This evening I managed to get a breakthrough while I was working on the wordclock. The clock kept showing a completely wrong time now and then, followed by the correct time a minute later.
So I loaded the test program for the DCF receiver again and monitored the output on the serial console over a longer time (one or two hours or so). I noticed that the receiver actually gave incomplete information once in a while. To rectify this I did two things:
First, I payed more attention to the actual electronic parts and made sure the two circuit boards, the Arduino and the antenna of the DCF receiver wouldn’t short out anything. This actually reduced the amount of errors quite a bit, but there was still the wrong time now and then.
So I took a closer look on the false information. The receiver doesn’t only get the time, it also gets the current date and the timezone. This data was not really “wrong”, it was rather incomplete. Sometimes fields would just be empty or zero. So, Second, I added some plausibility checks. If one of the date fields, day, month or year, was 0 (which, if I’m not completely mistaken, shouldn’t be possible anymore these days) the clock wouldn’t update the time and just keep showing the last known time. I added the same for the timezone fields.
These two simple changes did the trick, the clock always shows the correct time. Now I only have to improve the illumination of the words and mount it in the frame I got for it. This should be relatively easy.
Last, but not least, I want to share a short video I made while I was working on the clock. I must have accidently created a short somewhere.
Since it currently seems to be impossible to embed a YouTube video without it loading tracking scripts the video can only be viewed directly on YouTube.