Why would a Lithium AA battery make a clock run slow

[tjloeb]

I have an early SEIKO battery powered quartz clock. It has worked perfectly for 26 years requiring a single AA battery replacement once every 2 years or so.



I decided to see what kind of battery life I would get with a new Lithium AA cell.



The day after changing the battery the clock had lost 5 minutes. Everyday it lost another 5 minutes.



I change to another brand of battery, still loosing time.



Went back to standard alkaline battery and the time loss stopped!



Anyone have an idea why this was happening?

 

[mediabeing]

Your Seiko runs more slowly with the lithium rechargeable because he rechargeable does not hold a voltage of 1.5, but 1.2 volts.
In low drain rate situations like yours, it makes a difference.

 

[Timothybil]

Although the OP doesn't say it explicitly, I am assuming he was referring to an L91 Lithium primary, not a rechargeable 10400. As such, I believe the voltage would actually be higher than an alkaline AA. It would help if we knew if the clock was an electro-mechanical (ticks every second or so and the second hand jumps instead of progressing smoothly) or an LCD/LED. An electro-mechanical works on some sort of timing circuit driving a stepper mechanism, and from experience appears to be very voltage insensitive until the cell is drained enough it can no longer drive the stepping mechanism. On the other hand, an LCD/LED works with a timer circuit driving a display driver. Given the design criteria, I can't see such a circuit being very voltage sensitive either, given that the clock basically drains the cell to extinction over a period of a year or more. Others have stated that in such a low current drain application alkalines will usually outperform lithium primaries, there might be something going on in that area. Personally, unless this clock is in an area where temps regularly approach freezing or lower, I can't see any advantage L91s would have over alkalines, and the increased cost would be a negative factor.

Besides, 5 minutes in 24 hours is only an error rate of less than 0.4%, or an accuracy of over 99.6%. Makes it easy to see the old engineer aphorism of 'make it work, then see how many nines you can tack on!'

 

[jtr1962]

It's actually pretty well known that some quartz oscillator circuits are sensitive to input voltage but I've never heard of one this bad before. 0.4% is HUGE here. Typically, the 32.768 kHz crystals used in clocks and watches have a maximum tolerance of at worst 30 ppm (0.003%), or ~2.6 seconds per day. I'm going to guess once the battery voltages get much above the ~1.58 volts of a new alkaline battery some nonlinearities in the oscillator circuit cause the crystal to oscillate far below its natural resonant frequency. For what it's worth, I've carefully checked more than my share of time pieces. Besides the variations with temperature, there is often some variation with battery voltage. However, this is generally very small. A clock might be 1 second per day faster or slower with a fresh battery, as opposed to a nearly depleted one. 5 minutes a day is totally off the charts. That reminds me of the days when we had mechanical clocks and watches. I recall if I could get a mechanical watch to keep time to a minute a day it was doing pretty good. Nowadays the better watches keep time to a second per month.

 

[germanium]

I found that initially anyway that when a battery gets low the clock actually speeds up before it slows down. This is due to an effect similar to what you see with a ping pong ball. When you first drop a ping pong ball it bounces pretty high & takes a while to come down then as it looses energy the ball bounces less high & because it bounces less high it returns faster & faster to the surface. The higher voltage of the lithium L91 battery causes the crystal to vibrate with more vigor & consequently ever so slightly slower much like the high bouncing ping pong ball as the stronger the vibration the longer it takes to return to its at rest position