Barr & Poole 1927 Poole Electromagnetic Clock - Can it be regulated to less than 1 minute deviation per month?

[Uhralt]

I have recently acquired and serviced a 1927 Poole Executive electromagnetic clock. The vintage advertising for this clock claimed that these clocks would be precise to 1 minute a month. So, I'm trying to regulate my clock to achieve this. It means that the clock shouldn't gain or loose more than 2 seconds a day in average. Is this achievable? Have others been able to replicate this claim? The clock has a rather heavy bob and a short pendulum. It is regulated with a nut that has teeth so that it is easier to observe the amount of change made.

I've been trying for some weeks, and have made the following observations:
After adjusting for a very small amount (1 tooth or less), the clock may run within less than 2 seconds after 24 hours. After that it continuously slows down and after 4 days it may loose more than 5 seconds a day.

I think that may be due to the bob slowly completely settling into the new lower position

When I then move the bob up for half the amount that I had lowered it previously, the clock would run fast.

So I keep adjusting.....

The next I will try is to ignore what happens during the first two days and adjust only according to results that have been observed after the third, fourth, and fifth day.

Does this make sense? Has anybody successfully fine regulated this clock? Please share your experience.

Or was this just an unsubstantiated marketing claim?

Uhralt

 

[jmclaugh]

I can't help with your question but a cynic might say it is a bit like the claim on paint that is guaranteed for 5 years for whatever which doesn't of course mean in this case it can't.

 

[Uhralt]

I can't help with your question but a cynic might say it is a bit like the claim on paint that is guaranteed for 5 years for whatever which doesn't of course mean in this case it can't.

Yes, that's what I'm trying to find out. In one of the three articles of the NAWCC bulletins about Poole clocks the author stated that it should be possible to achieve the 1 minute in a month goal with "very careful adjustment" but no confirmation was given that this has actually been achieved.

Uhralt

 

[Willie X]

You will need to regulate at 1 month intervals to do your test. It would be pointless to do otherwise.

Well after a few seconds thought, it might just be pointless, period.

Willie X

 

[Tim Orr]

Good evening, Uhralt!

I am inclined to agree with Willie here. In fact, I might suggest an approach similar to what is sometimes done with chronometers: Just let the thing run for several months, checking at monthly intervals to see where timekeeping is. No adjustments. Just recordkeeping. Forget accuracy, and just go for precision. Then you can answer the question whether or not performance is consistent. If it doesn't remain stable from month to month, I don't think adjustment is going to "bring it in." It also would be good to keep a record of how many seconds it runs between weight actuations. If that isn't consistent, it may signal some irregularity that also could affect precision.

I have one of these (which has a switch problem right now, and won't stay running), and the other issue is that the pendulum is so short (and uncompensated metallic) that adjustment is going to be tricky to do under the best of circumstances. An old boss of mine used to warn me against "chasing," by which he meant constantly adjusting and almost invariably overshooting the adjustment each time. An engineer I once knew used to talk about putting a tiny screwdriver on a pot and then just "inhaling," not even trying to turn the screwdriver. Sometimes, just touching the screw moved it enough to change the setting.

Don't know if 2 seconds a month can be achieved, but it's almost a Heisenberg problem, wherein just fiddling with the thing may keep you from achieving the result you want.

Keep us posted, please. Once I get mine working again, I'd like to see what it can do.

Best regards!

Tim

 

[Uhralt]

Thanks Willie X and Tim, these are all good points. It is exactly what I thought, just touching that thing disturbs the pendulum enough that it will settle into a different position over time. I will try to get close to , say plus 5 seconds a day and let it run anticipating that the bob will settle in a deeper position over time. Just observing and documenting. What I haven't seen so far is what I was hoping to achieve, namely a couple of seconds fast on one day and accouple of seconds slow another. Right now, over a week it was either consistently slow or consistently fast. That's why I think once I found the right spot (probably by chance), it might work.

I agree with Tim that the time between impulses is also an important factor. It seems the longer the interval is, the faster the clock runs. Right now the clock runs consistently with 33 to 35 oscillations between impulses.

