betaFeedback

WorksEngineering

A Hybrid Digital/Analog Polyphase Filtering Approach for High-Speed RF-DACs

Abstract

This paper proposes a hybrid polyphase interpolation for RF-DACs that removes the need for full-rate digital commutation by using passive combining on the RF-DAC itself. A P-branch polyphase FIR filter is factored into M streams, each formed by a (P/M)-way digital commutator at fRF/M and driving a dedicated RF-DAC slice, giving either partial digital (1<M<P) or fully analog (M=P) recombination. We show that the slice array acts as a single full-rate DAC with an M-times longer zero-order hold, derive yield laws for slice gain, timing and unit-capacitor mismatch, and find that with retiming the binding requirement is sub-picosecond matching of the fRF clock arrival, so the best choice is the smallest M that closes digital timing. Two 28-nm digital macros, designed for a 12.8GSps output rate and verified by post-layout simulation, illustrate a 4-phase variant that closes timing up to 14.4GSps with its digital logic at fRF/4 and an 8-phase variant that closes timing up to 16GSps with its digital logic at fRF/8. The analog recombination is evaluated with a behavioral model of the slice array. The architectures target wideband software defined radio (SDR) platforms, including cognitive radio, electronic warfare, and satellite communications.

Introduction

Wideband software defined radios (SDRs) are central elements for modern wireless communications systems, including those in commercial and military usage. They enable frequency-agile transceivers that can accommodate a wide variety of waveforms from single hardware platforms [1]. SDRs have become relatively commonplace below 7 GHz (frequency range 1 (FR1)) [2], owing to the increasing sampling rate that is available for data converters. The advent of the radio-frequency digital-to-analog converter (RF-DAC) is a primary enabler for wireless transmitters (TXs), in that it can directly synthesize baseband (BB) waveforms and output them directly at radio frequency (RF) [3, 4, 5].

However, in both commercial and military networks, higher frequency spectrum, such as that used in both frequency range 2 (FR2) and frequency range 3 (FR3) bands is being proposed to offset the severe congestion that exists in FR1 bands [6]. A key challenge to scaling hardware systems to higher frequency spectrum is that the BB signals typically use

All authors are with the Bradley Department of Electrical and Computer Engineering, Virginia Tech, Blacksburg, VA 24061 USA. (Corresponding author: Jeffrey S. Walling, e-mail: [email protected]). This work was supported by the National Science Foundation (NSF) under grant #2314813. Simulation code and models available at https://github.com/noyades/Non-uniform-Sampling-RF-DAC.

wider bandwidths and in the case of RF-DACs the BB signals must be interpolated to higher sample rates to help suppress spectral images that arise from the process of sampling.

In conventional RF wireless systems (e.g., digital-BB + digital-to-analog converter (DAC) + RF front-end), the sampling images can be suppressed using active, high-order filters that serve as the reconstruction filter of the DAC. These are impractical when using an RF-DAC, where the reconstruction filter is typically a low-order passive RF filter acting as an impedance matching network. Instead, the sample rate must be increased as much as possible, and ideally as high as the output frequency, which enables easier suppression of the images. This is similar to the concept of oversampling in BB DACs.

Advanced CMOS process nodes are capable of supporting high sampling rates [7, 8, 9]. However, these are still traditional baseband architectures that require separate up-conversion chains and hence do not offer the flexibility of the RF-DAC. Furthermore, the designs feature highly customized digital architectures to meet stringent timing requirements. Time-interleaved ΔΣ\Delta\Sigma RF-DAC architectures [10] and direct-RF TXs with embedded FIR-like impulse responses [11, 12, 13, 14, 15, 16] have explored related techniques, but remain constrained in filter reconfigurability and achieve limited tap depth.

The challenge in purely analog interpolation methods is that they provide only a limited number of filter taps. In a time-interleaved (TI) converter array [17], the number of taps equals the number of interleaved phases. In a time-approximation filter (TAF), the limitation is on the available shaping of the clock pulse [15, 16]. Additionally, the shaping on the clock pulse can have other adverse effects such as spur generation. In analog FIR filters, the limitation is on the number of individual analog processing cores [13, 14]. This limits the out-of-band (OOB) rejection and adaptability to new waveforms. Hybrid approaches, as shown in Figure 1 that merge analog and digital filtering capabilities, are an alternative approach. A single-phase RF-DAC with a digital interpolation filter would rely on the filter operating at output speeds up to the RF carrier frequency for maximal image suppression. This is practical if operation is limited to FR1 frequencies (below 7 GHz), but becomes impractical as frequencies scale into FR3 (7–24 GHz) and FR2 (above 24 GHz), where even full-custom digital FIR filters have only approached 10 GHz [18].

RF-DAC interpolation architectures

Figure 1. RF-DAC interpolation architectures. (a) Conventional direct-digital FIR: requires logic running at fRFf_{\mathrm{RF}}. (b) Polyphase hybrid analog-digital FIR: MM lanes, each driving an RF-DAC slice, reduce the maximum digital rate to fRF/Mf_{\mathrm{RF}}/M; this paper implements M=4M = 4 and M=8M = 8.

This paper shows that the recombination of a polyphase interpolator can be split between digital commutators and the RF-DAC itself, giving either partial digital recombination followed by analog recombination, or fully analog recombination, and implements one variant of each (Figure 2, middle and right). A Farrow fractional-rate filter in software/FPGA decouples the fixed baseband data rate from the RF clock, and the multi-GHz integer-rate interpolation is decomposed into polyphase form on a dedicated mixed-signal ASIC macro. The 4-phase variant needs only four clock phases, easily derived from the RF local oscillator, but its digital logic must close timing at fRF/4f_{\mathrm{RF}}/4; the 8-phase variant needs a more complex multiphase clock network but halves the digital rate to fRF/8f_{\mathrm{RF}}/8.

Detailed polyphase interpolation filter block diagrams

Figure 2. Detailed polyphase interpolation filter block diagrams. (left) Conventional polyphase FIR integer rate resampling, with recombination in the digital domain. Output delay and sum operations run at fRFf_{\mathrm{RF}}. (middle) 4-phase polyphase decomposition: eight sub-filters running at fRF/8f_{\mathrm{RF}}/8, before parallel branches are recombined at fRF/4f_{\mathrm{RF}}/4, each driving a dedicated RF-DAC slice. (right) 8-phase polyphase decomposition: eight sub-filters running at fRF/8f_{\mathrm{RF}}/8, each driving a dedicated RF-DAC slice.

Section II describes the generalized recombination and the system partitioning. In Section III the recombination model, along with its mismatch, clock and in-band noise and distortion budgets is analyzed. Next, Section IV provides guidance on the optimal choice of the number of clock phases, MM. Implementation and post-layout results are presented in Sections V and VI. Finally, Section VII concludes the paper.

System Architecture and Partitioning

The block diagram of Figure 1(b), and the detailed signal-flow graphs of Figure 2, illustrate the proposed hybrid architectures for implementing a high-rate interpolation filter for an RF-DAC output. To achieve frequency agility while maintaining physical implementation viability, the architecture employs a distinct partitioning. The system relies on an initial Farrow filter structure for efficient fractional-rate resampling of the baseband signal to a rate that is an integer fraction of the desired RF output frequency. Farrow filters are well-suited for this role due to their ability to perform arbitrary-rate interpolation with a fixed set of filter coefficients [19, 20]. Given its mathematical complexity and requirements for runtime parameter adjustment, this fractional block is designated for flexible off-chip deployment (e.g., on an FPGA front-end).

The output of this off-chip Farrow filter then feeds the custom mixed-signal ASIC macro, which houses a polyphase FIR bank that generates multiple parallel hardware output streams, each operating at a manageable fraction of the target output frequency. As shown in Figure 2 (left), a pure-digital polyphase decomposition requires a digital commutator running at fRFf_{\mathrm{RF}} to reconstruct the output sequence. The proposed hybrid architectures move all or part of this recombination into the RF-DAC, where it is performed passively. This can be done using either current- or voltage-mode RF-DACs, by summing currents [21] or by controlling charge division on capacitor arrays [22], with the recombination controlled by interleaved clock phases.

Generalized Digital/Analog Polyphase Recombination

A length-NN FIR filter H(z)H(z) with coefficients h[n]h[n] is decomposed into PP polyphase sub-filters [23, 24, 25],

H(z)=∑k=0P−1z−kEk(zP),(1)H(z) = \sum_{k=0}^{P-1} z^{-k} E_k(z^P), \tag*{(1)}

where each sub-filter, with h[n]=0h[n] = 0 for n≥Nn \ge N, is defined as:

Ek(z)=∑n=0⌊N/P⌋−1h[nP+k]z−n.(2)E_k(z) = \sum_{n=0}^{\lfloor N/P\rfloor-1} h[nP+k]z^{-n}. \tag*{(2)}

Each sub-filter runs at fRF/Pf_{\mathrm{RF}}/P. The recombination in (1) is what must run at fRFf_{\mathrm{RF}}. Set D=P/MD = P/M, and index the sub-filters

M=1M = 1 (digital)M=2M = 2 (this work)M=4M = 4 (this work)M=8M = 8 (this work)
Sub-filter rate fRF/Pf_{\mathrm{RF}}/P1.6 GHz1.6 GHz1.6 GHz1.6 GHz
Digital commutators ×\times ways1×81 \times82×42 \times44×24 \times2none
Highest digital rate fRF/Mf_{\mathrm{RF}}/M12.8 GHz6.4 GHz3.2 GHz1.6 GHz
RF-DAC slices1248
Clock phases (spacing)—2 (180°)4 (90°)8 (45°)
Analog hold Tb=MTRFT_b = MT_{\mathrm{RF}}78.1 ps156 ps312.5 ps625 ps

Table I. Rates and resources of the generalized recombination, P=8P = 8, fRF=12.8 GHzf_{\mathrm{RF}} = 12.8\,\mathrm{GHz}

as k=ℓ+Mdk = \ell+ Md with ℓ=0,…,M−1\ell= 0, \ldots, M - 1 and d=0,…,D−1d = 0, \ldots, D - 1. The same filter then factors into two nested recombinations,

H(z)=∑ℓ=0M−1z−ℓFℓ(zM),(3)H(z) = \sum_{\ell=0}^{M-1} z^{-\ell} F_{\ell}(z^M), \tag*{(3)}

where Fℓ(z)F_{\ell}(z), the ℓ\ell-th MM-phase component of H(z)H(z), is assembled from DD of the sub-filters:

Fℓ(z)=∑d=0D−1z−dEℓ+Md(zD).(4)F_{\ell}(z) = \sum_{d=0}^{D-1} z^{-d} E_{\ell+Md}(z^D). \tag*{(4)}

The sum in (4) interleaves DD sub-filters into one stream at fRF/Mf_{\mathrm{RF}}/M: a DD-way digital commutator. The sum in (3) interleaves the MM streams yℓ[r]y_{\ell}[r] produced by the FℓF_{\ell} into the full-rate output, and it is this stage that the RF-DAC performs, with one slice per stream and MM clock phases spaced 360∘/M360^\circ/M apart at fRF/Mf_{\mathrm{RF}}/M [22, 26]. The pair (P,M)(P, M) thus spans a design space whose regimes are shown in the three columns of Figure 2:

  • M=1M = 1 (D=PD = P): fully digital recombination. A single commutator runs at fRFf_{\mathrm{RF}} and drives one DAC (Figure 2, left).

