Cheapo clock adjustment

[andrew b]

I bought a cheapo quartz clock, you know the ones from the dime store? about 40mm square and use 1 "AA" cell. The one I got gains 2-3 seconds an hour, not a lot but after a week!! :mad:

I was wondering how to adjust it, at the moment I have tried twisting a couple of inches of fine insulated wires together and soldering them across the Xtal to form a small load capacitor. Can anyone tell me if I am wasting my time by trying the load the Xtal frequency down by 50-60HZ?. I believe the Xtal frequecy is 32768Hz

[rich smith]

You might need more than a few turns of wire ("gimmick"). I had to add several pf to get one in line. Most seem to be dead on.

[Almaz]

I don't have a good solution for you but If I were you I would just buy another clock from a different store. On the other hand you need oscilloscope.

[F-111 John]

Most quartz clocks use a trimmer capacitor somewhere near the xtal to adjust the frequency. It should look something like:

http://www.trimmercapacitor.com/images/5mm-trimmer-capacitor_s.gif

If yours is missing, that would explain why it gains so much.

[andrew b]

Thanks for the contributions. F-111 John, a trimmer capacitor isn't normaly fitted to this "quality" of clock. I thought severy pf or tens of pf, but low pf capacitors are in short supply these days, thats why I tried the twisted wire.. I will keep my eyes open for any dumped junk radio gear that has very small disc ceramics inside. I will leave the clock running for 24 hours and see how much it gains now.

[jeffs555]

Don't know how fine of wire you used, but that might be your problem. Capacitance is proportional to common surface area and inversely proportional to separation. Using thicker wire might get you enough capacitance.

If you just wanted to do it for the fun of it, you could try building a compression trimmer or a tubular trimmer. A compression trimmer is just two or more flat spring metal plates with a screw thru them to adjust the separation(there are generally thin insulators to prevent the plates from touching). Two half inch square plates separated by a piece of scotch tape might get you 20pf. Here is a calculator for parallel plate capacitors. http://www.ajdesigner.com/phpcapacitor/parallel_plate_capacitor_equation.php

A tubular trimmer is just a metal tube with a screw going through the middle of it. The screw is insulated from the tube and one connection goes to the tube and one to the screw. You adjust the capacitance by turning the screw in and out of the tube. Here is a calculator for cylindrical capacitors. http://www.ajdesigner.com/phpcapacitor/cylindrical_capacitor_equation.php

I did a google search to see if there was a calculator or estimator for gimmick capacitors(ie your twisted wire), but didn't find anything.

[Brandano]

you could make a capacitor out of bi-adhesive tape and tin foil and literally trim away the excess...

[Ron W3FJW]

Use copper foil (get at most any craft shop) and solder your leads on first so the tape isn't melted while soldering.

[BushmanLA]

Anyone here know why they use 32768Hz the clock freq. It kinda nifty once you realize it.

[F-111 John]

Anyone here know why they use 32768Hz the clock freq. It kinda nifty once you realize it.
Because it's easy to divide down to 1 Hz using cascaded flip-flops?

[BushmanLA]

Because it's easy to divide down to 1 Hz using cascaded flip-flops?

Pretty much.

Just like dividing by 10 in decimal is like shifting a number to the right, dividing by 2 in binary is like shifting a number to the right.

So while numbers like 10000 and 100 are easy to work with in decimal, numbers like 32678 and 256 are easy to work with in binary.

Sorry for getting all nerdy on you guys.

[Richard Ingram]

Can anyone tell me if I am wasting my time by trying the load the Xtal frequency down by 50-60HZ?.

In my opinion you will not get it like you want it to be. I bought a clock recently at WalMart for $4. I am using the electronics for a 1 hz timebase for digital clocks. The ones I've used are very accurate for the most part.

Here's some info:
http://www.josepino.com/circuits/?one_second_timebase.jpc

[andrew b]

The twisted wire seems to have slowed it down a bit, not a lot, just a bit. There isn't any more space inside for longer length of wire so I am keeping an eye open while passing dumpsters for any old TV junk that I can "recycle" :D .

I might try the copper foil and tape but there isn't much space in there, like 8 mm square or so
Only problem would be unpredictable changes in the capacitance with the assembly being open to the atmosphere and this would affect the accuracy of this >superb and exquisitely crafted< timepiece. :)

I have a radio controlled LCD clock that I can use as a "master calibration" clock.

The clock has a Xtal, a "black epoxy blob" chip and a solenoid, so theres nothing really to go wrong with it. If a divider had packed up then it would run at least twice as fast, so I figure the Xtal is buzzing away a bit too fast.

