Ice Tube Clock Runs Slow

[mwa]

Hi, I'm new to the forum.

About a month ago I assembled the Ice Tube clock, purchased through Maker Shed. No problems with assembly; the display works fine, as do all clock functions. But it runs noticeably slow.

More specifically: I'm using (several) radio-controlled clocks (i.e. WWVB "atomic clock") as reference. I set the Ice Tube on 15 December against the reference. Tonight, I checked the clock against the reference, noting the Ice Tube clock drift. Over a period of 19 days, 5 hours, the Ice Tube lost 1 minute 18 seconds (- 01:18) over that period. I believe that works out to approximately 47ppm on the slow side.

I'm pretty sure that the crystal is the sole component determining clock accuracy. The specs indicate +/- 20 ppm accuracy, so it looks like the crystal is out of spec.

Okay here's my question: what's the best remedy? Replace the crystal? Adjust capacitor values to pull the crystal into shape? Is there a software coefficient that can be easily adjusted?

Thanks in advance,

-- Marty

[jarchie]

I believe this is normal behavior and suspect that it is due to the crystal caps (1, 2). Although not supported by the original firmware, a popular modification is drift correction to make periodic time adjustments to compensate for drift.

[adafruit_support_bill]

These clocks do drift, some more than others. Swapping parts might get you closer, but it would be a lot of trial & error to find the magic combination.

Several users have developed drift-correction as mentioned above by jarchie. There are also mods that use a GPS module for a precise time sync. http://learn.adafruit.com/ice-tube-clock-kit/mods

[jarchie]

After rereading this post, I wanted to answer wma's questions more directly.

what's the best remedy?

This depends on what you're comfortable with. I believe the most popular solution is to use an alternative firmware. My xmas-icetube is a popular choice. Two other alternatives I can recommend are William's firmware and Jeremy's firmware. But xmas-icetube has good documentation and is the most feature rich. All of those projects support drift correction, but only xmas-icetube and William's firmware support the GPS mod that adafruit_support_bill mentioned. If you do go with the GPS mod, you might also be interested in this post describing how to install the Adafruit Ultimate GPS module.

If you're interested in xmas-icetube, but do not want to mess with microcontroller programming yourself, I would be willing to preprogram a microcontroller and mail it to you if you reimburse my expenses ($3 plus shipping). Just email me privately if that interests you.

Replace the crystal? Adjust capacitor values to pull the crystal into shape?

If you decide to go the hardware trial-and-error route, I suspect the crystal is fine. I would suggest desoldering both crystal caps and using a 10 pF cap on the XTAL2 side of the crystal, leaving no cap on the XTAL1 side. If that doesn't give good results, I would try soldering a second 10 pF cap on the other side. (If you do want to try the hardware route, I can elaborate here...)

Is there a software coefficient that can be easily adjusted?

Not in the official Adafruit firmware, but if you switch to William's firmware or Jeremy's firmware, you can manually adjust a drift correction constant through the menus to make the clock run slower or faster. My xmas-icetube takes a different approach, and automatically updates the drift correction whenever you change the time; the full method is described in another post. The disadvantage of the automatic method is that drift correction is initially inactive; you must allow the time to drift by at least fifteen seconds and then set the correct time. After that, the timekeeping is pretty much dead-on.