Inside our anechoic chamber: how we test what patients never see

The first thing you notice is that your own voice sounds wrong. The second is the foam: two-foot pyramids of it, blue-gray, covering every surface including the floor, which you cross on a suspended wire grid. Nothing in the room reflects anything. That is the entire point.

This is the semi-anechoic chamber on the lower level of the Minneapolis lab. It is where every Nexaform device goes before it goes anywhere near a person, and it exists to answer two questions that sound simple and are not.

Does this device emit anything that could interfere with other equipment? And does anything other equipment emits interfere with this device?

Why a room with no echo

Radio measurement in an ordinary room is close to meaningless. Signals bounce off walls, benches, and the person holding the clipboard, and arrive back at the antenna slightly late and slightly out of phase. What you measure is the device plus the room, and you cannot separate the two.

An anechoic chamber removes the room from the measurement. The foam absorbs incident energy across the frequency range we care about, the shielded shell keeps the outside world out, and what the antenna sees is the device and nothing else. Our chamber is semi-anechoic, which means the floor is deliberately reflective under the wire grid, because that is what the relevant standards call for. A ground plane is part of the test.

The chamber, in numbers

Internal dimensions

7.0 × 3.9 × 3.6 m

Absorber

610 mm pyramidal, walls and ceiling

Frequency range

30 MHz to 6 GHz

Shielding effectiveness

>100 dB at 1 GHz

Ambient noise floor

At least 6 dB below the lowest limit we test to

Turntable

1.5 m, 0.5° positioning

A day of testing

A device arrives on a Monday with a test plan that has usually been argued about for a fortnight. The plan matters more than the equipment, because a chamber will happily give you a beautiful, precise, entirely irrelevant number if you point it at the wrong thing.

Emissions testing comes first. The device sits on the turntable, running a script that exercises every radio it has and every mode it can enter, while the turntable rotates and the antenna sweeps in height. The measurement is not one number; it is a surface, and what you are looking for is the worst point on it.

Then immunity, which is the more interesting half. Here the chamber generates fields and we watch what the device does. Radiated RF across the band. Electrostatic discharge, which is a person in a dry corridor in February touching a device. Electrical fast transients, which is what a motor starting on the same circuit looks like from the device's point of view. Surge. Magnetic fields, which matter enormously for anything that will be within a meter of an MRI suite door.

The failure mode that matters

Most people imagine electromagnetic interference as something dramatic: a screen going blank, an alarm going off. Those are real, and they are the easy case, because they are obvious and they get reported.

The failure we spend the most effort chasing is subtler. A device under a strong field does not crash. It keeps operating and its measurements drift, or its timestamps slip, or a handful of samples go missing and get silently interpolated over. Everything downstream looks normal. The chart looks normal. Nothing raises an alarm because from the system's point of view nothing has gone wrong.

Catching that requires knowing what the device should be reporting during the test, at sample-level resolution, and comparing against it continuously rather than eyeballing a screen. Every immunity run in this chamber is scored against a reference recording. A pass is not "it kept working". A pass is "every sample it reported during the exposure matches what it should have reported, within tolerance, and we can show you the file".

Two devices in a corridor

Standards testing puts one device in a room. Reality puts a patient wearing our sensor next to a patient wearing someone else's pump, in a corridor, next to a service lift.

So after the standards work is done, we run what the team calls the corridor set: our device alongside a collection of other equipment, at realistic separations, in combinations that the standards do not require anyone to test. Infusion pumps. Insulin pumps. Hearing aids. A cheap phone charger of the kind that is genuinely everywhere and is genuinely awful.

Nothing in that set is required. All of it has, at some point, told us something the required tests did not.

The boring result is the good result

A successful week in this room produces a folder of data in which nothing happens. No excursions, no resets, no drift, no missing samples. It makes for a dull report, and there is no way to demonstrate it to a visitor, because the entire achievement is an absence.

That is the trade. Everything this room does is invisible when it works. The only time anyone outside the building would ever learn that a device had an electromagnetic compatibility problem is if it shipped with one.

The chamber is booked most weeks. If you are a customer running your own device alongside ours and you want to understand how the two behave together, that is a conversation worth having, and occasionally it ends with your equipment on our turntable.