32768 = 2^15

65536 = 2^16 or FF in hex

Still it won't cost me anything to play with it and as #1 founder member of the official tightwads club then I hate to throw 50cents away. :p


UHoh, I went looking at Maplins for trimmers, I found a 22pF costing 62p :eek: or a new Xtal at 47p :eek: , what should I do? will they accept a return Xtal if it isn't dead on 32768Hz

Being a tightwad has serious dilemas

[andrew b]

I don't have a good solution for you but If I were you I would just buy another clock from a different store. On the other hand you need oscilloscope.

I have an antique dual beam Hewlet Packard 100 Mhz beast, but as I want to swing the Xtal frquency down by 0.08% or just 26Hz. It seems the 32768 Hz Xtal is actually 32794 Hz. How could I use my scope to measure such a change? Maybe I should explore the other pads on the PCB with the scope and see if there are other divider signals present, perhaps one with an output of 1024 Hz.

[F-111 John]

I have an antique dual beam Hewlet Packard 100 Mhz beast, but as I want to swing the Xtal frquency down by 0.08% or just 26Hz. It seems the 32768 Hz Xtal is actually 32794 Hz. How could I use my scope to measure such a change? Maybe I should explore the other pads on the PCB with the scope and see if there are other divider signals present, perhaps one with an output of 1024 Hz.

Andrew,

You might be able to see the difference in frequency by measuring the period of one cycle of the xtal frequency. However, the frequency would most likely be off just by the loading of the scope probe.

What you need is a frequency counter. Any other divider output on the board, such as your 1024Hz example, will also be off by the same percentage as the 1 Hz signal is, since they would both derive from the same original xtal frequency.

If this is indeed a mental exercise, I'd spend the extra pence on a trimmer cap and put it in parallel with the xtal. Chances are if you bought a new xtal, you'd just end up with very nearly the same error.

[rich smith]

John is right. A scope don't work. Not just loading but you need more accuracy. And more than a few digits on that freq counter too!

[Ron W3FJW]

Is it for sure a parallel resonant crystal? Or could it be series resonant?

[andrew b]

If this is indeed a mental exercise

It sure is, nothing but pure grey matter exercise. I am now in the quandry of "what range" of trimmer, a 10pF max, a 22pF max or a 60pF max, always remebering that there is a also a "min" capacitance. I think I should look up "Xtal pulling" using capacitors and buy the smaller trimmer AND a fixed fixed capacitor in parallel.


I am waiting for some chap on here to suggest using a PIC chip conected to a temperature sensor, a pressure sensor and a battery voltage monitor outputting to a varicap diode across the Xtal to fully compensate for all external calibration variables.


This ISN'T meant as a slight against the excellent technicians who post in this section of the forum, it's meant as " I know that some of you are fully capable" of implementing error correction using a microCPU and homewritten code.

My hat is off to you good gentleman because in the 3+ years as a member here I have seen "wonders from beyond the stars" produced here.

[rich smith]

I am waiting for some chap on here to suggest using a PIC chip conected to a temperature sensor, a pressure sensor and a battery voltage monitor outputting to a varicap diode across the Xtal to fully compensate for all external calibration variables.


LOL! That does sound like something I would do!

PS Note that these type circuits are on the bare edge of oscillating. I've actually seen cases where moisture from blowing on them will "stop the clock". I kid you not.

[gigelus2k3]

virtually all 32768Hz CMOS oscillators use a Pierce topology, consisting in a simple amplifier (e.g. an inverter with some kind of DC feedback - a large resistor between the input and output is enough) and the crystal in a PI-type feedback network, therefore it's likely to be parallel resonance.

[andrew b]

Another little toy I have been "abducted" to repair is my nephews "bait boat" Basically it's a model boat with an brushed ESC (working), a bait drop hatch solenoid (working) and a standard 1.5mS servo for direction .

The root of the problem was that the servo which uses an L4940V5 regulator (rated 1,5 amps on a good day) to drop it's main supply (a 12 volt 7.5 A/Hr dryfit lead acid) to 5 volts for the servo's power supply and 3 *big* LED's (so you can see the boat in low light)

I got the thing apart and the regulator has "gone home" with 0 Volts on the O/P of the regulator, I chose a L78S05CV 2 Amp (non low dropout) as a replacement., do you think this was a good choice?

[rich smith]

Sounds good to me. 12v to 5v don't need LDO. Only thing better might be switcher but much more expensive. And remember, a little extra piece of aluminum goes a long way with those linears.

