Design preview · not built yet

32-channel EEG, without the gel.

Dry silver contacts, pressed gently into place by air, with short petal tips that reach between hairs to the scalp. Up to 32 of them sit in a removable helmet liner, wired to one small board on your forearm.

The EEG design at a glanceA head in profile wearing a close-fitting liner dotted with dry silver contacts linked by air lines, under a clear helmet shell. A magnified circle shows one contact: a small air cushion presses a cluster of rounded silver petals down between the hairs to the scalp. One shielded cable runs from the back of the liner to a slim 32-channel board strapped to the forearm. Removable linerup to 32 dry silver contactsShielded cableone plug, detachablePetal tips reach the scalppressed by a small air cushion32-channel boardworn on the forearm
  1. Removable liner with up to 32 dry silver contacts, inside a helmet shell
  2. Petal tips reach between hairs to the scalp, pressed by a small air cushion
  3. One shielded cable
  4. 32-channel board, worn on the forearm

Today's EEG vs this design

Research and clinical EEG uses gel caps: good contact, slow and messy to fit. Consumer headbands are dry and quick, but sit only where there is no hair. This design aims for whole-head coverage without the gel.

Gel capResearch and clinical standard
Today's gel capA head in profile wearing a fabric cap with about a dozen electrodes. Each electrode has a blob of gel and its own wire; the wires gather into a cable at the back. A syringe fills one electrode with gel, a clock stands for the long fitting, and drops of gel are left in the hair. 123
  1. Gel is squeezed into every electrode, one at a time.
  2. Fitting usually needs a trained person, and it takes a while.
  3. Gel dries over a long session and has to be washed out of the hair.
Dry headbandConsumer
Today's dry headbandThe same head wearing a slim headband across the forehead with an arm behind the ear, carrying a few sensor pads on bare skin. The hair on the top and back of the head is left uncovered. 45
  1. A handful of sensors, on the forehead and behind the ears.
  2. Nothing reaches the top or back of the head, where there is hair.
This designDry, air-set, up to 32 channels
This designThe same head wearing a close-fitting liner with dry silver contacts spread over the forehead, crown and back of the head. Air lines link the contacts, and one shielded cable leaves the back of the liner. 12345
  1. Dry silver contacts: nothing to squeeze in.
  2. Air presses the contacts into place together, not site by site.
  3. Nothing to dry out and nothing to wash out.
  4. Up to 32 sites across the forehead, crown and back of the head.
  5. Short petal tips part the hair to reach the scalp.
What mattersTodayThis design
Putting it onGel caps: gel at each site, usually applied by a trained person.Put on the helmet liner and inflate. The air seats every contact; there is no gel.
CoverageHeadbands: a few sites on bare skin. Gel caps: the whole head.The whole head: 16 contacts to start, up to 32.
Through hairHeadbands avoid hair. Dry pin electrodes press rigid pins through it.Clusters of short, rounded petals, each sprung on its own, part the hair. Very dense or tightly curled hair may need a longer face.
Pressure on your headRigid pins rely on firm pressure to make contact.Designed to start below 1 N per contact, set by air and capped by a mechanical stop.
After a long sessionGel dries out, and hair needs washing.Dry contacts use the skin's own moisture. Nothing to clean out of your hair.
Signal qualityGel gives the lowest, steadiest contact resistance. It is the benchmark.Dry contacts are generally noisier and less steady than gel. That is why force and fit are controlled per contact, and every contact batch is compared with purchased standard silver/silver-chloride electrodes before anything is claimed.

Three design targets

32

channels from one liner, across the whole head. Start with 16 and add more.

<1 N

starting force per contact. Set by air and capped by a stop, so a contact presses without digging in.

No gel

Dry silver contacts. Nothing to apply before, nothing to wash out after.

How a contact reaches through hair

Each contact sits in its own cassette a few millimetres thick. Air pushes it toward the scalp, a flexible plate lets it follow the curve of your head, and a stop keeps it from pressing too hard.

