Wikipedia:Reference desk/Archives/Computing/2017 May 2

= May 2 =

Digital Clocks and their rates
I have a wall and a nightstand digital clock. Both are of same brandname (no other digital clock or watch of this brand in our house). I have noticed time and again that these both gain 1/2 an hour each month. Can something be done about them, like some manipulation in their circuits, or changing their oscillators etc.? Jon Ascton   (talk)  10:05, 2 May 2017 (UTC)
 * Are they quartz or line frequency clocks? If two line frequency clocks are used, they'll both track the same line - although it's unusual for them to gain that much. The problem would be with additional noise on the line, as assorted glitches could each make them count ahead a little.  You can try using a mains filter to trim these glitches, otherwise talk to the power company?
 * Quartz clocks are effectively unadjustable and certainly not to that extent - but then that's an exceptional amount of error and it would be unusual for two to go out by the same large amount. Andy Dingley (talk) 10:35, 2 May 2017 (UTC)
 * Yeah, there's something odd going on here. No brand of clocks that went out of sync that quickly would stay in market for long. Is the OP sure that nobody is manipulating the times? There's a subset of people who feel like the only way to ensure they're on time is push the clocks around them ahead so that their rushing will counteract their laziness and get them to appointments on schedule. I would talk to your spouse/kid/room-mate, etc. Matt Deres (talk) 11:23, 2 May 2017 (UTC)


 * Re: "No brand of clocks that went out of sync that quickly would stay in market for long." While true, all the company has to do is keep changing the brand.  Many companies do this, such as in the early days of cell phones.  It's true that a detailed search of reviews under either the old or new brand will show either bad reviews or few reviews, but most people don't spend this time before buying a cheap clock.  Also, the company can just buy out formerly reputable brand names from bankrupt companies.  StuRat (talk) 15:08, 3 May 2017 (UTC)


 * I agree. That amount of error sounds more like human intervention, though I suppose there is a possibility of some nearby electronic circuitry triggering extra cycles in a mains-synchronous clock. Here in the UK, most deviations from standard time are corrected by small adjustments in mains frequency at the generator site, so "mains" clocks are seldom more than a few seconds out. Try moving one or both to a different location in your house to see if some interference is affecting the standard frequency.   D b f i r s   11:26, 2 May 2017 (UTC)


 * Thanks for the answers, but they have nothing to do the mains etc, as both are run on 1.5v DC batteries. : OP


 * In that case, I suggest replacing the batteries.  D b f i r s   14:19, 2 May 2017 (UTC)


 * In my experience, when the batteries are running out, the clocks run slow - and these are running fast. Wymspen (talk) 19:39, 2 May 2017 (UTC)


 * Are you sure they are supposed to normally run on those batteries, as opposed to just using them as a backup when mains power is unavailable ? Look for a cord in the battery compartment.  The clocks I've seen with backup batteries (usually 9V, in my experience), do keep dismal time when running on them. StuRat (talk) 20:37, 3 May 2017 (UTC)


 * It's not clear when you say digital whether these are LCD (or other) display or whether they have sweeping hands. Are you using the wrong type of battery, such as rechargeable 1.2 V instead of carbon 1.5 V, or vice versa? Akld guy (talk) 20:00, 2 May 2017 (UTC)


 * Half-an-hour per month equates to around 1 second per day. I habitually wear a cheap quartz digital watch of a particular brand (Casio) and design, which after several years begins to lose its metallic coating to expose the plastic of the body, at which point I replace it. I have noticed that all these successive watches (including the current one) gained/gain, as exactly as observable, 1 second per day. Around once a month or less I re-set the watch according to the radio time signal ("The pips", broadcast on BBC Radio 4 hourly).
 * I speculate that this consistent 1-second-per-day gain in my watches (and the OP's clocks, if inexpensive) is actually a design feature, to allow for variations in manufacture and err on the side of gaining, which for a casual user is less problematic that losing. Being a minute or less early for, say, catching a bus is harmless, whereas being a minute or thereabouts late may mean missing the bus, for example. {The poster formerly known as 87.81.230.195) 2.122.60.183 (talk) 13:26, 3 May 2017 (UTC)


 * Check your math. A month is approximately 30 days. Half an hour is 30 minutes. 30 minutes over 30 days is 1 minute per day, not 1 second per day. 209.149.113.5 (talk) 13:55, 3 May 2017 (UTC)


