clock frequency range, SSC mode and depth, modulation rate, jitter definition and clock architecture.
- Identify the actual clock path and the devices that must accept the modulated input. Check configuration and operating modes, not just component names.
- Translate the proposed percentage into minimum and maximum frequency and period. Include the applicable source accuracy and other tolerances using the required method.
- Check the generator’s exact parameter units and profile. Confirm whether a software “depth” field means half-span or full span in that mode.
- Program only a supported configuration, read it back where possible and measure the output. Verify startup, lock and the required operating transitions.
A downstream PLL may track some modulation or filter it according to its design and operating conditions. Do not assume that modulation applied at a base clock reaches every multiplied clock unchanged. Conversely, do not assume that every PLL removes it. The exact architecture and documented behavior determine what must be measured.
If a board exposes an SSC option in firmware or BIOS, identify the clock domain and manufacturer’s intended setting before changing it. An overclocking anecdote is not a qualified configuration for a production interface.
Spread spectrum clock generator: verify the setting, not just the feature name
SSC may be integrated into a processor clock block or generated by a dedicated device. Choose by output format, voltage, frequency range, supported modulation, startup/configuration behavior and the receiving system’s requirements. A longer list of SSC modes is not proof of a better fit.
For a concrete documentation example, TI’s CDCE(L)913 datasheet, revision SCAS849I of August 2024, lists center and down settings separately in Table 7-2. A setting shown as ±0.5% in center mode is shown as −0.5% in down mode. The corresponding total frequency spans differ, as the calculation above demonstrates.
That family also illustrates why part suffixes matter: CDCE913 and CDCEL913 have different output-supply options. Verify the exact part, output levels, pins, loading and configuration rather than treating a similar model name as a drop-in replacement. The example identifies a documentation issue, not a recommendation for every clock tree.
Keep the selected configuration file or register record with the hardware revision. If production programs nonvolatile settings, define how programming success is verified. A correctly assembled board with the wrong saved clock configuration can still be the wrong product.
Spread spectrum clocking PCIe: the clock architecture matters
Do not enable SSC on a PCIe reference simply because another PCIe board uses it. Common-clock and separate-reference arrangements have different requirements, and both link partners and the clock distribution must support the intended arrangement.
TI’s November 2023 application note SNAA386 describes a common-clock SSC example with 0% to −0.5% down spread and a 30–33 kHz modulation frequency. At an ideal nominal 100 MHz, that percentage corresponds to 99.5–100 MHz before other tolerances are considered. The same note distinguishes separate-reference architectures.
Use those figures as a scoped manufacturer explanation, not a replacement for the applicable PCI-SIG specification and device documentation. Verify the required generation, topology, reference architecture and allowed SSC settings for the actual link. In particular, do not infer independent-clock compatibility from a successful shared-clock example.
Validate the link using the prescribed timing and compliance method, relevant traffic and operating conditions. Successful enumeration is a useful first check, not complete proof of long-term margin or compliance.
EMI measurement with SSC: keep the receiver settings comparable
The observed benefit depends on the emitted spectrum and how the measurement receiver observes it. Record resolution bandwidth, detector, sweep/dwell behavior, operating mode, cables, enclosure and measurement geometry. Changing these between SSC-off and SSC-on tests can create a misleading comparison.
A simple energy-per-band estimate helps explain the mechanism. Suppose an idealized total power is distributed uniformly across 1 MHz and the observation bandwidth is 120 kHz. The ratio is 1,000/120 = 8.3333; 10 log10(8.3333) = 9.20819 dB. This is a teaching estimate under a uniform-power assumption, not a predicted compliance improvement.
Actual modulation has a profile and spectral structure; peaks are not necessarily uniform across that span. Receiver detector and timing behavior, source harmonics and the coupling path also matter. Analog Devices’ spread-spectrum discussion uses a similar simplified bandwidth argument and distinguishes estimates from measured peak behavior.
Measure the complete required emission range and operating modes with the applicable method. A reduced peak at the fundamental does not prove every harmonic or nearby band improved. Record the worst relevant results and margin; do not promise a fixed dB reduction from a percentage setting alone.
Spread spectrum clocking applications: verify the benefit and the cost
Display and digital clock domains. SSC can be a supported way to reduce clock-related peaks. For example, TI publishes an AM62x display-subsystem procedure tied to that SDK and clock configuration. Confirm the panel/interface and receiver allow the resulting frequency excursion, then check the required display timing, image stability and operating transitions as well as emissions. Do not copy a platform-specific setting into an unrelated SoC.
Serial interconnects. Use the allowed reference-clock architecture and modulation profile. Verify both link partners, timing measurements and relevant traffic/error behavior. Preserve the clock settings in the test record. For measurement definitions, use the site’s jitter measurement guide; an intentional low-frequency modulation should not be compared with an unlabeled RMS jitter number.
