0

Chronodot drift
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Chronodot drift

by JakeConhale on Fri Jun 22, 2018 10:21 pm

I am working on my own variant of a BTTF clock, using the chronodot for time accuracy. Lately, however, I've noticed the device isn't keeping good time. Like I start the machine and set it and 24-48 hours the clock is 20+ minutes slow. My intention was to simply have the chronodot maintain time, periodically query it, and just display whatever time it was, so if my board did experience latency, it would all balance out as I'd always display the correct time at that moment. This does not seem to be happening.

Sequence of events:
- I power on the clock
- set the clock time to the NIST website and my watch, doesn't have to match but should give me an idea of relative drift.
- I either leave it alone or play around with different inputs (set different Destination or Past times, just play around with the musical keypad tones, switch between 12 and 24 hour modes) all without reloading the board, just using the latest instance.
- after a while, I come back and it's off.

I haven't been scientific such as to compare relative drift if I just leave it there rather than me fussing with the inputs, but the inputs should all be on the Uno and shouldn't affect the chronodot, should they? The time is only set in two places: during the initial setup() function and when I explicitly set the input, even then that wouldn't explain why it slowly deviates, I'd just find the time completely off.

I don't have a reliable multimeter, so I'm not able to currently confirm the voltage or current levels, but could some form of over-voltage or current affect it? I use a lot of the LCD character backpacks and thought I needed more power than my Uno could provide, so I'm running my circuitry off a 9V 2A power supply (which goes to the Uno) and through the u-beck step down converter to bring it to something like 5v 2A. The displays STILL flicker but that's another issue and one I'll probably just have to live with.

Could I be over-querying the chronodot? That is, could my requests for time be introducing latency which is somehow throwing off the dot? My current setup just queries for the .now() value, but alternate modes ask for the Unix time so I can easily add or subtract hours for different time zones, would that operation affect the dot?

My current plan of action is to return my circuitry to Uno power and see if the issue still exists, but was curious if someone could lead me to a more specific diagnosis.

Thanks!

JakeConhale
 
Posts: 4
Joined: Sun Feb 18, 2018 12:51 pm

Re: Chronodot drift

by adafruit_support_bill on Sat Jun 23, 2018 6:51 am

The Chronodot time should not be affected by polling the device. And assuming that you have a good backup battery installed, it should not be affected significantly be any voltage dips.

My current plan of action is to return my circuitry to Uno power and see if the issue still exists,

That is a reasonable plan. If you are still seeing drift in that configuration, we can assume that the problem is in the Chronodot.

adafruit_support_bill
 
Posts: 69809
Joined: Sat Feb 07, 2009 10:11 am

Re: Chronodot drift

by JakeConhale on Sat Jun 23, 2018 10:09 pm

Alright, guess I knew the solution coming in.

I first removed the external power supply and used the Uno power outputs, synched the clock to my watch, and proceeded to throw the sink at it. I changed times, visited different mode screens, played little jingles with the keypad such as Close Encounters from the Third Kind, Mario, Ode to Joy, and the Pacific Rim riff and just did everything I could to alter the system. After several hours, it was still within a second of my watch.

I altered my circuit, running the chronodot, a .56 quad backpack, and a couple other devices off the Uno power and left the 11 other backpacks and 5 neopixels off the external power source. Still solid.

JakeConhale
 
Posts: 4
Joined: Sun Feb 18, 2018 12:51 pm

Re: Chronodot drift

by adafruit_support_bill on Sun Jun 24, 2018 7:21 am

Hmmm. Good news for the Chronodot I guess. But that still doesn't explain the drift you were seeing in your clock. Do you have any more detail on the clock build? If you post some diagrams and/or photos it might give us a clue about the drift.

adafruit_support_bill
 
Posts: 69809
Joined: Sat Feb 07, 2009 10:11 am

Please be positive and constructive with your questions and comments.