sense wall clock 32khz crystal frequency without shifting it

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
as you can understand it's not a serious subject, considering that i wanna "adjust" the frequency of a stoopid wall clock to make it precise:), it's a classic quartz arm clock with an AA batt and goes like 5 mins slower per month

i think i can make some sort of circuit with an op amp or something to output the 32768hz signal and compare it to a precise (yhea, it is) DS3231 but i suppose that any circuit i arrange it will add some capacitance or load to the oscillator and shift the frequency enough to make any adjusment actually random

i'm thinking about a 10Mohm resistor (soldered on air) on an TL082 input but i also think it's a crappy idea, detecting magnetic\electromagnetic field vould be perfect but we are talking about very low currents and i'm not on this level:eek:

ideas?
 

MrChips

Joined Oct 2, 2009
30,720
5 min per month works out to about 1:10000 which is crappy.
A typical 32768Hz crystal should be at least 10 times better, about 10ppm or better, or about 30 seconds per month.
Almost anything you do will upset the frequency and give you an incorrect measurement.
Most clocks have a trimmer to adjust the frequency. The best you can do is adjust the trimmer and check the time after a month.
 

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
it's exacly the "after a month" approach that i don't want to do LOL

nope... no trimmers whatsoever, it's a german made "long shaft" generic mechanism for "decorated ceramic gimcrackery" but it's anyways an "el cheapo" clock
 

DickCappels

Joined Aug 21, 2008
10,152
You can compare the phase of one second ticks from the clock with one second ticks from a reliable time source and adjust your 32768 Hz oscillator so that the phase between the two signals is constant. That should be a lot faster.

An accurate time reference can he derived from a GPS receiver as well as from standard time and frequency broadcasts.
https://avrlab.com/upload_files/1.PDF

The frequency of the crystal oscillator can be derived by loosely coupling to the crystal and amplifying it. One such method is to monitor the current through the crystal, as discussed in this application note by Jim Williams.
www.linear.com/docs/25519

 

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
You can compare the phase of one second ticks from the clock with one second ticks from a reliable time source and adjust your 32768 Hz oscillator so that the phase between the two signals is constant.
mh.. sounds easier\faster, i'll give it a try, i just need to know the two source's deadtime difference to do an acurate compare in a couple of minutes (i think)... THANKS
An accurate time reference can he derived from a GPS receiver as well as from standard time and frequency broadcasts.
https://avrlab.com/upload_files/1.PDF
good to know for the next times :) but i already have a pair of DS3231 modules that both shift something like 1\4 of a second in 10 days when compared to http://www.time.gov/ (reloading the page several times and making an average:rolleyes:)
The frequency of the crystal oscillator can be derived by loosely coupling to the crystal and amplifying it. One such method is to monitor the current through the crystal, as discussed in this application note by Jim Williams.
www.linear.com/docs/25519
yhea!.. i also tought about a DC or maybe AC curent probe but this one is quite expensive (but it truly hits the spot :))
 

ian field

Joined Oct 27, 2012
6,536
as you can understand it's not a serious subject, considering that i wanna "adjust" the frequency of a stoopid wall clock to make it precise:), it's a classic quartz arm clock with an AA batt and goes like 5 mins slower per month

i think i can make some sort of circuit with an op amp or something to output the 32768hz signal and compare it to a precise (yhea, it is) DS3231 but i suppose that any circuit i arrange it will add some capacitance or load to the oscillator and shift the frequency enough to make any adjusment actually random

i'm thinking about a 10Mohm resistor (soldered on air) on an TL082 input but i also think it's a crappy idea, detecting magnetic\electromagnetic field vould be perfect but we are talking about very low currents and i'm not on this level:eek:

ideas?
My solution was to replace all my clocks with MSF/DCF77 time signal type - including my wristwatch.

Probably the least pulling on the existing oscillator, is to build a second oscillator around a CMOS gate - increase the feedback resistor until it just fails to oscillate. The existing oscillator might excite it into action if you just place it near. The existing crystal is most likely strung between the output and input of an inverter inside the chip - so one pin will have a stronger signal than the other, find out which and give it a Cm or two of antenna wire. If that isn't enough coupling, you can use a "gimmic" capacitor - Twist two short lengths of insulated solid core wire together for a very small capacitor that you can tune.

