Showing posts with label Accuracy. Show all posts
Showing posts with label Accuracy. Show all posts

Rate and Accuracy in Vintage Watches

I often go to a fair amount of effort to get even what were originally inexpensive antique watches to keep time well. I think I may stop doing this, or at least not as much. And this post is about the reasons... But before going further a few points about mechanical watches in general, and vintage mechanical watches in particular, are in order. A few points about mechanical watches...

Hands

The hands on a mechanical watch are geared together. They can not, if everything is functioning correctly, move independently. Their movement relative to each other is mechanically dictated by gear ratios. For example, the minute hand must advance one minute each time the seconds hand completes one trip around. Likewise the hour hand must advance one minute each time then minute hand completes one trip around the dial. It can not be otherwise as the hands are mechanically joined by gears.

Rate

We expect the minute hand to complete one trip around the dial in one minute, as read by some external reference. However, the hands actually do not move in a smooth motion. All the hands jump forward several times a second. Between these jumps, all the hands are still. How many of these jumps occur in one obsolete, correct minute is what dictates how well the watch reads time. How fast the hands move, the speed of the tip of the hand, when the the hand is actually moving, is almost always irrelevant.

One can see these jumps in the movement of the seconds hand on a mechanical watch. The other hands are also jumping, all at the same instant.

Escapement

Mechanical timekeeping is possible because of an escapement. The escapement "escapes" power into the gear train many times a second. At each beat of the escapement, the gears turn. Then they go still again in between. The escapement controls the frequency of the hands advancing, and thus the accuracy of the watch over time.

What is time keeping?

The "accuracy" of a watch boils down to how fast the hands move around the dial according to some reference time source. But all mechanical watches (really, all, even new ones) have certain issues that impact this.

Mainsprings impart more power when first wound than when almost run down. So the rate of a watch wound in the morning varies as the day passes. Watches may have some means of compensating for this. Most older ones do not.

Gravity effects the hairspring. The hairspring and the balance wheel dictate the frequency of the escapement, and so alter the reading of time on the hands. A wristwatch is in many different orientations as it is worn, so the rate varies over the day, moment to moment. Again, some watches have the means to reduce this variation.

These are just two examples. There are more many factors that also alter the watch escapement from moment to moment under normal operating conditions, on a watch that is functioning correctly. Because these issues exist, watchmakers of old did not even bother to measure a watches rate moment to moment. They were only interested in a watch that read good time after running for some period without being reset (a day, or several days). Sometimes during the run interval the rate may gain, and sometimes lose. Watch "adjustment" is setting up everything so that plus and minus errors (the ones we can do nothing about) cancel each other out over time.

Age

In vintage watches there is another factor; simply the passage of time. Old style, steel hairsprings change their metallurgic properties with the passage of decades. Simply put, a spring just isn't as springy anymore. As a result, the frequency of the escapement is impacted. The solution would be the installation of a newly manufactured spring. But this is not usually possible as they do not exist.

In these cases, there are several adjustments that can be made. One can adjust the mass of the balance wheel. The balance screws provide this mass. They can be replaced, in opposite pairs, with lighter or heavier screws.

This is time consuming work. Once the screws are disturbed, the rate will usually be way off, out of the ballpark. Pulling it back in takes a lot of basically trial and error.

The Regulator

All of the above applies to watches that have been correctly serviced and repaired and are functioning correctly. Watch designers selected mainspring strength, gears, hairspring strength and balance mass to result in a watch that would be in the ballpark of the correct time after the passage of an interval (say 24 hours). Was a watch, assembled at the factory, ready to go and accurate? No. Manufacturing was not that good. It was close though. So what did they consider "in the ballpark"?

We actually have a pretty good idea. Mechanical watches feature a regulator that can be adjusted by a watchmaker, or the end-user if they like, to effectively lengthen or shorten the installed hairspring, and thus alter the frequency of the beat, and thus the rate of the watch. On vintage pocketwatches, the range of the regulator is about +/- 5 minutes per 24 hours (this does vary from design to design, it can be as much as 10 minutes) over the end to end sweep of its setting.

So, we know that watch designers knew that between variation in manufacturing precision, and variation in real world conditions, an error of +/- 5 minutes was to be expected.

A Proposition...

Within the range of error provided for by the regulator, a fully serviced watch is functioning correctly.

