Problems with Timesquare watch

[AK_Eric]

That's awesome, thanks.

[AK_Eric]

So, big sigh: I got the replacement watch, and finally had time to solder it myself (rather than my son): I wanted to make sure I really took my time, so all the solders were really solid. Long story short: I have the exact same problem as the first watch. At least the battery that came with this one was fully charged. But the behavior (like the first watch) is really weird, and all over the place:
Put in battery, switch it to binary mode: I can see the seconds counting normal (yay, it works!). I set the time, and suddenly, the seconds stop advancing. (what the....)
Pull out \ put in the battery, switch to binary mode: Seconds aren't moving at all. (saddness...)
Pull out \ put in the battery, switch to binary mode: Seconds are moving normal again. (uh... success?) Set the time. Time stops advancing. (saddness...)
I go back to the watch a few hours later, and the time has advanced maybe 5 minutes: It's once again keeping time super.... slow....
I went back and resoldered the tops of all the IC's jut to make sure they're all getting good connections (but they all had good flowthrough to begin with)... but the behavior hasn't changed.
I've soldered a lot of stuff successfully in my day (most recently from Adafruit specifically the motorshield & waveshield)... I have a hard time thinking I have a bad solder. I just can't believe I've got the exact same problem twice, on two different kits.
I'm guessing there probably isn't a resolution to this... by some freakish coincidence I've received two bad watch kits. Adafruit was great by sending me out a replacement, but I think at this point I'm going to write it off as a solid attempt, that wasn't meant to be.

[Zener]

Is there a schematic somewhere?

[adafruit_support_rick]

@Zener - schematic is here. Not much there to go wrong. The DS1337 datasheet claims it will keep time down to 1.3V. No external caps or resistors required on the oscillator, either.

If I'm reading this thread right, the watch functions normally apart from the time slowth, correct? That would seem to rule out some sort of firmware issue on the 168.

The only thing left would be some sort of failure on the I2C communication. However, the firmware uses the standard arduino RTCLib and, if I'm reading that code right, you would just get garbage or zeros if the I2C read failed - you wouldn't just keep getting the results of the last successful read. So kind of rules out I2C as a source of 'frozen' time.

I can't see anything in the firmware that could account for 'slow' time.

The 168 does go into power-down when the display isn't running, so there exists the possibility of it waking up into a screwy state. But I can't see any way for that to affect timekeeping on the RTC, so long as the voltage doesn't totally brown-out.

[Zener]

Are pins 4 to 13 of the DS1337 connected to ground?

[AK_Eric]

@driverblock: Yes, everything seems completely normal other than the fact it doesn't actually keep the time correctly: I just checked it: it's advanced 16 minutes since I set the time yesterday. But all the modes work just fine other than that. I've attached a pic of the new watch just so people won't think I'm completely nuts Some of the solder on the IC legs looks a bit messy since I did that after I had the whole thing put together. The back side is solid though.
I took out \ put in the battery several more times: When I first soldered it up a few times the seconds would advance at the correct rate, but now it's in constant slowmo, about ten real-world seconds to one watch second. I think the previous watch was about 8:1.
Eh, as I said, I'm giving up on this. But I appreciate the feedback.

[adafruit_support_rick]

I'm sorry it's not working for you. I've asked someone else here to have a look at this thread. Hopefully, he'll have some ideas.

If not, please contact support@adafruit.com with a link to this thread; we'll refund you the cost of the kit.

[Zener]

Sorry my previous post was way off. Yeah, only 8 pins.. All I can think of is the Maxim part might not like that particular crystal. I have seen 32khz circuits be picky before. Maybe verify the crystal shipped in the kit matches the pF and ESR called out in the data sheet. Maxim tech support is helpful if the trouble is with their chip.

[adafruit_support_rick]

@Zener - you may be right. I mean, really, what else could it be? So far, this is the only report of this problem we've seen. If it happens again, we'll see what Maxim has to say.

[Chuckz]

@Zener - you may be right. I mean, really, what else could it be? So far, this is the only report of this problem we've seen. If it happens again, we'll see what Maxim has to say.

Has anybody's watch been running slow?

[Smeeon]

I completed soldering together my watch sunday, and within the first hour I noticed it was running slow, then for 3-6 hours it was spot on time, now a few days later it is 10 mins behind consistently. I was wondering if it was the crystal too, please advise.

[adafruit_support_rick]

now a few days later it is 10 mins behind consistently.

Please reset it to the current time. Check it every day for a week and let us know how much it's losing every day...