123 Main Street, New York, NY 10001

Submit Your BOM

IF-Sampling ADC Design for Radar and Test Instruments

IF-Sampling ADC Design for Radar and Test Instruments

← Back to:Analog-to-Digital Converters (ADCs)

This page explains how to design and choose IF-Sampling chains for mid-band signals (tens to hundreds of MHz): how to plan IF frequency and sampling rate, control jitter and images, build the IF front-end and filters, and select suitable ADCs so radar and instrumentation systems meet their SNR, SFDR and bandwidth targets.

What this page solves – IF-Sampling scenarios & pain points

This page focuses on mid-band IF-Sampling in radar and test instruments: signals that are first mixed down from RF to an intermediate frequency (IF) in the 5–500 MHz range and then digitized by an ADC using Nyquist-zone planning.

The goal is to give a clear system-level view of how RF front-end, mixer, IF filter and IF-sampling ADC work together, and to show how to choose sampling rate, jitter performance and front-end filtering so that the IF band is captured cleanly without images and aliases falling back into the useful bandwidth.

Typical IF-Sampling use cases

  • Superheterodyne radar receivers with IF at 10.7 MHz, 70 MHz, 140 MHz or 250 MHz.
  • Spectrum analyzers and communication service monitors using fixed or tunable IF bands.
  • General RF instruments that avoid direct RF-Sampling by converting to a convenient mid-band IF.

Why use IF-Sampling instead of pure RF or baseband sampling

  • Easier analog filtering and image rejection around a well-defined IF band.
  • Access to a wider range of mature, high-resolution ADCs at moderate sampling rates.
  • Compatibility with established superheterodyne RF front-ends and existing IF hardware.

Typical pain points this page addresses

  • Sampling rate selection is unclear, so images and aliases fall back into the IF band.
  • Jitter requirements cannot be quantified, making clock selection guesswork and over-designed.
  • IF filter and ADC driver design are vague, leading to poor flatness, ringing and elevated in-band noise.
  • Radar and instrument SNR/SFDR budgets are not traced through the chain, so bottlenecks are hard to locate.
System-level chain from RF front-end to IF-Sampling ADC Block diagram showing antenna or sensor, RF front-end, mixer with LO, IF band-pass filter and an IF-sampling ADC, with RF, IF and digital regions highlighted. RF IF Digital Antenna / Sensor RF Front-End LNA + BPF Mixer LO → IF IF BPF IF ≈ 70 MHz IF-Sampling ADC Nyquist-zone capture DSP / FPGA DDC & analysis

IF-Sampling basics and Nyquist zone concept

IF-Sampling refers to digitizing a band-limited signal whose center frequency lies in a higher Nyquist zone of the ADC (for example in the second or third zone), such that the band folds into the 0 to Fs/2 range after sampling and can be observed as if it were at baseband.

The signal of interest is centered at tens to hundreds of megahertz with a bandwidth that is much smaller than its center frequency. By choosing the sampling rate intentionally, the IF band is placed fully inside a chosen Nyquist zone and then folded into the first Nyquist zone without overlap from images or aliases.

Nyquist zones and bandpass undersampling

For an ADC with sampling rate Fs, the first Nyquist zone spans 0 to Fs/2, the second spans Fs/2 to Fs, and the third spans Fs to 3Fs/2. IF-Sampling is a form of bandpass (undersampling) where a narrowband signal is intentionally placed in one of these higher zones and then folded back into the first zone when sampled.

When the IF bandwidth is contained within a single Nyquist zone and the sampling rate is chosen correctly, the folding is one-to-one and the original spectrum can be reconstructed without distortion, even though Fs is lower than twice the highest RF or IF frequency.

Baseband sampling vs IF-Sampling vs RF-Sampling

  • Baseband sampling: signal energy is concentrated near DC up to a few megahertz; Fs is slightly above twice the highest baseband frequency. Typical for precision voltage, current and low-frequency sensor measurements.
  • IF-Sampling: signal center frequency is in the tens to hundreds of megahertz in a single Nyquist zone; the ADC samples at a moderate rate and folds the band into 0 to Fs/2. Typical for radar and RF instruments.
  • RF-Sampling: the ADC digitizes RF directly at hundreds of megahertz to gigahertz, with very wide analog bandwidth and strict clock phase-noise requirements. This is handled in the RF-Sampling dedicated page.
IF band folding from higher Nyquist zone into the first zone Illustration of Nyquist zones along the frequency axis with a narrow IF band in the second or third zone folding into a band in the first Nyquist zone. 0 Fs/2 Fs 3Fs/2 Nyquist zone 1 Nyquist zone 2 Nyquist zone 3 IF band Folded band Bandpass undersampling

Planning IF bandwidth, image locations and sampling rate

Once the IF center frequency and bandwidth are chosen, the next step is to place this band in a suitable Nyquist zone, understand where aliases and images will appear after sampling, and select a sampling rate that keeps unwanted bands away from the useful IF spectrum.

