What kind of atomic clocks do GPS satellites carry?

Mostly rubidium standards, with caesium units flown on several earlier generations of satellite. Each spacecraft carries more than one and runs a single clock at a time, keeping the others as spares, because a clock failure would otherwise put an entire satellite out of service.

Why does GPS need relativity corrections?

Because orbital clocks run at a different rate from clocks on the ground. Weaker gravity makes them gain about 45 microseconds a day and orbital speed makes them lose about 7, a net gain of 38 microseconds. Uncorrected, that would add roughly eleven kilometres of position error daily.

How much position error does a nanosecond cause?

About 30 centimetres, because that is how far a radio signal travels in a billionth of a second. A microsecond of error corresponds to roughly 300 metres. This conversion is why satellites carry atomic clocks and why ground stations upload fresh clock corrections regularly.

Do I need four satellites for a position fix?

Yes, in the normal case. Three would be enough if the receiver had a perfect clock, but it does not, so the receiver's own clock error becomes a fourth unknown solved alongside latitude, longitude and altitude. This is also why navigation receivers make excellent time references.

Atomic Clocks in GPS Satellites

Satellite navigation looks like a problem in geometry and is really a problem in timekeeping. The atomic clocks in GPS satellites are the whole system: a receiver on the ground works out where it is by measuring, to a few billionths of a second, how long each signal took to arrive. Get the timing wrong by a nanosecond and the position is wrong by about thirty centimetres.

This article explains why positioning is really timing, which atomic clocks each satellite carries, the relativistic corrections that have to be applied before any of it works, and what a nanosecond of error costs on the ground. The scale the satellites keep can be compared with civil time on the TAI and GPS time page.

Why Positioning Is Really Timing

Each satellite broadcasts a signal stamped with the moment it left the antenna. A receiver compares that stamp with its own clock, multiplies the difference by the speed of light, and obtains a distance. Repeat with several satellites and only one point in space fits all the distances at once.

The complication is that the receiver in a phone has a cheap quartz oscillator, hopelessly inadequate for measuring nanoseconds. The system turns that weakness into a feature. With four satellites in view there are four measurements and four unknowns, the three coordinates of position plus the receiver’s own clock error, so the arithmetic solves for the error along with the position. That is why three satellites are not enough for a fix, and why a navigation receiver is also an extremely good clock: having solved for its own offset, it knows the time to within tens of nanoseconds. Telecoms networks, power grids and financial systems exploit this, using satellite receivers as a cheap route to atomic clock accuracy at the top of their timing chains.

Which Atomic Clocks in GPS Satellites Are Used

Every satellite carries several atomic clocks, and only one runs the spacecraft at a time. Rubidium standards dominate the current constellation, with caesium units flown on a number of earlier blocks, and the spares exist because a clock failure would otherwise end a satellite’s useful life.

Rubidium is chosen for orbit rather than caesium for the reasons that make it popular on the ground: it is small, light, undemanding of power and mechanically robust. Its weakness, a slow drift as the vapour cell ages, does not matter here, because the ground segment measures every satellite continuously and uploads fresh corrections. An atomic clock in orbit does not need to be right; it needs to be predictable, and a predictable clock plus a broadcast correction is as good as a perfect one. The differences between the families are set out in caesium, rubidium and optical clocks. Other constellations made other choices, and Galileo satellites carry passive hydrogen masers alongside rubidium units.

What Does Relativity Do to the Atomic Clocks in GPS?

It changes the rate, by two competing effects that do not cancel. A clock roughly twenty thousand kilometres up sits in a weaker gravitational field and runs fast by about 45 microseconds a day, while its orbital speed of nearly four kilometres a second slows it by about 7 microseconds a day.

The net result is a clock gaining roughly 38 microseconds every day relative to one on the ground. That is not a small correction: left alone it would add about eleven kilometres a day to the position error, and the system would be useless within minutes of switch-on. The fix is applied before launch. The oscillator frequency is deliberately set slightly low, so that once in orbit it runs at the intended rate as observed from the ground. A second, smaller relativistic term cannot be handled that way, because the orbits are slightly elliptical and a satellite therefore speeds up and slows down over each revolution; every receiver applies a periodic eccentricity correction of up to a few tens of nanoseconds to account for it.

Why It Cannot Simply Be Ignored

Satellite navigation is the most familiar system whose everyday operation depends on both of Einstein’s theories being quantitatively correct. The corrections are not a refinement applied for the sake of elegance: without them the atomic clocks in GPS satellites would run measurably wrong within minutes, and the errors would compound at a rate anyone could notice on a single car journey.

What Does a Nanosecond Cost in Metres?

About thirty centimetres. Radio waves travel at the speed of light, which covers 299,792,458 metres per second, so one nanosecond of timing error translates into 29.98 centimetres of range error, and a microsecond into roughly 300 metres.

That conversion explains every design decision in the system. It is why the satellites carry atomic clocks rather than good quartz, why the broadcast clock corrections are updated regularly, and why receivers bother to model the delay a signal suffers passing through the ionosphere and the troposphere. It is also why the biggest error in a typical consumer fix is not the clocks at all. The atomic clocks in GPS satellites contribute perhaps a metre or so between them, while atmospheric delay and signals arriving after bouncing off buildings contribute considerably more. Precision users remove the atmospheric term with a second receiver at a known location, which is how surveying achieves centimetre results with the same constellation.

1 nanosecond: about 30 centimetres of range error, the scale the onboard clocks are managed at.
10 nanoseconds: about 3 metres, roughly the accuracy an ordinary phone achieves in the open.
1 microsecond: about 300 metres, and the limit within which satellite time is held to civil time.
38 microseconds: the daily relativistic gain, worth about eleven kilometres if left uncorrected.

Keeping the Constellation on Time

A network of monitor stations tracks every satellite continuously and reports to a master control station, which measures how far each clock has wandered, predicts where it will be over the coming hours, and uploads coefficients that the satellite then broadcasts as part of its navigation message.

Receivers apply those coefficients before using any timestamp, which is how an ageing rubidium atomic clock still delivers nanosecond performance. The whole constellation is steered to a scale called GPS time, itself held close to the reference kept by the United States Naval Observatory. GPS time is continuous and has never taken a leap second, so it now runs 18 seconds ahead of civil time, and the navigation message carries that offset separately so receivers can display the correct hour. The consequences are covered in TAI vs UTC vs GPS time, and the physics of the onboard standards in how atomic clocks work.

Conclusion

The atomic clocks in GPS satellites do not tell you where you are; they let a receiver measure how long a signal took to reach it, which amounts to the same thing at 30 centimetres per nanosecond. Rubidium atomic clocks do most of the work in orbit, relativity is compensated by detuning the oscillator before launch and by a small correction in every receiver, and the ground segment keeps the whole constellation honest with regular uploads. To see the resulting timescale against civil time, open the GPS and TAI comparison, check your own device on the live atomic time display, or start at the atomicclock.now homepage.

Atomic Clocks in GPS Satellites: Why Timing Is Position