Course Description
EEE 4377 Mixed Signal ICs is the course at the boundary between the analogue and digital halves of an integrated circuit — the boundary where nearly every real system has to work, and where most of its hard problems live. Sensors, radios, instruments and control systems all take continuous physical signals into a digital processor and send digital commands back out, and the circuits that make that crossing are what this course designs.
The Statewide Course Numbering System titles it Mixed Signal ICs and describes it as an “introduction to mixed signal processing using analog and digital integrated circuits. Topics include fundamentals of sampled data systems, nonlinear and dynamic analog circuits, Nyquist-rate data converters, over-sampling data converters, and digital filters. Includes use of computer-aided-design programs.” The statewide prerequisite is EEE 4313 or EEE 4376.
Florida A&M University carries the course at 3 credits under the statewide title. The FAMU–FSU College of Engineering bulletin lists it as Mixed Signal ICs and describes it in matching terms.
⚠ The prerequisite is a genuine fork, and it has consequences. A student may enter this course from either the digital IC design side (EEE 4313) or the analogue IC design side (EEE 4376). Both routes are legitimate, but mixed-signal design needs both halves, so whichever route you took, the other half is the one you will be acquiring during the term. The Special Information section says what to expect from each direction.
Learning Outcomes
Required Outcomes
- Apply sampled-data theory to circuit design: sampling and the Nyquist criterion, aliasing, anti-aliasing filter requirements, reconstruction, and the z-domain description of a sampled circuit.
- Analyse quantisation as a noise process, derive the signal-to-quantisation-noise ratio and the 6.02N + 1.76 dB relationship, and explain the conditions under which the white-noise model is valid.
- Specify and interpret data converter performance metrics: resolution, INL and DNL, monotonicity, offset and gain error, SNR, SNDR, SFDR and effective number of bits.
- Analyse switched-capacitor circuits: the equivalent-resistance argument, non-overlapping clocking, charge injection, clock feedthrough, and the design of switched-capacitor integrators and filters.
- Design and analyse sample-and-hold circuits, including aperture uncertainty and droop.
- Analyse Nyquist-rate DAC architectures — binary-weighted, R-2R, current-steering, charge-redistribution — and evaluate matching requirements against resolution.
- Analyse Nyquist-rate ADC architectures — flash, pipelined, successive-approximation, integrating — and select one for a stated speed, resolution and power requirement.
- Explain oversampling and noise shaping, analyse a first- and second-order delta-sigma modulator, and relate the oversampling ratio and modulator order to achievable resolution.
- Design decimation and interpolation filters for oversampled converters.
- Analyse comparators, including offset, hysteresis, metastability and kickback.
- Analyse the practical problems of mixed-signal integration: substrate coupling, supply and ground noise, clock jitter, and the layout and floorplanning practices that mitigate them.
- Verify mixed-signal designs using computer-aided design tools, including mixed-mode simulation and spectral analysis of converter output.
Optional Outcomes
- Design continuous-time delta-sigma modulators and compare them against discrete-time implementations.
- Analyse phase-locked loops, clock generation and jitter.
- Apply digital calibration and correction to analogue imperfections.
- Design digital filters for implementation in the converter data path.
- Analyse converter testing methods, including histogram and FFT-based characterisation.
- Analyse sensor interface and data-acquisition chains end to end.
- Complete a design project taking a converter block from specification to simulated verification.
Major Topics
Required Topics
- Fundamentals of sampled data systems — sampling and reconstruction, the Nyquist criterion, aliasing and anti-aliasing filtering, sample-and-hold behaviour and its sinc response, and the z-transform description of discrete-time circuits.
- Quantisation — the quantiser transfer characteristic, quantisation error and its noise model, SQNR, dither, and the limits of the white-noise assumption at low signal levels.
- Converter specifications — static metrics (offset, gain, INL, DNL, monotonicity, missing codes) and dynamic metrics (SNR, SNDR, SFDR, THD, ENOB), and how each is measured.
