What's Actually On Your WiFi? The Hidden Components Behind Dropped Connections and Erratic Blood Pressure Readings

Your WiFi drops in the kitchen. Your blood pressure monitor reported 128/80 yesterday morning and 116/74 today. You've reset the router, bought a newer one, replaced batteries, and started measuring at a different time of day. The randomness didn't go away.

I've spent the last four years reviewing component quality for Murata Manufacturing Co., Ltd. Roughly 200+ unique part numbers cross my desk every year, and in 2024 I rejected 14% of first-pass deliveries for deviations that most datasheets would technically allow. The gap between "technically within spec" and "actually reliable in practice" is wider than most engineers want to admit.

This isn't a generic "components matter" sermon. It's a look at the specific components behind two very common problems—WiFi that won't stay connected and blood pressure readings that don't make sense—and why those problems get blamed on everything except the actual cause.

It's Not the Router. It's Not the Monitor.

The surface problem is the one you can see. Your WiFi drops, so you open the router app and check what is on my wifi. Everything looks normal: laptop, phone, TV, thermostat, smart speaker. Signal strength is decent. The router was rebooted twice this month. So the problem must be the router.

It's not.

When a blood pressure monitor disagrees with itself, the usual suspects are cuff position, battery level, or the algorithm. Sometimes those are real factors. But in my experience—looking at return data, teardowns, and the failures I've personally tested—the root cause is usually inside the device, in components chosen for cost.

What's Actually On Your WiFi? The Part That Doesn't Show Up

Here's what never appears in a network scan: the RF environment around your network is crowded with noise. Microwaves on 2.4 GHz, Bluetooth, cordless phones, your neighbors' networks, IoT gadgets, even some LED drivers emit interference. The answer to "what is on my wifi" is honest: your devices, plus a hidden layer of noise that no router app will ever display.

Every one of those noise sources is a potential drop event. And how well your device survives that noise is decided by components at the front end of its radio.

This is where a murata filter enters the story. A surface acoustic wave (SAW) filter sits between the antenna and the radio chip. Its job: let the signal you want pass, reject the frequencies you don't. The difference between a filter with tight insertion loss and strong stop-band rejection versus one with merely okay specs isn't visible when you're streaming video two meters away from the router. It shows up in the kitchen, on 2.4 GHz, at the exact moment the microwave turns on and your signal is already marginal.

FCC Part 15 sets limits on how much interference a device can emit. It doesn't say anything about how well a device should hear through noise. That part is left to the designer—and the components they choose.

I remember testing two batches of WiFi modules side by side. Same PCB, same antenna, same firmware, same production line. The only difference: one batch used a filter with a tighter spec, the other with a wider one. At 10 meters through a wall, the drop rate went from 0.4% to 11%. Same board, same everything—except the part that decides what counts as a signal.

The filter is effectively the doorway. A quality filter lets the right signal through with minimal loss and shuts the door on everything else. A cheaper filter, with wider tolerances and shallower rejection, lets more noise in. Bottom line: your device's real-world performance is partly a filter quality issue.

The Enclosure Is Doing More Than You Think

The second hidden layer is the enclosure. The plastic or metal shell around a device is not just aesthetics. It's the first line of defense against electromagnetic interference. And honestly, a lot of enclosures are designed for thermal performance and drop tests, while shielding effectiveness gets little attention.

Now consider a blood pressure monitor. The pressure sensor produces a small analog signal—a few millivolts. That signal is a direct representation of the pulse in the cuff. If electrical noise leaks into the enclosure and couples to that signal, the reading drifts. No algorithm can fully fix noise that enters after the sensor.

One example that stuck with me: a blood pressure monitor whose reading jumped 8 mmHg every time the pump motor turned on. Not a sensor failure, not an algorithm problem. The sensor was placed too close to the motor driver on the PCB, and the enclosure didn't contain the motor's noise. A properly shielded enclosure—or simply moving the sensor—fixed it. But in the field, nobody sees that. They just see a device that's "unreliable."

This is why the same monitor can give different readings minutes apart. The patient's blood pressure didn't change. The noise environment did. Phone charging nearby, laptop plugged into the same power strip, HVAC compressor cycling—all irrelevant if the enclosure keeps noise out, all measurement variables if it doesn't.

Everything I'd read about consumer blood pressure monitors said accuracy was primarily an algorithm problem. In practice, after testing hundreds of units and reviewing return data, the biggest variance traced to sensors and signal integrity, not algorithms.

