Hacker News Reader

Comment by mbertschler | Hacker News Reader

> But for some applications, accuracy is less important than precision. Precision has to do not with delineating the perfect second but rather with creating extremely regular ticks, or oscillations.

I don’t follow here, doesn’t accuracy follow if you have a precise oscillator? What is the difference between an accurate or a precise clock?

View on HN

No. A precise oscillator ticks as close to the second mark as it can, but may be biased in one direction or another consistently. A precise clock can be thought of as a ruler with graduations down to millimetres, where every millimetre is 1% too short. In an accurate clock (like atomic clocks) every millimetre may be 10% too long or too short, but in a random direction.

In the precise ruler if you measure out a 1000 milimeters you will be 1 milimeter short. In the accurate ruler you will be 1/10th of a millimetre long or short.

Both are useful. If you're a carpenter measuring a space to put a cabinet in and you use the same ruler to measure the openings as to cut the cabinet you'd rather the ruler be precise than accurate. If you email those measurements to someone else you'd rather the ruler be accurate than precise.

Simpler version: _accuracy_ is how close the mean of your measurements is to the true value, _precision_ is the variance of your measurements.

It implies that the "tick" intervals in atomic clocks aren't always the same length in time, they just average out very well. While the new sapphire clocks have a more consistent tick interval.

So.... Cesium clock is "guaranteed" 10 ticks in X timespan, but the distribution is not perfectly equidistant? Do I understand that correct?

Right. Actually, from what I understand, the situation is even more dire and Cesium clocks are (would be) complete rubbish at timing consecutive ticks, so when people say cesium clock they actually mean a quartz oscillator (ultrapure, double ovenized, binned, aged, etc -- but still quartz) to create the output and then a control loop to push/pull the quartz oscillator and frequency lock it to the cesium absorption line.

A similar trick is more commonly used to lock, say, a 5.8GHz on-chip oscillator to a 10MHz quartz oscillator, except in this case the quartz oscillator plays the role of long-term reference and the on-chip VCO plays the role of short-term reference, whereas in the case of an atomic clock it's the other way around.

The overall game is that different filters/oscillators are better at different timescales, so you use control loops to synthesize the best parts of each of them into a clock that is good at all timescales.

Cesium beam atomic clocks are actually just a very selective band pass filter / detector. There is a DC output in microamperes which is proportional to how well centered the input frequency is in relationship to the resonance of the Cesium atoms in the ambient conditions as they drift across the tube. To use this filter as a clock, it is slaved to a quartz oscillator which is them multiplied and phase locked, with a small bit of FM at 137 hz prior to being fed into the tube. If the frequency is too low, the output will be in phase with the FM, if the frequency is too high, the phase will be opposite, and if it is centered, there will be a 274 hz "2x" output, which is used to indicate lock.

Thus, the cesium isn't actually sampled at ~9.192 Ghz, but rather a much slower rate. The actual loop maintaining phase is as slow as possible to keep phase noise down, which is part of why they take a while to lock on startup.

Further complicating things is the need for a small persistent magnetic field in the tube, as getting zero in all 3 dimensions is a much harder problem. This bias keeps things stable, but also changes the frequency slightly, but is offset out in the divider chain.

Well, in other electronic device, say one type of temperature sensor, let's say microchip's MCP9701A you can have accuracy of +/-2 Celcius with a precision of 0,01V per Celcius.

In this case Accuracy mean the reference as in: this device reads 10C but could actual temperature could be 8 or 12. Because it drift according to multiple factors.

Precision means the linearity or if you prefer resolution (I think). Datasheet for the device says +/-0,5C meaning you can't have a reading more precise than 1C with that device. I think the term linearity apply if you try to mesure below that value: the reading may become non-linear because of fluctuaction, noise or other errors in the device.

Long time ago I saw an audio analog to digital converter where the resolution was 22 bits or something. But the accuracy was 0.5% or something.

Don’t you just need to determine the scaling factor and now you have precise and accurate?

If you try to talk yourself through how you'd actually accomplish that then you'd quickly see that it is turtles all the way down. Every time you discipline the clock for accuracy you trade some precision to do so - because that is the only way to move the needle. This isn't a big deal for most, but you'd definitely notice NTP slewing the oscillator frequency in the middle of something like a logic trace running at 10s of MHz - which is why labs generally prefer a precise local time standard that is costly (in dollars and hassle) over a cheaper GPS reliant solution offering better absolute accuracy.

Reply on news.ycombinator.com