Black Lives Matter - Action and Equality. ... Adafruit is open and shipping.
0

Ice Tube Clock loosing time
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Ice Tube Clock loosing time

by ki4mmm on Mon Sep 14, 2009 9:47 am

My clock is loosing about 4 seconds a day. I even tried wiggling my crystal, as in a current, related thread (about the atmega168 clock), but with no change. Anybody else notice this, or am I alone on this one? Any suggestions?

Thanks,
Greg
ki4mmm
 
Posts: 19
Joined: Sun Aug 16, 2009 7:05 am

Re: Ice Tube Clock loosing time

by adafruit on Mon Sep 14, 2009 12:13 pm

dont 'wiggle' the crystal it will damage or break the leads. the clocks will lose/gain a few seconds depending on the humidity/temp/etc. you can try replacing the capacitors with 18pF.

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm
Location: nyc

Re: Ice Tube Clock loosing time

by sparr on Mon Sep 14, 2009 1:22 pm

4 seconds a day... kids these days! I remember when a GOOD digital wristwatch lost as much. That's only one minute every 2 weeks.
sparr
 
Posts: 196
Joined: Tue Nov 04, 2008 5:21 pm

Re: Ice Tube Clock loosing time

by adafruit on Mon Sep 14, 2009 1:35 pm

well, its more than the clock is 'spec'd' for - we did tests on our prototypes and found that it lost/gained 2 s a day which is as good as it gets. you may want to perhaps clean off the solder flux? that could be adding a pf or two which can cause the crystal to be unstable

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm
Location: nyc

Re: Ice Tube Clock loosing time

by ki4mmm on Mon Sep 14, 2009 1:57 pm

Okay, thanks for the information! I am new to clocks and just naively presumed it would be '"real accurate", but I realize now that my other clocks are no more accurate. It seems that I have gotten spoiled on WWV clocks. I'll watch it, and maybe play with the caps (I have all of the common low-pF NPO's); otherwise, I'll just enjoy it, and reset it once in a while, as I do for my other clocks :shock:
ki4mmm
 
Posts: 19
Joined: Sun Aug 16, 2009 7:05 am

Re: Ice Tube Clock loosing time

by mpeg2 on Mon Sep 14, 2009 4:38 pm

What about using the powerline cycles for the timing element? This would mean using an AC brick instead of DC (and putting some of the powersupply inside) - but the accuracy is much better than a few seconds a day...

Rich
mpeg2
 
Posts: 30
Joined: Fri Sep 11, 2009 9:45 pm

Re: Ice Tube Clock loosing time

by adafruit on Mon Sep 14, 2009 5:59 pm

AC adapters are tougher to find than DC and also we would have to deal with different countries' frequency. mostly its the former. we can source 220V DC adapters but not AC and this was an issue for x0x customers that we didnt want to repeat

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm
Location: nyc

Re: Ice Tube Clock loosing time

by mpeg2 on Mon Sep 14, 2009 6:01 pm

Good point - hadn't thought about those aspects...
mpeg2
 
Posts: 30
Joined: Fri Sep 11, 2009 9:45 pm

Re: Ice Tube Clock loosing time

by richms on Wed Sep 16, 2009 9:15 am

AC adapters are getting hard to find, since they don't meet the efficiency requirements that are in place now due to them using an iron transformer. Also the cost is almost always more then a cheap switch mode one.

Also power line only has long term accuracy - during the day its not uncommon to see errors of +/- 1 min depending on the loading on the network. That's only a 0.1% error but still enough to really annoy you.

I have cheap wall clocks that are far far worse for timekeeping, really the error isnt anything to get too worried about IMO.

richms
 
Posts: 558
Joined: Tue Jan 20, 2009 3:05 am
Location: New Zealand

Re: Ice Tube Clock loosing time

by Abbarach on Wed Sep 16, 2009 11:27 am

I know that some of the nixie clocks are set up to use a serial GPS receiver to time-sync. I'm far from an expert in these types of things (Software developer by trade, trying to expand and learn the hardware/microprocessor end of things), but it may be possible to hack on the appropriate components and make some code modifications to allow GPS as a time source.

Now I'll sit back and wait for the experts to comment on just what would be involved!

