Budget Smartwatch Physics: How Sensors Measure Your Body

When you hold a phone flashlight against your fingertip, the tip glows red because the skin spreads the red light away from the exit. The same physics sits inside every pulse-reading wearable on the shelf. The gap between a watch that tracks a week of workouts and one that misreads a running session is not a magic algorithm. It is a better physical setup, better signal handling and more honest assumptions about where the device is being worn.
A budget smartwatch like the Soudorv P97 Smart Watch runs the same sensing stack as a far pricier optical heart-rate unit. It just packs that stack into a smaller case, where there is less room for the light, the inertial chips and the radio to do their jobs. The rest of this article works through the physical parts that decide what a budget smartwatch can and cannot measure: the green light that beats against your wrist, the red and infrared pair that estimates blood oxygen, the microchips that turn arm movement into a step count, the gyroscope that tells a brisk walk from an anxious fidget, and the 2.4 GHz link that carries it all to your phone without eating the battery.
None of that needs a laboratory to follow. It needs a clear picture of what each component actually measures, and where its limits are.

How a Wrist Sensor Reads a Heartbeat

Turn a watch with an optical heart-rate sensor around and you will see a cluster of tiny green diodes glowing against the skin. Those diodes are the emitters of a PPG, or photoplethysmography, sensor, and the light they use sits near 530 nanometres. That wavelength is chosen because it is the colour that red blood absorbs most strongly. Hemoglobin, the protein that carries oxygen, has a sharp absorption peak in the green part of the spectrum, so the more red blood that is sitting near the skin, the more green light is swallowed before it can bounce back.
With every beat the heart pushes a fresh volume of blood through the small vessels under the skin, and that extra volume absorbs a little more green light. Between beats the volume falls and a little more light returns. A photodiode sitting beside the diodes tracks that up-and-down oscillation, the processor turns the beat frequency into beats per minute, and the number lands on the display. The whole measurement is a light pulse and a reflected pulse, counted across a minute.
The light is green rather than red because, at the wrist, green gives a stronger contrast between the pulse wave and the background. Red spreads through skin more evenly and is absorbed less sharply by the small vessels near the surface, so the pulse rides on a dimmer and noisier carrier. Green stays closer to the skin, where the blood volume changes fastest, which is what the photodiode needs to read a clean signal.
No wearable sits perfectly still, and movement turns the light it sends into noise. When the arm moves, the sensor shifts, the optical path changes and the reflected pattern shifts in a way that can look just like a heartbeat. Cheap optical sensors are more sensitive to that motion artifact than a lab pulse oximeter, so a quick hand gesture mid-run can drag a reading off. The mitigation is an accelerometer cross-check plus adaptive filtering, which is why a budget watch will ask you to keep the arm still for a blood-oxygen reading, while it tolerates more movement for heart rate.

Measuring Blood Oxygen with Two Wavelengths

Blood-oxygen saturation is a second, smaller optical channel on the same sensor package. It uses a red diode near 660 nanometres and an infrared diode near 940 nanometres, alongside the green heart-rate emitter. Oxygen-rich hemoglobin absorbs more infrared and lets more red through. Deoxygenated hemoglobin does the reverse, absorbing more red and passing more infrared. The processor divides the two absorption ratios, and that ratio is the saturation estimate. The higher the number, the more oxygen the blood is carrying.
That physics is identical in a clinic fingertip oximeter and in a wrist tracker. The wrist is just a harder place to work. A finger held still between two diodes and a detector gives a strong, stable light path. A watch clamped to the wrist, where the skin over the artery is thicker and the sensor sits slightly off the vessel, gives a weaker and noisier one. So a budget watch keeps the heart-rate channel sampling to build a running average, and runs the blood-oxygen channel only when you hold still. When a saturation reading looks low, the best move is to re-take it with the strap snug and the arm resting, because a tight, still wrist is the best configuration and a loose, moving one is the worst.

The Accelerometer That Turns Motion into Steps

A MEMS accelerometer, etched onto the chip next to the processor, does the step counting. It is not a moving spring balance. It is a microscopic mass, only a fraction of a millimetre across, suspended on tiny beams. When the arm accelerates on a stride, that mass lags a few nanoseconds behind the case and the beams deflect by a tiny amount. The deflection changes a capacitor built from a fixed and a moving electrode, and the electronics read that change as a voltage. The chip runs this on three axes at once, so it knows not just that the arm moved, but in which direction it moved.
A step is not a single spike in the data. It is a characteristic burst: a heel strike, a roll through mid-stance, and a toe-off. The counting algorithm looks for that three-part pattern in the fused three-axis signal and increments a counter on each confirmed cycle. That is also why a repeated arm rhythm that is not walking, like brushing teeth or chopping vegetables, can add a false step. The sensor sees a repeated acceleration and, without another cue, treats it as a gait. The cue that lets a better processor discount those false positives is the rotation signal from a gyroscope.

