ORRA / Engineering

The document a factory would actually read.

Concept art does not get quoted. What gets quoted is a layer stack, a component list with real part numbers, a block diagram, a bill of materials and a test plan. Here is what exists, and what is still missing from it.

The stack

Six layers in fifteen millimetres

The ORRA Core, exploded. Diameter is set by the coil, thickness by the Peltier's need to move heat somewhere.

Dimensioned engineering blueprint of the ORRA Core: 75 mm diameter, 15 mm tall, with each layer called out against a millimetre scale. — Working concept sheet from the ORRA development thread, published whole and unretouched. Open it to read the annotations at full size. They are working notes rather than verified findings, and several state things this site does not claim: nothing is patent pending, no protein simulation has been run, and no certification named on any sheet is held. What is true and what is not.

The drawing above is the dimensioned version of the same stack, at 75 mm diameter and 15 mm tall against a millimetre scale. It is a design intent drawing. It is not a schematic, not a board layout and not mechanical CAD, and a factory cannot quote from it — that gap is the single largest item on the list below.

The family as drawn

Nine units, one architecture

Every unit in the family was drawn against the same six layers before any of them was costed. Each product page carries the annotated working sheet it came from; this is the set as it would sit on a shelf.

The ORRA Core: a matte black anodised disc 75 mm across and 15 mm thick, with a micro-perforated top face, a cyan light seam around its rim, and red emitters glowing through from inside.

The ORRA Node: a brushed titanium pendant about 45 mm wide with a polished ceramic contact face, on a braided steel cord with a machined clasp.

The ORRA Dome: a matte black structured five-panel cap with a charcoal drive module on the rear strap and red light glowing from the forehead liner onto the underside of the brim.

The ORRA Neural Suit: a matte black short-sleeve compression top with a fine silver conductive filament woven in a branching pattern across the chest and shoulders, and a charcoal drive module at the base of the collar.

The ORRA Neural LS: a matte black long-sleeve compression top with the silver filament weave carried to both cuffs and rings of magenta-red emitters at the shoulders and biceps.

The ORRA Shorts: matte black compression shorts with red emitter panels over both thighs, cyan coil weave banding the quadriceps, and a charcoal drive module clipped into the back of the waistband.

The ORRA Pants: full-length matte black compression tights with magenta mesh panels on the thighs and calves, red emitter rings at both knees, and silver coil weave running hip to ankle.

The ORRA Glove: a matte black knit glove with a reinforced knuckle weave, magenta filament coils across the back of the hand and wrist, grey conductive fingertips, and a round drive module at the wrist cuff.

The ORRA Strides sock: matte black compression knit with magenta coil weave over the Achilles and along the arch, and a charcoal disc module set into the outside of the ankle cuff.

Generated product visualisation, not a photograph. No ORRA unit has been manufactured, so no photograph of one exists. These images were built from the dimensions and components on the concept sheets and are replaced with real photography when the first units are made.

The unified ORRA family sheet. — Working concept sheet from the ORRA development thread, published whole and unretouched. Open it to read the annotations at full size. They are working notes rather than verified findings, and several state things this site does not claim: nothing is patent pending, no protein simulation has been run, and no certification named on any sheet is held. What is true and what is not.

Components

Chosen once, reused across the family

Every active part below is specified by manufacturer part number, not by category. Picking a family standard early means one qualification, one firmware driver and one supply relationship instead of nine.