Switching-power synchronization. Some regulators support frequency spreading or an appropriate modulated synchronization input. Check the regulator’s allowed synchronization range and tracking behavior, then test regulation, ripple, transients and the required emissions. A modulation profile suitable for a digital reference is not automatically suitable for a power converter.
Precision sampling and spectral measurements. Be cautious when the sampling clock’s spectrum directly affects the measured signal. ADI AN-1386 explains how clock spectral imperfections influence ADC output spectra. Establish whether deliberate spreading is compatible with the converter and measurement task, then verify the relevant SNR, spurs, resolution and timing behavior. There is no blanket rule that every ADC must use SSC or that none can; the system requirements decide.
In each application, the release decision has two sides: did the required emission result improve, and does the product still meet its functional and timing requirements? Passing only one side is not a completed design change.

Spread spectrum clocking checklist: release a measured configuration
Use this worksheet before enabling SSC in a design release. The first row is a worked numerical check; the other rows require actual device and test evidence.
| Input or evidence | Decision and next action |
| Example: nominal 100 MHz, center ±0.5%, 30 kHz triangle | Calculate 99.5–100.5 MHz, 1 MHz span and 33.3333 µs modulation period. Confirm the receiving path allows these conditions before programming. |
| Receiver and interface documents | Record applicable versions, architecture and allowed profile/range. If SSC support is not established, do not assume it. |
| Clock-tree and configuration revision | Identify which outputs are modulated and how downstream devices behave. Read back settings and measure the relevant output nodes. |
| EMI comparison record | Keep detector, bandwidth, setup and product mode comparable. Record actual worst relevant results, not just one improved trace peak. |
| Functional and timing results | Check the application-specific link, display, regulator or sampling requirements under the required conditions. Resolve degradation before release. |
If increasing spread reduces a spectral peak but breaks a receiving clock constraint, use a smaller supported setting or another EMI-control method. Do not trade an unexplained functional failure for a better-looking emissions screenshot.
Spread Spectrum Clocking FAQs
Does SSC reduce the total energy generated by a clock?
Its basic purpose is redistribution over frequency, not a universal reduction of total emitted energy. The measured peak benefit depends on the source, coupling path and receiver settings.
Are center ±0.5% and down 0.5% the same spread?
No. At 100 MHz, the center example spans 99.5–100.5 MHz, while the down example spans 99.5–100 MHz. Compare actual endpoints and the manufacturer’s parameter definition.
Does 30 kHz modulation mean the clock runs at 30 kHz?
No. It means the frequency-sweep pattern repeats every 33.3333 µs. A nominal 100 MHz clock still has clock periods near 10 ns.
Does down spread guarantee that the real clock never exceeds nominal?
The ideal modulation profile has that upper endpoint, but the actual clock also has specified accuracy and other tolerances. Evaluate the complete allowed frequency range, not just the nominal profile.
Can I enable SSC in BIOS without checking the clock domain?
Do not assume the option affects only the intended branch. Use the system manufacturer’s documentation and validate the affected interfaces and operating modes.
Will a downstream PLL remove all SSC?
Not necessarily. Its tracking and filtering behavior depend on the design and settings. Check the documented path and measure the relevant output.
Does the 9.20819 dB calculation guarantee that EMI improves by that amount?
No. It assumes uniform distribution over 1 MHz and a 120 kHz observation band. Actual profile, spectral peaks, detector behavior and coupling require measurement.
Is an operating PCIe link enough to release an SSC change?
No. Confirm the specified clock architecture, timing/compliance requirements and required operating/error tests. Initial link operation is only one check.
Custom High-Speed PCBs with Engineering DFM Support
Need custom PCBs for a timing-sensitive product? EBest Circuit supports multilayer and high-frequency PCB manufacturing with engineering DFM review. Reviewing the stackup and controlled-impedance geometry before fabrication helps align the manufactured construction with your released design; PCB assembly support keeps board and component requirements connected.
Send your Gerber or ODB++ files, stackup, material and impedance requirements, quantity and delivery needs to sales@bestpcbs.com. Add the BOM and placement data for assembly. Request a manufacturing review and project-specific quotation with construction and inspection requirements clearly identified. Clock configuration and product EMC qualification remain part of the system validation plan.
Technical References
- TI SPRABY0A, spread-spectrum parameters and configuration, revised November 2017.
- Analog Devices: Clock Generation with Spread Spectrum, spread definitions and measurement-bandwidth reasoning.
- TI CDCE(L)913 datasheet, SCAS849I, August 2024, Table 7-2.
- TI SNAA386: Clocking for PCIe Applications, November 2023, section 3.3.
- TI AM62x SDK: SSC for the Display Subsystem, platform-specific configuration.
- Analog Devices: Spread Spectrum Frequency Modulation Reduces EMI, power-converter and detector considerations.
- Analog Devices AN-1386, sampling-clock spectrum and ADC measurements.






