  • 1<M<P1 < M < P: partial digital recombination. MM commutators run at fRF/Mf_{\mathrm{RF}}/M, and their outputs are recombined in the analog domain by MM RF-DAC slices (Figure 2, middle, P=8P = 8, M=4M = 4).

  • M=PM = P (D=1D = 1): fully analog recombination. Every sub-filter drives its own slice with no digital commutators (Figure 2, right, P=M=8P = M = 8).

Table I lists the resulting rates. It is noteworthy that the digital commutators are synchronous logic whose correct timing is guaranteed by static timing analysis at sign-off, so they contribute no amplitude or timing error of their own, and every analog non-ideality (e.g., the staggered hold, the slice mismatch and the clocks of Section III) depends on MM alone. Additionally, the fastest digital clock required is fRF/Mf_{\mathrm{RF}}/M for any PP, so it is MM, and not PP, that sets the required digital timing closure; P≥MP \ge M only sets how parallel the FIR arithmetic is [27]. The optimal choice of MM is analyzed and quantified in Section IV.

Polyphase FIR Architecture

To demonstrate the effectiveness of the proposed techniques, we designed an interpolation filter targeting 802.11be-like 320 MHz signals at FR3 frequencies. Though no monolithic standard currently exists, this particular combination of wide signal bandwidth (320 MHz) and ultra-high output frequency (12.8 GHz nominal, up to 16 GSps signed off) serves to demonstrate the capabilities of the proposed technique for high-dynamic-range, wideband signals.

Filter Specification

The integer-rate interpolation from fRF/32f_{\mathrm{RF}}/32 to fRFf_{\mathrm{RF}} is split into a ×4\times4 pre-interpolator, from fRF/32f_{\mathrm{RF}}/32 to fRF/8f_{\mathrm{RF}}/8, followed by a ×8\times8 stage. The ×4\times4 pre-interpolator is designed to a 160 MHz passband, a 240 MHz stopband, and >80> 80 dB stopband attenuation (before coefficient quantization). The ×8\times8 stage keeps the same 160 MHz passband and rejects the images of the fRF/8f_{\mathrm{RF}}/8 stream, centered at multiples of 1.6 GHz.

To achieve these characteristics, the ×8\times8 stage cascades a ×2\times2 FIR filter, which compensates the passband droop, with the response of a 6-stage ×4\times4 cascaded integrator-comb (CIC) filter [28]. The sinc6^6 response of the CIC places its nulls at the image centers; its droop is removed by the compensation filter, which also enforces the stopband attenuation [29]. No CIC structure is built: the cascade is implemented as an equivalent direct-form FIR for polyphase decomposition.

Polyphase Decomposition and Coefficient Quantization

The coefficients of the equivalent ×8\times8 cascade were derived using the impulse response of the full structure, quantized to 16 bits for hardware execution, and decomposed according to (1) and (2) with P=8P = 8, giving 165 taps. The ×4\times4 pre-interpolator is a 95-tap Nyquist(4) filter, also with 16-bit coefficients, implemented as four polyphase branches. After quantization, the pre-interpolator attenuates its stopband by 66.8 dB, and the ×8\times8 cascade attenuates the images of the fRF/8f_{\mathrm{RF}}/8 stream by more than 90 dB, both well above the 55 dBc spurious-free dynamic range (SFDR) target of Section III.

To enable a rigorous comparative analysis between the 4- and 8-phase physical macro variants, we utilize a single set of master filter coefficients and apply the same polyphase decomposition framework to both. The 4-phase architecture combines interleaved pairs of the 8-phase branches in the digital domain, i.e., (3) and (4) with D=2D = 2, while the 8-phase architecture directly implements all eight physical hardware branches. This approach ensures both variants maintain an identical underlying filter response, isolating the post-layout evaluation to differences in digital logic power, clock routing, and layout congestion.

For the M=8M = 8 hardware instantiation, the input bitstream data is fed in parallel to all eight sub-filters on the macro, each operating at 1.6 GHz and outputting one sample per clock cycle. These eight outputs map to successive time offsets kTRFkT_{\mathrm{RF}} (k=0,1,…,7k = 0, 1, \ldots, 7, where TRF=1/fRFT_{\mathrm{RF}} = 1/f_{\mathrm{RF}}). These parallel streams directly drive the input ports of the eight RF-DAC slices, sequencing their updates across the 45°-spaced clock phases to passively reconstruct the 12.8 GHz analog output waveform.

Architecture Trade-offs

A single FIR at fRF=12.8f_{\mathrm{RF}} = 12.8 GHz (Figure 1(a)) exceeds what even full-custom digital FIR filters have achieved; these have only approached 10 GHz, by relying on power-of-two coefficients, bit-level pipelining and supply overdrive [18]. A fully digital polyphase decomposition (M=1M = 1) still needs an output commutator at fRFf_{\mathrm{RF}}. Both implemented variants avoid these constraints. The 4-phase variant (P=8P = 8, M=4M = 4) needs four 90°-spaced phases at fRF/4f_{\mathrm{RF}}/4, which a divide-by-4 counter clocked by fRFf_{\mathrm{RF}} provides directly, but its four two-way commutators must close timing at 3.2 GHz. The 8-phase variant (P=M=8P = M = 8) needs eight 45∘45^\circ-spaced phases, a more complex clock network, in exchange for removing the commutators, so its fastest digital clock is the 1.6 GHz sub-filter rate.

Recombination Model and Mismatch Effects

Staggered Zero-Order Hold

This section analyzes the analog stage of (3), the recombination of the MM streams on the RF-DAC. As argued in Section II-A, the digital commutators contribute no error of their own, so every result below depends on MM alone and holds for any D=P/MD = P/M, i.e., for partial and fully analog recombination alike.

The recombination performed on the slice array is not a one-hot commutator. On each slice update-clock edge all MM stream words are held, then loaded into the array sequentially, one per successive fRFf_{\mathrm{RF}} cycle. Once loaded, a slice continues to switch at fRFf_{\mathrm{RF}} with that data until its next update, so every slice drives the array for a full slice update period Tb=MTRFT_b = MT_{\mathrm{RF}}. If the slice weights are equal, the slices differ only in when their data changes, and the array output is a staggered zero-order hold,

y(t)=∑k=0M−1∑ryk[r]pTb(t−rTb−kTRF),(5)y(t) = \sum_{k=0}^{M-1} \sum_r y_k[r]p_{T_b}(t-rT_b-kT_{\mathrm{RF}}), \tag*{(5)}

where y(t)y(t) is the baseband-equivalent output, pTb(⋅)p_{T_b}(\cdot) is a TbT_b-wide unit pulse, and yk[r]y_k[r] is the stream that drives slice kk, i.e., the output of FℓF_\ell in (4) with ℓ=k\ell= k. Since yk[r]=v[rM+k]y_k[r] = v[rM+k], where v[n]v[n] is the output of the full-rate filter H(z)H(z), substituting n=rM+kn = rM+k in (5) gives

y(t)=∑nv[n]pTb(t−nTRF).(6)y(t) = \sum_n v[n]p_{T_b}(t-nT_{\mathrm{RF}}). \tag*{(6)}

Hold Response

By (6), the array is equivalent to a single DAC updated at fRFf_{\mathrm{RF}} whose zero-order hold lasts Tb=MTRFT_b = MT_{\mathrm{RF}} instead of TRFT_{\mathrm{RF}} [30, 31]. Its frequency response is the usual sinc of a zero-order hold, PTb(f)=Tbsinc⁡(fTb)P_{T_b}(f) = T_b \operatorname{sinc}(fT_b), but with nulls at multiples of fb=fRF/Mf_b = f_{\mathrm{RF}}/M rather than of fRFf_{\mathrm{RF}}; at RF they lie at ±mfb\pm mf_b from the carrier. The MM-times narrower sinc also increases the passband droop, but only to 0.04 dB (M=4M = 4) and 0.14 dB (M=8M = 8) at the 160 MHz band edge, which the FIR can pre-compensate if required. A sample-level simulation of the slice array agrees with (6) to within 2×10−152 \times10^{-15}, i.e., double-precision rounding, for both variants.

