[time-nuts] Re: Power and heat re: Heathkit WWV clock / where are the good oscillators?

[Bob kb8tq]

Hi

Backing up a bit …..

A month might be 30 days, that’s about 2.5 million ( 2,592,000) seconds. 

A second a month is  about 0.4 ppm ( 0.3858).

There are a *lot* of oscillators out there that will hold much better than a
tenth of a ppm over a month. That’s for all effects and not going with the
somewhat odd “chronometer” rating approach. 

If you want to spend some money you can find devices that get into a tenth 
of a PPB ( so 1,000 X better) and still not be over $100 on eBay. No, they
won’t run on a AA cell for a couple years. 

For a bit more money, various outfits will sell you TCXO’s that are in the 
PPB per month sort of range. They will run on low-ish power. Finding them
on eBay? not so much. Getting them to sell you less than a couple hundred?
also not so much. Still, if you have an OEM sort of “need” the parts are
available. 

RTC modules are targeted at very low power. They sell into a market that 
has a “good enough” approach to time, but values low power a lot. If you 
want something more accurate, there are parts out there to build up what
you want. This DIY approach is pretty much guaranteed to use a bit of 
power. That seems to work for the OEM’s who need it …..

Bob

> On Aug 6, 2022, at 12:34 PM, Joe Duarte via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> Hi all – I've been reading up on the *Heathkit GC-1000 Most Accurate Clock*
> from the 1980s, which syncs with WWV. I've seen numerous reports of the
> flawed power supply and regulator, and the intense heat it generates in the
> chassis, and I'm stumped. Why does it need so much power that it's getting
> hot? There's hardly any computation involved in syncing with WWV, decoding
> its BCD bitstream, etc. Can I expect similar issues if I build my own clock
> instead of restoring a GC-1000?
> 
> Quartz watches can sync with WWVB and run for a couple of years on, what,
> ≈200 mAh coin cells? I'm amazed that some are able do GPS, which is far
> more computationally taxing, but I think most of those are solar with a few
> months of battery backup. I don't understand why a WWV clock isn't sipping
> milliamps in cool silence like a watch. Is it something to do with the WWV
> medium band frequencies (5-15 MHz) compared to WWVB's long wave (60 KHz)?
> (Why did Heathkit sync to WWV and not WWVB? I thought the latter has less
> propagation delay.)
> 
> The oscillator runs faster than a standard 2^15 Hz movement, but so do lots
> of high accuracy quartz watches. I've looked for an Omega quartz ship's
> chronometer from the early 80s, but they're too rare apparently. It has a
> 2^22 Hz oscillator and runs for a couple of years on two AA batteries.
> (Scroll down to the ship's version:
> https://en.wikipedia.org/wiki/Omega_Marine_Chronometer) And the new Citizen
> Caliber 0100 movement runs at 2^23 Hz, in a watch... (
> https://www.hodinkee.com/articles/citizen-eco-drive-caliber-0100-review)
> 
> By the way, why am I not finding any RTCs better than 20 or 15 seconds per
> month accuracy? I've looked on Mouser and Digikey. It's like there's been
> no progress since the 80s. That Omega was good for maybe 0.4 sec/month
> drift, worst case, and the new Citizen is unbelievable at under 0.1
> sec/month. What performance can we expect in a disciplined oscillator like
> the GC-1000's? I haven't found any specs on its *endogenous* accuracy after
> some break-in period with enough disciplining. I'd like to have less than
> 0.1 sec/day between syncs – syncing isn't possible at all hours, and GPS
> won't work for my application. So a 3 sec/month clock on its own. The world
> is full of quartz watch movements more accurate than that, without needing
> ovens, many dating to the 1970s, so I'm confused by why movements that sit
> on a desk and never move are so subpar compared to watch movements. Has
> anyone leveraged a watch movement in a desktop chassis? I wonder about the
> interfaces, since they're all used in analog watches and I don't know if
> they express/output time in a way that can be used by a controller.
> 
> Thanks,
> 
> Joe
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com