| AlphaChannel | Posted: 8 Apr 2026, 07:19 PM |
|---|---|
 Member Posts: 2 Joined: 6-April 26 |
My clock is working well! Thanks for making it! Now for the minor bugaboo. When I have a GPS fix, the colons toggle in sync with the seconds digit changing — mostly. Occasionally however, the colon toggle is late by a tenth of a second. I filmed it with the high-speed mode of my phone with the seconds and deciseconds digits, a colon, and the GPS fix LED on the back all visible. The fix LED blinks perfectly in sync with the digit change, but the colon toggle can be late without any pattern that I can see. It can turn on promptly or late, and turn off promptly or late. It does seem to be always either prompt or exactly 10 centiseconds late. I have a Mark III. Rev. 3b on the back of the PCB. I have a hypothesis or two and I am comfortable compiling/assembling a custom build and flashing it, if it turns out software can fix this. But I'll wait to hear what others might think before posting wild hypotheses. ------------- |
| [top] | |
| mit | Posted: 8 Apr 2026, 08:44 PM |
 yeah whatever Admin Posts: 670 Joined: 4-May 16 |
Hello - this is actually intentional behaviour for the Mk III clock. The delay to the blinking was added as a debug signal while I was developing the firmware. It signals whether the internal oscillator is currently fast or slow. The oscillator is disciplined to the GPS pulse, with the firmware tweaking it faster or slower each second. If it's alternating between delay and no delay every few seconds, that means the internal oscillator is correctly locked to the GPS pulse. Off the top of my head, I can't remember if there's an easy way to disable the effect if it's not wanted. I would need to look through the software again. On the Mk IV of course, the colon blinking is excessively customisable... ------------- |
| [top] | |
| AlphaChannel | Posted: 9 Apr 2026, 05:01 AM |
|
Member Posts: 2 Joined: 6-April 26 |
Aha! I'm glad I asked. I'm probably going to explore building a custom patch the software and flash it myself. Is github: mitxela/PrecisionClockMkII the correct repository for Mk III software? I don't see a closer one. ------------- |
| [top] | |
| mit | Posted: 9 Apr 2026, 11:06 AM |
|
yeah whatever Admin Posts: 670 Joined: 4-May 16 |
Yes, the Mk II and III use the same firmware. The colons are wired up as the decimal points of the last two digits. One of the tricky parts, and partly why I used this as a convenient display, is that if the oscillator is fast, it will roll over before the PPS status is known. The timing adjust routine at the very end of the firmware stores 0x80 in the variable "fix", the interrupt routine at the start of the firmware caches it in a variable "fixDisplay" to keep both colons in sync. But if it was fast, the deciseconds won't be updated until the centiseconds roll over. You could potentially re-send the serial data for the deciseconds every interrupt. Or, you could force the toggle of the colons outside of the timingAdjust, and accept the possibility of a one second delay after it loses a fix before the colons stop blinking. ------------- |
| [top] | |
Sign in to post a reply.