Contact cassette cross-section Cross-section drawn at about sixteen times scale. Under the helmet shell and liner, a micro-bladder presses a flexure plate and a silver/silver-chloride contact toward the scalp. A pressure port with a restrictor feeds the bladder, a lip on the carrier acts as a force stop, and the insulated lead leaves through a strain relief toward the loom. SCALP PETALS REST ON HAIR HELMET SHELL LINER 5 mm 123456

Petals that part the hair. A dense cluster of short, rounded petals, 2–4 mm long and each independently compliant, shares one electrical node. The design prefers many short compliant tips over a few long rigid tines.

Air on: the cushion is inflated and the face meets the scalp, starting below 1 N per contact.

Cross-section drawn to the design's target dimensions.

  1. Air comes in.A small restrictor and pressure port let each contact's force be set and checked.
  2. An air cushion pushes.A thin bladder, 1.0–1.5 mm flat and 2–3 mm inflated, spreads a gentle force over a region of the head.
  3. A flexible plate follows your head.It is 0.5–1.0 mm thick, so each face settles on its own patch of scalp.
  4. A stop limits the push.The helmet shell takes the reaction force, and a mechanical stop keeps any contact from being overdriven.
  5. Silver meets skin.A chlorided pure-silver face: flat on bare skin, a cluster of short rounded petals where there is hair.
  6. The wire leaves gently.A strain relief anchors the lead, so a tug on the cable is not a tug on the contact.
Contact positions on the head, seen from above Top view of the head with the nose at the top. Up to 32 contact positions in rings; the 16-contact layout uses a subset. Three forehead positions use flat faces; the rest use petal clusters for hair. Leads converge at the back of the head where the loom leaves. Illustrative distribution, not a selected clinical montage. FRONT BACK · CABLE EXIT
Flat face · bare skinPetal cluster · hair

The whole head, not only the forehead

Start with 16 contacts or fill the liner with 32. Flat faces sit on the bare forehead; petal clusters go everywhere there is hair. Each contact is measured against one common reference, with a separate bias contact.

Select a contact

Every position has a job.

Choose any contact on the map to see its face type and cassette depth. Wires from every position run to one exit at the back of the liner, where the cable leaves.

Positions are illustrative. The exact layout, reference and bias sites are set during testing.

A tiny signal, kept clean

Brain signals are small, so everything else on the wire matters. Each kind of interference has its own answer in the design. Pick a source to see the problem, then switch on the design's response.

Mains & radio pickup
Illustrative trace with an interference source A synthetic brain-like waveform with the selected interference added. Switching the design response on shows how the matching design choice changes it. Illustrative, not recorded EEG and not a measured noise level. ILLUSTRATIVE · NOT RECORDED EEG 1 s

Shielded loom, isolated contacts. The loom is shielded, and a removable silver-textile RF layer can screen the garment. The shield is kept isolated from the contacts, so it guards the signal without becoming part of it.

Engineering detail

For readers who want the parts, the path and the numbers behind the pitch.

Signal path, scalp to screen
  1. Contact. A chlorided pure-silver face meets the scalp dry. Every contact is measured against a common reference, with a separate bias contact.
  2. Lead. Each insulated lead is anchored by a strain relief. The back of every contact is fully encapsulated, with no sweat access to copper, nickel, solder or conductive adhesive.
  3. Shielded cable. One detachable shielded loom carries every contact, the reference and the bias to the forearm. Nothing on the head is powered: the helmet holds contacts, air lines, passive wiring and the liner.
  4. Board. One IronBCI-32 board digitizes all channels: 32 channels on AD7771 converters, planned at 500 samples per second per channel.
  5. Software. The repository already holds Rust code for sliding-window feature extraction, a streaming server and a WebSocket viewer, written first for a 4-channel consumer headband. Connecting it to the 32-channel board is open work.
The board on your forearm

The helmet has no slot for electronics. The board is about 70 × 70 mm and is expected to need a 14–17 mm protective enclosure before battery allowance. It is easier to reach and service on the forearm, and the cable detaches so liner and electronics separate. Its channel count, sample rate, noise, software support and fit come from its published description and are confirmed against a purchased unit before the design depends on them.