In IF-Sampling, every spectral line or band repeats at integer multiples of the sampling rate Fs and then folds into the 0 to Fs/2 range. A simple way to estimate where an IF component appears after sampling is to use the mapping falias = |k·Fs ± fIF|, where k is an integer that represents the repetition index along the frequency axis.

Example: 70 MHz and 140 MHz IF with moderate sampling rates

For an IF at 70 MHz with Fs = 200 MSPS, the first Nyquist zone extends to 100 MHz. The 70 MHz band sits inside the first zone and is observed around 70 MHz after sampling. Aliases of other interferers or images are repeated around multiples of 200 MHz and may fold back into the first zone if there is insufficient filtering.

For an IF at 140 MHz, Fs = 200 MSPS places the center in the second Nyquist zone between 100 MHz and 200 MHz. After sampling, this band folds into the first zone near 60 MHz. A different choice such as Fs = 250 MSPS shifts the folded location and moves images and aliases, which can be used to pull interfering bands away from the desired IF channel.

Practical guidelines for choosing Fs in IF-Sampling

  • Place the IF bandwidth near the middle of the chosen Nyquist zone rather than close to its edge, to reduce sensitivity to filter roll-off and to ease jitter and anti-alias filtering requirements.
  • Evaluate where strong interferers, harmonics and mixer images will fold by applying the aliasing relationship, and avoid sampling rates that cause these unwanted bands to land inside or directly adjacent to the useful IF band.
  • Consider slightly increasing or decreasing Fs to move aliases into regions where analog and digital filtering can provide more stopband attenuation without excessive filter order.
  • Prefer simple integer or rational relationships between the LO frequency and Fs when possible, since these relationships simplify digital downconversion, calibration and multi-channel synchronization.
Sampling rate choice and image planning for IF-Sampling Two frequency-axis examples showing an IF band and image bands for two different sampling rates, with arrows indicating how aliases move when the sampling rate changes. Scenario 1 — Fs1 0 Fs1/2 Fs1 IF band Image Scenario 2 — Fs2 0 Fs2/2 Fs2 IF band Image Adjust Fs to move image bands away from the useful IF band

Clock jitter and its impact on IF-Sampling SNR

In IF-Sampling, clock jitter sets a fundamental limit on the achievable signal-to-noise ratio at a given input frequency. Higher IF and wider bandwidth make the system more sensitive to sampling edge uncertainty, especially in radar and test instruments that demand high dynamic range.

A commonly used approximation for the jitter-limited SNR of a sinusoidal input at frequency fin is SNRjitter ≈ −20·log10(2π·fin·σjitter), where σjitter is the RMS clock or aperture jitter. This relationship shows that SNR falls as input frequency increases or as jitter becomes larger.

Comparing low and high IF under the same jitter

With a fixed jitter level, a low IF such as 10 MHz can support very high SNR, while a higher IF such as 70 MHz or 200 MHz experiences a significant reduction in the jitter-limited SNR. The same clock source that is adequate for low-frequency precision measurements may be insufficient for wideband IF-Sampling in radar or RF instruments.

When SNRjitter is converted to an effective number of bits, ENOB ≈ (SNR − 1.76) / 6.02, a drop of tens of decibels translates into several bits of lost dynamic range. This is why higher IF front-ends demand low-jitter clock generators, clean PLLs and careful clock distribution.

Jitter sources and practical estimation

  • External sampling clock source phase noise and period jitter, often dominated by the reference oscillator and PLL.
  • Additional jitter introduced by clock dividers, fan-out buffers and distribution networks that feed the ADC inputs.
  • ADC internal aperture jitter associated with the sampling switch timing and internal clock circuitry.
  • Combined RMS jitter from these contributors sets an upper bound on SNR at the IF frequency and therefore on the maximum ENOB that can be achieved in the system.

Detailed clock-tree and phase-noise budgeting is handled in the clocking and jitter design topic. The main focus here is to help estimate the jitter quality required for a given IF and SNR target and to highlight how strongly SNR depends on input frequency in IF-Sampling applications.

