This article is within the scope of WikiProject Radio Stations, a collaborative effort to improve the coverage of radio stations on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
This article is within the scope of WikiProject Time, a collaborative effort to improve the coverage of Time on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
I noticed that another person removed this text that someone else added. I don't have enough knowledge on WWVB signal to know if this is important or not, so I copied it here. • Sbmeirow • Talk • 03:34, 29 January 2014 (UTC)
Shortcomings of the current signal format
Unlike MSF and DCF77, WWVB does not have a built in tick around the minute mark to discipline clocks that are in continuous reception mode.
What MSF does (and almost identical DCF77) to indicate the start of a minute
If each second is considered as ten 100 ms pieces, the minute marker is transmitted as 1111100000, while all other seconds are transmitted as 1AB0000000.
The first statement is not true. WWVB transmits two consecutive periods of reduced-carrier signal, each 800 ms long and starting on the second, to indicate the top of the minute. The start of the second of these periods is the exact top of the minute. -- Denelson83 03:49, 29 January 2014 (UTC)
I'm the one who removed it and I of course agree with Denelson83 re the first statement. Re removal of the rest, we wouldn't put details of the WWVB format in the MSF or DCF77 articles; nor do their details belong in the WWVB article. Jeh (talk) 04:54, 29 January 2014 (UTC)
Removals were appropriate. Glrx (talk) 00:11, 30 January 2014 (UTC)
I strongly object to this edit. Yes, some modems are "many bits per symbol"; that's the point! "symbol" is not "less informative," it is more correct.
In the AM code one bit is not sent per second. If you consider only the data bits, it's an average of less than one bit per second, since only 53 data bits are sent per minute. But that is an incorrect analysis as you can't interpret the data bits without recognizing the marker symbols. Fact is there is one symbol per second, each symbol having three possible values, not two. That's about 1.6 bits per symbol (1.6 being the log base 2 of 3), so the effective bit rate would be 95 b/s! Better to just say "symbol per second" and explain what a symbol is. Note that we're already using "symbol" in the description of the AM format. Jeh (talk) 09:57, 7 December 2015 (UTC) it.
The first reason I prefer the "bit" language is that it's describing both amplitude and phase modulation, and for phase modulation, "one bit" is exactly correct.
Regarding the amplitude modulation, I agree that from a low-level perspective, there are three possible symbols and thus it's not binary.
However, from a high level perspective, at most one payload bit is sent per second, so it very much is a one bit per second code.
I note that NIST's own documentation describes the AM format as "broadcast continuously at a rate of 1 bit per second" (p. 17).
There are many line codes which use additional low-level symbols for framing, spectrum shaping, forward error correction, or other purposes, but are still described in terms of their payload data rates.
The one that leaps immediately to mind is alternate mark inversion as used on T1 lines. There are three line levels used, yet a T1 line is universally referred to as "1.544 Mbit/s", because the additional state is not used for data transmission.
Another example is USB 1.x. Frames are terminated with a "single-ended zero" which is distinct from the usual 1 and 0 bits. Yet it's called 12 Mbit/s, the specification talks about "bit times", and so on.
Straying further afield, modems routinely use additional states for forward error correction. Most commonly, n bits of payload and 1 bit of FEC are bundled into an n+1-bit constellation. But the modems all talk about the payload' data rate.
Other codes use violations of the usual coding rules for framing purposes. Modified frequency modulation and AES3 (=S/PDIF) are both called rate-1/2 codes, even though there are additional sync symbols which may be sent through the channel, making the rate technically greater than 0.5.
Likewise, 8b/10b encoding actually transmits 8.0444 bits per 10 bit times if you include the additional control codes. But nobody does.
If WWVB used the extra values to transmit more data, in the style of 4B3T, I'd call it a ternary code. But as long as they're framing bits which actually preclude sending a time code bit in those slots, I think the high level "one bit per second" is correct.
The other is detrimental pedantry. It's being technically correct (and even then, only from one specific point of view) at the expense of intelligibly.
The article includes the full details for someone who wants to understand. It's not a secret that there are three possible "bits". But for someone skimming the heading (and the location of the edit is an introduction), "one bit per second" is more informative and more useful.
I can try to rephrase it to make the high-level perspective clearer and remove any implications to the contrary.
(BTW, nice edit getting rid of the "modulation depth" subsection. That definitely cleaned things up!)
P.S. Another source to look at is the IRIG timecode specification that WWVB is modeled on. And section 3.2.3 of http://irigb.com/pdf/wp-irig-200-04.pdf (p.3-1 of doc, p. 17 of pdf) says "Each pulse in a time code word/sub-word is called a bit." Even though a "bit" can be a 0 bit, a 1 bit, or a "position identifier". 3.2.4 specifically calls the position identifier at the beginning of the frame "a reference bit".
I suppose I could cite that and the NIST spec as reliable sources for the factual assertion that that the code is "one bit per second", but I think that's *also* obnoxious pedantry. I honestly think saying "bit" makes for better exposition. 184.108.40.206 (talk) 11:31, 7 December 2015 (UTC)
"Three possible bits" - you mean three possible symbol values. :) (Awhile ago I edited this, and/or WWV, to get rid of the term "marker bit". If it's a marker, it's not one of the two possible binary digits. To say otherwise contradicts the definition of "bit". Ok, the IRIG doc has its own special definition of "bit", but the reader of these articles is pretty unlikely to know that.)
I don't feel that saying "one symbol per second" is detrimental or obnoxious. On the contrary I think saying "one bit per second" for the AM code is. I know that Wikipedia is supposed to be edited for a general audience, but I'm tired of seeing it watered down on the assumption that the "general audience" is content with approximations and generalities. To put it another way, I hate the "lying to children" concept. (Some of them don't like it when they find out.)
Hmm... How about "approximately one bit per second"? This is more correct than simply "one bit per second", invites the reader to read more if they want the details, but is good enough if they don't. Then the stuff about markers being a possibility doesn't have to be in this intro section at all, it can just appear in the AM subsection. Any mention of "symbols" can be shoved to the AM subsection too.
And likewise BTW, the rest your edits in that session were all good - thanks! Jeh (talk) 11:35, 7 December 2015 (UTC)
To me, it's like replacing a clear word with a vague euphemism. "Boyfriend" means something; "partner" requires extra context to interpret. As I said, from an information-carrying point of view, they are bits; the frame markers are a lower-level detail. Would it be better if I tried to phrase it that "in some seconds, a marker is transmitted in lieu of a bit" without being too awkward?
We're not lying to anyone because the details are right there. Nor is it talking down to anyone. In the field of line coding, special escape codes, comma symbols, or whatever you want to call them are common things, and practitioners call them "bits" without blinking.
Maybe it's a software thing. Hardware engineers are quite familiar with three-state logic. For overkill, see IEEE 1164: nine possible values for one bit!