Effective Resolution

(6) also bounds the achievable resolution. Slices carrying successive time samples add full-scale output and update rate but never subdivide the unit element, so MM slices of BsB_s bits do not yield Bs+log⁡2MB_s+\log_2 M bits. Each slice quantizes independently, so the in-band signal-to-noise and distortion ratio (SNDR) improves over a single slice by only 10log⁡10M10\log_{10} M, i.e., 12log⁡2M\frac{1}{2}\log_2 M bits (+1.0 bit for M=4M = 4, +1.5 bit for M=8M = 8), which simulation confirms. The array therefore falls 12log⁡2M\frac{1}{2}\log_2 M bits short of a (Bs+log⁡2MB_s+\log_2 M)-bit DAC, even with perfectly matched slices. Section III-B treats unequal slice weights and timing skew, whose spurs at f0+mfbf_0+mf_b fall on the skirts of the sinc nulls.

Impact of Timing and Gain Mismatch on Spurious Outputs

With a per-slice gain error gk=1+ϵkg_k = 1+\epsilon_k and timing error/skew τk\tau_k, the output spectrum is

Y(f)=PTb(f)∑k=0M−1gkAk(f)e−j2πf(kTRF+τk),(7)Y(f) = P_{T_b}(f)\sum_{k=0}^{M-1}g_kA_k(f)e^{-j2\pi f(kT_{\mathrm{RF}}+\tau_k)}, \tag*{(7)}

where Ak(f)A_k(f) is the branch spectrum, that is periodic with fb=1/Tbf_b = 1/T_b, and PTb(f)=Tbsinc⁡(fTb)P_{T_b}(f) = T_b\operatorname{sinc}(fT_b). Because every slice holds for exactly TbT_b, PTb(f)P_{T_b}(f) factors out of the sum. Hence, mismatch never moves the location of the null, it only reduces the cancellation it achieves across the null’s skirts.

For a single tone at f0f_0 each Ak(f)A_k(f) is a comb at fm=f0+mfbf_m = f_0+mf_b. Within ±fRF/2\pm f_{\mathrm{RF}}/2 of the carrier, the output has only the carrier (m=0m=0) and the polyphase spurs m=−M/2,…,M/2m=-M/2,\ldots,M/2, m≠0m\ne0. With x≜f0Tbx \triangleq f_0T_b, the hold weights the spurs relative to the carrier by the exact ratio:

PTb(fm)PTb(f0)=(−1)mxm+x.(8)\frac{P_{T_b}(f_m)}{P_{T_b}(f_0)} = (-1)^m\frac{x}{m+x}. \tag*{(8)}

To a first order in ϵk\epsilon_k and τk\tau_k, the spur-to-carrier amplitude is then:

Y(fm)Y(f0)≈(−1)mM[xm+xEq−j2πf0Tq],(9)\frac{Y(f_m)}{Y(f_0)} \approx\frac{(-1)^m}{M}\left[\frac{x}{m+x}\mathcal{E}_q-j2\pi f_0\mathcal{T}_q\right], \tag*{(9)}

where q=m mod Mq = m\bmod M, WM=e−j2π/MW_M=e^{-j2\pi/M}, and Eq=∑kϵkWMqk\mathcal{E}_q=\sum_k\epsilon_kW_M^{qk} and Tq=∑kτkWMqk\mathcal{T}_q=\sum_k\tau_kW_M^{qk} are the MM-point discrete Fourier transforms (DFTs) of the slice gain and timing errors [32, 33]. A skew τk\tau_k gives slice kk a phase error of 2πfmτk2\pi f_m\tau_k at the spur frequency fmf_m, and the hold attenuates that spur by x/(m+x)x/(m+x) from (8). Since fmx/(m+x)=f0f_m x/(m+x)=f_0, the product is 2πf0τk2\pi f_0\tau_k for every mm, so the timing sensitivity depends on the signal frequency f0f_0 and not on fRFf_{\mathrm{RF}} [34]. Skew of the fRFf_{\mathrm{RF}} clock itself behaves differently, because it also rotates the carrier of the slice; Section III-C shows that its per-spur weight is M/(1−x)M/(1-x) times larger, i.e., 12.5 dB at M=4M=4 and 19.0 dB at M=8M=8, which tightens the median skew budget by 10.6 and 14.9 dB (Table II). The gain error is attenuated by x/(m+x)x/(m+x) alone, least at m=−1m=-1, where ∣m+x∣=1−x\lvert m+x\rvert=1-x is smallest.

To turn (9) into a budget, let the slice errors be independent, zero-mean Gaussian random variables. The gain (or weighting) error is ϵk∼N(0,σg2)\epsilon_k\sim\mathcal{N}(0,\sigma_g^2), while the timing error is τk∼N(0,στ2)\tau_k\sim\mathcal{N}(0,\sigma_\tau^2), where σg\sigma_g and στ\sigma_\tau are the respective standard deviations of the errors. For a real white Gaussian sequence, the DFT bins q=1,…,M/2q=1,\ldots,M/2 are mutually independent, and bin M−qM-q is the complex conjugate of bin qq. The spurs at m=±qm=\pm q therefore have the same random amplitude, and we refer to them as group qq. For q<M/2q<M/2 the bin is a circular complex Gaussian of variance Mσ2M\sigma^2, so the spur power is exponentially distributed. The bin q=M/2q=M/2 is real, so its spur power is a scaled χ12\chi_1^2 variable. Taking the larger spur of each group, the mean spur power of group qq is

μq(τ)=(2πf0στ)2M,(10)\mu_q^{(\tau)}=\frac{(2\pi f_0\sigma_\tau)^2}{M}, \tag*{(10)}
μq(g)=σg2M⋅x2(q−x)2.(11)\mu_q^{(g)}=\frac{\sigma_g^2}{M}\cdot\frac{x^2}{(q-x)^2}. \tag*{(11)}

and the two mechanisms add in power. Timing skew contributes equally to every group, whereas gain error falls mainly on q=1q=1. The SFDR is set by the worst spur. Because the groups are independent, its cumulative distribution is the product of the group distributions,

F(t)=∏q=1M/2−1(1−e−t/μq)⋅erf⁡t2μM/2.(12)F(t)=\prod_{q=1}^{M/2-1}\left(1-e^{-t/\mu_q}\right)\cdot\operatorname{erf}\sqrt{\frac{t}{2\mu_{M/2}}}. \tag*{(12)}

For a single mechanism, every μq\mu_q is proportional to σ2\sigma^2, where σ\sigma is either στ\sigma_\tau or σg\sigma_g. Let η\eta denote the yield, i.e., the fraction of fabricated chips whose worst spur lies below a level tηt_\eta, so that F(tη)=ηF(t_\eta)=\eta. This level can then be written as tη=κησ2t_\eta=\kappa_\eta\sigma^2, and a yield η\eta at an SFDR target of SS dBc is achieved if [35]:

σ≤σmax⁡(S,η)=10−S/10/κη.(13)\sigma\leq\sigma_{\max}(S,\eta)=\sqrt{10^{-S/10}/\kappa_\eta}. \tag*{(13)}

The yield cost in SFDR is the gap between tηt_\eta and the largest group mean μˉ=max⁡qμq\bar{\mu}=\max_q\mu_q. For gain error, q=1q=1 dominates and the worst spur behaves like a single exponential variable: it lies 1.1–1.4 dB below μˉ\bar{\mu} at the median and 4.8 dB above it at 95% yield. For timing skew, all M/2M/2 groups have the same mean, so the worst of them grows with MM: 0.7 and 2.7 dB above μˉ\bar{\mu} at the median for M=4M=4 and 8, and 6.4 and 6.9 dB at 95%.

A given SFDR target follows from the slice resolution of the subsequent RF-DACs. In our example each slice is a sign-magnitude DAC with an 8-bit magnitude (255 unary units plus sign, i.e., 511 levels), whose quantization-limited signal-to-noise ratio (SNR) for a full-scale sine is 55.9 dB. Recombining MM slices adds 10log⁡10M10\log_{10}M, as shown in Section III-A2, i.e., 61.9 dB at M=4M=4 and 64.9 dB at M=8M=8, over the fbf_b Nyquist band for a full-scale sine. Hence, a reasonable limit for the SFDR target is set by the quantization-limited SNR of a single slice. For example, mismatch spurs at or below 55 dBc remain below the quantization noise of any one slice. For a 55 dBc SFDR with the tone at the band edge, f0=160f_0=160 MHz, (13) gives a median budget of στ≤3.26\sigma_\tau\leq3.26 ps and σg≤7.7%\sigma_g\leq7.7\% at M=4M=4, and στ≤3.66\sigma_\tau\leq3.66 ps and σg≤5.3%\sigma_g\leq5.3\% at M=8M=8. At 95% yield the budgets tighten to 1.70 and 2.27 ps, and 3.9% and 2.6%. For a different target SS, e.g., with higher-resolution slices, the error budgets scale by 10−(S−55)/2010^{-(S-55)/20}, i.e., they halve for every 6 dB of additional SFDR. To verify the closed-form estimates, a 2×1042\times10^4-trial Monte Carlo of the exact, non-linearized (7) was run, and it agrees with them to within 0.1 dB for every SFDR above 45 dBc; beyond στ≈7\sigma_\tau\approx7 ps the first-order expansion becomes conservative, by at most 1.25 dB at 20 ps. The yield law also explains why doubling MM relaxes the median timing budget by only 12% rather than the 2\sqrt{2} that the single-spur mean (10) suggests: the 1/M1/M reduction of each spur is largely offset by the larger number of independent spurs competing for the worst case. The gain budget tightens with MM because x=f0Tbx=f_0T_b doubles.

In the case of a modulated stimulus of bandwidth BB, each spur is spread over the band, at baseband offsets ν\nu between −B/2-B/2 and B/2B/2. Averaging ν2\nu^2 over a flat band gives ⟨ν2⟩=(B/2)2/3\langle\nu^2\rangle=(B/2)^2/3, so to leading order f0f_0 in (10) and (11) is replaced by B/2B/2, and the spur power drops by a further factor of 3. For gain this is 25–31 dB below the conventional interleaving result σg2/M\sigma_g^2/M. A Monte Carlo simulation with a 320 MHz modulated stimulus, evaluated on the worst spur band, gives budgets of 5.7 ps (M=4M=4) and 6.6 ps (M=8M=8) for timing skew and 14.5% and 10.0% for gain error. These are all looser than the single-tone budgets above, hence the band-edge tone sets the requirement.

