The 41 second clock failure syndrome

[anorlunda]

Instead of removing the second hand maybe just cutting it down to a very short nub will fix the problem and yet still allow you to see if it's moving or not.

Didn't the experiment of hanging it upside down prove that it wasn't just a question of torque?

 

[Dullard]

I have a cheap clock in my office that does exactly this when the batteries are shot. I always attributed it (strictly) to torque. The 'upside-down' result is unexpected. I'm wondering if that is an artifact of the (previous) excessive wear on the gear teeth at the 'stall' position. In other words: I would predict that a clock which has never operated in 'stall' for an extended time would not show a difference in performance when upside-down.

 

[Baluncore]

The 'upside-down' result is unexpected. I'm wondering if that is an artifact of the (previous) excessive wear on the gear teeth at the 'stall' position.

It could be wear to plastic on the uphill run, or redistribution of the lubricant.
Rotating the second hand, relative to the gear that directly drives it, was not tested.

 

[anorlunda]

I'm wondering if that is an artifact of the (previous) excessive wear on the gear teeth at the 'stall' position.

Did you read post #1? I started with an old clock stuck at 41s. I bought a new clock. It stuck at 41 s the first day.

 

[JT Smith]

Didn't the experiment of hanging it upside down prove that it wasn't just a question of torque?

Yes, but it could still be partly a question of torque. What is the explanation for why it doesn't stick when upside down?

If it's [edit: not] got to do with the hand itself then removing it or balancing it (or shortening it) won't fix it. But I'm not sure anybody really knows why it's sticking.

Could you try removing the second hand as an experiment?

 

[hutchphd]

If you remove the second hand, it will not tell you anything.

1. It won't stop because torque from the final drive is sixtyfold more important.
2. The second hand has a shaft attached...if removed nothing visible turns. Only an internal gear (maybe can see through the back?)

The battery is by far the heaviest piece in the motor and it, too, sits very asymmetrically. Changes in orientation will cause distribution of its weight on chassis of gears to shift. The clock ran for 4 years without stopping.

On a slightly less technical plane, and as a person of some time on-planet, I offer a more worrisome supernatural explanation from a favored musician:

 

[Dullard]

Did you read post #1? I started with an old clock stuck at 41s. I bought a new clock. It stuck at 41 s the first day.

I did. The '40 - 45 second' failure mode is common when available torque is 'just' insufficient. I don't believe that it's necessary (or correct) to assume that the failure of the 2 clocks have anything in common (except gravity). You apparently do? I do find the 'upside-down' behavior interesting.

I'll assume that you didn't mean to be as offensive as your post reads.

 

[sophiecentaur]

The second hand is very likely to be a friction fit on the shaft. Pull it off carefully and replace it 180 degrees different.
OR, take off the hand an put a tiny blob of paint on the shaft.

Do you still get the 41 minute effect?

 

[hutchphd]

The second hand is very likely to be a friction fit on the shaft

In my experience the second hand is rigidly attached to a ~1mmx8mm shaft which is friction fit through front hole into gear. Easy enough to pull straight out and reseat arbitrarily if you so desire. As mentioned (#26) I spent some time playing with these. With good bearings (from small Syma helicopter) these were just capable of rotation torque I needed when well balanced.

 

[JT Smith]

If you remove the second hand, it will not tell you anything.

1. It won't stop because torque from the final drive is sixtyfold more important.

You know for certain that removing it prevents the clock from stopping? That's what I wanted to know.

 

[Baluncore]

Another test. Wait for the clock to stop at about 41 sec.
Then subject the clock to maximum torque and identify the stopping point by:
Start it again, as you place the clock on edge in front of you on a table.

Roll the clock slowly towards the left so seconds hand always points left.
What time does it stop? 41 sec = gear wear. Anytime = hand balance.

Next rotate seconds hand by 30 seconds and repeat the roll with seconds hand pointing left.
What time does it stop ? 11 sec = gear wear. Anytime = hand balance.

 

[anorlunda]

I'll assume that you didn't mean to be as offensive as your post reads.

A apologize if it sounded offensive. I said that because I said the 2nd clock was new, and you wrote about excessive wear. Those two things would not be consistent.

Another test.

Why additional roll tests? We already saw that with a 180 degree roll, it doesn't stop at all.

