Access Official Technical PDF

Share Summary:LinkedInEmail

Low-power mixed-signal circuits must manage competing demands: capturing small signals accurately, processing data within tight energy budgets, and integrating analog and digital functions into compact systems. These challenges arise in specialized biomedical devices as well as semiconductor memory and radio-frequency systems. Chul Kim’s engineering and research work offers a view into several of these design problems, from neural recording circuits and wireless power to through-silicon-via (TSV) memory.

Low-power mixed-signal circuits across applications

Mixed-signal design brings together analog circuitry, which handles continuous signals, and digital circuitry, which represents and processes information in discrete form. In a complete system, the two domains must work together without allowing noise, power consumption, or interface requirements to undermine the intended function.

Kim’s experience spans both industry and academic research. At SK Hynix, he worked on 3D-stacked dynamic random-access memory (DRAM) using TSV technology. His responsibilities included power-management units, thermal sensors, power distribution networks, and TSV test circuitry. Later research at UC San Diego focused on integrated biomedical systems, including neural recording, stimulation, wireless power, and radio-frequency signal processing.

The applications differ, but both areas place importance on integrating circuit functions and addressing power and signal constraints at the system level. The appropriate design choices depend on what the device must measure, communicate, or support.

Neural recording: resolving small signals under tight power budgets

Neural recording electronics must acquire electrical activity while keeping circuit noise low enough to preserve useful information. They also need to fit within the system’s power and size constraints. Kim’s research included a 16-channel neural-recording chip described as achieving sub-microvolt noise, a 92 dB dynamic range, and sub-microwatt power per channel. A related published design used predictive digital autoranging to support neural recording with transient recovery.

These design elements address different parts of the measurement problem. Low input-referred noise supports the capture of small signals, while dynamic range describes the span of signal levels a circuit can represent. Autoranging adjusts the measurement range in response to changing inputs. In combination, these techniques are relevant to systems that must capture signals without making the recording electronics prohibitively power-intensive.

The work also included a fully integrated electrocortical interface for 16-channel recording and stimulation in a volume of less than 3 mm3. This system combined functions that otherwise require coordination across separate circuit blocks. Integration can therefore be understood not simply as placing components together, but as designing their power, signal, and physical interfaces to operate as one system.

Wireless power and data links for compact devices

Supplying energy and returning data are central challenges for small wireless biomedical devices. One approach represented in Kim’s research used a single inductive link for both wireless power and data telemetry. The work also explored energy-recycling telemetry and integrated wireless power receivers, including a receiver with an on-chip inductor-capacitor tank for miniaturized neural interfaces.

Combining functions around one link makes coordination between power delivery, communication, and the receiving circuitry especially important. The research included a resonant regulating rectifier and, in a published telemetry design, simultaneous power delivery and backward data transmission over an inductive link. These examples illustrate how circuit architecture can be shaped by a system’s physical scale and the need to support multiple functions.

For engineers evaluating a wireless design, the practical lesson is to treat the power receiver and communication path as connected parts of the architecture. Choices in one block can affect how the overall system is organized and integrated.

Signal separation and mixed-signal processing

Mixed-signal methods also support radio-frequency signal processing. Kim’s research included a four-channel multiple-input-multiple-output (MIMO) baseband receiver designed for cognitive radio and full-duplex operation. Reported circuit features included a capacitive spectral filter for harmonic rejection and a matrix-multiplication spatial filter for signal separation.

In this context, filtering and spatial processing address different interference challenges. The design combined analog processing structures with baseband receiver functions, showing how mixed-signal architecture can support operations that would otherwise require separate stages. The published receiver was reported with 65 dB harmonic rejection and 48.5 dB spatial signal separation; those figures describe that specific design rather than a general expectation for MIMO systems.

Lessons from 3D-stacked memory engineering

Industry experience adds a different perspective on integration. At SK Hynix, Kim worked on TSV technology for 3D-stacked DRAM, including a project involving 44 nm 2G DDR3 memory. TSVs provide electrical connections through silicon to link stacked structures. The work included power distribution, thermal sensing, and test circuitry, all of which are important considerations when designing a stacked memory system.

This experience highlights why packaging and circuit design cannot always be treated as separate concerns. Power delivery, temperature monitoring, and test access were part of the design responsibilities alongside the memory technology itself. For system designers, integration means accounting for these supporting functions as well as the primary processing or storage elements.

Practical implications for engineering teams

Across neural interfaces, wireless receivers, signal processing, and stacked memory, several design considerations recur:

  • Set system constraints early. Signal quality, power, size, and integration requirements influence circuit architecture.
  • Design interfaces deliberately. Analog, digital, power, and communication blocks must work together within the complete device.
  • Evaluate metrics in context. Noise, dynamic range, rejection, and power figures refer to particular circuits and operating purposes.
  • Include supporting functions. Power management, thermal sensing, and test circuitry can be part of the core integration challenge.

For industrial automation engineers and technical decision-makers, these examples are useful as design principles, not as direct performance claims for automation equipment. The underlying work demonstrates the importance of matching circuit architecture to application constraints and considering system integration from the outset. Whether a project involves sensing, communications, memory, or embedded electronics, careful coordination of signal integrity, energy use, and physical integration remains central to building a functional system.

Access Official Technical PDF

About The Author