[Saturday’s run: 6.88 in Starkville]
While I was in Dubuque I worked on a project for receiving the time of day from the low frequency WWVB broadcast on 60 kHz. This is the same signal that is received by “atomic” clocks you can buy at Wal-mart or where ever.
I don’t think my solution works nearly as well as the cheap commercial clocks. So I’ve been thinking about some improvements.
There are two different signals to be decoded. One is based on amplitude and the other by phase change. My solution and the cheap clocks both work on the amplitude signal.
There are three different tokens or amplitude signatures. You can read all about it on the wikipedia page for WWVB. They call the three tokens 0 (zero), 1 (one), and M (mark). A token is sent each second, 60 tokens for the minute. And some of the tokens are laid out in a fixed pattern. My solution has two main parts. First is to detect which token is the most likely for each second that elapses. That involves collecting amplitude data 100 times per second, and trying to match that to the expected amplitude signature for each token. Second is to take that stream of tokens and try to decode a complete message (one per minute).
I don’t think my token detection is very good. Also I think my method of decoding the messages is too strict; I require two consecutive minutes of 100% reception in order to say I am “locked”.
I’ve been thinking about trying to match portions of the token stream to the expected patterns and see if I could decode year alone, or some other parts. And building up my information from multiple messages, knowing the passage of time as I do this, it seems to me I could eventually make some inference of the complete time string.
For example, if I successfully decoded 2013 as the year, that would give me an anchor in the correct positioning of the flow of tokens, where I am at in relation to the start of the minute. And with that I could then know when to expect other message elements and have some predictive ability which might help me crack Day, Hour, or Minute.
I have been able to get a lock with my current algorithm both in Dubuque and here in Mississippi. So it will eventually work. But it is slow in getting a lock, particularly here where I have a 100 watt radio signal transmitting about 100 feet away 24×7.
And I would someday like to try the phase detection. I’m told that will work more reliably then the amplitude detection. There are supposedly some commercial radios you can buy which do have the phase detection ability. I do not have one.