Jitter-limited SNR versus input IF frequency A simple curve showing jitter-limited SNR decreasing with input frequency, with markers at 10 MHz, 70 MHz and 200 MHz for a fixed jitter value. Input frequency (MHz) SNR jitter (dB) 0 50 150 250 60 70 80 90 100 10 MHz 70 MHz 200 MHz σ jitter – constant (for example 100 fs)

IF front-end and anti-alias filter design

The purpose of the IF front-end in an IF-Sampling system is to preserve the desired IF band with good flatness and linearity, while providing enough attenuation at image and alias frequencies before the signal reaches the ADC input. The front-end sets the loading, gain, noise and distortion environment seen by the converter.

Typical IF front-ends include a mixer output stage, one or more IF gain blocks, a band-pass or narrow low-pass filter, and the ADC input network. The design must consider impedance levels (for example 50 Ω single-ended, 100 Ω differential or high-impedance voltage-mode), filter topology and order, and the ability of the driver amplifier to deliver the required swing into the ADC over the full IF bandwidth.

Impedance, filter topology and stopband requirements

The IF stage can be implemented as a 50 Ω environment carried from the RF side, a conversion from 50 Ω to a high-impedance differential ADC input, or a fully high-impedance voltage-mode chain. Band-pass filters are often used for fixed IFs such as 70 MHz or 140 MHz, while narrow low-pass sections can complement mixer and RF filters to form an overall band-pass response.

The filter order and stopband attenuation are chosen based on the locations and strengths of image and alias bands derived from the sampling-rate planning. Image frequencies that would otherwise fold back into the first Nyquist zone must be suppressed by tens of decibels to keep their contribution below the required SNR and SFDR targets for the application. Higher sampling rates allow more relaxed filters by placing aliases farther away from the IF band.

IF driver amplifier selection

The IF driver amplifier must provide enough bandwidth, linearity and output swing to drive the ADC input network. A practical rule is to choose a small-signal bandwidth of at least three times the IF center frequency so that gain and phase remain well controlled across the IF band. For a 70 MHz IF, a driver with 200 MHz or higher bandwidth is typically preferred.

Linearity metrics such as SFDR, HD2 and HD3 must support the system dynamic range, since distortion products generated in the driver appear at the ADC input as spurs. Noise figure and input-referred noise density of the driver contribute to the total noise at IF, especially when the IF gain is concentrated in this stage. The driver output swing, common-mode level and load must match the ADC input structure and any series RC or transformer network used for anti-alias filtering.

IF front-end design checklist

  • Define IF center frequency and bandwidth based on the RF front-end and required signal bandwidth.
  • Use sampling-rate and Nyquist-zone analysis to identify image and alias frequencies that must be attenuated before the ADC.
  • Choose a band-pass or narrow low-pass filter topology and order that can deliver the required passband flatness and stopband attenuation at the image and alias frequencies.
  • Select an IF gain block with sufficient bandwidth, SFDR and noise performance, and verify that its output swing and common-mode level are compatible with the ADC input network.
  • Simulate the combined mixer output, driver, filter and ADC input network to verify stability, gain flatness and the overall anti-alias response at IF and image frequencies.
IF front-end chain with amplifier, band-pass filter and ADC input network Block diagram of mixer, IF amplifier, IF band-pass filter and ADC input network, with a small frequency response sketch showing passband and image stopband. Mixer RF → IF IF Amplifier Gain IF Band-Pass BW & image rejection ADC Input RC / XFMR ADC IF response Passband & image stopband

Image and spur suppression with analog filtering and digital downconversion

After the IF front-end and anti-alias filter have attenuated large image and adjacent bands, the ADC output still contains residual out-of-band energy and aliases. Digital downconversion (DDC) completes the job by shifting the IF band to baseband and applying narrowband filtering and decimation to further suppress images, noise and unwanted spurs.

The DDC path typically consists of a numerically controlled oscillator (NCO), digital mixers that generate I and Q components, and a cascade of decimation filters. The analog filter provides coarse image rejection before sampling, while the digital filters define the final channel bandwidth and stopband attenuation in the baseband I/Q domain.

Roles of analog filtering and digital DDC

  • The analog IF filter removes strong out-of-band signals and images before sampling, preventing them from folding directly into the Nyquist band and overloading the ADC.
  • The digital DDC chain shifts the IF band to baseband, shapes the passband and stopband response, and uses decimation to reduce data rate while maintaining the required in-band SNR and SFDR.
  • Combined analog and digital filtering must deliver the total image and spur rejection required by the radar or instrument application across the complete IF and baseband processing chain.