Function	Part	Why this one
MCU and radio	Nordic nRF52840	Low-power Bluetooth with enough headroom to run the timing loop on-chip. Industry default, so the toolchain and the certification path are well travelled.
Thermal driver	TI DRV8833	Dual H-bridge. Reverses polarity across the Peltier, so one part gives both the cold jolt and the warm face.
Coil driver	TI CSD18504Q5A	NexFET MOSFET fast enough to keep a square wave square at 40 Hz into an inductive load.
Power management	TI BQ25895	Handles the burst current the Peltier draws without browning out the MCU.
Wireless charging	TI BQ51013B	Qi receiver. The device that started as a question about a charging pad charges on one.
Bio-impedance	ADI MAX30001	Clinical-grade front end. The calibration argument fails with a hobby-grade sensor.
Motion	Bosch BMI270	Detects grip or wear so a session cannot start on a bedside table.
Emitters	OSRAM GH CSSRM4.24 (660 nm), SFH 4715S (810 nm)	Binned parts with real radiometric data, which the photobiological safety file will need.
Thermoelectric	TEC1-04903 class, 40 x 40 mm	Commodity module. The engineering is in the heat path around it, not in the part.
Cell	Custom flat LP604040, 2500 mAh	Flat pack, because 15 mm of total thickness does not allow a cylindrical cell.

Cost concept for the Core at 100 units. The enclosure dominates because CNC aluminium at that volume is expensive, and it is the line that drops most at 500.

$116 per unit at 100 units, roughly $95 at 500 as the chassis moves off CNC. These are estimates built from component pricing and comparable job costs. No factory has quoted this. The first real number will come back higher, because they always do.

Route to a part

Direct to the factory, not through a distributor

A large distributor adds real value on a programme with a hundred variants and a supply team to manage. On a first run of a hundred units it mostly adds margin and a layer of translation.

The route here is: specify the exact manufacturer part numbers, source them against a component search engine, send the board files and the bill of materials to a turnkey assembly house that buys the parts and ships finished boards, and take the enclosure to CNC and cast silicone rather than paying for injection-mould tooling on a hundred units.

What that route does not buy is engineering judgment. Nobody in that chain will tell you the tooth pitch is wrong or the thermal path is too short. That work stays in-house, which is the reason this page exists rather than a quote request.

Step	Who does it	What has to be handed over
Board fabrication and assembly	Turnkey PCBA house	Gerbers, pick-and-place file, costed BOM with approved alternates
Enclosure	CNC shop, cast silicone for the skin	STEP files, tolerance and finish notes, an IP-rating target
Box build	Same PCBA house	Assembly sequence, torque and adhesive notes, thermal interface spec
Test	Factory floor, to our procedure	A pass/fail script per unit: field strength, plate temperature, emitter output, radio
Certification	An accredited lab	Golden samples and a full technical construction file

Test plan

What every unit has to pass before it is boxed

Test	Criterion
Field strength	Measured flux at the contact face within tolerance of 1.8 mT at the rated drive
Thermal	Contact face reaches 10 degC within the jolt window; internal face never passes its cut-off
Emitter output	Both wavelengths within radiometric tolerance at the contact face
Cut-off	Deliberately drive the plate past limit and confirm the hardware watchdog stops it
Contact loss	Break the impedance path mid-session and confirm output stops
Radio	Bluetooth link and RF output inside the pre-scan envelope
Ingress	Sealed to the rated level under a dynamic, not static, test
Thermal cycling	A sample from each batch runs the jolt cycle to failure or to the cycle target, and the housing is inspected for micro-fracture

Alongside the per-unit script, a golden sample defines what a correct unit looks and feels like, so a batch dispute has a reference to point at rather than an argument.

Gaps

What is genuinely not done

No schematic. A block diagram is not a schematic. Nothing has been drawn at net level.
No board files. No layout, no stack-up, no Gerbers.
No mechanical CAD. No STEP files for the chassis, which is what a CNC shop needs before it can quote.
No firmware. The mode table is a specification, not code.
No thermal model. The heat path from the Peltier's hot face to the outside is the single biggest unknown in the Core, and it is currently a paragraph rather than a simulation.
No textile programme. Every worn unit assumes a conductive weave that carries a coil, survives washing and does not break at a flex point. That is its own multi-year development.

Ready to look at the engineering?

The layer stack, the component list with manufacturer part numbers and the mode table are written down. Ask and you get the same document a contract manufacturer would.

Request the data package Read the architecture

See what it would cost

Concept retail against a real bill of materials, with the margin shown rather than implied.

Pricing →

Start with the science

Three physical layers, six set points, one calibration baseline. Ninety seconds of reading.

How it works →