To summarize, inter-slice amplitude matching is not the critical constraint; the fRFf_{\mathrm{RF}} clock arrival at the slices is (Section III-C), while intra-slice unit matching sets the in-band SNDR (Sections III-D and III-E).

To this point, the skew τk\tau_k considered is only at the instant when a slice’s data changes, and Section III-C shows that it is not the dominant timing error.

Clock Consideration on the Timing Budget

Each slice is timed by two clocks: its phase ϕk\phi_k of the slice update clock, at fbf_b and delayed by kTRFkT_{\mathrm{RF}}, which delivers its new input code, and the fRFf_{\mathrm{RF}} clock that switches its current sources or capacitors. In the proposed implementation the baseband word is retimed by the fRFf_{\mathrm{RF}} clock at the slice, so the update instant follows the local fRFf_{\mathrm{RF}} edge and ϕk\phi_k only has to land inside the retiming window. τkc\tau_k^c is the arrival skew of the fRFf_{\mathrm{RF}} clock at slice kk [36]. The slice output bk(t−τkc)c(t−τkc)b_k(t-\tau_k^c)c(t-\tau_k^c), with c(t)c(t) the carrier, has the baseband equivalent bk(t−τkc)e−j2πfRFτkcb_k(t-\tau_k^c)e^{-j2\pi f_{\mathrm{RF}}\tau_k^c}: besides delaying the data, the skew rotates the slice’s carrier by a constant phase across the hold duration. A carrier rotation is a complex slice gain, so the skirt weights it like ϵk\epsilon_k, and (9) becomes

Y(fm)Y(f0)≈(−1)mM[xm+x(ϵq−jΘq)−j2πf0Tq],(14)\frac{Y(f_m)}{Y(f_0)}\approx\frac{(-1)^m}{M}\left[\frac{x}{m+x}(\epsilon_q-j\Theta_q)-j2\pi f_0T_q\right], \tag*{(14)}

with Θq\Theta_q the DFT of θk=2πfRFτkc\theta_k=2\pi f_{\mathrm{RF}}\tau_k^c. A carrier skew therefore acts as a quadrature gain error of σθ=2πfRFσc\sigma_\theta=2\pi f_{\mathrm{RF}}\sigma_c, and its budget is the gain budget divided by 2πfRF2\pi f_{\mathrm{RF}}. Its weight 2πf0M/(m+x)2\pi f_0M/(m+x) exceeds that of the data-update term by M/(1−x)M/(1-x), i.e., 12.5 dB at M=4M=4 and 19.0 dB at M=8M=8. With retiming, Tq=Θq/(2πfRF)T_q=\Theta_q/(2\pi f_{\mathrm{RF}}) and the two terms combine into the weight 2πf0(M+m+x)/(m+x)2\pi f_0(M+m+x)/(m+x) of a whole-slice time shift, worst at m=±1m=\pm1. The phase clocks ϕk\phi_k, in contrast, cause no spur when their skew stays inside the retiming window w≤TRF/2w\leq T_{\mathrm{RF}}/2, and all MM slices clear it with probability η\eta for

σϕ≤wΦ−1(12(1+η1/M)).(15)\sigma_\phi\leq\frac{w}{\Phi^{-1}\left(\frac{1}{2}\left(1+\eta^{1/M}\right)\right)}. \tag*{(15)}

Beyond this timing window, a slice will latch the wrong fRFf_{\mathrm{RF}} cycle and the worst image rises to −40.0 dBc at M=4M=4 and −45.4 dBc at M=8M=8.

  • Assumed in the proposed design (retiming at each slice’s ϕk\phi_k from (15) with w=TRF/2w=T_{\mathrm{RF}}/2).

Shown in Table II are the analytically obtained timing skew error budgets to achieve the 55 dBc SFDR target. Each agrees to within 1% with a 2×1042 \times10^{4}-trial Monte Carlo of (7) in which each slice gain gkg_k also includes the carrier phase e−j2πfRFτke^{-j2\pi f_{\mathrm{RF}}\tau_k}. What is observed is that the slice update clocks are not the critical elements. Because each slice retimes its data on the local fRFf_{\mathrm{RF}} edge, in the median case, the M=4M = 4 and M=8M = 8 slice update clocks can be skewed by up to 28 and 23 ps rms before a slice latches the wrong fRFf_{\mathrm{RF}} cycle. The critical clock is the fRFf_{\mathrm{RF}} clock itself, because its skew also rotates the carrier of each slice. Its arrival at the slices must match to 0.79 ps (M=4M = 4) and 0.67 ps (M=8M = 8) rms at the median, and to ≈0.4\approx0.4 ps for 95% yield. Only differential skew between the slices is consequential; common skew to all of the slices does not result in a spur. Since the carrier rotation is a fixed phase per slice, it could be measured and removed by a digital phase rotation of each stream, as in the background calibration of time-interleaved ADCs [37]. In that case, only the data-update error would remain, with the few-picosecond budgets of (10).

Skew sourceM=4M=4M=8M=8
Data update, not retimed3.26 / 1.703.66 / 2.27
fRFf_{\mathrm{RF}} carrier only0.96 / 0.480.66 / 0.33
fRFf_{\mathrm{RF}} arrival, retimed*0.79 / 0.420.67 / 0.35
Phase clock ϕk\phi_k, retimed*27.7 / 15.722.5 / 14.3

Table II. RMS SKEW FOR 55 dBc SFDR, SINGLE TONE AT f0=160f_0=160 MHz, MEDIAN / 95% YIELD (PS)

In degrees of the slice update clock, a skew τ\tau is δ=360∘fbτ\delta= 360^\circ f_b\tau. The phase-clock budgets of Table II are therefore 32∘32^\circ (M=4M = 4) and 13∘13^\circ (M=8M = 8) rms at the median, and 18∘18^\circ and 8.3∘8.3^\circ at 95% yield; without retiming they would be 3.8∘3.8^\circ and 2.1∘2.1^\circ. Since fb=fRF/Mf_b = f_{\mathrm{RF}}/M, the fixed slip window shrinks in degrees as MM grows, and the 45∘45^\circ generator must be 1.8 to 2.5 times more accurate than the 90∘90^\circ one.

Unit-Capacitor Mismatch

Section III-B treats each slice as a single weight with gain error ϵk\epsilon_k. Inside a slice, however, the weight is built from unit capacitors, and their mismatch produces a code-dependent error. In the RF-DAC assumed for this design, each slice has a sign-magnitude I- and Q-array [38], each with NuN_u thermometer-coded unit capacitors; the implemented macros drive Nu=255N_u = 255 units per array through their binary-to-thermometer decoders. Unit ii of array c∈{I,Q}c \in\{\mathrm{I}, \mathrm{Q}\} in slice kk has a fixed relative error σuek,c,i\sigma_u e_{k,c,i}, where the ek,c,i∼N(0,1)e_{k,c,i} \sim\mathcal{N}(0,1) are independent and σu=σΔC/C\sigma_u = \sigma_{\Delta C/C} is the rms unit-capacitor mismatch [39].

Unit ii is switched on whenever the code magnitude is at least ii, and it then contributes ±1/Nu\pm1/N_u according to the sign of the code. Its error therefore adds σuek,c,i\sigma_u e_{k,c,i} times this on/off pattern to the output, and the output spectrum at frequency bin jj is

Yj=Yj(0)+σu∑k,c,iek,c,iBk,c,i(j),(16)Y_j = Y_j^{(0)} + \sigma_u \sum_{k,c,i} e_{k,c,i} B_{k,c,i}(j), \tag*{(16)}

where Yj(0)Y_j^{(0)} is the spectrum without mismatch, limited only by quantization, and Bk,c,i(j)B_{k,c,i}(j) is the spectrum of the on/off pattern of unit ii, held by slice kk as in (7). Because YjY_j is a weighted sum of Gaussian errors, its real and imaginary parts are jointly Gaussian, centered on Yj(0)Y_j^{(0)}, with the 2×22 \times2 covariance matrix Cj=σu2∑k,c,i[ℜBk,c,i;ℑBk,c,i][ℜBk,c,i;ℑBk,c,i]TC_j = \sigma_u^2 \sum_{k,c,i} [\Re B_{k,c,i};\Im B_{k,c,i}][\Re B_{k,c,i};\Im B_{k,c,i}]^{\mathsf{T}}. Rotating to the eigenvectors of CjC_j, with eigenvalues λ1\lambda_1 and λ2\lambda_2, splits the spur power into two independent terms,

Sj=λ1(ξ1+b1)2+λ2(ξ2+b2)2,ξ1,2∼N(0,1),(17)S_j = \lambda_1(\xi_1+b_1)^2 + \lambda_2(\xi_2+b_2)^2,\qquad\xi_{1,2} \sim\mathcal{N}(0,1), \tag*{(17)}

where b1,2b_{1,2} are the components of Yj(0)Y_j^{(0)} along the eigenvectors, divided by λ1,2\sqrt{\lambda_{1,2}}. This is a noncentral generalized χ2\chi^2 distribution, and its cumulative distribution FjF_j can be evaluated by a one-dimensional numerical integral.

The dominant spur is the image at −f0-f_0 caused by I/Q imbalance. If the I and Q arrays have slightly different total gains, a tone at +f0+f_0 leaks an image at −f0-f_0 whose amplitude is proportional to the gain difference, averaged over the MM slices. The phase of this image is fixed by the signal, and only its size and sign are random. One of the two terms in (17) therefore vanishes (λ2/λ1<10−7\lambda_2/\lambda_1 < 10^{-7}), and the image power is the square of a single Gaussian variable, i.e., a scaled χ12\chi_1^2 variable, as for the group q=M/2q = M/2 in Section III-B. Averaging over the slices makes its variance fall as 1/M1/M. Compared with the exponential distribution of a spur with random phase, the χ12\chi_1^2 distribution has a long tail, which is why the 95%-yield budget below is about three times tighter than the median.