Minimal DDC structure: NCO, mixer and decimation filter

A minimal DDC path for IF-Sampling starts with an NCO that generates digital sine and cosine waveforms at a frequency aligned with the folded IF center. The ADC output stream is multiplied by these waveforms to produce I and Q components, effectively translating the IF band to baseband in the digital domain.

Following the digital mixer, decimation filters such as CIC and half-band or FIR stages provide additional stopband attenuation for images and aliases and reduce the sampling rate. The passband is tailored to the desired signal bandwidth, while the stopband fill in the rejection that the analog IF filter does not provide. Detailed filter architecture and coefficient design are typically handled in the digital backend and DDC design topic.

Practical considerations for DDC-based image suppression

  • Choose the NCO frequency in line with the LO and IF plan so that the translated baseband is centered where the digital filters have maximum control and minimal sensitivity to frequency offsets.
  • Size the digital filter passband to fully cover the IF bandwidth with adequate margin, and allocate stopband attenuation to complement the analog filter so the combined rejection meets system requirements.
  • Select decimation factors that avoid new aliasing within the baseband and that provide a balance between oversampling margin, filter complexity and downstream processing load.
Digital downconversion chain after IF-Sampling Block diagram showing ADC output feeding an NCO-driven I/Q mixer and decimation filters to create baseband I and Q streams, with image and noise shaping in the DDC block. Image & noise shaping in DDC ADC Output NCO Digital Mixer I / Q Decimation Filter CIC / FIR Baseband I Baseband Q

Typical architectures: radar IF and instrument IF-Sampling

IF-Sampling is widely used in radar receivers and RF measurement instruments. In both cases, the RF signal is converted down to an intermediate frequency, filtered and amplified, and then digitized by an IF-Sampling ADC. Digital downconversion and DSP complete the chain to extract range, Doppler or spectrum information.

The following examples show where IF-Sampling ADCs sit in typical pulsed and FMCW radar receivers and in spectrum analyzers or communication service instruments. The focus is on the IF path and Nyquist-zone sampling, not on waveform design or higher-level radar and instrument algorithms.

Radar IF-Sampling architecture

In pulsed and FMCW radar, the RF front-end receives echoes with very large dynamic range: strong clutter and very small target returns can be present at the same time. The front-end converts the RF to an IF band, typically tens to hundreds of megahertz, where an IF-Sampling ADC and DDC chain provide the bandwidth and SFDR needed for range and Doppler processing.

The radar IF chain can be summarized as RF front-end, mixer, IF amplifier and filter, IF-Sampling ADC and a radar DSP stage. The IF-Sampling ADC must support high SFDR so that clutter and spur levels do not mask weak targets, and low jitter so that the phase of the IF signal is stable enough for accurate range and Doppler estimation.

IF-Sampling in spectrum analyzers and communication instruments

Modern spectrum analyzers and communication instruments also use IF-Sampling to achieve wide observation bandwidths and high dynamic range at practical sampling rates. The RF input is conditioned, mixed to one or more IF stages and then digitized by a wideband ADC. Digital LO, DDC and FFT processing provide flexible resolution bandwidths and measurement functions.

IF-Sampling allows these instruments to use high-performance mid-speed ADCs rather than ultra-high-speed RF-Sampling converters. The IF bandwidth and sampling rate are chosen to balance span coverage, resolution bandwidth, spur performance and the complexity of the analog front-end and clocking.

Typical IF-Sampling chains in radar and RF instruments Two block diagrams showing a radar IF-Sampling chain and a spectrum analyzer IF-Sampling chain, both highlighting the IF-Sampling ADC in the signal path. Radar IF-Sampling chain Instrument IF-Sampling chain Radar Front-End Mixer IF Chain Gain + BPF IF-Sampling ADC Radar DSP Range / Doppler RF Input / Attenuator RF / IF Front-End Mixer IF Chain IF-Sampling ADC Analyzer DSP FFT / measurements

Engineering checklist: key IF-Sampling design parameters

An IF-Sampling design must satisfy system-level performance targets while respecting practical constraints in sampling, front-end design, converter choice and clocking. The following checklist can be used as a self-review list or as a template when sending requirements and questions to semiconductor vendors or distributors.

Grouping parameters into system, sampling, front-end, ADC and clocking categories helps keep the discussion structured and ensures that important IF-Sampling details such as Nyquist-zone placement, image rejection and jitter performance are not overlooked.

System-level parameters

  • IF center frequency fIF and IF bandwidth BWIF (for example 70 MHz, 20 MHz).
  • Target SNR, SFDR and overall dynamic range required by the radar or instrument use case.
  • Required frequency or range resolution (for example radar range bin spacing or spectrum analyzer RBW).