Another little toy I have been "abducted" to repair is my nephews "bait boat" Basically it's a model boat with an brushed ESC (working), a bait drop hatch solenoid (working) and a standard 1.5mS servo for direction .

The root of the problem was that the servo which uses an L4940V5 regulator (rated 1,5 amps on a good day) to drop it's main supply (a 12 volt 7.5 A/Hr dryfit lead acid) to 5 volts for the servo's power supply and 3 *big* LED's (so you can see the boat in low light)

I got the thing apart and the regulator has "gone home" with 0 Volts on the O/P of the regulator, I chose a L78S05CV 2 Amp (non low dropout) as a replacement., do you think this was a good choice?

[vpatron]

Hi andrew_b,

Re: 32 kHz, I can pull off some xtals from some old boards for you. I think they are 50 ppm crystals. A good clock shouldn't need trim caps; just a good crystal to begin with. 50 ppm would get you about 4 seconds per day accuracy. They are really small SMT parts so it's best to have a stereo microscope and really fine soldering tips.

When we needed super accurate clock at my old work, we had a 10 MHz VCTCXO oscillator phase-locked to GPS 1 Hz to get very accurate reference. If you need accurate 1 Hz, can you use GPS?

Re: Regulator, if you're cheap, can't you just use one LM7805 pulled from some junk or even two in parallel?

-Vince

[Alan Hopper]

Hi andrew_b,

Pic and varicap, complete overkill, when all you have to do is put the clock in the correct orbit and let relativity sort it out :)

ps don't forget you can make small caps by putting big ones in series.
Alan.

[andrew b]

Sounds good to me. 12v to 5v don't need LDO. Only thing better might be switcher but much more expensive. And remember, a little extra piece of aluminum goes a long way with those linears.

Yes, it has a 2" X 2" heatsink all to itself, I think the original died due to being run too close to it's tolerances (and in the hands of a 17 year old), although when I took the casing apart I found the GND wire adrift from the pin on the regulator, it might have been me that pulled it off while taking the casing off or it came off "naturally" with help of the nephew :rolleyes: Either way when I resoldered it I still had almost 0 volts on the O/P pin, so it's a fair safe bet that it's goosed.
I will use a 555 rig up to test the servo to ensure that I don't smoke another regulator up.

[Jsutherland8]

Yes, it has a 2" X 2" heatsink all to itself, I think the original died due to being run too close to it's tolerances (and in the hands of a 17 year old), although when I took the casing apart I found the GND wire adrift from the pin on the regulator, it might have been me that pulled it off while taking the casing off or it came off "naturally" with help of the nephew :rolleyes: Either way when I resoldered it I still had almost 0 volts on the O/P pin, so it's a fair safe bet that it's goosed.
I will use a 555 rig up to test the servo to ensure that I don't smoke another regulator up.
Do You have the circuit drawing for this 555 rig, and can you post it ?

[andrew b]

Do You have the circuit drawing for this 555 rig, and can you post it ?

I have been using this one HERE (http://www.drivecalc.de/circuits/Servotester.html) . It's a standard design and cost pennies(cents) to build. Turn the knob and the servo disc follows suit.

[rich smith]

I have been using this one HERE (http://www.drivecalc.de/circuits/Servotester.html) . It's a standard design and cost pennies(cents) to build. Turn the knob and the servo disc follows suit.

I built that one some time back. Worked like a charm. Ultimately went to micro due to 1/10th as many parts and ability to do fancy patterns. Programming is a pain if not setup for it though.

[andrew b]

Hi Rich, yes, theres nothing to go wrong with it.

Might be fun to build a REALLY tiny one using SMT parts, see if it's possible to go sub 1/2" square inch for the board and components.

To get back to the boat. I put the regulartor in and got the 5 volts back on the supply.

As suspected the servo was goosed when tested, just jumped about and twitched, so slipped in a new futaba and all seems to funtion on the receiver, lights flash, servo, ESC and baitdrop work OK, just radio has a short range.

As the ariel was bust off the TX, I made one up to *about* the right length, but I think the final PA tranny has died or gone leaky due to operating without a proper ariel.

RX is simple enough inside, no RF stage, ANT straight onto mixer, a dual conversion using an MC3361 IF/demod chip (SO16 size), out put from this is amplified and then to an 8 pin PIC for the servo, ESC and bait drop solenoid. Motor ESC is a single IRF/BUZ type powerfet.

Clock repair has been put on hold while I fix this boat, still I can claim 3 seconds overtime every hour :D

By the way, how easy is it to get SMT sized transistors if I do more small work?. SMT project building looks fun.