These days - there may be LV CMOS logic that will just about run off the AA cell. Once you've calibrated the oscillator - you can leave everything in place so you aren't removing any capacitance effect.
 

AlbertHall

Joined Jun 4, 2014
12,345
If it is an analogue display you should be able to use a microphone to pick up the ticking and then measure that period accurately.
 

dannyf

Joined Sep 13, 2015
2,197
Depending on how far you want to go.

The easiest is to get a better clock.

The 2nd easiest is to use a better crystal.

The 3rd easiest is to use a temperature controlled oscillator. And apply its output to one of the two pins the crystal was connected.

More fancy solutions possible.

If you want to make it worse, use an ooamp.
 

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
My solution was to replace all my clocks with MSF/DCF77 time signal type - including my wristwatch.

Probably the least pulling on the existing oscillator, is to build a second oscillator around a CMOS gate - increase the feedback resistor until it just fails to oscillate. The existing oscillator might excite it into action if you just place it near. The existing crystal is most likely strung between the output and input of an inverter inside the chip - so one pin will have a stronger signal than the other, find out which and give it a Cm or two of antenna wire. If that isn't enough coupling, you can use a "gimmic" capacitor - Twist two short lengths of insulated solid core wire together for a very small capacitor that you can tune.

These days - there may be LV CMOS logic that will just about run off the AA cell. Once you've calibrated the oscillator - you can leave everything in place so you aren't removing any capacitance effect.
-i REALLY don't need a precise clock, i just want to do it for "challenging" my self
-an "empaty" circuit??? not bad!! but maybe the 1Hz solution is easier with a 32 bit MCU's registero_O

If it is an analogue display you should be able to use a microphone to pick up the ticking and then measure that period accurately.
i prefer to hook some basic circuit to the coil outputs (simplier, cleaner... i think)

If you are going to modify the frequency anyway, does it really matter if loading the oscillator affects its frequency slightly?
i don't now how "slightly" is actually big... i need to monitor its frequency and see the adjustment "realtime" or (as previously said) the adjustment could be random once i remove the "probe"
am i getting something wrong?o_O

Depending on how far you want to go.

The easiest is to get a better clock.

The 2nd easiest is to use a better crystal.

The 3rd easiest is to use a temperature controlled oscillator. And apply its output to one of the two pins the crystal was connected.

More fancy solutions possible.

If you want to make it worse, use an ooamp.
i really don't care about this clock:D... but i've already replaced the crystal obtaining the same behaviour
 

ian field

Joined Oct 27, 2012
6,536
If you are going to modify the frequency anyway, does it really matter if loading the oscillator affects its frequency slightly?
Hand capacitance alone can throw the frequency off just as you hit perfect calibration then take your hand away.
 

MrChips

Joined Oct 2, 2009
30,720
If you already are using a 32768Hz xtal, then buffer the output and measure the signal with an accurate frequency counter.
Leave the buffer in place.
Usually two load capacitors on the xtal oscillator are required, one being a trim capacitor for adjustment.
 

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
If you already are using a 32768Hz xtal, then buffer the output and measure the signal with an accurate frequency counter.
Leave the buffer in place.
Usually two load capacitors on the xtal oscillator are required, one being a trim capacitor for adjustment.
this is also good, assuming that not powering the buffer anymore doesn't affect it's caracteristics (but it'll likely works)
 

Thread Starter

BulbChangeExpert

Joined Mar 26, 2016
54
anyways... thanks, it works
-these clocks have an H bridge output that has positive pulses on even seconds and negative pulses on odd seconds (i already knew it but i forgot... so 0.5Hz)
-the project was simpler than i tought
-the arduino due master clock ferquency and 32 bit register size happened to be "perfectly ENOUGH"
-there's a pretty good consistency between the "races"... so the measurement was good, i was not sure about this
here's the program i made.... and photos
Untitled.jpg DSCF5483.JPG DSCF5482.JPG
 

Attachments

Top