Sampling and timing

  • Planned sampling rate Fs and chosen Nyquist zone in which the IF band will reside before folding.
  • Folded IF band location within 0 to Fs/2 and the guard band margin to nearby aliases or images.
  • Allowed RMS clock jitter derived from the IF frequency and target SNR or ENOB, or a request for recommended jitter performance for a given IF and dynamic range.

Front-end and filtering

  • IF filter type and bandwidth (band-pass or low-pass), passband flatness and allowed ripple.
  • Required stopband attenuation at key image and alias frequencies identified in the sampling-rate plan.
  • Input and output impedance levels (for example 50 Ω single-ended, 100 Ω differential or high-impedance).
  • IF gain distribution and IF driver requirements, including bandwidth, SFDR and noise contribution.

ADC-level parameters

  • Nominal resolution and ENOB at the intended IF frequency and input level.
  • SNR, SFDR and IMD3 performance at IF, including spur levels at relevant tone spacing for the application.
  • Full-scale input range, interface type (differential or single-ended) and input impedance or recommended matching network.
  • Power consumption, supply options, package type and operating temperature range constraints.

Clocking and synchronization

  • Clock source type (external reference, on-chip PLL, clock generator) and acceptable phase-noise or jitter performance.
  • Clock distribution approach, including buffers and fan-out, and any available reference or synchronization signals.
  • Multi-channel synchronization requirements for phased arrays, MIMO or multi-channel radar, and supported sync mechanisms such as triggers or SYSREF-style alignment.
IF-Sampling engineering checklist categories Card-style diagram showing system, sampling, front-end, ADC and clocking as key groups of IF-Sampling design parameters. IF-Sampling engineering checklist System f_IF, BW_IF SNR, SFDR Sampling Fs, Nyquist zone jitter, folded band Front-End Filter BW, ripple Image rejection, gain ADC Resolution, ENOB SNR, SFDR, IMD3 Clocking Phase noise, jitter Sync, distribution

IF-Sampling ADC selection logic

IF-Sampling ADCs are chosen based on IF frequency, signal bandwidth, dynamic range and practical constraints such as sampling rate, power and interface type. Rather than revisiting converter architectures, this section focuses on how to map an IF-Sampling requirement to suitable ADC resolution, sampling rate and data sheet parameters, and then narrow down candidate device families.

The selection process starts from the application and IF specifications, then moves through bandwidth and dynamic range requirements to define target SNR, ENOB and sampling rate. Candidate ADC families are then evaluated based on IF-Sampling capability, input bandwidth, ENOB and SFDR at IF, along with interface, power and packaging constraints.

Key selection dimensions for IF-Sampling ADCs

  • System-level requirements: IF center frequency fIF, IF bandwidth BWIF, target SNR, SFDR and overall dynamic range.
  • Sampling plan: selected sampling rate Fs, Nyquist zone placement, folded IF band location and margin to nearby aliases and images.
  • Dynamic range capability: ENOB, SNR and noise floor at IF, together with SFDR and IMD3 performance at the intended input frequency and level.
  • Interface and implementation: input interface (differential or single-ended, full-scale range, impedance), output interface (CMOS, LVDS, JESD204), power, package and temperature range.

Typical IF-Sampling ADC operating ranges

For radar and broadband instrumentation IF paths, converters are typically in the 14–16 bit range with sampling rates around 125–500 MSPS. These devices are often pipeline, hybrid or pipelined-SAR converters with IF-Sampling specifications up to several hundred megahertz and SFDR in the 80–90 dBc range at representative IF frequencies.

For narrower-band communication IF and industrial IF-Sampling, 14-bit converters in the 65–250 MSPS range offer a good balance between power, noise floor and dynamic range. These devices may be based on low-power pipeline or high-speed SAR architectures and emphasize IF-Sampling capability, low noise spectral density and simpler clocking and layout.

In harsh and space-grade environments, radiation-tolerant pipeline converters with high input bandwidth and robust packaging are often used to implement IF-Sampling with long-term reliability and wide temperature range support.