Spurious tones in different bins share the same unit errors; hence, they are correlated, but the distribution of the worst spur is bounded by 1−∑j(1−Fj)≤P(max⁡jSj≤t)≤min⁡jFj1-\sum_j(1-F_j) \le P(\max_j S_j \le t) \le\min_j F_j, with the product ∏jFj\prod_j F_j as the point estimate. Only the 64 bins with the largest error variance or nominal level are kept; in a 2000-trial Monte Carlo simulation of the full spectrum the worst bin never fell outside of this set. For a 55 dBc SFDR the analysis gives a median budget of σΔC/C≤10.5%\sigma_{\Delta C/C} \le10.5\% (M=4M = 4, bounded by 10.2 and 10.8%) and 15.0% (M=8M = 8, bounded by 14.8 and 15.3%) and a 95%-yield budget of 3.7% and 5.3%, for M=4M = 4 and M=8M = 8, respectively. In this case, the bounds collapse because one spur is dominant. The median worst spur lies within 0.3 dB of the median I/Q image. Against the Monte Carlo, every point falls inside the 95% confidence interval of the simulated quantile, with a largest deviation of 0.4 dB.

In-Band SNDR Budget

The non-idealities in the prior subsections result in OOB spurs. However, with a modulated signal, the in-band SNDR is also important. The non-idealities are split into two classes. With static slice errors the array remains periodic in TbT_b, so by (7) their errors land at f+mfbf + m f_b, m≠0m \ne0, outside the channel for B<fbB < f_b, apart from a common gain and delay that any receiver removes. With these removed, a Monte Carlo simulation at the median budgets of Section III-B and Table II, with a 320 MHz orthogonal frequency-division multiplexing (OFDM)-like stimulus (10 dB peak-to-average power ratio (PAPR)), gives an in-band SNDR above 80 dB in 95% of the trials. Inter-slice mismatch is therefore an OOB emission problem only. Quantization, clock jitter and unit-capacitor mismatch, in contrast, place error inside the channel, and all of them add in power,

SNDR−1=SNRq−1+SNRj−1+SNDRu−1.(18)\mathrm{SNDR}^{-1} = \mathrm{SNR}_q^{-1} + \mathrm{SNR}_j^{-1} + \mathrm{SNDR}_u^{-1}. \tag*{(18)}

Each slice quantizes at fbf_b, so a fraction B/fbB/f_b of its noise falls in band, and recombination adds 10log⁡10M10\log_{10} M (Section III-A2). The resulting SNR is:

SNRq=SNR1+10log⁡10fRFB,(19)\mathrm{SNR}_q = \mathrm{SNR}_1 + 10\log_{10}\frac{f_{\mathrm{RF}}}{B}, \tag*{(19)}

where SNR1\mathrm{SNR}_1 is the Nyquist-band SNR of a single slice. The in-band quantization noise is thus set by fRFf_{\mathrm{RF}}, not by MM: for the aforementioned signal and resolution, simulated with the stimulus peak at 0.95 of full scale, SNR1=45.5\mathrm{SNR}_{1}=45.5 dB and SNRq=61.4\mathrm{SNR}_{q}=61.4 dB for both variants.

Clock Jitter

1) Clock Jitter: Unlike skew, jitter changes from edge-to-edge; hence, it raises the noise floor rather than producing a spur. As in Section III-C, the critical clock is fRFf_{\mathrm{RF}}. If its edge in RF cycle nn at slice kk is displaced by jk[n]j_k[n], the held input word hk[n]h_k[n] of that slice is rotated by θk[n]=2πfRFjk[n]\theta_k[n]=2\pi f_{\mathrm{RF}}j_k[n], and to a first order the output error is −j∑kθk[n]hk[n]-\mathrm{j}\sum_k\theta_k[n]h_k[n]. The jitter is split into a common-mode (e.g., the source and the shared clock tree), with an rms σcm\sigma_{\mathrm{cm}} value, and an independent mode due to each slice’s local buffer, with rms σloc\sigma_{\mathrm{loc}}. The common part rotates the whole output, whereas the local parts add in power, and since ∑k∣hk∣2≈∣y∣2/M\sum_k\lvert h_k\rvert^2\approx\lvert y\rvert^2/M for an oversampled signal they are averaged by MM. For white jitter, a fraction B/fRFB/f_{\mathrm{RF}} of the error falls in band, giving:

SNRj=fRF/B(2πfRF)2(σcm2+σloc2/M).(20)\mathrm{SNR}_{j}=\frac{f_{\mathrm{RF}}/B}{(2\pi f_{\mathrm{RF}})^2\left(\sigma_{\mathrm{cm}}^2+\sigma_{\mathrm{loc}}^2/M\right)}. \tag*{(20)}

For SNRj=55\mathrm{SNR}_{j}=55 dB this allows σcm=140\sigma_{\mathrm{cm}}=140 fs, or σloc=280\sigma_{\mathrm{loc}}=280 fs (M=4M=4) and 396 fs (M=8M=8), and (20) agrees with a Monte Carlo simulation to within 0.16 dB. The common term defines the phase-noise requirement of any RF-DAC at fRFf_{\mathrm{RF}}; when it is not white, its contribution is the phase noise integrated over the channel. At fRF=12.8f_{\mathrm{RF}}=12.8 GHz, σcm=140\sigma_{\mathrm{cm}}=140 fs corresponds to a white phase noise of −140-140 dBc/Hz, i.e., to −55-55 dBc integrated over the 320 MHz channel. Only the local term is specific to the architecture, and it is relaxed by M\sqrt{M}. The fRFf_{\mathrm{RF}} jitter is required to be at femtosecond levels, whereas the static fRFf_{\mathrm{RF}} skew of Table II may reach 0.7 ps, because static errors leave only a common gain and delay in band, while jitter is white and falls in band directly. Because the data are retimed, the same edge also moves the data transition, but this adds only M(B/2fRF)2/3M(B/2f_{\mathrm{RF}})^2/3 of the carrier term, at most −33.8-33.8 dB (M=8M=8). The jitter of the phase clocks ϕk\phi_k adds to their skew in (15) and is otherwise absorbed by the retiming.

Unit-Capacitor Mismatch

2) Unit-Capacitor Mismatch: The unit-capacitor errors of Section III-D produce, besides the I/Q image, code-dependent distortion, and with a modulated signal both lie in band. Since (16) is exactly linear in the unit errors, the in-band error after projecting out the common complex gain with Π⊥\Pi_{\perp} is a quadratic form,

PuS=σu2tr⁡Que,Qu=1Sℜ{BHΠ⊥B},(21)\frac{P_u}{S}=\sigma_u^2\operatorname{tr}Q_u\mathbf{e},\qquad Q_u=\frac{1}{S}\Re\{\mathbf{B}^{H}\Pi_{\perp}\mathbf{B}\}, \tag*{(21)}

where SS is the in-band signal power, e\mathbf{e} collects the 2MNu2MN_u unit errors ek,c,ie_{k,c,i}, and the columns of B\mathbf{B} are their in-band spectra Bk,c,i(j)B_{k,c,i}(j). As for the spurs, this is a weighted sum of squared Gaussian variables, and its CDF follows from Imhof’s integral [40]. Two properties set the budgets. First, the mean error power, σu2tr⁡Qu\sigma_u^2\operatorname{tr}Q_u, falls as 1/M1/M, because the errors of different slices are independent and average out. Second, a single term dominates: the I/Q image carries 75% of the error power, so the in-band error inherits the long tail of the image spur, and the 95%-yield budget is much tighter than the median one. For SNDRu=55\mathrm{SNDR}_u=55 dB, the unit capacitors must match to σΔC/C≤5.3%\sigma_{\Delta C/C}\leq5.3\% (M=4M=4) and 7.5% (M=8M=8) at the median, and to 2.4% and 3.4% at 95% yield. An independent time-domain Monte Carlo simulation of 2000 trials reproduces the mean to within 1% and both quantiles to within 6% in power. These budgets are 1.6 to 2 times tighter than the SFDR budgets of Section III-D: the SFDR counts only the largest spur, whereas the in-band error collects all of the distortion and, relative to the signal, grows with the back-off. In-band SNDR, not SFDR, therefore sets the size of the unit capacitors.

TermM=4M=4M=8M=8
Quantization SNRq\mathrm{SNR}_q61.4 dB61.4 dB
Common fRFf_{\mathrm{RF}} jitter σcm†\sigma_{\mathrm{cm}}^{{\dagger}}140 fs140 fs
Local fRFf_{\mathrm{RF}} jitter σloc†\sigma_{\mathrm{loc}}^{{\dagger}}280 fs396 fs
σΔC/C†\sigma_{\Delta C/C}^{{\dagger}} for median / 95% yield5.3 / 2.4%7.5 / 3.4%
Inter-slice mismatch∗^{*}>80>80 dB>80>80 dB
Total SNDR, (18)51.5 dB51.5 dB

Table 3. In-band SNDR budget, for the 320 MHz OFDM stimulus

†^{\dagger}Allowed value for 55 dB from that term alone. ∗^{*}At the median SFDR budgets, 95% of trials; gain and delay over-rode.

Table III summarizes the in-band budget. Like the SFDR budgets, the jitter and unit-capacitor budgets scale by 10−(S−55)/2010^{-(S-55)/20} for a different target SS. With quantization at 61.4 dB and jitter and unit-capacitor mismatch each at 55 dB, (18) gives 51.5 dB.

Optimizing the Number of Phases, M

By using the yield laws derived in Section III along with the timing overheads of a target process and standard cell library, we can determine how every requirement scales with MM, before any implementation, by first establishing digital and analog design budgets and constraints.