- Nonlinear and dynamic analogue circuits — comparators and their design, latches, translinear circuits, analogue multipliers, and circuits whose behaviour is deliberately non-linear.
- Switched-capacitor circuits — the switched-capacitor resistor equivalence, two-phase non-overlapping clocking, parasitic-insensitive integrators, charge injection and clock feedthrough and their cancellation, and switched-capacitor filter design.
- Sample-and-hold — open-loop and closed-loop architectures, acquisition time, aperture jitter, droop, and pedestal error.
- Nyquist-rate DACs — binary-weighted and R-2R resistor ladders, current-steering with thermometer and segmented coding, charge-redistribution DACs, glitch energy, and the matching requirement implied by a target resolution.
- Nyquist-rate ADCs — flash (and its exponential comparator count), two-step and pipelined architectures with inter-stage gain and digital error correction, successive-approximation with charge-redistribution, and integrating converters; the speed–resolution–power design space.
- Oversampling converters — oversampling gain, noise shaping, first- and second-order delta-sigma modulators, the noise transfer function, stability of higher-order modulators, multi-bit quantisers, and the trade of speed for resolution.
- Digital filters in the converter path — decimation filters and the sinc/CIC structure, interpolation filters, and the division of labour between analogue and digital.
- Mixed-signal integration problems — substrate and supply coupling, separate analogue and digital supplies and grounds, guard rings, clock jitter and its effect on SNR, and floorplanning.
- Computer-aided design — mixed-mode and behavioural simulation, long transient runs, FFT-based output spectrum analysis, coherent sampling and windowing.
Optional Topics
- Continuous-time delta-sigma modulators; excess loop delay and clock jitter sensitivity.
- Phase-locked loops, delay-locked loops and low-jitter clock generation.
- Digitally assisted analogue: background and foreground calibration.
- Time-interleaved converters and their mismatch problems.
- Converter testing: histogram (code density) testing, FFT testing, servo-loop methods.
- Sensor interfaces, instrumentation front ends and data-acquisition systems.
- A design project on a converter or switched-capacitor block.
Resources & Tools
- CMOS Mixed-Signal Circuit Design (R. Jacob Baker) is the standard undergraduate-to-graduate text for this exact course and is closely matched to the statewide topic list. Analog Integrated Circuit Design (Johns and Martin) is the other common adoption and is particularly strong on switched-capacitor circuits and delta-sigma modulation. Understanding Delta-Sigma Data Converters (Schreier and Temes) is the reference for the oversampling material.
- Data Conversion Handbook (Analog Devices, Kester ed.) is freely available and is the practical industry reference for converter specification and testing — genuinely useful and often overlooked by students.
- EDA tools — Cadence Virtuoso with Spectre and AMS for mixed-mode simulation is the industry standard; Synopsys HSPICE and Siemens EDA tools also appear. Mixed-signal simulation is a specific skill: analogue blocks and digital control must be co-simulated, and choosing the right abstraction for each is part of the work.
- MATLAB is central here, more than in most circuits courses. The Delta-Sigma Toolbox (Schreier) is the standard tool for modulator design and is free; MATLAB is also used for behavioural converter modelling and for FFT-based output analysis.
- Python with NumPy and SciPy for the same purposes; the
deltasigma package ports much of the Schreier toolbox.
- Open-source flow — Xschem, ngspice and the SkyWater SKY130 PDK support mixed-signal coursework without commercial licences.
- Reference material — the IEEE Solid-State Circuits Society, the IEEE Journal of Solid-State Circuits, the ISSCC digest, and Boris Murmann’s ADC Performance Survey, a freely available compilation of every published converter’s speed, resolution and power. That survey is the field’s shared reference for what is achievable, and knowing it exists is worth as much as a chapter of a textbook.
Career Pathways
- Mixed-signal IC design engineer — the direct destination, and a role that commands a premium precisely because it requires competence on both sides of the analogue–digital boundary.