Example IF-Sampling ADC families by application tier

Application tier Part number Resolution / Fs (max) IF-Sampling notes
Radar / broadband instruments AD9250 14-bit, 250 MSPS, dual IF-Sampling up to hundreds of MHz, JESD204B, high SFDR for multi-carrier and radar IF
Radar / broadband instruments AD9467 16-bit, 250 MSPS, single High dynamic range, wide input bandwidth, suited for high-IF radar and instrumentation
Radar / broadband instruments ADS42JB69 16-bit, 250 MSPS, dual High-linearity IF-Sampling, JESD204B output, often used in communications test and radar
Narrowband comms / industrial IF ADC3564 14-bit, 125 MSPS, single Low-noise IF-Sampling, very low noise spectral density and power for comms and control IF
Narrowband comms / industrial IF ADC3644 / ADC3664 14-bit, 65–125 MSPS, dual Dual low-power IF-Sampling channels with good dynamic range for comms and industrial IF
Narrowband comms IF ADC14x250 family 14-bit, 250 MSPS High-IF sampling for base-station and multi-carrier receivers in a 14-bit class
Narrowband comms / instruments ADS6145 14-bit, 125 MSPS, single Low-power IF-Sampling ADC for mid-band communication and instrumentation IF
Space / harsh IF-Sampling ADS5474-SP 14-bit, 400 MSPS, single Radiation-tolerant pipeline ADC with high input bandwidth for space-grade IF-Sampling

Data sheet parameters to prioritize for IF-Sampling

  • ENOB and SNR at IF: effective number of bits and SNR measured at the intended IF and input amplitude, rather than only at low frequency.
  • SFDR and IMD3 at IF: spur-free dynamic range and intermodulation performance at IF, critical for clutter-plus-target radar scenes and high-dynamic-range spectrum analysis.
  • Noise floor and noise spectral density: determines the noise floor in wideband FFTs and multicarrier applications, and should be compatible with system dynamic range targets.
  • Input bandwidth and IF-Sampling specification: confirms that the ADC maintains gain flatness and linearity at the chosen IF and that the device is characterized for IF-Sampling.
  • Aperture jitter and recommended clocking: supports the jitter budget derived from IF frequency and required SNR, including any clocking solutions suggested by the vendor.
  • Interfaces, power and package: differential or single-ended inputs, CMOS/LVDS/JESD204 outputs, total power, thermal characteristics and package options appropriate for the target environment.

Simplified IF-Sampling ADC selection flow

A practical selection flow starts from the application and IF specifications, then refines the requirements into target bandwidth, dynamic range, sampling rate and ENOB. Candidate ADC families are filtered by IF-Sampling capability and measured performance at IF, and the final choice incorporates front-end architecture, interface, power and packaging constraints.

  • Start from the application type (radar, broadband instrument, narrowband communication or industrial IF) and define fIF, BWIF, SNR and SFDR targets.
  • Use IF bandwidth and dynamic range to determine an approximate Fs range and required ENOB, and select a Nyquist zone and folded band placement that provide margin to images and aliases.
  • Filter candidate ADCs by IF-Sampling support, ENOB, SNR, SFDR and noise floor at the intended IF, then check input bandwidth and aperture jitter against the sampling plan.
  • Finalize the choice based on front-end compatibility, input and output interfaces, multi-channel synchronization requirements, power budget and packaging or environmental constraints.
IF-Sampling ADC selection decision flow Simple decision-flow diagram from application and IF specifications through bandwidth and dynamic range, to sampling rate and ENOB, and then to candidate IF-Sampling ADC families. Application & IF spec Radar / Instrument / Comms Required BW & DR f_IF, BW_IF, SNR, SFDR Sampling rate & ENOB plan Fs, Nyquist zone, jitter, ENOB Candidate IF-Sampling ADC family 14–16 bit @ 125–500 MSPS Pipeline / Hybrid / SAR

Request a Quote

Accepted Formats

pdf, csv, xls, xlsx, zip

Attachment

Drag & drop files here or use the button below.

IF-Sampling FAQs

This FAQ section closes remaining IF-Sampling questions around bandwidth and sampling-rate planning, jitter limits, image rejection, multi-channel clocking, bench verification and when to move from IF-Sampling to true RF-Sampling architectures. Each answer stays focused on IF-Sampling and does not re-derive ADC architectures, full clock-tree design or complex digital modulation schemes.

What is the difference between IF-Sampling and RF-Sampling in practice?

In an IF-Sampling receiver, the RF signal is first downconverted to an intermediate frequency (IF), typically from a few tens to a few hundred megahertz. The ADC then samples this IF band in a higher Nyquist zone, and digital downconversion moves it to baseband. The analog front-end includes mixers, IF gain and IF filters, and the ADC speed can remain in the low hundreds of megasamples per second.

In an RF-Sampling architecture, the ADC directly samples the RF band of interest, often at several hundred megahertz up to gigahertz. This simplifies or removes analog mixing stages, but demands much higher sampling rates and significantly tighter clock jitter and front-end bandwidth requirements.