Digital Lower Bound

The fastest digital clock runs at fRF/Mf_{\mathrm{RF}}/M (§ 2-A), so each register stage has a cycle of MTRFMT_{\mathrm{RF}}, where TRF=1/fRFT_{\mathrm{RF}}=1/f_{\mathrm{RF}}. Out of this, every stage requires a fixed overhead that does not depend on MM: the flip-flop clock-to-Q and setup times, tovt_{\mathrm{ov}}, and the jitter tjt_j and skew tskt_{\mathrm{sk}} of the clock. The time left for combinatorial logic is then:

L(M)=MTRF−(tov+tj+tsk),(22)L(M)=MT_{\mathrm{RF}}-(t_{\mathrm{ov}}+t_j+t_{\mathrm{sk}}), \tag*{(22)}

which grows linearly with MM. Pipelining cannot recover a negative LL, because every added register requires the same overhead. Hence, once tov+tj+tsk≥TRFt_{\mathrm{ov}}+t_j+t_{\mathrm{sk}}\geq T_{\mathrm{RF}}, a direct-digital implementation (M=1M=1) cannot close timing at any pipeline depth. Even a positive LL is not sufficient: the datapath contains stages that cannot be split further, and the longest of them, which no further pipelining can shorten, sets a minimum logic delay Dmin⁡D_{\min}. Requiring L(M)≥Dmin⁡L(M)\geq D_{\min} gives the lower bound

M≥fRF(tov+tj+tsk+Dmin⁡),(23)M\geq f_{\mathrm{RF}}(t_{\mathrm{ov}}+t_j+t_{\mathrm{sk}}+D_{\min}), \tag*{(23)}

or, equivalently, for a fixed MM timing can be closed up to fRF,max=M/(tov+tj+tsk+Dmin⁡)f_{\mathrm{RF,max}}=M/(t_{\mathrm{ov}}+t_j+t_{\mathrm{sk}}+D_{\min}). The bound depends only on the technology/library for the implementation and the longest indivisible stage, not on the filter length, and on PP only through whether M<PM<P (commutator crossing) or M=PM=P (adder tree).

Cost of Raising MM on Analog/RF

The analog budgets of Section III depend on fRFf_{\mathrm{RF}} only through the ratio x=f0M/fRF=f0/fbx = f_{0M}/f_{\mathrm{RF}} = f_0/f_b (i.e., the position of the band edge relative to the nulls from the zero-order hold at mfbm f_b). Hence, for a given signal bandwidth and fRFf_{\mathrm{RF}}, the dimensionless budgets (gain, unit-capacitor, and phase-clock skew in degrees) are functions of MM alone; the fRFf_{\mathrm{RF}} skew and jitter budgets additionally scale as 1/fRF1/f_{\mathrm{RF}}. Increasing MM relaxes three of the specifications: the unit-capacitor budget grows as M\sqrt{M}, because the I/Q image averages over MM independent slices, the local fRFf_{\mathrm{RF}} jitter budget also grows as M\sqrt{M}, and the data-update skew budget increases, although retiming absorbs that skew in any case. The common jitter budget is independent of MM, and all other budgets tighten, as shown here for a 55 dBc SFDR at median yield with f0=160f_0 = 160 MHz and fRF=12.8f_{\mathrm{RF}} = 12.8 GHz (x=M/80x = M/80):

  • gain matching, from 7.7% at M=4M = 4 to 1.8% at M=32M = 32, because the images move up the hold’s skirt as xx grows;

  • the fRFf_{\mathrm{RF}} arrival, from 0.79 to 0.23 ps;

  • the multiphase generator, from 32° to 2.4° rms, because its slip window is fixed in time while its period shrinks in degrees.

Earlier analysis focused on specific cases of MM. However, Monte Carlo simulations of other values of MM also support the yield laws. A 2×1042 \times10^4-trial Monte Carlo at M=2M = 2 and M=16M = 16, with each mismatch set to its predicted median budget, lands within 0.13 dB of the 55 dBc target. The digital energy and area also grow with the number of slices, as each slice carries its own decoder and output drivers.

The unit-capacitor relaxation should not be extrapolated far. The slice-differential spur sits 18 dB below the I/Q image at M=4M = 4, 11 dB at M=8M = 8, 6 dB at M=16M = 16, and reaches it at M=32M = 32. The image falls as 1/M1/M while the differential spur grows with MM. For M≥32M \ge32, the differential spur sets the unit-capacitor budget.

Design Rule

The analysis above shows that, while increasing MM relaxes the digital timing, it tightens most of the analog requirements. Hence, the best design choice is the smallest MM that satisfies (23), rounded up to the next divisor of PP (a power of two for P=8)P = 8),

M∗=2⌈log⁡2fRF(tov+tj+tsk+Dmin⁡)⌉.(24)M^* = 2^{\left\lceil\log_2 f_{\mathrm{RF}}(t_{\mathrm{ov}}+t_j+t_{\mathrm{sk}}+D_{\min}) \right\rceil}. \tag*{(24)}

As an example, consider the 28-nm library used in Section V. In the slow-slow (SS) corner the flip-flop overhead is tov≈104t_{\mathrm{ov}} \approx104 ps, and with budgets of tj=50t_j = 50 ps and tsk=20t_{\mathrm{sk}} = 20 ps for clock jitter and skew, every stage requires 174 ps. The longest indivisible stage of the 4-phase datapath, the crossing from the FIR branches into the digital commutators, takes Dmin⁡≈98D_{\min} \approx98 ps. At fRF=12.8f_{\mathrm{RF}} = 12.8 GHz, (23) then requires M≥3.5M \ge3.5, so M∗=4M^* = 4. This choice holds for 7.4<fRF≤14.77.4 < f_{\mathrm{RF}} \le14.7 GHz, and M=8M = 8 is required up to 29.4 GHz. For M>PM > P there are no digital commutators, and Dmin⁡D_{\min} is set instead by the sub-filter adder tree (Section VI); the 29.4 GHz figure, like the fRF,maxf_{\mathrm{RF,max}} column of Table IV, therefore assumes the combiner Dmin⁡D_{\min} and is an estimate. Since the FR3 range targeted in Section II-B extends beyond 14.7 GHz, both variants are implemented in Section V: the 4-phase variant as the choice at 12.8 GHz, and the 8-phase variant as the candidate for the upper FR3 range (signed off here at 16 GSps) and to quantify the digital cost of raising MM at a fixed fRFf_{\mathrm{RF}}. Table IV summarizes the resulting requirements for M=2M = 2 to 32 at fRF=12.8f_{\mathrm{RF}} = 12.8 GHz.

MML(M)L(M) (ps)fRF,maxf_{\mathrm{RF,max}} (GHz)σg\sigma_g (%)σc\sigma_c (ps)σϕ\sigma_\phi (°)ΔC/C\Delta C/C (%)
2−187.414.51.26867.4†^{\dagger}
413814.77.70.793210.5
845129.45.30.671315.0
16107658.93.40.445.521.2†^{\dagger}
3223261181.80.232.4n/a‡^{\ddagger}

Table 4. Requirements Versus MM for 55 dBc SFDR at fRF=12.8f_{\mathrm{RF}} = 12.8 GHz. Median-yield budgets. σc\sigma_c: fRFf_{\mathrm{RF}} arrival skew, retimed. σϕ\sigma_\phi: phase-clock skew in slice update-clock degrees. †^{\dagger}Extrapolated as M\sqrt{M}. ‡^{\ddagger}Set by the slice-differential spur, not by M\sqrt{M} scaling. P=MP = M assumed for M>8M > 8. fRF,maxf_{\mathrm{RF,max}} uses the 4-phase combiner Dmin⁡=98D_{\min} = 98 ps; for M≥PM \ge P, Dmin⁡D_{\min} is set by the sub-filter adder tree and fRF,maxf_{\mathrm{RF,max}} is lower (Section VI).

Circuit Implementation and Layout

To validate the physical viability of the proposed hybrid architectures, both macros were implemented up to the layout level: the 4-phase variant, the choice of (24) at 12.8 GHz, and the 8-phase variant, to quantify the cost of raising MM. The off-chip Farrow front-end resamples the modem stream to fRF/32=400f_{\mathrm{RF}}/32 = 400 MSps, and the macro performs the high-speed integer-rate interpolation, to be paired with a custom switched-capacitor power amplifier (SCPA) acting as the RF-DAC for the passive analog recombination [22]. On the macro, an integer-rate ×4\times4 FIR interpolator first raises the sample rate from fRF/32f_{\mathrm{RF}}/32 to the interleaved rate fRF/8=1.6f_{\mathrm{RF}}/8 = 1.6 GSps. Then, the polyphase FIR bank completes the ×8\times8 interpolation. For the 8-phase variant, the data is streamed out of the macro at fRF/8=1.6f_{\mathrm{RF}}/8 = 1.6 GSps, to be recombined on the RF-DAC. For the 4-phase variant, an additional digital commutator recombines the signals into 4-phases operating at fRF/4=3.2f_{\mathrm{RF}}/4 = 3.2 GSps, to be recombined on the RF-DAC.

The digital polyphase FIR filter branches were generated as RTL from the Simulink model with HDL Coder and synthesized into a standard-cell gate-level netlist, optimizing the critical timing paths to ensure strict timing closure at 3.2 GHz (fRF/4f_{\mathrm{RF}}/4) for the 4-phase variant and 1.6 GHz (fRF/8f_{\mathrm{RF}}/8) for the 8-phase variant. The 8-phase macro was signed off at a 2 GHz slice update clock (16 GSps effective output rate), 25% above the nominal 1.6 GHz, with propagated clocks in the SS corner (0.81 V, 125°C), using low-threshold-voltage (LVT) standard cells only. The 4-phase macro was signed off at 3.6 GHz (14.4 GSps effective output rate), 12.5% above the nominal 3.2 GHz, in the same corner, using ultra-low-threshold-voltage (ULVT) standard cells only; its polyphase FIR branches run at half this rate. Figure 3 shows the placed-and-routed layouts of both digital macros.

Placed-and-routed layouts of the digital macros in 28-nm CMOS

Figure 3. Placed-and-routed layouts of the digital macros in 28-nm CMOS. (a) 4-phase macro, 776 × 652 µm (0.51 mm²). (b) 8-phase macro, 1024 × 716 µm (0.73 mm²).