I had a contact issue in between, meaning that there were double impulses from time to time. This slowed down the clock considerably. I cleaned the contacts carefully with a pencil eraser first, then alcohol, followed by Deoxit (a contact cleaner/antioxidant with good reputation). This was about a month ago and I didn't notice any double impulses since then. I think if these occur it will throw the whole timing off.

Overall I think that the manufacturer's claim, though maybe possible, was very rarely achieved in practice, if at all. Who had the time and patience to try to regulate a new clock for several months and then have to start over by a contact issue or the batteries running out of power?

Uhralt

 

[Uhralt]

Don't know if 2 seconds a month can be achieved

The claim is no more than 1 minute a month, which would be 2 seconds a day (not counting fluctuation within a month).

Uhralt

 

[Andy Dervan]

Willie and Tim make outstanding points.

Even if the clock achieve the stated precision originally, it is almost 100 hours old and components have aged. that will hinder its performance.

I prefer weight driven timepieces and have a number running from period ones 1810 -1860 through modern reproductions. I am pleased if they keep 1-2 minutes week accuracy and amazingly most do it on regular basis - unaffected by seasonally changes.

Let is run and weekly record its performance. Do you ability to monitor the house temperature and relative humidity, I would be curious if those change impact its performance?

Andy Dervan

 

[Uhralt]

Willie and Tim make outstanding points.

Even if the clock achieve the stated precision originally, it is almost 100 hours old and components have aged. that will hinder its performance.

I prefer weight driven timepieces and have a number running from period ones 1810 -1860 through modern reproductions. I am pleased if they keep 1-2 minutes week accuracy and amazingly most do it on regular basis - unaffected by seasonally changes.

Let is run and weekly record its performance. Do you ability to monitor the house temperature and relative humidity, I would be curious if those change impact its performance?

Andy Dervan

Thank you Andy, another good point. I can monitor temperature and also humidity. I expect that temperature will have an impact but it will be minimized by the fact that air-conditioning will keep the temperature rather constant, maybe 72 F in Winter and 75 F in Summer. We will see if that has an impact. I don't expect humidity will have a significant impact because the clock is all metal (no wooden pendulum rod) and is sitting under a glass globe. Because the globe is not removed to wind the clock, the humidity inside shouldn't change much, at least in the short term.

I have a ST #2 regulator which keeps very good time, better than 1 minute a week. The clock has a wooden pendulum rod. When the humidity increases in summer, it will loose time. I have to lift the bob about 1/4 turn of the regulating screw to compensate the humidity effect. So I have a "Winter setting" and a "Summer setting". Under the conditions of my home I didn't observe a temperature effect.

I will do the weekly testing, without changing anything and report from time to time.

Uhralt

 

[Willie X]

Uhralt,
Please report back if you do the 'untouched' monthly test that Tim mentioned. This stuff is always interesting to me and I'm to lazy to do stuff like this on my on. Ha. Willie

 

[Tim Orr]

Good afternoon, all!
"
Of course, that was a mistake of mine to quote "2 seconds a month" instead of "2 minutes a month."

Unless I'm mistaken, the Poole and Barr clocks may actually be said to be "weight-driven," inasmuch as a "falling" (or "rocking") weight gives the, essentially, "free" pendulum a little kick from time to time, ideally, about every 30 seconds or so. And temperature, as I said, would definitely be a factor.

When I was working in analytical chemistry, I was taught to distinguish between "accuracy" and "precision" in this manner: Accuracy is a determination of how much a measurement differs from a known standard. Precision is a determination of how much successive measurements differ from each other. A clock that gains 2 minutes a month, every month, without fail, for month after month after month is clearly not accurate, but very precise. Because of that, it can be brought into accuracy.

A clock that gains or loses different amounts of time over time is neither precise nor accurate, and really can't be brought into accuracy, ever. If you have precision, you can get accuracy. But if you don't have precision, you'll never get accuracy.