IF-Sampling offers a compromise: it leverages mature 14–16 bit ADCs in the 100–500 MSPS range, while RF-Sampling is favored only when very wide instantaneous bandwidth or highly reconfigurable frequency coverage is needed and the system can tolerate the cost and complexity of very high-speed converters and clocking.

How should the IF frequency and sampling rate be chosen to keep images out of the useful band?

The IF frequency and sampling rate must be selected together. The IF band should fall well inside a chosen Nyquist zone so that, after folding into 0 to Fs/2, there is margin between the passband and the Nyquist edges. This reduces sensitivity to component tolerances and simplifies filter design.

A practical approach is to choose Fs and fIF such that aliased images of strong out-of-band signals land outside the folded IF passband, leaving room for analog and digital filters to attenuate them. IF bands placed near the middle of a Nyquist zone usually enjoy better robustness than bands located close to Nyquist boundaries.

Once a candidate pair of Fs and fIF is chosen, the main alias and image frequencies can be mapped, and the IF filter requirements can be derived from how much suppression the system needs at those locations.

How much oversampling is recommended for a 20 MHz IF bandwidth?

The theoretical minimum for a 20 MHz IF bandwidth is a sampling rate slightly above 40 MSPS. In an IF-Sampling design, more oversampling is typically used so that the IF band occupies only part of the Nyquist zone and filters have room to roll off.

As a rule of thumb, choosing Fs so that the occupied IF bandwidth is a modest fraction of the folded Nyquist band (for example an effective oversampling of several times the 20 MHz) can provide reasonable guard bands for images and transition regions for analog and digital filters. The exact oversampling factor depends on how aggressive the image and alias suppression requirements are and how much filter complexity is acceptable.

Oversampling should be sized jointly with IF frequency, Nyquist zone choice and filter design, rather than set as a single fixed number independent of the rest of the chain.

How does ADC aperture jitter limit SNR at a 70 MHz IF?

Aperture jitter introduces uncertainty in the instant at which the IF waveform is sampled. For a sinusoidal input at frequency fin, the jitter-limited SNR is often approximated by:

SNRjitter ≈ −20·log10(2π·fin·σjitter)

At 70 MHz, the product 2π·fin is several times larger than at 10 MHz, so the same RMS jitter will produce several decibels more SNR degradation. As IF frequency and bandwidth increase, the allowed clock and aperture jitter must decrease to keep SNR within the target range.

In practice, the total SNR is set by both jitter and the ADC’s own quantization and thermal noise. The jitter calculation provides a budget for how clean the sampling clock and aperture timing must be at a given IF.

Can a ΣΔ ADC be used for IF-Sampling, or is pipeline/SAR always preferred?

For wideband IF-Sampling at tens to hundreds of megahertz, pipeline, hybrid or high-speed SAR ADCs are usually preferred because they provide the required sampling rates and IF bandwidth with well-characterized SNR and SFDR. Many of these devices are explicitly specified for IF-Sampling up to a certain frequency.

ΣΔ converters are more often used for lower-frequency or narrowband applications where oversampling and digital filtering are used to achieve high resolution in a limited bandwidth. Some specialized ΣΔ solutions integrate downconversion and can handle low-IF scenarios, but they are less common for the wideband IF-Sampling ranges discussed in this context.

When evaluating a ΣΔ device for IF-Sampling, the key check is whether the data sheet explicitly supports the intended IF frequency and bandwidth, including clear SNR and SFDR specifications at those conditions.

How much image rejection is needed from the IF filter before the ADC?

The required image rejection depends on how much image and alias energy can be tolerated in the final baseband or IF spectrum. The IF filter and analog front-end should attenuate strong out-of-band signals enough that, after folding, they do not dominate the noise floor or appear as significant spurs within the useful band.

A common strategy is to allocate rejection between analog and digital filtering. For example, the IF filter may provide tens of decibels of attenuation at key image frequencies, while the DDC and decimation filters add further stopband attenuation. The combined rejection should align with the system’s SFDR and blocker requirements.

The exact number is derived from a system-level spur and noise budget. As a starting point, many IF-Sampling designs aim for the analog IF filter to deliver sufficient attenuation so that residual image energy is comfortably below the ADC’s in-band noise floor after digital filtering.

What happens if the IF band is placed too close to the Nyquist edge?

When the IF band is located near the Nyquist edge of the chosen zone, the system becomes more sensitive to clock frequency tolerances, LO drift and component variations. Small changes can push part of the signal out of the effective passband or cause additional overlap with aliases.