Edit: Also, it might be possible to hack on an ethernet connection and do an NTP time-sync. I imagine that this would probably be more complicated than the USB, but it may also be cheaper. IIRC most serial GPS receivers are around $50 USD, and you have to be near a window or other area that is relatively radio-transparent.
Abbarach
 
Posts: 15
Joined: Mon May 04, 2009 8:58 pm

Re: Ice Tube Clock loosing time

by richms on Wed Sep 16, 2009 12:57 pm

There are probably loads of other time sources that could be used, but really once you have the clock assembled, thats the beauty of an open source project like this, you can implement it and then publish it.

I saw that the usart pins are not used, so thats great if you were to want to hook a gps up to it. Otherwise if you have still got a landline you could hook up an old modem and use it as a big caller id box that syncs off that for the time (accurate to within a minuite since they dont send seconds)

I cant get one for at least 7 weeks, but I am seriously thinking of it rather then more xbee stuff since its cooler and I can leave it out and people will see it. Probably the most presentable as a household item kit of the lot IMO..

richms
 
Posts: 558
Joined: Tue Jan 20, 2009 3:05 am
Location: New Zealand

Re: Ice Tube Clock loosing time

by adafruit on Wed Sep 16, 2009 11:15 pm

a gps module would be perfect, simply use a '406 which is 5V compliant. we have them in the adafruit shop ;)

adafruit
 
Posts: 12151
Joined: Thu Apr 06, 2006 4:21 pm
Location: nyc

Re: Ice Tube Clock loosing time

by Abbarach on Fri Sep 18, 2009 4:45 pm

adafruit wrote:a gps module would be perfect, simply use a '406 which is 5V compliant. we have them in the adafruit shop ;)


Well, I think I have a project on my hands...

... once I can get my hands on an Ice Clock (so sad I missed the first batch! :( ), and a 406, and get my other projects out of the way (waiting on nixie clock parts to arrive from overseas), and find some free time!
Abbarach
 
Posts: 15
Joined: Mon May 04, 2009 8:58 pm

Re: Ice Tube Clock loosing time

by fat16lib on Sat Sep 19, 2009 10:10 am

First this kit is a masterpiece!

One way to improve the accuracy of clocks with 32.768 kHz watch crystals is to use a trimmer capacitor as part of the crystal load. If you could replace c8 or c9 with a trimmer, mounted in a secure way, you could adjust the accuracy

I built a clock for a data logger using a PCF8583 clock calendar. This part is designed to accept a trimmer for higher accuracy.

I was able to get under one second per day drift. The data sheet said you should be able to get 300 seconds per year or better with a trimmer. I used this trimmer http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=SG3003-ND. It is not special, I just had it and it matched the range specified by the data sheet for the PCF8583.

Most watch crystals are designed to have an inverted parabolic temperature curve with the zero coefficient at 25°C. So it is best if the crystal is near 25°C. One second per day is about 10 ppm.
tempcoeff.jpg
tempcoeff.jpg (29.17 KiB) Viewed 3323 times
fat16lib
 
Posts: 593
Joined: Wed Dec 24, 2008 1:54 pm

Re: Ice Tube Clock loosing time

by hinermad on Sat Sep 19, 2009 10:58 am

The power line frequency in most of North America is a quite good timing reference over the long term, but it does drift a bit. Back in the late 70s I had a kit-built line synchronous clock (any of you oldsters remember Poly Paks?) that I would synchronize to a WWV standard time transmission every day. The clock was never off by more than 4 seconds from WWV, but some days it would be fast and other days it would be slow.

The problem with power line sync is that if the power goes off you lose your timing reference. Inexpensive digital alarm clocks with battery backup have a cheap RC oscillator to provide a backup timing reference, but their accuracy is terrible. They're good enough to keep you from being late for work if the power goes off in the night, but that's about it.

Some GPS receivers provide an accurate digital pulse-per-second (PPS) output. It's just a pulse; there's no time value encoded on it. But it'd be simpler to hack a clock kit to use that as a timing reference and make the user set the correct time once than to make the clock receive and understand a time code stream from the GPS.

But I guess if we wanted simple we wouldn't be here, now would we? (Grin)

Dave

hinermad
 
Posts: 56
Joined: Sun Sep 13, 2009 1:25 pm
Location: Upstate New York, USA

Please be positive and constructive with your questions and comments.