 

[Baluncore]

Why additional roll tests? We already saw that with a 180 degree roll, it doesn't stop at all.

That is one test, an enecdote. I feel uneasy about it.
How long did the inverted test run?
Did you give it a fair chance to settle down and fail at maybe 11 or 41 sec?
You have branched out in the tests, but not closed the loop to verify a conclusion.
Things do not add up, so a firm conviction beyond reasonable doubt has not yet been made.
The hand position relative to the driven gear has not been changed. The shaft could be eccentric or bearings worn.

 

[anorlunda]

That is one test, an enecdote.

Let me see if I can do better. I no longer have the old clock, just the new one. I am not sure if I can locate the old weak battery. But I'll do some tests and post the raw data.

But I resist the idea worn parts. The thread was triggered because the new clock started getting stuck in the very first hour of use, and the symptoms seemed the same as the old clock with ~35k hours of use.

 

[Rive]

Many of these cheap clocks has movements like this, I think.
The 'axle' and 'bearing' of the stepper looks pretty lame and their prowess is probably susceptible to gravity.
The connecting gears may cause the asymmetry required for the no-problem-upside-down phenomenon.

That's all I can think of without having the thing in my hands.

 

[anorlunda]

That is one test, an enecdote. I feel uneasy about it.
How long did the inverted test run?
Did you give it a fair chance to settle down and fail at maybe 11 or 41 sec?

OK, I promised to run more tests. It took me a while to locate that weak battery. But I found it and did 20 tests. The raw data is below in the chronological order I measured them. So here's the experiment.

1. Put in the weak battery.
2. Hang the clock with a certain angle. I express that angle in terms of the hour pointing up. Angle 12 is the normal orientation for a clock, angle 3 is 90 degrees on its side, angle 6 is 180 degrees upside down ...
3. Measure how long for the second hand to get stuck. Some tests I terminated at 30 minutes. In those cases it might have continued indefinitely without sticking.

Trial	Angle	Time Until Stuck
1	12	30 seconds
2	3	> 30 minutes
3	6	> 30 minutes
4	9	1 second
5	12	100 seconds
6	9	1 second
7	11	100 seconds
8	10	40 seconds
9	8	30 seconds
10	7	30 seconds
11	6	7 seconds
12	6	1 second
13	5	20 seconds
14	4	2 seconds
15	3	40 seconds
16	2	45 seconds
17	1	45 seconds
18	12	1 second
19	9	45 seconds
20	11	45 seconds

Here is a plot, sorted by angle. I capped the maximum time to 600 seconds. Some tests were repeated, so there is more than one dot at the same angle.

Unfortunately, the data doesn't really support my theory that a flaw in the gears made it angle sensitive. There does seem to be slight angle sensitivity. However, I think the battery got weaker as I conducted the tests, and that swamped out other factors.

By the way, no matter the orientation angle of the clock. When the second hand did stick, it always did so in the 3rd quarter (180-270 degrees). That gives credence to the theory of increased torque to lift the weight of the second hand.

Note that this brand new clock started sticking on its very first day of service, so worn out parts should not be a factor.

 

[NTL2009]

Thanks for the update. Another test (if you are game), based on these results:

Add or subtract weight from the second hand so it is balanced, then try again. I'm actually surprised these hands (at least on large face clocks) are not balanced. Many styles of clock hands look like they were meant to be balanced (the short side of the hand is wide), and I think that is partially aesthetics, and partially historical. I'd bet that old large clocks had to be balanced to operate, but...

OK, I did a google image search, and it is not obvious that old, large clock hands were balanced. Some seem to be, but it's not universal, and doesn't appear to fully balance them. But that balance could also be inside the clock, behind the face.

 

[anorlunda]

Another test (if you are game), based on these results:

Add or subtract weight from the second hand so it is balanced, then try again.

I did one better. I removed the sweep second hand. Now, the clock runs indefinitely with the weak battery.

 

[Baluncore]

We have considered static balance, but not impedance and time matching through the seconds drive train. The seconds hand moves in steps. By adjusting the time constant of the seconds-hand mass system, the reflection from the hand impedance may oppose or support the next step in one second.

I would consider making the seconds hand into a balanced rotary pendulum with a period of one second. That might require the addition of a spring into the drive train to dominate the restoring force.