- Data converter design engineer — a specialisation within mixed-signal design, concentrated at Analog Devices, Texas Instruments and the large system companies.
- Analog IC design engineer and digital IC design engineer more broadly.
- RF and wireless transceiver designer — modern radios are mixed-signal systems, and the converter and clocking material is directly applicable.
- Sensor interface and instrumentation engineer; power management IC designer, where digital control of analogue converters is now standard.
- Test and characterisation engineer for mixed-signal parts — converter testing is a specialised discipline with its own career track and is unusually well served by this course.
- ⚠ A note on positioning. Mixed-signal is the specialisation employers most often report difficulty filling, because it requires two skill sets that students usually acquire separately. A graduate who can genuinely work on both sides is rarer than one who is strong on either alone, and the market reflects that.
- Florida employers include L3Harris (Palm Bay — mixed-signal and radiation-hardened design), Lockheed Martin (Orlando), Northrop Grumman (Melbourne), onsemi and SkyWater Technology (Kissimmee). ⚠ Most mixed-signal design employment is outside Florida — Texas, California, Massachusetts, North Carolina and Arizona — so plan on a national search and use internships as the entry route.
Special Information
⚠⚠ Two entry routes, and the half you did not take
The statewide prerequisite is EEE 4313 (CMOS Digital IC Design) or EEE 4376 (Introduction to Analog IC Design). That “or” is generous, and it means the class arrives split. What to expect:
| Entering from EEE 4313 (digital) | Entering from EEE 4376 (analogue) |
|---|
| Comfortable with: CMOS switching, timing, clocking, logic, layout rules | Comfortable with: matching, current mirrors, differential pairs, op-amps, noise, compensation |
| Will need to acquire: device matching, noise analysis, op-amp design, settling and stability | Will need to acquire: clocking disciplines, digital logic in the control path, timing analysis |
| Hardest part: the switched-capacitor and op-amp material in the first half | Hardest part: the digital decimation filters and clocking in the second half |
The honest advice: take both prerequisite courses if your schedule permits. Mixed-signal design is where the two disciplines meet, and a student with only one of them spends part of the term reading the other course’s textbook. Where that is not possible, identify your weaker side in week one and address it deliberately rather than hoping it will not matter — it will.
⚠ Course-code variation across Florida
Mixed-signal design is not widely offered at undergraduate level in Florida, and where it appears the numbering varies:
- EEE 4377 — Florida A&M University. The FSU bulletin lists the course with a prerequisite of “EEL 4313 or EEL 4376C”, which appears to be a bulletin typographical error for the EEE numbers — confirm with the department rather than assuming either reading.
- EEE 5378 Mixed Signal ICs — the graduate counterpart, carried at FAMU, FSU and UCF.
- EEE 4404 Mixed Signal IC Testing I — University of Florida, which approaches the same devices from the test side rather than the design side; it requires EEE 3308C and EEL 3701C.
- Much of this material appears elsewhere only at graduate level, or inside a broader analogue IC course.
SCNS equivalency does not cross course numbers. Because the subject is offered narrowly, expect this course to articulate as a technical elective rather than as a named requirement, which is usually what a student wants from an elective in any case. Carry the syllabus and, if you completed a design project, keep it.
Position in the curriculum
EEE 4377 is a senior-level elective and, at the FAMU–FSU College of Engineering, close to the terminal undergraduate course in the integrated-circuit track. It leads naturally into EEE 5378 at graduate level and into research in data converters and mixed-signal systems.
It pairs well with EEE 4510 Digital Signal Processing, which supplies the filter-design machinery that the decimation and noise-shaping material assumes. Students who have taken DSP find the delta-sigma section substantially easier, because noise shaping is a filtering argument dressed in circuit clothing.
Difficulty and time commitment
This is a demanding senior elective. Its characteristic difficulty is holding two mental models at once: the sampled-data, frequency-domain view and the transistor-level circuit view. A delta-sigma modulator has to be understood simultaneously as a noise-shaping filter and as a set of switched capacitors, comparators and clocks, and a student who can only do one of the two cannot design it.
