Jitter: Difference between revisions
Jump to navigation
Jump to search
mNo edit summary |
mNo edit summary |
||
(One intermediate revision by the same user not shown) | |||
Line 1: | Line 1: | ||
<!-- | <!-- | ||
When you have periodic events, jitter is the timing deviation from that true/presumed periodicity. | When you have periodic events, jitter is the timing deviation from that true/presumed periodicity. | ||
(Below 10Hz or so, it is sometimes called wander instead) | (Below 10Hz or so, it is sometimes called wander instead, probably more so when the underlying reason accumulates over time) | ||
This idea can be applied to varied things, but largely to electronic signals | This idea can be applied to varied things, but largely to electronic signals | ||
People also sometimes use this to indicate noise in sample values. | |||
--> | |||
====Some more applied jitter terms==== | |||
<!-- | |||
'''Sampling jitter''' in DACs or ADCs refers to the fact that the clock signal in use (which might come from clock recovery) is imperfect to a certain degree | |||
is the variation in end-to-end delays. For example, when you [[ping]] someone else on the internet, | |||
you may have an RTT | |||
'''Latency jitter''' tends to refer to networked communication (so a.k.a. '''packet jitter''' ), which has | |||
: latency - some | |||
: variation in that latency | |||
' | In one situation there may be reasons it's 10ms late with very little variation, | ||
in another it might be anything between 1 and 100+ms with little consistency | |||
This is something of a misnomer in that there is no reason to expect a true periodicity, | |||
but it's still a useful way to point at "the variation on top of the average". | |||
Line 30: | Line 40: | ||
'''Video jitter''', '''image jitter''' refer to horizontal lines being displaced, usually due to corruption of timing signals (not to be confused with [[rolling shutter]] details). | |||
"micro jitter" is a term that was quickly made vague, by people using to mean anything they want | |||
--> | |||
' | ====When you're talking specifically about clock signals==== | ||
<!-- | |||
'''Clock recovery'''[https://en.wikipedia.org/wiki/Clock_recovery] (which might more clearly be called clock ''discovery'') | |||
refers to communication where the timing isn't sent separately, but extracted from a serial data stream itself - the precision of which is naturally limited by the undferlying jitter. | |||
--> | |||
====Deterministic or random?==== | |||
<!-- | |||
Random jitter caused by electronics's [[thermal noise]] is likely to be unpredictable and usually has a gaussian distribution. | |||
It is also likely to be very small | |||
Random jitter makes for broadband noise, deterministic jitter tends to have a more defined spectral envelope. | |||
Jitter can be a lot more deterministic - in that it can be predictable that it's there, and well bound, | |||
because you know why it happens (e.g. ). | |||
One reason is that you know it's correlated to the data (e.g.), | |||
though that is not the only type/reason. | |||