And, the other thing I forgot to mention is: What would a clock owner in 1927 use as a standard to determine whether or not his Poole clock was keeping time to 2 minutes per month? No radio time signals that I know of then. I suppose you could use a good quality pocket watch as a "comparing watch" and check it against Western Union telegraph-synchronized time at your local train station or telegraph office. But really, who would do that?

Best regards!

Tim Orr

 

[Uhralt]

Uhralt,
Please report back if you do the 'untouched' monthly test that Tim mentioned. This stuff is always interesting to me and I'm to lazy to do stuff like this on my on. Ha. Willie

That's the plan. To be able to see fluctuations within the month I will also look at weekly results without touching anything. Will report in (roughly) monthly intervals.

Uhralt

Good afternoon, all!
"
Of course, that was a mistake of mine to quote "2 seconds a month" instead of "2 minutes a month."

Unless I'm mistaken, the Poole and Barr clocks may actually be said to be "weight-driven," inasmuch as a "falling" (or "rocking") weight gives the, essentially, "free" pendulum a little kick from time to time, ideally, about every 30 seconds or so. And temperature, as I said, would definitely be a factor.

When I was working in analytical chemistry, I was taught to distinguish between "accuracy" and "precision" in this manner: Accuracy is a determination of how much a measurement differs from a known standard. Precision is a determination of how much successive measurements differ from each other. A clock that gains 2 minutes a month, every month, without fail, for month after month after month is clearly not accurate, but very precise. Because of that, it can be brought into accuracy.

A clock that gains or loses different amounts of time over time is neither precise nor accurate, and really can't be brought into accuracy, ever. If you have precision, you can get accuracy. But if you don't have precision, you'll never get accuracy.

And, the other thing I forgot to mention is: What would a clock owner in 1927 use as a standard to determine whether or not his Poole clock was keeping time to 2 minutes per month? No radio time signals that I know of then. I suppose you could use a good quality pocket watch as a "comparing watch" and check it against Western Union telegraph-synchronized time at your local train station or telegraph office. But really, who would do that?

Best regards!

Tim Orr

I am a retired chemist myself so I know exactly what you are talking about. The fact that my clock, once it has settled for a few days after manipulation, seems to gain or loose about the same amount of seconds per day speaks for some precision, so I hope I can achieve some accuracy. Maybe not within the range of the 1 minute/month claim but maybe, maybe, close. We'll see.

I was thinking the same: How in the world would someone in 1927 be able to determine how accurate (that's the claim) the clock is? I figured that, if the clock was for example 5 minutes off after 6 month, that would be called within the claim. If it was off 7 minutes after 5 month, it was not within specs. These differences could probably be observed by a lay person.

Uhralt

 

[Phil G4SPZ]

Hi Uhralt,

Interesting and elegant clock! I wonder whether you've tried regulating in one direction only, i.e. from going fast towards going slow, and if you 'overshoot', start over again? I was advised to regulate a 400-day clock in this manner, and it worked very well, although with a conventional pendulum clock I find it more reliable to regulate from slow towards fast, as moving the bob a consistent and tiny amount upwards is more easily achievable than getting the bob to consistently come down. It seems that if you try to reverse the direction in which you're regulating, the slop or hysteresis inherent in many pendulum mechanisms can cause an overshoot in the opposite direction. As Willie says, as you get close to the correct rate, try regulating no more frequently than monthly.

Tim, I don't know about the United States, but in Britain the BBC broadcast time signal 'pips' from 1922 onwards, and I imagine many US broadcasters were doing likewise around 1927.

Phil

 

[jmclaugh]

How in the world would someone in 1927 be able to determine how accurate (that's the claim) the clock is? I figured that, if the clock was for example 5 minutes off after 6 month, that would be called within the claim. If it was off 7 minutes after 5 month, it was not within specs. These differences could probably be observed by a lay person.

Uhralt

How could the differences be observed by a lay person and with reference to what?