The second difficulty is practical: mixed-signal simulation is slow. Establishing a converter’s SNR requires a long transient run followed by an FFT, and getting that right — coherent sampling, an appropriate window, enough cycles — is itself a skill. Runs take hours, so the design iteration cycle is measured in days.
Plan on ten to twelve hours a week, and start any design project immediately. Simulation time cannot be compressed at the end of a term.
Articulation and transfer
SCNS records EEE 4377 as guaranteed to transfer to an institution offering the same course. One Florida institution carries the undergraduate number, so read that narrowly. The course is upper-division, carries 3 credits, and has no general-education or Gordon Rule designation.
FE exam relevance
The NCEES Fundamentals of Engineering (Electrical and Computer) exam covers sampling, analogue-to-digital and digital-to-analogue conversion and digital filters within its Signal Processing and Electronics topic areas — at a considerably lower level than this course, but with genuine overlap. A student who has taken this course should find the FE’s conversion and sampling questions straightforward.
AI Integration
This course sits at an interesting point with respect to AI: the circuits it designs are the ones that feed machine learning systems, and the design task itself is one that automation has partially but not fully solved.
Where AI is genuinely used in the discipline. Machine learning is applied to analogue and mixed-signal sizing — given a topology, finding device dimensions meeting a specification — and to design-space exploration, and commercial tools now ship these features. Digitally assisted analogue design is a more fundamental connection: modern converters routinely correct analogue imperfections in the digital domain, and adaptive and learned calibration is an active research area. A student in this course is learning the architecture that makes that possible.
There is also a direct application link worth naming: the data converter is the first stage of every machine learning system that touches the physical world. The quantisation, jitter and linearity limits taught here set the ceiling on what any downstream model can recover, which is a useful corrective to the assumption that model quality is the binding constraint.
Where a general-purpose assistant helps in coursework. Explaining noise shaping or why a pipelined ADC needs inter-stage gain; generating MATLAB behavioural models of converters; writing FFT analysis and windowing scripts; setting up corner sweeps; and interpreting mixed-mode simulator errors, which are notoriously opaque.
⚠ Where it fails, and why the failure coincides with this course’s subject. The characteristic error of an AI tool asked a data converter question is to quote the ideal SNR relationship — 6.02N + 1.76 dB — as though it described a real converter. That formula assumes perfect quantisation and nothing else: no thermal noise, no jitter, no mismatch, no distortion. The entire content of this course is what happens to that number in practice, where a 16-bit converter achieves perhaps 13 effective bits and the gap is the design problem. A confident answer that returns the ideal figure has skipped the course.
A second failure is specific and consequential: models generate converter output spectra and SNR figures from simulation setups that are wrong in a way that inflates the result. Non-coherent sampling without windowing spreads a tone across bins and can make distortion disappear; too few FFT points hides the noise floor; measuring SNR over the wrong bandwidth flatters an oversampling converter enormously. These are exactly the measurement errors the course teaches you to avoid, and none of them produces an obviously wrong-looking plot.
Third: models routinely omit jitter from a high-speed converter analysis. Clock jitter sets the achievable SNR at high input frequencies regardless of resolution, and it is the constraint that dominates modern high-speed converter design — and it is absent from the textbook formulae a model reaches for first.
The engineer’s responsibility. A converter specification is a claim about what a system can measure, and everything downstream inherits it. The habit worth forming here is to ask of every reported SNR or ENOB: at what input frequency, over what bandwidth, with what clock, and measured how? An FFT plot with its sampling and windowing conditions stated is evidence; a number is not.
Academic integrity. FAMU and FSU both maintain academic honour policies covering AI-generated work. Design justification and analysis are normally expected to be your own even where scripting and simulation assistance is permitted. Generated simulation results or converter spectra are data fabrication and are treated more seriously than plagiarism. Ask before you rely on a tool, and disclose its use where the syllabus requires it.