As the mechanism changes temperature the gear backlash may change which might change the hysteresis and reflection time.

 

[anorlunda]

I would consider making the seconds hand into a balanced rotary pendulum with a period of one second. That might require the addition of a spring into the drive train to dominate the restoring force.

I think I'm going to apply for a NSF grant to finance this research. Enough to provide for 3 grad students and a workshop. We can call it Studio 41.

 

[Baluncore]

I think I'm going to apply for a NSF grant to finance this research.

Well, I understand that the wall clock is immersed in a flow of time, in the same way that a water wheel is immersed in a river. If your research group, Studio 41, can increase the efficiency of the clock, to the point that it slows down at a lower rate than the Earth's rotation slows due to tidal drag, then it should be possible to charge the clock battery with the difference. And that does not even require PM.

 

[jedishrfu]

Perhaps this is where Douglas Adams got his answer to everything is 42 when clock stopped at 41.

https://en.wikipedia.org/wiki/42_(number)

versus 41

https://en.wikipedia.org/wiki/41_(number)

 

[Not anonymous]

I can't believe it! I thought it was only my old wall clock that was jinxed by 40-41 seconds. Now I know it is was not alone :)

 

[Baluncore]

I can't believe it! I thought it was only my old wall clock that was jinxed by 40-41 seconds. Now I know it is was not alone :)

My wall clock sped up over the last week, then two days ago, it stopped at 40 seconds. I replaced the battery, and it is now keeping time again. I think clocks are organised and stop in sympathy.

 

[sophiecentaur]

The evidence is about clocks that have run for a long time. It won't only be to do with the angle of the second hand at the time it actually stops, I'm sure. Every rev of the second hand produces a maximum load at round about the same place on the dial - say 42s. As the movement passes this point the wear of the bearings and the gears is likely to be be highest here. (the whole mechanism is built as cheap as poss) So rotating the clock by 180 degrees will change the points on all the gears where the wear is now greatest. So, if the clock then keeps going on an almost flat battery, it would not be too surprising. With a new battery, it could be that it will then carry on until it now stops at (say) 12s, when a new set of worst wear coincides with failing to drag the second hand beyond the same value of around 225degrees relative to vertical.

So the answer to the universe and everything could, in fact be 225.

 

[Baluncore]

I can accept that the answer may approach 42 asymptotically, and I can understand why in a digital system, it may fall short given insufficient observation time, but where is zero?

 

[sophiecentaur]

https://en.wikipedia.org/wiki/Fusee_(horology) can accept that the answer may approach 42 asymptotically, and I can understand why in a digital system, it may fall short given insufficient observation time, but where is zero?

It's not really relevant that the clock uses a digital system to provide pulses to to drive a motor. The same thing could happen with a mechanical clock with a spring. I would pursue this further but I don't have the time.
Or is this whole thread just a wind up?

 

[jedishrfu]

According to some sources, the meaning of 42 is rooted in the ASCII character set where 42 represents the code point for ‘*’ the asterisk which is used to reference every file or everything.

 

[anorlunda]

In any event, it's fun to separate the sf literate from illiterate using 42.

 

[gmax137]

In any event, it's fun to separate the sf literate from illiterate using 42.

Or, the musically literate:

On a slightly less technical plane, and as a person of some time on-planet, I offer a more worrisome supernatural explanation from a favored musician:

I was fortunate to see Doc Watson play a few times, just amazing. I'm not a guitar player but even I can hear how clear each note is.

The only other bit I have to add is this, normally seen in barber shops.

 

[DaveC426913]

In any event, it's fun to separate the sf literate from illiterate using 42.

I worked it out when I was a young lad:

6 x 9 does equal 42 ... In Base 13.

 

[hutchphd]

I was fortunate to see Doc Watson play a few times, just amazing.

I saw him play in the early 70's with his son Merle (at Antioch College in Ohio). What an extraordinary man.

 

[sophiecentaur]

42 represents the code point for ‘*’

If he'd used 2A then everyone would have got it!

 

[hutchphd]

Literati devotees of Kurt Vonnegut will recognize the "*" as Fundamental

Spoiler: Warning Literary Content

 

[DrGreg]

6 x 9 does equal 42 ... In Base 13.

For what it's worth, 42 days is exactly 10! seconds. (Ten factorial.)