The physical routing of this multi-phase clock network presents a primary design challenge, particularly for the 8-phase variant, which distributes eight 45°-spaced clock phases.

Because each slice retimes its data on fRFf_{\mathrm{RF}}, skew in this network causes no spur as long as it stays inside the slip window of (15), i.e., 22.5 ps rms at M=8M = 8 at the median (Table II). A symmetric, balanced H-tree routing topology was used across the layout, isolating the clock lines to minimize cross-coupling and differential phase skew [41]. The fRFf_{\mathrm{RF}} distribution, which sets the binding sub-picosecond requirement, resides in the RF-DAC and is outside the scope of these macros.

Post-Layout Validation and Results

Both macros were placed and routed in Cadence Innovus, and their parasitic resistances and capacitances (R+C) were extracted with Quantus. The implementation results of this section come from sign-off analysis of the extracted post-route netlists: static timing in Tempus, with propagated clocks, and power in Voltus, with switching activity from a gate-level Verilog simulation of the post-route netlist. To exercise the datapath with realistic activity, an 802.11be-like 320 MHz signal, chosen for its bandwidth and PAPR, was processed through the behavioral MATLAB/Simulink model of the Farrow filter, and the resulting bitstreams drove the gate-level simulation. The output bit patterns of the post-route netlist, ideally reconstructed, were identical to those of the Simulink model over the full 40,799-sample burst, confirming that both macros are bit-exact. This check is a zero-delay simulation and verifies function only; timing is covered by the static sign-off.

The impact of the non-idealities of Section III on the SFDR is evaluated with a behavioral model of the 4- and 8-phase recombiners, with slice timing and gain errors that model clock-routing asymmetries and unit-capacitor mismatch.

Figure 4 verifies the yield law of Section III-B: for each mechanism, the worst spur of 2×1042 \times10^4 trials of the exact, non-linearized (7), normalized to the largest group mean μˉ\bar{\mu}, follows (12) for both variants. For gain error, the M=4M = 4 and M=8M = 8 curves nearly coincide, because the q=1q = 1 group dominates. For timing skew, all M/2M/2 groups have the same mean, so the M=8M = 8 curve lies to the right, which is the yield cost of a larger MM.

CDF of the worst mismatch spur for timing skew and gain error

Figure 4. CDF of the worst mismatch spur for (a) timing skew and (b) gain error, for a single tone at f0=160f_0 = 160 MHz. The spur power is normalized to μˉ=max⁡qμq\bar{\mu} = \max_q \mu_q, the mean power of the strongest spur group across chips, from (10)–(11). Markers: exact Monte Carlo of (7) over 2×1042 \times10^4 trials; black lines: (12). Dotted lines mark the median and 95% yield.

Figure 5 shows the output spectrum for a modulated stimulus with both errors applied. With matched slices, only the signal band and the full-rate images at ±fRF\pm f_{\mathrm{RF}} remain. Mismatch adds a copy of the band at every mfbm f_b, about 60 dB below the in-band power spectral density (PSD) at its edges for the draw shown. Each copy straddles a hold null, so its center is notched and its band edges set its level, as assumed in Section III-B.

Simulated output spectrum for a 320 MHz flat, random-phase multitone

Figure 5. Simulated output spectrum for a 320 MHz flat, random-phase multitone at fRF=12.8f_{\mathrm{RF}} = 12.8 GHz for (a) M=4M = 4 and (b) M=8M = 8, with slice timing skew στ=2\sigma_{\tau} = 2 ps and gain error σg=3%\sigma_g = 3\%; the median-SFDR draw of 1001 Monte Carlo draws is shown. The PSD is normalized to the in-band PSD. Grey: hold envelope sinc⁡(fTb)\operatorname{sinc}(fT_b), with nulls at mfbmf_b, fb=fRF/Mf_b = f_{\mathrm{RF}}/M. Each mismatch image is the signal band shifted by mfbmf_b, so it is centered on a null that notches its middle; with matched slices these images vanish. The bands at ±fRF\pm f_{\mathrm{RF}} are images of the recombined full-rate DAC, not mismatch products.

For unit-capacitor mismatch, the output error splits exactly into a common-slice part, which by (6) is a single full-rate DAC carrying the errors averaged over the MM slices, and a differential-slice part, the only one that can break the polyphase cancellation. The worst resulting spur is the I/Q image of the common part; odd harmonics lie about 7 dB lower, and the differential part stays below the image up to M=16M = 16 (Section IV). Averaging over the slices therefore gains about 10log⁡10M10 \log_{10} M over a single array, and since standard digital I/Q calibration removes the image, the budgets of Section III-D are conservative [42].

Table V provides a quantitative post-layout performance breakdown contrasting the conventional direct-digital approach against the implemented 4-phase and 8-phase hybrid variants.

MetricConventional Direct-Digital4-Phase Variant8-Phase Variant
Nominal Output Rate12.8 GSps12.8 GSps12.8 GSps
Sign-off Rate (SS)n/a14.4 GSps16 GSps
Nominal Digital Clock12.8 GHz3.2 GHz1.6 GHz
RF-DAC Slices (MM)148
Clock Phase SpacingSingle PhaseQuadrature (90°)Multi-Phase (45°)
Timing Closure (SS)Not achievable by (23)Met at 3.6 GHz ¶Met at 2 GHz §
Macro Energy / Sample (TT) ∥—27.2 pJ ¶37.0 pJ §
Power TT / SS / FF (mW) ∥—391 / 414 / 459592 / 558 / 710
Clock Tree, TT—30.0%29.2%
Leakage, TT / SS—6.3 / 27.4%2.6 / 15.5%
Digital Macro Area—0.51 mm2^20.73 mm2^2 §
Routing Congestionn/aMediumHigh

Table V. Post-Layout Performance and Complexity Comparison

The benefit of a larger MM is timing feasibility rather than TT: 0.9 V, 25°C; SS: 0.81 V, 125°C; FF: 0.99 V, −40°C.

§ Implemented 8-phase macro, signed off and characterized at a 2 GHz slice update clock (16 GSps output).

¶ Implemented 4-phase macro, signed off and characterized at a 3.6 GHz slice update clock (14.4 GSps output).

∥ Whole digital macro, including the ×4 interpolator.

a reduction in arithmetic, as predicted by (23). The 4-phase macro closes at 3.6 GHz (14.4 GSps), 12.5% above nominal and just below the 14.7 GHz bound of Section IV; its critical path, the crossing from the half-rate FIR branches into the full-rate digital commutators, has 6 ps of slack. The 8-phase macro closes timing at 2 GHz, where its critical adder-tree stage meets setup with 17 ps of slack in the SS corner without additional pipelining.

Although the FIR arithmetic is the same for both variants, the implemented 8-phase macro carries about 1.9× the datapath cell area of the 4-phase macro, excluding the ×4 interpolator (1.4× in die area, 0.73 versus 0.51 mm2^2): it drives twice as many RF-DAC slices, each with its own thermometer decoder, and it must span the larger analog core of the 8-phase array, with correspondingly longer interconnect from the FIR branches to the lanes. Its energy per output sample, each macro at its own sign-off rate in the TT corner, is nevertheless only 1.4× higher: 37.0 versus 27.2 pJ for the whole macro (Table V), or 34.5 versus 24.5 pJ for the FIR datapath alone, excluding the ×4 interpolator. It closes timing with LVT standard cells, whereas the 4-phase macro requires ULVT cells throughout. At their sign-off rates, the 8-phase variant thus buys timing margin for about 1.4× the energy per sample, consistent with the 4-phase choice of (24). A linear fit through the two macros gives about 17.4 + 2.45MM pJ per sample and 0.29 + 0.056MM mm2^2; with two parameters for two points, mixed threshold-voltage flavors and sample rates, it indicates the trend rather than validating it.

Post-route power of both macros was annotated with switching activity from a zero-delay gate-level simulation of the post-route netlist driven by the 802.11be-like 320 MHz stimulus (100% of nets annotated; glitch power not included, so the reported energies are lower bounds), with each macro at its sign-off rate (Table V). In the TT corner, the ×4 interpolator consumes 40 mW (6.8%, 8-phase) and 37.5 mW (9.6%, 4-phase), and the clock tree about 30% in both, so both macros are dominated by dynamic power and benefit directly from the reduced slice update clock. Leakage rises to 27.4% of the 4-phase total at 125°C, because closing timing at 3.6 GHz required ULVT cells throughout.

Conclusion

This paper generalized polyphase interpolation for RF-DACs by splitting the recombination between digital commutators and the RF-DAC itself, which removes the full-rate output commutator and confines digital timing closure to fRF/Mf_{\mathrm{RF}}/M. This is particularly useful for wideband RF-DACs, which cannot use high-order active reconstruction filters to suppress OOB emissions. The slice array is equivalent to a single full-rate DAC with an MM-times longer hold, and the yield laws, confirmed by Monte Carlo simulation, show that with retiming the binding requirement is sub-picosecond matching of the fRFf_{\mathrm{RF}} clock arrival across slices, and that the best choice is the smallest MM that closes digital timing.

Both implemented 28-nm macros close timing above their nominal rates in the SS corner: the 4-phase macro at 3.6 GHz (14.4 GSps) and the 8-phase macro at 2 GHz (16 GSps). The choice between them is a trade between timing feasibility and cost, not a power saving. The 8-phase variant consumes about 1.4× the energy per output sample and die area of the 4-phase variant and needs eight 45°-spaced clock phases, but closes timing with LVT cells and has lower leakage at 125°C. The 4-phase variant therefore suits platforms that favor energy and only four clock phases, and the 8-phase variant suits those that need timing margin, high-temperature operation, or headroom to raise the output rate.

