I'm running a classic kit with version 55 firmware and an RTC clock module. This is my second clock. One thing I notice is that the blink is out of sync with the changing of the seconds.
Just to be clear, I'm using 4 neons, 2 between the hours and minutes and 2 between the minutes and seconds. The neons pulse at a rate of 30Hz...1 second of increasing intensity and 1 second of decreasing intensity. When the intensity of the neon gets to the lowest and the highest, the "seconds" also change...this is how it is in my older build. However, with this new build, it is a split second off and that split second will vary a little from minute to minute.
The other thing I notice is that at the top of minute...when it goes from :59 to :00, sometimes the time spent at :00 is very short and sometimes changes to :01 almost instantly and sometimes, a second later (as it should be), or somewhere in between.
The clock does seem to keep good time though. I set it several days ago and throughout the day, the time varies from 0 to 2 second difference from my smart watch which is synced to my cell phone.
It seems that it could be a difference between the internal clock and the RTC module. Reviewing the code, it seems the RTC is checked every minute which might explain the update at the top of the minute and why the time seems to be holding true.
I'm wondering if an update to version 56 will fix this or if perhaps there is an issue with my oscillator. Any thoughts?
My first thought is to check the crystal and the 22 pf caps surrounding it. If the crystal oscillator is off frequency it could produce results like you describe.
Try re-melting the solder on those parts. If that doesn't help, you may need another crystal. You may have difficulty changing any of those parts because of the plate-through holes and the size being so small.
That's what I was thinking. Here's an update. I updated the FW to version 56 and it seems to have solved the 1st part of my problem...where the sync between the change in blink intensity correlates with the change of the 'seconds' digit.
However, there is still a correction going on every few minutes. I called a friend of mine who has his own build of the clock and as we were talking, he noticed the correction too. Could be that the oscillator is off. I wonder what the accuracy should be. Every 7-8 minutes, I notice that a second is advanced at the top of the minute. It goes from :00 to :01 almost instantly. So it would seem that the internal clock is losing 1 second every 450 seconds or 0.2%. So if my oscillator should measure 15.968MHz or there about. I'll see if I can measure it.
I'm pretty sure everything is soldered well. I inspected everything with a microscope after I assembled the board...and, I know someone else who observes the same issue. The oscillator I did source very inexpensively from China, so that might also be an issue too. Thanks for your suggestion.
What? you suspect a bad crystal from China, where they pay their highly skilled workers a bag of rice a month and put nets under the windows to catch the suicides? /sarcasm
Yeah, it's likely you got a bad crystal. The problem with measuring it is that as soon as you connect a frequency counter to an oscillator circuit, the added capacitance of the probe will lower the frequency substantially.
I bought a bag of 10 or 20 crystals when I built mine. so far the ones I've built run OK. If you could source another crystal of known quality that would be ideal.
Thanks for your reply. You're right, if I probe it, it would likely slow it down more, but I think it would be measurable. I might play with it, but probably not immediately.
The only reason I noticed the second skipping was because of the first issue listed, but I've now noticed this on the first clock I built. Poking around the internet yesterday, I was looking for the expected accuracy of the crystal resonator used. Although I don't have specs for this particular one I purchased, I found that the expected accuracy (at least from one internet site's perspective) is 0.5%. Mine is actually better than that at 0.22%. So I don't think my crystal is bad. Even a more precise oscillator is going to be off and will eventually show a sign of re syncing.
So I'm just going to ignore it...it is fine and I'll eventually forget about it. However, I think the issue would be better fixed by FW. Basically the FW can learn the discrepancy and adjust the loop time accordingly to prevent a resync...but again, how important is this...not very.
Thanks for your help. I appreciate having a place to share ideas.
I had the impression that your clock had a bad stutter. A small stutter is normal. I think when I build my next clock I'll use 15pf-18pf caps on the crystal to make it run slightly fast. What I expect is a long second when it stutters - not quite as noticeable as a short second.