The people that designed the watch did their job as well as they could, with the tools and technologies of their time, and within their business requirements. If they could be assured that each watch would leave the factory more accurate than they did, they would have reduced the range available on the regulator. Or perhaps left the regulator off altogether. In short, the range of the regulator represents the expected range of error in a perfectly working watch.

Antiques

All of the above points that impact the rate of a mechanical watch are general. I'd like to add one more that is specific to antiques; history.

Whether a watch was an economy 7 jewel model, or a high end 23 jewel watch that originally cost 15 times as much, after 100 years they are all individuals. It doesn't matter what sort of watch it is if it's been through two loads of laundry, a fire, and stored in a damp and dirty junk drawer in the garage with the oily rags for 30 years. Watches are not magical. They are constructed (mostly) of machined pieces of relatively soft metal. Countless damages can happen to these parts over the decades that can not be undone.

What's more, a 100 year old watch has likely been serviced and repaired dozens of times, with varying levels of appropriateness. It is extremely common to see parts in a watch that have been replaced with the best thing available at the time - an incorrect or handmade part, and the rest of the watch permanently altered to make an incorrect part work.

From time to time I get an objection from a watch owner that their once high-end railroad watch should keep better time because of what is was when it left the factory. In a few of these cases over the years, I confess I had to laugh (fortunately my communication with watch owners is almost never face to face) because the watch was in such horrendous condition that I was pretty pleased with myself for getting it to even tick. They would have been hard pressed to find another watchmaker that would have even taken on the challenge.

In short, it doesn't matter what the watch once was. It matters what has happened to it since it left the factory.

Adjustment

Now we come to the heart of the matter. Just how accurate should we make an antique watch? I'll  make several points about this.

1) It doesn't matter.
What's important is that an antique is in good condition, functioning correctly, rust-free, correctly lubricated, stored in a clean, dry manner, and preserved for future generations. Even setting aside family heirlooms, how well a watch keeps time (once it is functioning correctly), among serious collectors, is not even a secondary concern. A watch that keeps time to +/- 30 seconds a day is not automatically "better" that a different instance of the same watch that runs +/- 1 minute a day, In fact, the opposite may well be true as so many other factors are more important.

2) Diminishing returns for the effort
If someone wants to pay me for the 40 hours (or more) that it might take to bring a hypothetical watch from +/- 10 seconds a day to +/-  6 seconds a day, in multiple positions, reliably, let me know. They did that sort of thing at the factory, but on dozens of brand new, mint watches, all at the same time.

3) Why?
Why worry about time keeping at all? With basic effort, and service by appropriate procedures, many (no, not all) watches fall with a minute or two of error fresh off the bench before I even touch the regulator. How much time and effort should be devoted to fine tuning?

Well, before answering, keep in mind that the watch doesn't "care", at all. Once it it operating correctly, there is no difference at all, for the watch, if the error is two minutes or 10 seconds a day. So the only reasons to do adjustment, are for time keeping, or for fun. As for the former, if you want to know what time it is accurately, and that is important, get a cheap quartz watch or use your mobile phone. These are a 1000 times more accurate than any mechanical watch will ever be, new or old.

So how much effort? I try to strike a reasonable balance. Almost all get to less than 10 seconds a day of error in at least dial up position, and many less than +/- 5 seconds. I consider that excellent. These watches come from a time when it was not unusual for people to be walking around with watches that read +/- 10 minutes a day.

I spend more effort reducing the beat error than the rate though, since beat error is actually a fault causing inefficiency, regardless of the rate. But it does take quite a bit of time to work on the rate, especially if the balance has to be adjusted in anyway. I think perhaps time might be better spent getting more watches to run well, rather than seeing how good each one can possibly keep time.

What do you think?

More about watch accuracy here...


Watch Accuracy in 1950

The following articles appeared in the Journal of Horology in 1950. They provide insight into what watchmakers were telling their customers about accuracy at that time.
And...



Why Does a Timing Machine Sometimes Not Work on a Vintage Watch?

Q: Why Does a Timing Machine Sometimes Not Work on a Vintage Watch, even while the watch keeps good time?

A: The "tick" of a watch, of a lever escapement, is actually a repeating cycle of five separate impacts in the mechanism. The machine has to be able to pick up on this rhythm to tell anything at all. Sometimes there is too much other noise in there from something rubbing, the roller being too small and knocking, the pallet hitting the banking pins too hard, hairspring "singing", wheel missing a tooth, worn escape, excessive end-shake, whatever... and the machine can't lock on. In general extra noise, rubs, impacts, friction, etc, or all inefficiencies that extract a little power with each cycle. But a watch can run very inefficiently and still keep OK time, over the course of a day.

