Inaccurate Clock, But Why?

  2018 February 2     Tech,     Electronics


Sometimes you just want a simple electric clock.  No satellite uplinks, no Wi-Fi, none of that.

A clock this simple should last forever, shouldn't it?  I mean, there's not much to it.

Or... maybe there is.

Get a nice cup of tea and sit down for a read.  We're going to tear apart some electronics.  (Caution: some of this might be dangerous.  Disclaimer.)

In This Article

Something's Off About The Time

Possible Reason

Fix vs. Throw Away

Components To Look For

Clock Chip

About That Oscilloscope


Something's Off About The Time

You loved that LED clock.  1980's-vintage Ken-Tech.  No frills, not even a radio.  Just these big, simple, red LED numbers.

Remember the first calculators like that?  They did math!

For thirty years, your LED clock was keeping pretty good time. 

But then one day, you notice it's gaining.  And gaining.  You set the clock to 12 Noon, just to see what happens.  In ten minutes it's already 3 AM!


Possible Reason

Cheap AC clocks use line-frequency to get the time signal.  60Hz line frequency is actually very stable over the long term.  In fact, a cheap AC digital clock can be more accurate than many other timepieces... but only because there's a lot of care that goes into keeping the line frequency at 60 Hz.

For a cheap item like this, a Rigol or Siglent might seem like overkill.  But think of all the electronics you could test with it.  Every time I tear apart a radio or something, I keep wishing I had a digital oscilloscope. 

A clock circuit can become inaccurate for different reasons.  If it has a voltage-controlled oscillator, look for a capacitor across two of the pins on that.  Standard "Real-Time Clock" circuits are 32.768 KHz;  crystal-based circuits may use a capacitor to adjust or "pull" the crystal frequency. 

Bad cap = wrong frequency. 

The better digital circuits can handle frequency variation, yet they still keep accurate time.  But what about a cheap LED clock from 1985?  Maybe not.

I know for a fact that on certain clock circuits, a bad cap can cause inaccurate time.  Sometimes really inaccurate. 

Here's how I learned this.  I was repairing a vintage clock radio once, and I didn't have the right capacitor for a certain one, so I used what was available.  The capacitance was different by a factor of 10.

When I tested the clock, it ran 10 times too fast. 


Fix vs. Throw Away

Hong Kong, 1985:  some guy (or lady) had to sit there and solder like ninety-seven different connections in a hot factory.  Or maybe ninety-seven people each soldered one thing.  But the good news is, at least it uses traditional components and regular solder joints.  So it's fixable, in theory. 

Taking this thing apart now:

Clock Circuitry

There must be at least thirty or thirty-five solder joints that hold the circuit board in place.  You can't get to the capacitors without de-soldering all these.  (Well, there might be a way...)

At least if this repair doesn't work out, it's cheap and easy to buy a new one.  This one even has indoor temperature. 

Also, a reminder about safety:  those little metal bumps on the back of a circuit board-- or anything in there that's metal-- could be energized with enough juice to be lethal.  Even when it's unplugged.  330 uF at 16 volts probably won't kill you, but at first you should assume there's some high-power capacitor you overlooked. 

This is another reason to get a good multimeter and learn to use it, before doing anything else. 

I'll just say this reminder for whomever it may help: "Don't work on electronics if you don't absolutely, positively know how to avoid electrocution."  I'm assuming you have the basic skills, and if you don't.... a nice new LED clock can be had right here


Components To Look For

Before I took this apart, I was thinking in terms of a crystal oscillator.  Perhaps with a "pulling" capacitor in parallel. 

There are a couple ways, at least, to build a digital clock.  Crystal oscillator is one option.  Another way is to sync with line frequency, using a chip to convert that to a meaningful time signal. 

When a clock chip has a pin that requires a 60 Hz input signal... that would be a pretty good indicator that it's getting its time base from the mains frequency.  Since this is a cheap LED clock, that makes sense.

So, let's see what there actually is.


Clock Chip

Instead of desoldering all those pins, here's what I did.  I broke off two plastic tabs, removed the whole board assembly, then folded the one board away from the other. 

The way these things are built, you almost can't "not" break something when taking them apart.  Even in their day they were cheap mass-produced items.  It's not that they were designed to be unserviceable;  corporations hadn't gone that way yet, for the most part.  But at the same time, they didn't go out of their way to make stuff like this easily-servicable, either. 

That said, the plastic looks like polystyrene, so you could probably bond it with cyanoacrylate. 