To be fair, no amount of shielding can save a device with an unstable sensor. But in my experience, the sensor usually isn't the first thing to fail. It's the noise that reaches the sensor through weak filtering and inadequate shielding.

The Cost of "Close Enough" Specs

In Q1 2024, a supplier shipped us a batch of filters where insertion loss at -30°C was 2.1 dB against our 1.5 dB spec. Normal tolerance is 0.3 dB. The supplier called it "within industry standard." We rejected the batch.

Why does 0.6 dB matter? Because a device that performs at 25°C and degrades in cold weather is the quietest kind of failure. It works long enough to earn trust, then drops out for no apparent reason. The user blames the router. The manufacturer tests it in an air-conditioned lab and finds nothing. The component was the culprit the whole time.

The supplier redid the run at their cost. They adjusted the material set, met the spec at -30°C, and shipped. They could have done that on the first run. It wasn't a capability problem. It was a willingness problem—they saw the spec as a suggestion rather than a contract.

That's what separates component quality at murata manufacturing co ltd from "good enough" suppliers. Not magical technology, but process discipline. Material control, test conditions, batch consistency, and the willingness to say no to an out-of-spec lot even when the schedule hurts.

You can't inspect quality into a product. It has to be built in from the start.

There's a cost side to this that rarely shows up in procurement spreadsheets. Let me run the numbers I usually run.

Our line uses roughly 50,000 pieces of a given filter per year. A cheaper alternative saves $0.02 per unit—about $1,000 annually. If that cheaper component has a 3% higher failure rate, that's 1,500 extra returns per year. At $25 per return in processing, support, and logistics, that's $37,500. The savings flow to the P&L. The cost hides in a different spreadsheet, and someone else has to explain it to the customer.

I'll be straight with you: I've gone back and forth on this math myself. There was a batch of sensors that passed all electrical specs but had inconsistent solder quality. The line was down, the schedule was tight. Passing them would have protected the deadline. Rejecting them meant weeks of delay and a very unhappy sales team. I rejected them anyway. The vendor reworked the solder, and the batch that finally shipped was indistinguishable from the best batch I'd ever seen. I still don't know if it was the right call, economically, in the short term. But I know it was the right call for every customer who opened a box with our name on it.

For medical devices, the math gets worse. A blood pressure monitor that occasionally reads 10-15 mmHg high can send someone to urgent care over a false alarm. A reading that reads low can make them ignore a real problem. The device doesn't look defective. It just—occasionally—lies. That's the most dangerous failure mode there is, and it's never caught by a final functional test because the test environment is quiet. The real environment is not.

Standards like IEC 60601-2-30 and AAMI SP10 exist to define acceptable performance for automated blood pressure monitors. But no standard can guarantee a device will behave in every noisy, messy, real-world environment. That margin is determined by components and design choices—by filters, sensors, enclosures—and by whether the people who made them took the spec seriously.

What I'd Check If I Were On the Buying Side

If you're an engineer or a procurement manager choosing components for a device that transmits or senses RF, here's what I'd verify:

First, the filter's full-temperature behavior. Insertion loss and stop-band rejection at 25°C are table stakes. Ask for data at -30°C and +85°C. If the supplier hesitates, that's a red flag.

Second, the enclosure's shielding story. Ask how the housing connects to ground. Where are the seams? How much noise reaches the PCB in the 100 MHz to 2.4 GHz range? A beautiful enclosure that doesn't close electrically is an antenna, not a shield.

Third, sensor long-term stability. For pressure sensors in blood pressure monitors, baseline drift over temperature and aging matters more than initial accuracy. A sensor that starts perfect but drifts 2 mmHg after six months is worse than one with 1% error that never changes.

A serious supplier—one with reliability data, process logs, and a track record of rejecting borderline batches—will answer these questions directly. Not because they're nice. Because their reputation depends on it.

It took me three years and hundreds of batch reviews to understand that reliability isn't a single pass/fail. It's a distribution. Every component has variance. The difference between a quality manufacturer and a marginal one is how narrow that distribution is, and how carefully each batch is watched before it leaves the factory.

The next time your WiFi drops for no visible reason, or your blood pressure monitor gives you a number that doesn't feel right, consider the invisible chain: the filter, the sensor, the enclosure. The brand matters. The software matters. But the components inside, the ones nobody advertises, are where reliability actually gets decided.

And the next time you ask yourself "what is on my wifi"—the honest answer, once you count the noise, is way more than the device list shows.