Improving a slow/fast digital clock

[eccerr0r]

Re: Improving a slow/fast digital clock

Note that 50ppm error is already about 30 seconds per week ... so don't get the cheap ones

[redwire]

Re: Improving a slow/fast digital clock

Tuning fork crystals are delicate. Maximum drive level of 1μW for 32.768kHz parts so they need a big series resistor ~300k is usually missing.
So most circuits overdrive them, and they age rapidly or fracture.

The oscillator is almost impossible to work on, because scope probe capacitance a few pF upset things.

Hard to tell what is causing the clock to be slow/fast, it can be VCC or temperature moving around too much, or a bad oscillator design that needs tuning (caps) optimized. I would try a new crystal - if it worked when new.

pic from https://cdn.badcaps-static.com/pdfs/...326956ddf3.pdf

[eccerr0r]

Re: Improving a slow/fast digital clock

That thing looks like it uses a "tuning fork" crystal oscillator though you need to closely inspect it to be sure. Then go buy a bunch of them and replace them every week until you find one that's accurate enough

[petehall347]

Re: Improving a slow/fast digital clock

Your (cell) phone should resync with the network time servers each time it acquires signal.

only if its set to do that .mine isnt as i dont trust it

[Curious.George]

Re: Improving a slow/fast digital clock

this is also interesting on how to fix this for me as i have a battery powered digital lcd alarm clock. it uses three triple A batteries as opposed to ac line voltage and no surprises, its china made and the time drift on it is quite terrible. its like 10 secs faster every day.

You can purchase inexpensive clocks that will sync to WWV/CHU. I keep one for my bedside alarm. Unfortuantely, the clock isn't smart enough to know (or be told!) that we DON'T observe DST so, twice a year, I have to tell it we're in a different timezone.

[Curious.George]

Re: Improving a slow/fast digital clock

my mobile phone gains time .

Your (cell) phone should resync with the network time servers each time it acquires signal.

[petehall347]

Re: Improving a slow/fast digital clock

my mobile phone gains time .

[ChaosLegionnaire]

Re: Improving a slow/fast digital clock

this is also interesting on how to fix this for me as i have a battery powered digital lcd alarm clock. it uses three triple A batteries as opposed to ac line voltage and no surprises, its china made and the time drift on it is quite terrible. its like 10 secs faster every day.

after about a week, its 1 min fast and i have to keep setting it back and its getting quite annoying to do that every week. well, on the bright side, i wont be late for work or my appointments with it running fast and making me move faster! lol!

i guess the chinese tuned it for their colder temperate climate and when its brought into singapore, due our hotter tropical climate, the higher temperature causes a drift in the circuit, making the seconds tick by at a faster rate.

[Dannyx]

Re: Improving a slow/fast digital clock

Oh yeah, that's another thing I forgot: temperature. I took the timer inside after swapping it with the other one, since it's normally plugged into an outlet in a balcony where it can get pretty cold when winter comes around, since there's no radiator there, though the behavior occurred during summer time too, so something is drifting there due to various reasons...

[stj]

Re: Improving a slow/fast digital clock

the 2 white ceramic caps tune the crystal,
so it or the caps may be temperature sensitive.

[Dannyx]

Re: Improving a slow/fast digital clock

Here's some more pics of the inside of this thing. I took one with a light source behind the board to make the traces more visible like an x-ray. Sadly, nothing too interesting going on in there. I managed to work out what all those pins are for and that the clock is indeed crystal-based.

I popped off the board which holds the buttons and the display to the face plate and I was able to spot the crystal tucked under the corner of the display there. The proprietary epoxy blob chip is directly under the LCD as well, so no need to take that out as well, since it might not line up properly again.

That J pin we saw connects to the base of the S8050 transistor, so it's the control signal for the relay. The PWR pin is a sense wire connected directly to the unregulated rectified DC via a resistor and tells the chip whether or not the thing is actually plugged in, since it won't pull the relay when it's not, even if you manually set it to "ON". VCC and GND are self explanatory: GND is GND and VCC is a zener-regulated supply for the IC AND the battery, so now we know what the battery does too: it's simply placed in parallel with VCC and has a ceramic cap between it and GND. There IS a resistor feeding its + terminal from VCC, but it's got a single black band on it, so it's zero ohms used as a jumper, isn't it ? The relay's VCC comes from the unregulated DC as well.