Clock Circuitry 2

Now we see it's a basic clock-chip circuit.  Texas Instruments TMS3450NL, made in 1984.  It's equivalent to Sanyo LM8560. 

The board also has two capacitors, three diodes, a Zener, and ten resistors.  Fairly simple.

Line-Frequency Input Pin

So there we have Pin 25, which takes the stepped-down voltage at 60 Hz;  (There's a way to jumper Pin 26 so it looks for 50 Hz, but I don't know how.)  So, what would cause this clock to gain hours per day? 

That green, 0.01 uF tantalum capacitor is probably OK.  (I have to stop calling these things "tantalum"; they're not.)  They don't usually go bad.  Electrolytics do, though. (And actually, so do tantalum caps... quite a lot.)

Look at that one resistor.  Is that a 680 ohm?  It depends on whether that last band was originally supposed to be brown... or black, in which case it would be 68 ohm.  I see what looks like possible heat damage. 

680 or 68 Ohm Resistors

It is very common for electronics to take a surge through a low-value resistor.  You don't always see much of a difference in appearance, but expansion cracks and discoloration can sometimes help locate which one got fried.

But I still think it's that electrolytic cap.  After 30+ years, it's probably way out of spec.  Without a schematic, I would guess it's some type of filter cap to decrease ripple. 

There are not many components here.  On a board like this, I could probably desolder and replace them in about the same time it takes to desolder and test them.  Here is the order in which I would start replacing stuff:

1.  The 330 microfarad electrolytic capacitor;  it's 34 years old.
2.  That discolored 680-ohm resistor, or whatever the ohms are supposed to be (desolder, test).
3.  The TMS3450NL / LM8560 clock chip.  Un-fun, because many solder joints. (Wonder if an IC socket would add too much height?)
4.  One of the 1N4001 diodes;  it looks kind of crumbly at one end.
5.  The other two 1N4001 diodes (might as well;  they're cheap).
6.  Oh yeah, should have checked this one FIRST... look for loose wires or cold solder joints......

If the clock chip is bad, three or four other components would not be that much extra work.  You've already got to desolder 28 pins anyway.

I still think it's that capacitor, and we'll find out.  But if it is, and it's been letting full-amplitude square wave or sine wave where it shouldn't be, then the clock chip could also be damaged.  Ripple is not good for electronic components.


About That Oscilloscope

Yes, a good oscilloscope would be overkill for this.  I can fix this cheap clock with a 25-cent part.  (Maybe it's 85 cents.)  But if you wanted to be able to troubleshoot almost everything you might encounter.... a digital oscilloscope like that Rigol or that Siglent would be nice to have.  It should let you detect slow pulses, such as a once-a-minute output to an LED tile.  I think perhaps you'd want to use Peak Detect, maybe with a wide timebase.  Some of the fancier multimeters can also detect transient-voltage peaks, I believe.  In other words they are not part of the signal; they're incidental.

Just so I don't forget, I also noticed this little device, an o-scope which I think is good through at least 100 or 200 KHz.  That would be good for some things.

A Real-Time Clock circuit runs at 32.768 KHz before the conversion step, but I don't think this clock chip is going to emit any detectable signals at 32.768.  I don't think that's how this one works.  You have a stepped-down 60Hz AC voltage going to one of the input pins;  then the chip does some logic functions on it;  then it sends the Hours and Minutes signals to the LED display.

So, we might not find a 1 Hz signal in this circuit, either.  I think there would have to be 1/60th Hz pulses at some point.  I don't know, but that low-cost oscilloscope sure looks like fun. 



OK, guys, it was the part that I thought it would be.  I didn't have to replace the other stuff. 

So far it's keeping accurate time now. 

I still don't know how they fine-tuned the clock frequency, because I didn't see any trim capacitors or potentiometers in there.  Maybe the clock chip was that good that it's unaffected by small variations.  One thing's for sure, though:  a bad cap can really throw it off. 

This was just a quick electronics teardown and repair.  I learned something about electric LED clocks, and hopefully you did, too. 

If you found this article helpful or entertaining, please help me keep this website on-line by purchasing any of your stuff through these links.  Gear, clothing, anything.  It's the only way I can continue bringing you helpful articles like this one.  Thanks!!

Another Article You Might Like:  Fixing The Solar #1002 Battery Charger... Sort Of

Contact me:

3 p o.t o . 1 2 0 s t u d i o.. c o m

This won't directly copy and paste.  Please manually type it into your mail program.
No spaces between letters.

Home Page

Site Map

What's New!


Copyright 2017

Back to Top of Page