Materials
  • Contacts: 99.99% silver foil, formed, then chlorided electrochemically with the charge recorded. Solid silver avoids plating that can peel or expose base metal. A plated route stays as a later option.
  • Air cushion: heat-welded TPU or molded silicone, 20–30 mm across.
  • Flexure: a nonconductive 0.5–1.0 mm polymer plate with a load-spreading face, so the only exposed metal on your skin is the contact.
  • Cassette depth: 4–7 mm over bare skin and 6–10 mm over ordinary hair, before liner padding.
  • Shielding: a shielded cable and an optional silver-plated textile layer, both kept electrically separate from every contact. The shield is never the signal reference.
How contacts will be judged

Every custom batch is compared with purchased sintered silver/silver-chloride electrodes: DC offset, drift over an hour, impedance at 1, 10 and 100 Hz, noise with the intended amplifier, and recovery after a bounded test pulse. Pass marks are set only after measuring the purchased control.

Where this goes next

The dry EEG liner is the first instrument in a wider brain-interface research program. The lanes below are separate research ideas at an earlier stage than the EEG design. None has been built, and each has its own open questions.

  • Light instead of wireCarry each electrode's voltage over plastic optical fiber, so no wire connects the head to the recorder.
  • Pulse-timing linksTurn one contact's signal into the timing of light pulses with a cheap oscillator chip. Whether this measures EEG is unverified.
  • Near-infrared light (fNIRS)Follow blood-oxygen changes in the outer cortex from the forehead. A trend instrument, not a medical oximeter.
  • Atomic magnetometrySense the brain's magnetic field with laser-pumped rubidium vapor, built up on a bench against a commercial reference.
  • Magnetic reference sensorsCheaper magnetic sensors that track the surrounding field. Useful as references, not as brain sensors.
  • Active head loopDrive a known, gentle signal through a coil or resonator near the head and look for changes. Whether anything neural remains is the open question.
  • Printed sensor fabricsMany sensing sites in a fabric, read through a few ports with a calibrated decoder. A long-term idea.
  • FeedbackTurn measured signals into sound, light or touch. Stimulation research stays separate and alternates measuring with output.

Where it stands

Exists

A detailed design: contact cassette dimensions, contact shapes for bare skin and hair, wiring, shielding and the choice of board. Software from an earlier headband project: feature extraction, a streaming server and a viewer.

Unproven

Nothing has been built or measured. Comfort, contact quality and full-system noise are unknown. Dry contacts are generally noisier than gel, and the board's specifications come from its maker and still need confirming.

Next

Make pure-silver contacts and compare them with purchased electrodes on a saline phantom. Bench-test noise, pressure against contact resistance, overload recovery and battery brownout. Then fit, comfort and motion on the head.

Cerebro by Redgold is an engineering design, not a product you can buy. Nothing here is a medical device or a measured result.

Cost

Not priced yet.

The core system (board, 16–32 contact cassettes, shielded cable and air supply) has no qualified price yet. The only figures so far are engineering budgets for the separate fNIRS experiments, $50–$1,000 depending on stage, which are never added together or to the EEG system.

Full component list

Questions

Can it read thoughts or diagnose a condition?

No. EEG records small voltage differences shaped by brain activity and by many other things: contact quality, movement, muscles, eye motion and the electronics. No diagnostic, cognitive or clinical capability is claimed or established.

If gel gives a cleaner signal, why go dry?

Because something you wear often has to go on and come off without paste, cleanup or drying time. The trade-off is real: dry silver/silver-chloride relies on natural skin moisture and pressure, and generally has higher, more variable contact resistance than gel. That is why the design controls force, compliance and fit at every contact, and compares every batch with purchased standard electrodes before claiming anything.

Will it work with thick or curly hair?

Ordinary hair is what the petal faces are for. Dense or tightly curled hair may need a longer replaceable face or deliberate parting, because air pressure cannot push a contact through an arbitrary layer of hair. The design names this as a fit constraint up front.

Why is the board on the forearm?

Keeping electronics off the head keeps the helmet simple: contacts, air lines, passive wiring and a removable liner. The board is easier to reach, check and replace on the forearm, and the detachable cable lets the two separate.

Can I buy one?

Not yet. Nothing has been built, measured or certified. The next milestones are contact fabrication and bench measurements, then fit and comfort on the head.