Anti-alias and reconstruction filters also have a limited transition region. Placing the IF band too close to Fs/2 reduces the available space for the filter roll-off, often forcing higher-order filters with tighter tolerances. Any residual filter droop or group-delay variation near the edge can distort the useful band.

For these reasons, IF-Sampling designs normally keep the folded IF band away from Nyquist boundaries and reserve some margin between the signal band and the edges of the Nyquist zone.

Can IF be directly digitized without an analog band-pass filter in front of the ADC?

In principle, a sufficiently wideband ADC and a clean, well-controlled environment might allow direct digitization of an IF signal with minimal analog filtering. In practice, most IF-Sampling systems benefit strongly from an analog band-pass or low-pass filter in front of the ADC to control out-of-band energy.

Without an analog filter, strong adjacent channels, LO leakage, mixer products and wideband noise can reach the ADC and fold into the Nyquist band. This can raise the noise floor, generate apparent spurs and even cause ADC overdrive when high-level interferers are present.

A practical IF-Sampling design therefore includes at least a modest analog IF filter stage to limit the spectrum seen by the converter, even when most of the channel selection and shaping is done digitally.

How can one IF-Sampling clock be shared across multiple radar channels?

Multi-channel radar systems require tight phase and time alignment between channels. To share one IF-Sampling clock, the usual approach is to generate a single low-jitter clock source and distribute it to all ADCs through well-matched clock buffers and routing with controlled skew.

In addition to the sampling clock, most systems use synchronization signals such as triggers, frame clocks or SYSREF-style references to align sampling phases, digital downconversion phases and data-capture frames across channels. Any remaining timing skew translates into phase error and can degrade beamforming and MIMO performance.

Clock-tree topology, buffer selection and PCB layout practices are therefore an integral part of the IF-Sampling design when multiple radar channels share the same sampling clock.

How can IF-Sampling performance be verified on the bench (SNR and SFDR)?

Bench verification typically starts with single-tone IF tests. A low phase-noise RF or IF generator produces a clean tone at the intended IF, which is routed through the mixer and IF front-end into the ADC input network. The ADC output is captured, and FFT analysis is used to measure SNR, SFDR and harmonics at different frequencies and input levels.

It is important that the test equipment performance exceeds the expected system performance, so that generator phase noise, clock quality and cabling do not dominate the measured SNR or spur levels. Repeating measurements over IF frequency, sampling rate and front-end gain settings gives a realistic picture of the IF-Sampling chain behavior.

For more complex use cases, multi-tone or modulated-signal tests can be added to verify intermodulation, blocker handling and effective noise floor in conditions closer to the final application.

Why does the digitized IF spectrum show unexpected spurs after sampling?

Unexpected spurs in the digitized IF spectrum can originate from several sources. Common contributors include aliased interferers that were not sufficiently suppressed by the IF filter, residual LO leakage and mixer products, clock spurs mixing with the IF signal, and nonlinear distortion in front-end amplifiers or the ADC itself.

Digital downconversion and filtering can also introduce or reveal spurs if NCO frequencies, gain settings or filter configurations are not aligned with the sampling and IF plan. PCB layout, grounding and coupling between clock, digital and analog traces can further inject periodic disturbances visible as spurs.

Systematic spur analysis usually involves varying IF frequency, sampling rate, LO frequency and clock configuration to see how spurs move. Changes in spur position with these variables often point to a specific mechanism such as aliasing, intermodulation or clock mixing.

When should an IF-Sampling architecture be replaced by a true RF-Sampling architecture?

An IF-Sampling architecture is well suited when a single or limited set of IF bands can cover the required channels and bandwidths with manageable analog front-end complexity. It becomes less attractive when multiple IF stages, many mixers and complex filter banks are required to support wide or reconfigurable frequency coverage.

Moving to an RF-Sampling architecture is often considered when wide instantaneous bandwidth, flexible spectrum capture, or dense multi-standard operation is needed and suitable RF-Sampling ADCs and clocking solutions are available at acceptable cost and power levels. Removing IF stages can simplify calibration and improve flexibility at the expense of higher-speed conversion and tighter jitter requirements.

The transition point is therefore application specific and should be based on a trade-off between analog front-end complexity, converter performance, clocking feasibility and overall system cost and flexibility.

icnavigator

ICNavigator helps engineers find verified ICs, alternatives, and fast quotes.

Explore
Categories
  • Power Management
  • MCU
  • Automotive ICs
  • Sensors
  • Interface & Transceivers
  • Analog Front-End
Get in Touch
  • Email: hello@icnavigator.com
  • WeChat/WhatsApp: +86-xxxx

© 2025 ICNavigator. All rights reserved.