 

[Uhralt]

Hi Uhralt,

Interesting and elegant clock! I wonder whether you've tried regulating in one direction only, i.e. from going fast towards going slow, and if you 'overshoot', start over again? I was advised to regulate a 400-day clock in this manner, and it worked very well, although with a conventional pendulum clock I find it more reliable to regulate from slow towards fast, as moving the bob a consistent and tiny amount upwards is more easily achievable than getting the bob to consistently come down. It seems that if you try to reverse the direction in which you're regulating, the slop or hysteresis inherent in many pendulum mechanisms can cause an overshoot in the opposite direction. As Willie says, as you get close to the correct rate, try regulating no more frequently than monthly.

Tim, I don't know about the United States, but in Britain the BBC broadcast time signal 'pips' from 1922 onwards, and I imagine many US broadcasters were doing likewise around 1927.

Phil

Hi Phil,
Yes, I did try to regulate in one direction only. Tried it going slower only or faster only. What I found is that the disturbance I caused by manipulating was greater than the effect of the minimal change I made. It took the clock a while to settle. I will now try the monthly adjustment as suggested.
Uhralt

 

[Uhralt]

How could the differences be observed by a lay person and with reference to what?

A lay person could set a good watch by looking a standard, maybe a Jeweler's regulator. Go home and compare his watch with the clock.
I heard that in England some people made a living by going from house to house with a good watch and tell the people the correct time so they could set their own clocks. The watch would be set to a good reference every day.

Uhralt

 

[THTanner]

Or perhaps the accuracy claim was as much a marketing ploy as today's claims of mileage in a car?

 

[Tim Orr]

Tim, I don't know about the United States, but in Britain the BBC broadcast time signal 'pips' from 1922 onwards, and I imagine many US broadcasters were doing likewise around 1927.

Good point, Phil!

I hadn't thought of that. I did see that Milton Biow's advertising agency, which subsequently was bought by three guys named "Foote," "Cone" and "Belding," and known as just "FCB" today, began doing "B-U-L-O-V-A, Bulova Watch Time" in about 1926 here. Of course, I couldn't find what they used for their standard. I did see one source suggesting that WINS in New York could easily be off by several seconds with its time tones.

My guess is that use of synchronized "Self Winding Clocks" from Western Union probably began in about 1922 with the beginning of radio networks in the USA. If a signal was coming down the equalized (phone) line to you, you needed to know what time it was, to avoid a long delay (Dead Air!) or "stepping on" (talking over) the signal from the network. Either one was a huge embarrassment to a broadcaster, especially in the days when radio announcers wore suits and ties to work and spoke in "pear-shaped tones." Not sure how many people had radios in those days (aside from "experimenters").

There's a fantastic story of how Ed Murrow used synchronized time to create the illusion of an on-air interview between Murrow (in New York) and Bill Shirer (in Berlin) right before WWII. They used shortwave to feed Shirer to NY, and carefully scripted and timed speeches to make it sound as though Murrow was actually interacting with Shirer. It's in a big biography of Murrow that came out some years back.


Ad claims also probably played a part. When I worked in photo processing, we had an operator-assisted film splicing machine that claimed it could do 200 rolls of film in an hour. Nobody in our lab could ever match that productivity claim. So, one night, I stepped into the darkroom with 200 rolls and stepped out again when I had completed the job. It was an hour and some minutes, and I was damn near ready to go to the hospital with nervous exhaustion. (No way I'd ever have been able to maintain that pace for an 8-hour shift.) I later learned that the manufacturer of the splicer had stopwatched the cycle time of the machine for a single roll, then extrapolated to an hour's productivity. As an old sales trainer I once knew used to say, "Tell 'em anything. Who checks?"

Best regards!

Tim

 

[Uhralt]

The first 4 weeks of undisturbed running are over. This is what I've found: First, I've been chasing my tale by trying to regulate the clock on a daily basis. This is clear from the First 3 weeks of running. After the first week, the clock had lost 12 seconds, after the second week it had made up some of the loss and was only 7 seconds late. After the third week it was 25 seconds ahead of time. Then, during the fourth week it rapidly lost time (I had started daily observations to better capture variations) and at the and of week 4 it had lost 75 seconds. Then, a day later the clock had stopped due to a contact problem.