The Gyroscope That Separates Walking from Fidgeting

A MEMS gyroscope is the second inertial sensor, and it usually sits on the same chip as the accelerometer. It does not measure absolute rotation, it measures angular rate, using a vibrating structure and the Coriolis force. A walking gait has a clean, rhythmic wrist rotation. A fidget or a typing session does not. The gyroscope hands the activity classifier that rotation signal, so the tracker can split a brisk walk from a nervous tap on a desk.
The same rotation, fused with the accelerometer and sometimes a magnetometer, is what lets the watch tell a swimming stroke apart from a stroke in open water, and a running stride apart from a cycling cadence. A sports mode on a budget watch is not a separate algorithm per sport. It is a tuning of this fused-sensor pipeline. Each of the 113 plus sport presets retunes the detection thresholds and the fusion for a specific movement cadence, so the same silicon reads a swim, a run and a tennis swing differently. The hardware stays the same, the thresholds move.

The 2.4 GHz Radio That Carries It All

The Bluetooth 5.3 link is the bridge that turns the watch into a companion rather than an island. It lives in the 2.4 GHz industrial, scientific and medical band, the same crowded space that Wi-Fi, a microwave and a garage door opener use. Bluetooth keeps its data usable in that noise by frequency-hopping. The link moves across the 79 channels of that band and changes frequency 1,600 times a second, along a pseudo-random pattern both sides can predict. When a Wi-Fi burst lands on one channel, the link hops to the next, and the disruption is one hop rather than a lost packet. That hop rate, and the error correction pushed around it, is what keeps a Bluetooth call on a wrist watch stable in a room full of 2.4 GHz signals.
The LE part of the stack is the power story. A classic Bluetooth radio stays tuned and listening between packets. A low-energy radio sleeps most of the time, wakes for a few milliseconds to push a small notification, a heart-rate sample, a step burst, a call flag, and returns to sleep. Bluetooth 5.3 adds connection subrating, which lets the link flip quickly between a low-drain sync interval and a high-data call interval without tearing the connection down and re-syncing from scratch. For a color-screen watch meant to last a week, how the radio sits between its sync windows is one of the largest terms in the power budget, second only to the display.

What an IP68 Rating Covers

IP68 is an IEC rating, a two-digit code. The first digit, 6, means dust-tight, with no harmful solid ingress. The second digit, 8, means the enclosure is built for continuous immersion beyond one metre, under conditions the maker defines. For a watch rated for shallow swimming, that is both what it does and what it does not.
Water pressure is a still number in the test and a moving one in the pool. A freestyle stroke, with the arm breaking the surface at speed, spikes the pressure on the case for an instant past whatever the still-water rating was sized for, and that spike can find a seal that was built only for gentle submersion. Heat matters more. The gaskets that hold the seal are silicone or rubber, and their elasticity depends on temperature. A hot shower or a sauna cycles the case and the gasket at different rates, and that ageing thins the seal over time. A hot bath into a cold rinse in quick succession is the worst case. An IP68 watch is a swim-ready device, not a hot-tub-ready one.

The Power Budget Behind a Week of Wear

A budget color-screen watch meant to last seven days on one charge is a small engineering budget, not just a feature list. The biggest drain is the display, a color panel kept partly on for the clock, the always-on face and the step ticker. The next is the optical heart-rate sensor while it is kept sampling. A PPG channel that wakes every few seconds draws only a few milliwatts on average, but across a week it adds up. The radio stays cheap because it sleeps except through its sync windows. The processor doing the sensor fusion and the step counting is the least hungry of the three, since the chip is a sub-watt system on a single package.
Watch run time comes down to three levers. A wearer can turn the heart-rate sampling down, switch off the always-on display, and pick a simpler watch face. Each one trades a feature for run time. When a watch claims seven days, that figure assumes the default sampling cadence, a default face, and a radio that sleeps between notifications. Push any of those levers toward the maximum and the week drops to two or three days. That is not a flaw in the physics, it is the physics.

What the Numbers Can and Cannot Tell You

A wrist-worn optical tracker is a strong instrument for trends and a weak one for a single spot reading. A heart-rate average across a week is useful for spotting whether your resting heart rate is drifting down or up between training days. A single number taken mid-walk, mid-run, or against a loose strap is the worst case for accuracy, and it should not be read as a clinic value. Blood oxygen measured at the wrist is a trend, not a pulse-oximeter grade, and a few points of error is expected whenever the signal is weak.
The honest summary is this. A budget fitness watch is a good trend tool, not a swap for a lab instrument, and in the right use case it is also a better instrument than most people assume. Read the weekly heart-rate trend to check recovery between sessions. Read the step count and the activity recognition to check whether a day was moving enough. Treat the blood-oxygen channel as a spot check rather than a monitor, and hold the arm still for that reading. Read the seven-day battery claim as the default-setting number, not the best case. None of that asks the watch to be a laboratory device. It asks the person wearing it to know which reading is solid and which is noise, and that is the whole point of understanding what sits under the display.