 * Think we know what 2.122.60.183 meant, as we can do the math in our head. To get back to the OP question. When these clock chips reach the end of the production line they all get tested. Those that fall way outside the specifications are often sold off very cheaply. The OP  was careful not to mention the brand, but pointed out they were one of the same.  He may just have bought cheap and cheerful clocks that keeps time worse than an 100 year old Grandfather Clock (or long case clock to give it its proper name).  As far as I know, there is no way, one can recalibrate them.  It is down to the crystal oscillator. If that is off,  the frequency divider circuitry can't compensate because it is a hardwired  chip. Radio controlled clocks are a good idea but then again, you get what you pay for.  Some only last for a few years. In the long run it pays to buy   good quality timepieces that will serve not only throughout  your lifetime but you can pass them on to your grandchildren. Personally, I don't bother to wear a wrist watch anymore, as my wife tell me when I have to do anything. Aspro (talk) 17:25, 3 May 2017 (UTC)


 * Yes it should be obvious to the first IP that they were wrong since they say re-set it once a month or less. I'm guessing they're not only re-setting it when it's 30 minutes or more out of sync since that's a difference that is likely to be fairly annoying. Nil Einne (talk) 12:18, 4 May 2017 (UTC)

---

Here is a TI datasheet for a typical digital watch crystal:

An accuracy of ±20 ppm is typical, but you can buy crystals at ±10 ppm or ±5 ppm. That being said, look carefully at figure 6 of the above datasheet.

Here is a website that will convert ppm to seconds per day:

A Casio F-91W is specified to be accurate to ±30 seconds per month. Some of the high-end non-atomic casios are specified to be accurate to ±15 seconds per month.

The above claim ("It is down to the crystal oscillator. If that is off, the frequency divider circuitry can't compensate because it is a hardwired chip.") is incorrect. There are two ways to adjust the frequency of the crystal. First, you can use a variable trimmer capacitor, as detailed in figure 5 of the above TI datasheet. Second, you can program the chip to gain or lose time, storing the calibration value in flash memory. In theory you can also detect the temperature and adjust for that, but this adds expense and I do not believe any watch or low-cost digital clock does this.

You may also find this to be of interest: --Guy Macon (talk) 19:19, 3 May 2017 (UTC)


 * Your assuming that the manufacturer bothered to add a trimmer ( or external load) capacitor at extra expense. If they have done, one would expect the clock  to have been calibrated to keep better time before it left the factory. As I said, it probably has no available adjustment. Take  your average computer. Why do they not keep good time?– were is this trimmer on the mother board?  .  The OP is free to look and post an image of anything he may find. If an external cap is not present, then as I said – it is just a hardwired clock chip. Yet, I think we agree on the basic probable cause. Aspro (talk) 20:32, 3 May 2017 (UTC)


 * At the old school engineering, when a quartz device was in the size of a thumb, even radio amateurs opened it up and were grinding it down. Using toothpaste for grinding until it resonance became the preferred frequency which rises by grinding. The data sheets show oscillator circuits using a capacitor to ground at each pin of the quartz to keep it in it's origin resonance and prevent it to oscillate it a double of it's resonance. In older devices, one of this capacitors was an adjustable one. Adding mote solder causes a higher capacity and gets the frequency lower. Two causes result form the soldering. Frist, a ceramic capacitor restores at 120°C or higher temperature and increases it's capacity to nearly condition in new. (Btw. this it whats behind applying hot air to the Play Station motherboard, which restores the ceramic capacitors, but ages the electrolytic capacitors). Second, the additional solder behaves like the plate of a capacitor itself an decreases the frequency. The only solution might be to replace capacitors or the resonator. In manufacturing of electronic devices and components, samples were tested and selected when necessary. This would require You to setup a test circuit, independent of the clock, You are going to fix. In Laurens Hammond company, capacitors even used for the filters of the tone generators were carefully selected, to provide precise tone quality of each Hammond organ. Today CPU wavers are selected to target its best application. Capacitors are selected when using them at the border of specification or the application required least tolerance the manufacturer specified within his production. -- Hans Haase (有问题吗) 22:05, 3 May 2017 (UTC)

1/2 hour a month is completely wack. A few seconds a day, maybe. I had a cheap digital watch some years ago that in fact had a tiny trimmer cap, but don't know if they're still made like that. I'd expect during manufacturing if they use a cheap crystal and no trimmer, they can just run the oscillator for a few seconds comparing with a calibrated frequency standard, measure the error, and program digital compensation into the watch chip, whether with flash memory or just burning some lines on the chip with a laser (chip-on-board packaging can make that convenient). 173.228.123.121 (talk) 06:07, 4 May 2017 (UTC)