Research analysis

August 1, 2026 Population compaction and the recalibrated front end

Over weeks of iBCI typing practice, a participant's premotor manifold contracted in step with faster typing but not with decoder accuracy. For array instrumentation, a stable decode turns out to be a poor proxy for a stable acquisition.

July 31, 2026 Floating-gate drift and at-array compute

A floating-gate analog in-memory array recovers image-classification accuracy to within a few points after 60 days, but only with an adaptive read reference plus statistical recalibration. The harder question for an at-array recorder is the 37 degree environment, not the fix.

July 31, 2026 Timing metadata and the propagation artifact

A 4096-electrode organoid recording finds no signal propagation and a tenfold stimulation-driven collapse, but only after the true sample rate is recovered. The acquisition chain, not the biology, sets what the array is entitled to claim.

July 30, 2026 Cyborg cardiac tissue and electrode co-registration

A National Library of Medicine grant proposes mesh nanoelectronics distributed through the full volume of a cardiac organoid, linked cell by cell to single-cell RNA sequencing and driven by a closed-loop controller. Read as instrumentation, it promotes the electrode array from a voltmeter to a spatial index for a molecular readout, and makes co-registration fidelity the thing that decides whether any of it is trustworthy.

July 30, 2026 Dynamic range, events, and the array noise floor

An event-based wavefront sensor uses active illumination modulation to reach a theoretical 260 dB dynamic range and kilohertz tracking on a plain CPU. The lesson for microelectrode arrays is a hype-correction: dynamic range is not the bottleneck, the noise floor is, and the trick needs a carrier biopotentials do not have.

July 29, 2026 The active substrate that shapes the signal

A cluster-assembled zirconia film changes the calcium dynamics of the glia growing on it. For microelectrode arrays that raises an uncomfortable question about what the coating is doing to the tissue it records.

July 29, 2026 SpiNNaker2 as the closed-loop back end

SpiNNaker2 is a real, measured, low-power event-based many-core chip. Read as the compute end of a living-tissue loop, its strengths and its ceilings both come into focus, and neither is where the marketing points.

July 28, 2026 Adaptive routing under a fixed channel budget

Switch-matrix HD-MEAs expose far more electrodes than they can sample at once. A discounted Thompson-sampling policy reallocates the channel budget as the active sites move, and its limits teach as much as its gains.

July 28, 2026 Myelination and the extracellular signature

A new human tri-culture grant aims to build a myelinating neural model read out in part by microelectrode arrays. Myelin confines transmembrane current to the nodes of Ranvier, and that biophysics quietly changes what a planar array can and cannot see.

July 27, 2026 Backscatter power and the array telemetry ceiling

A battery-free implant harvests power at 13.56 MHz and returns neural data by 434 MHz backscatter. The measured system exposes where wireless recording stops scaling, and how the power carrier threatens the very signal it enables.

July 27, 2026 DOI hyperexcitability and the connectivity detector

A psychoplastogen pushes cortical cultures into a hyperexcitable, more integrated network on a 59-electrode array. The instrument question is how much of that network lives in the tissue and how much lives in the detector.

July 26, 2026 5,000 organoids in beads, and the electrode standoff

A vortex method yields more than 5,000 intestinal organoids per dish inside 250um Matrigel beads. Read against array hardware, the sealed lumen and closed epithelial shell, not just the standoff, defeat a contact electrode.

July 26, 2026 Neuromorphic tiles that have not yet met silicon

A four-block neuromorphic IP suite on SkyWater 130nm is honest that no silicon exists yet. The block that matters for array readout is the on-tile PVT sensor, and the threat is the noise it couples into a microvolt amplifier.

July 25, 2026 The averaging step that erased a signal

A human iPSC knockout study found no genotype effect in per-well MEA averages, yet a clear one in the pooled burst distribution. The gap is a lesson about where the microelectrode acquisition chain quietly decides what you can see.

July 25, 2026 When the array senses molecules, not spikes

Most microelectrode arrays measure voltage; a neurochemical-sensing array measures the oxidation current of transmitters like dopamine. The front end, the failure modes and the selectivity ceiling are all different, and that matters for what an organoid platform can and cannot claim to see.

July 23, 2026 I/O pad integrity and the tissue-to-silicon boundary

A tutorial chapter makes the case that the I/O pad frame, not the core, often decides a chip's fidelity and yield. We extend it to microelectrode readout silicon and the electrode-to-amplifier boundary.

July 23, 2026 Soft 3D electrode integration and the organoid wiring wall

A funded EFRI program proposes over a thousand soft electrodes integrated in three dimensions through a brain organoid, closing a sensing to stimulation loop. We read the plan as instrumentation and ask where it breaks.

July 22, 2026 Dual-threshold delta modulation and the cost of in-pixel compression

A 32-channel implant SoC pushes spike detection and decoding into the pixel to beat the wireless data wall. We read the measured specs skeptically and ask what a compressive front end does to recording fidelity.

July 22, 2026 Persistent homology and the spike-sorting bottleneck

A new analysis recovers robust loop topology from spontaneous organoid recordings at the modest unit counts real arrays deliver. We read it as an instrumentation result and ask what it demands of the front end.

July 21, 2026 Heart-on-a-chip as an integration bet on the acquisition chain

A new cardiac-slice platform grant wants microfluidics, optics, electrodes and control electronics fused on one chip. The interesting risk is not the array; it is moving the electronics next to living tissue.

July 21, 2026 Recording through the stimulus: closed-loop DBS as a front-end problem

A renewed Parkinson's DBS grant proposes evoked-response feedback and high-density array ground truth. The real bottleneck it exposes is analog: recording microvolts through your own volt-scale stimulus.

July 20, 2026 Polymer probes and the chronic-interface ceiling

A translational grant proposes to scale flexible polymer microelectrode arrays to deeper targets and larger animals. Its premise relocates the weak link in the acquisition chain from silicon to the tissue interface.

July 20, 2026 Spatial redundancy and the channel-count question

A masked regression framework quantifies how much of one electrode's signal can be rebuilt from the others. The result reframes how many channels an acquisition chain actually needs.

July 19, 2026 At-sensor adaptive encoding meets a noisy front end

A Zurich group fabricated a 180 nm front-end chip that adapts its event-encoding threshold to the noise floor, aiming to shrink neural data at the electrode. The silicon works, but a 68.72 microvolt noise floor and missing power and sparsity numbers bound what it shows.