References

  1. [1]J. Mitola, “The software radio architecture,” IEEE Commun. Mag., vol. 33, pp. 26–38, May 1995.DOI
  2. [2]A. O. Watanabe, M. Ali, S. Y. B. Sayeed, R. R. Tummala, and M. R. Pulugurtha, “A review of 5g front-end systems package integration,” IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 11, no. 1, pp. 118–133, 2021.
  3. [3]S. Luschs, R. Schreier, and H.-S. Lee, “Radio frequency digital-to-analog converter,” IEEE J. Solid-State Circuits, vol. 39, pp. 1462–1467, Sep. 2004.
  4. [4]S.-M. Yoo, J. S. Walling, E. C. Woo, B. Jann, and D. J. Allstot, “A switched-capacitor RF power amplifier,” IEEE J. Solid-State Circuits, vol. 46, pp. 2977–2987, Dec. 2011.
  5. [5]M. S. Alavi, R. B. Staszewski, L. C. N. de Vreede, and J. R. Long, “A wideband 2× 13-bit all-digital I/Q RF-DAC,” IEEE Trans. Microw. Theory Tech., vol. 62, pp. 732–752, Apr. 2014.DOI
  6. [6]D. Shakya et al., “Comprehensive rf(c) and fr3 lower and upper mid-band propagation and material penetration loss measurements and channel models in indoor environment for 5g and 6g,” IEEE Open Journal of the Communications Society, vol. 5, pp. 5192–5218, 2024.
  7. [7]A. Bhide and A. Alvandpour, “An 11-GS/s 1.1-GHz bandwidth interleaved ΔΣ DAC for 60-GHz radio in 65-nm CMOS,” IEEE J. Solid-State Circuits, vol. 50, pp. 2306–2318, Oct. 2015.DOI
  8. [8]E. Oljemann, A.-J. Annema, and B. Nauta, “An interleaved full Nyquist high-speed DAC technique,” IEEE J. Solid-State Circuits, vol. 50, pp. 704–713, Mar. 2015.
  9. [9]P. Caragiulo, O. E. Mattia, A. Arbabian, and B. Murmann, “A 2× time-interleaved 28-GS/s 8-bit 0.03-mm² switched-capacitor DAC in 16-nm FinFET CMOS,” IEEE J. Solid-State Circuits, vol. 56, pp. 2335–2346, Aug. 2021.DOI
  10. [10]J. J. McCue et al., “A time-interleaved multimode ΔΣ RF-DAC for direct digital-to-RF synthesis,” IEEE J. Solid-State Circuits, vol. 51, pp. 1109–1124, May 2016.DOI
  11. [11]D. Su and B. Wooley, “A CMOS oversampling D/A converter with a current-mode semidigital reconstruction filter,” IEEE J. Solid-State Circuits, vol. 28, pp. 1224–1233, Dec. 1993.DOI
  12. [12]D. Barkin, A. Lin, D. Su, and B. Wooley, “A CMOS oversampling bandpass cascaded D/A converter with digital FIR and current-mode semi-digital filtering,” IEEE J. Solid-State Circuits, vol. 39, pp. 585–593, Apr. 2004.
  13. [13]S. Taleie, T. Copani, B. Bakkaloglu, and S. Kiaei, “A bandpass ΔΣ RF-DAC with embedded FIR reconstruction filter,” in IEEE ISSCC Dig. Tech. Papers, 2006, pp. 2370–2379.DOI
  14. [14]R. Bhat, J. Zhou, and H. Krishnaswamy, “Wideband mixed-domain multi-tap finite-impulse response filtering of out-of-band noise floor in watt-class digital transmitters,” IEEE J. Solid-State Circuits, vol. 52, pp. 3405–3420, Dec. 2017.DOI
  15. [15]S. Su and M. S.-W. Chen, “A SAW-less direct-digital RF modulator with tri-level time-approximation filter and reconfigurable dual-band delta-sigma modulation,” in IEEE ISSCC Dig. Tech. Papers, 2020, pp. 174–176.
  16. [16]S. Su and M. S.-W. Chen, “A time-approximation filter for direct RF transmitter,” IEEE J. Solid-State Circuits, vol. 56, pp. 2018–2028, Jul. 2021.DOI
  17. [17]W. Black and D. Hodges, “Time interleaved converter arrays,” IEEE J. Solid-State Circuits, vol. 15, pp. 1022–1029, Dec. 1980.
  18. [18]J. Muller et al., “A 7-bit 18th order 9.6 GS/s FIR up-sampling filter for high data rate 60-GHz wireless transmitters,” IEEE J. Solid-State Circuits, vol. 47, pp. 1743–1756, Jul. 2012.DOI
  19. [19]C. Farrow, “A continuously variable digital delay element,” in Proc. IEEE Int. Symp. Circuits Syst. (ISCAS), 1988, pp. 2641–2645.
  20. [20]T. Hentschel and G. Fettweis, “Sample rate conversion for software radio,” IEEE Commun. Mag., vol. 38, pp. 142–150, Aug. 2000.DOI
  21. [21]M. Hashemi, Y. Shen, M. Mehrpoo, M. S. Alavi, and L. C. N. de Vreede, “An intrinsically linear wideband polar digital power amplifier,” IEEE J. Solid-State Circuits, vol. 52, pp. 3312–3328, Dec. 2017.DOI
  22. [22]W. Yuan and J. S. Walling, “A multiphase switched capacitor power amplifier,” IEEE J. Solid-State Circuits, vol. 52, pp. 1320–1330, May 2017.DOI
  23. [23]R. E. Crochiere and L. R. Rabiner, Multirate Digital Signal Processing. Englewood Cliffs, NJ, USA: Prentice-Hall, 1983.DOI
  24. [24]P. P. Vaidyanathan, Multirate Systems and Filter Banks. Englewood Cliffs, NJ, USA: Prentice-Hall, 1993.
  25. [25]F. J. Harris, Multirate Signal Processing for Communication Systems. Upper Saddle River, NJ, USA: Prentice-Hall, 2004.
  26. [26]R. Khoini-Poorfard, L. Lim, and D. Johns, “Time-interleaved oversampling A/D converters: Theory and practice,” IEEE Trans. Circuits Syst. II, vol. 44, pp. 634–645, Aug. 1997.
  27. [27]Y.-C. Tsao and K. Choi, “Area-efficient parallel FIR digital filter structures for symmetric convolutions based on fast FIR algorithm,” IEEE Trans. VLSI Syst., vol. 20, pp. 366–371, Feb. 2012.DOI
  28. [28]E. Hogenauer, “An economical class of digital filters for decimation and interpolation,” IEEE Trans. Acoust., Speech, Signal Process., vol. 29, pp. 155–162, Apr. 1981.DOI
  29. [29]A. Kwentus, Z. Jiang, and A. Willson, “Application of filter sharpening to cascaded integrator-comb decimation filters,” IEEE Trans. Signal Process., vol. 45, pp. 457–467, Feb. 1997.DOI
  30. [30]S. Balasubramanian et al., “Systematic analysis of interleaved digital-to-analog converters,” IEEE Trans. Circuits Syst. II, vol. 59, pp. 83–87, Feb. 2012.DOI
  31. [31]L. Duncan et al., “A 10-bit DC-20-GHz Multiple-Return-to-Zero DAC With >48-dB SFDR,” IEEE J. Solid-State Circuits, vol. 52, pp. 3262–3275, Dec. 2017.DOI
  32. [32]Y.-C. Jenq, “Digital spectra of nonuniformly sampled signals: Fundamental and high-speed waveform digitizers,” IEEE Trans. Instrum. Meas., vol. 37, pp. 245–251, Jun. 1988.DOI
  33. [33]N. Kurosawa, H. Kobayashi, K. Maruyama, H. Sugawara, and K. Kobayashi, “Explicit analysis of channel mismatch effects in time-interleaved ADC systems,” IEEE Trans. Circuits Syst. I, vol. 48, pp. 261–271, Mar. 2001.DOI
  34. [34]M. El-Chammas and B. Murmann, “General analysis on the impact of phase-skew in time-interleaved ADCs,” IEEE Trans. Circuits Syst. I, vol. 56, pp. 902–910, May 2009.
  35. [35]G. Leger, E. Peralias, A. Rueda, and J. Huertas, “Impact of random channel mismatch on the SNR and SFDR of time-interleaved ADCs,” IEEE Trans. Circuits Syst. I, vol. 51, pp. 140–150, Jan. 2004.DOI
  36. [36]B. Razavi, “Design considerations for interleaved ADCs,” IEEE J. Solid-State Circuits, vol. 48, pp. 1806–1817, Aug. 2013.DOI
  37. [37]J. Elbornsson, F. Gustafsson, and J.-E. Eklund, “Blind adaptive equalization of mismatch errors in a time-interleaved A/D converter system,” IEEE Trans. Circuits Syst. I, vol. 51, pp. 151–158, Jan. 2004.DOI
  38. [38]M. Mehrpoo, M. Hashemi, Y. Shen, L. C. N. de Vreede, and M. S. Alavi, “A wideband linear I/Q-interleaving DDRM,” IEEE J. Solid-State Circuits, vol. 53, pp. 1361–1373, May 2018.DOI
  39. [39]J.-B. Shyu, G. Temes, and F. Krummenacher, “Random error effects in matched MOS capacitors and current sources,” IEEE J. Solid-State Circuits, vol. 19, pp. 948–956, Dec. 1984.DOI
  40. [40]J. P. Imhof, “Computing the distribution of quadratic forms in normal variables,” Biometrika, vol. 48, no. 3/4, pp. 419–426, 1961.DOI
  41. [41]E. Friedman, “Clock distribution networks in synchronous digital integrated circuits,” Proc. IEEE, vol. 89, pp. 665–692, May 2001.DOI
  42. [42]L. Anttila, M. Valkama, and M. Renfors, “Frequency-selective I/Q mismatch calibration of wideband direct-conversion transmitters,” IEEE Trans. Circuits Syst. II, vol. 55, pp. 359–363, Apr. 2008.DOI

Paper details

vlnc:W_uaxhk5nv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Engineering

Not recorded for this paper

View statement
Looking for related papers…
Contents