So that's about it....not much I can do to improve this thing. Funny thing is that it's sat unplugged for a couple of days now and it seems to be running SLOW now - when plugged in, it would run too fast, though I haven't done an in-depth analysis and actually watch the thing carefully, but that's the way it feels...

[eccerr0r]

Re: Improving a slow/fast digital clock

I didn't know until fairly recently that crystal controlled (but not TCXO) clocks are fairly common nowadays. I have a few now. I haven't kept track of their drift.

And of course the WWVB clocks... I have one that drifts really badly when it can't pick up WWVB.

On another note, don't forget about all those analog clocks that are "quartz crystal" ... these too are of course use crystals. Don't need to wind them, but are they more accurate than mechanical detent clocks?

[Dannyx]

Re: Improving a slow/fast digital clock

Well, not necessarily, those old alarm clocks used RC oscillators to "keep" time when the power is out...they are HORRIBLE in accuracy when power is out, but better than if it didn't have an RC oscillator. It'd have a clock error equal to at least the amount of time the power has been out...

Yes, I can fully confirm this with two alarm clock radios I've got which have a 9v backup battery and "keep" the time when their power is cut, though like you said the accuracy is terrible over anything more than a few minutes at best

[eccerr0r]

Re: Improving a slow/fast digital clock

Well, not necessarily, those old alarm clocks used RC oscillators to "keep" time when the power is out...they are HORRIBLE in accuracy when power is out, but better than if it didn't have an RC oscillator. It'd have a clock error equal to at least the amount of time the power has been out...

[redwire]

Re: Improving a slow/fast digital clock

It would need to use a crystal for timekeeping from the battery. This unit looks like good quality.
I think there is no voltage regulator (maybe just zener Z2) and they wear out the NiMh battery because it's always charging. J might be the battery. It's big at 80mAh for some reason.

It could be the crystal needs tuning or that VCC moves up when the relay is off.

[eccerr0r]

Re: Improving a slow/fast digital clock

Ah okay, so it charges the battery and powers to the microcontroller, so that J wire remains a mystery that could be some sort of sync pulse. Need photos of the bottom side of the board, and/or see if there's a crystal on the LCD board...

[Dannyx]

Re: Improving a slow/fast digital clock

One thing that's weird is that the battery is on the power board, does the controller know whether or not AC is applied (when you unplug it, does the LCD go blank?)

No, it doesn't go blank when you unplug it, so it's a backup battery to save the clock and settings. However, when the battery finally dies out, like this one has after it's sat in a drawer for so long, the next time you plug it in, the display doesn't come on right away like you'd expect, so when I first bought it and brought it home I thought it was defective, but it turns out that battery must reach a minimum lever first before the thing can run, so maybe the capacitor dropper is only charging the battery and the IC runs ON the battery itself rather than on mains...I'll have to do a bit of a dig to find out how the traces are connected to find out for sure.

[eccerr0r]

Re: Improving a slow/fast digital clock

One thing that's weird is that the battery is on the power board, does the controller know whether or not AC is applied (when you unplug it, does the LCD go blank?)

[Dannyx]

Re: Improving a slow/fast digital clock

does the timer involve mechanical parts ?

No, it doesn't. In fact, here's some pictures of it...not much going on inside. There's a cap dropper, a relay and a couple of capacitors as far as the power side is concerned, so I reckon the brain of it all is on the board on the left where the buttons and display are. The flex cable going between the boards has its pins labeled "VDD, GND, J and PWR". Three of those make sense right away: VDD power for the chip, GND and I would surmise PWR stands for the signal going to the relay trigger circuit. That J is a bit of a mystery and although I haven't taken the board out to see where it leads, I think it might be where it's grabbing the 50Hz sync pulse for the clock ?

This particular one is not the one currently in service, but an identical one I had laying around for some unknown reason (I actually had no idea I bought TWO ), so I'll pop this one in to see if it drifts as well over, say, one month's time.

[petehall347]

Re: Improving a slow/fast digital clock

does the timer involve mechanical parts ?