I believe that the rapid loss during the last week was due to an developing contact problem. I had previously observed that when the contact goes bad there are double impulses that cause the clock to slow down. I will work now on improving the contact and will resume measurements thereafter.

I found no correlation between timekeeping and the number of oscillations between impulses. These were rather stable between 33 and 37 oscillations.

I monitored temperature but there were no significant temperature changes during the observation period.

The graph below shows the data.

Overall I believe that the clock will be capable to deliver a deviation of no more than a minute per month once the contact issue is solved. The first three weeks looked really promising.

Uhralt

 

[Uhralt]

Now, almost 6 months later, I think i have an answer to the original question. It took so long because of the erratic timekeeping that the clock showed originally. I replaced the contacts with pieces I found in an old telephone from the 1920s. I used a the phone contact that worked with the dial of the phone. It was almost identical to the Poole contact, only somewhat longer. I cleaned the contacts well and the problem of the double impulses disappeared and has not come back in 6 months.

However, the clock would still sometimes suddenly gain or lose up to 60 seconds in a day after a period of good timekeeping. After a long search for the reason, with much trial and error, I finally realized that the suspension spring had some small kinks in it. I now believe that the spring was sometimes flexing above such a kink and sometimes below. I replaced the spring and surprisingly, the number of oscillations between impulses went to 41 to 44, up from 33 to 37, which was not shabby to begin with.
At this point I started the "official" test. As can be seen in the first picture, over a period of 31 days, the daily deviation ranged from -8 sec/d to + 2 sec/d, with a mean of -3sec/d.

The second picture shows the actual timekeeping over 31 days in the upper panel. As can be seen, timekeeping was within 60 seconds (compared to my trusted iphone) for 21 days. However, when corrected for the mean loss of 3 sec/d, the time shown would be within an range of +5 and -10 seconds, well below the limit of 60 seconds. This is shown in the lower panel.

In conclusion, I believe that the clock can indeed keep time within 60 seconds within a month, if regulated perfectly. The problem is to set the pendulum length exactly right which is difficult because the clock shows the largest daily variability within the first 14 days after adjustment. It is very time consuming to be able to tell if you've got it just right or not. The adjustment for such a small change is tiny. So, overall I think in theory the advertised claim is possible, but in practice very hard to achieve. It takes a clock in "perfect" condition and a LOT of patience to regulate it right.
Uhralt

 

[Phil G4SPZ]

Hi Uhralt. This has been a very interesting story to follow. Replacement of the contacts with ones taken from an old telephone was probably a good move - I believe tungsten was used in some telephone contacts in the interests of long life and low maintenance.

Are you planning to adjust-out the three-second drift and see if you can regulate the clock perfectly?

Phil

 

[Uhralt]

Hi Uhralt. This has been a very interesting story to follow. Replacement of the contacts with ones taken from an old telephone was probably a good move - I believe tungsten was used in some telephone contacts in the interests of long life and low maintenance.

Are you planning to adjust-out the three-second drift and see if you can regulate the clock perfectly?

Phil

Hi Phil,
Thanks for your interest. Right now I will just correct the time setting and observe for another month if the 3 sec/d remain. We are just about to see winter temperatures and the clock may speed up when the night temperature inside the house falls. My thermostat shuts the furnace off for some time at night. That might have an effect. I think I will go to weekly observations now and may report back in another month or two.

Uhralt

 

[P.Hageman]

Uhralt, thanks for this interesting information. I have a Barr clock which is identical to the Poole? I only have it now for a few weeks but because it has frequently double impulses, I removed the batteries and changed them for new ones, hoping that would solve the problem. Without succes, so now its standing here without batteries. Cleaning the contacts would perhaps solve the problem? Which contacts and how to clean, please some advise. Cheers Peter

 

[Uhralt]

Uhralt, thanks for this interesting information. I have a Barr clock which is identical to the Poole? I only have it now for a few weeks but because it has frequently double impulses, I removed the batteries and changed them for new ones, hoping that would solve the problem. Without succes, so now its standing here without batteries. Cleaning the contacts would perhaps solve the problem? Which contacts and how to clean, please some advise. Cheers Peter

View attachment 626342

Hi Peter,
Yes, the Barr clocks are the somewhat later models of the Poole clock. The double impulses are a problem coming from the contacts that are closed by the toggle switch and trigger the electromagnet. They are on the back of the movement and should be two, wired in parallel. You can use a contact cleaner for electronics and Q-tips to clean them. I used a product called DeoxIT. Don't use abrasives because they are thinly coated. Also take a look at the contacs for the batteries. The bottom one is often corroded.
Uhralt

 

[P.Hageman]

Uhralt, thanks very much. I will clean them up then and see if it will have a positive affect.

 

[Uhralt]

Uhralt, thanks very much. I will clean them up then and see if it will have a positive affect.

Let us know how it goes.

Uhralt

 

[thesnark17]

I bought one of these a little over two months ago, not running. Some jury-rigged repairs got it running, and I gave it a rough adjustment (the pendulum had been adjusted far out of position) and have left it to see what it would do.

Starting condition: -40 seconds
I checked every day for the first week or so, and the clock gained about 2 seconds each day. I thought that was pretty good for a rough adjustment.
T+2 weeks: -16 seconds
After about two weeks, the clock began losing slowly.
T+4 weeks: -68 seconds
At this point, I considered adjusting the pendulum but decided to leave it alone and see what another month would do.
T+8 weeks: -40 seconds

I'm pretty happy with how that turned out!

Over two months, the clock has met the manufacturer's claim of being accurate within a minute. At no time did the clock vary by more than 30 seconds from the starting time.

The cycles per impulse gradually climbed as the clock ran, from an initial 29-32 to the current 36-39.

 

[TQ60]

For cleaning the contacts you burnish them.

Back in the day, relay service kits had burnishers and bending tools.

To make one is simple, cut a strip of good paper the width of the contacts.

Place it between them, close them on the paper with light pressure and pull out the paper.

We created a transistor switch today for our American clock.

It is to provide switching current in a manner to reduce the arcing when the contacts first close.

It also can limit the current if needed.

Bonus is no modification to the clock.

Circuit is described in the other post, here is the diagram.

It has a 10k resistor that limits the current to just a couple mA, this small current then turns on the transistor bypassing the resistor.

 

[TQ60]

Others already stated, will add.

Start the clock running SLOW.

Turn the nut to adjust, wait for some time, NOW record the actual time on the clock as the start time, and the error from whatever your standard is.

The pendulum is upset when you touch it, so give it some time to normalize before you check your accuracy.

 

[Robert Gift]

View attachment 583060 View attachment 583061 View attachment 583062

Love the clock.
Do temperature fluctuations not affecthe pendulum length?
Our house goes from 65º F. at night to hot in the summertime, so we canever expect any accuracy.

 

[Uhralt]

Love the clock.
Do temperature fluctuations not affecthe pendulum length?
Our house goes from 65º F. at night to hot in the summertime, so we canever expect any accuracy.

It certainly will have an effect on time keeping but the temperature fluctuations in a modern home with central heating and air conditioning will have a much smaller effect than in a house in the 1920s.

Uhralt

 

[Robert Gift]

It certainly will have an effect on time keeping but the temperature fluctuations in a modern home with central heating and air conditioning will have a much smaller effect than in a house in the 1920s.
Uhralt

We have central gas heating and air conditioning but being a registered tree-hugger and Greenie, we savenergy by letting the house cool to 65º F. by bedtime and not heating until 20 minutes before we get home. We sethe.rmostat from the cell phone!
Only quartz and electriclocks are are accurate.

 

[thesnark17]

During my two-month test, the temperature in the clock room varied between 58 and 74, averaging about 65. It's the coldest room in the house in the winter, and the warmest in the summer.

The temperature range is quite wide for a modern home, but still likely smaller than it would have been in the 1920s in an average home. Of course, these clocks were not exactly sold to the "average Joe"...