The machine also needs to detect the beat rate in order to get the timing. I have seen slow beat watches that have had a fast beat hairspring put in, then other parts changed to compensate (this is pretty common in older Elgins). The machine then reads wildly inaccurately, if it can work at all. Often it helps to turn the amplification down on the machine. Sometimes the machine can't detect the beat rate, and setting manually will help. But sometimes it just won't work.

When the rate is way off, the machine may not be able to detect anything, mostly because it can not get a handle on the beat rate. A watch that does not have a lever escapement, a cylinder or pin-lever escapement watch for example, is right out. The machine will not work because the sound of the watch is completely different.

I think timing machines are more limited in their use on the older movements. I use one as an initial reality check, but I test all the watches for several days in positions to do reasonable adjusting. The machine, even when it works fine, measures the rate moment to moment. But the rate actually varies a lot over the course of the day both because of positions, and because of the properties of the mainspring. Watchmakers of old did not much concern themselves with the rate at any given moment for this reason, but were more concerned with the drift over longer intervals of time.

Read more Q&A here!

I Heard...

I heard that a watchmaker that does very similar work to myself, mostly antique pocketwatches, tells people to expect an error as much as +/- 10 minutes per 24 hours. I was surprised, that seems like a lot. It doesn't take extreme effort to get to +/- 1 minute on the majority of watches, and the majority do less than +/- 30 seconds. Occasionally there's a good reason for a larger error, such as with an especially primitive or old watch. But even then +/- 2-3 is achievable in all but the most extreme cases, in my experience, without spending a lot of time. There is diminishing returns for ever greater efforts yielding ever smaller improvements.

More about watch accuracy and rates here...



Timing and Positional Rates of Watches

Watches run at a rate of course, but there is a difference between a watches accuracy over time, and the rate of the watch at any given moment.  In modern times, most people would think of "accuracy" as a reference to the rate, that is as how closely the passage of seconds on a watch matches the passage of seconds on a reference source.

However, this is not how watchmakers of old would think of accuracy in a timepiece.  In the days before quartz movements, which have an extremely consistent rate, accuracy in a watch was about how much it would drift away from a reference over a long period of time.  A "long period" being the intervening time between opportunities to set the watch to a reference source.  This could be a day, a few weeks, or even longer.  Accuracy would describe how far off a watch reads after a day, or a week, or a month.  It was important to build watches with this in mind because access to a good reference time source, a century ago, could be quite infrequent.

The rate at any given moment was less important, as a practical matter.  Railroad standards for example, refer to accuracy over time periods, but are silent on the question of rate.  And in fact during the course of a given day, the rate of a vintage watch varies by a surprising amount, even as over time, the accuracy may be quite good.

The rate at which a vintage watch runs varies for several reasons.  The most significant is that something is wrong, such as worn balance pivots, broken teeth, cracked jewels, unbalanced balance, distorted hairsprings, and so on...  Such problems in a watch typically cause large accuracy issues; errors observable within only one hour's time.  In my experience, once basic faults are corrected though, most mid-grade vintage watches are capable of accuracy of +/- one minute or less per 24 hours with little timing effort.

This discussion is about the discrepancies that remain after basic watch service.

Temperature

Temperature causes metal to expend or contract.  We see this impact watch timing because the balance wheel will literally get smaller of larger, thus moving its weight inward or outward and changing its period.  Compensating bi-metallic and split balance wheels were invented to help compensate for temperature changes.

Power

The rate of a mechanical watch varies over the course of running down from a full wind, to fully stopped.  The mainsprings of watches apply slightly more power when first wound, over when almost unwound.  Motor barrels and alloy springs were invented that reduce this effect, but it is not eliminated.

Gravity

All vintage watches will have a discrepancy in the rate in positions, especially between horizontal and vertical orientation in particular, due to gravity.

The root reasons for rate shift due to power and to gravity are changes in the center of gravity of the hairspring during its motion (here's the word of the day; isochronism, which is basically what the rest of this discussion is about).

Some watches have a flat hairspring.  But in some watches the outside end of the hairspring is bent up and across the spring in what's called an overcoil.  The overcoil helps make the motion of the spring more consistent, so watches with a flat hairspring will have greater rate errors, in general.  But the effects we're speaking of here are not eliminated by the overcoil, just reduced.  And the overcoil tends to make the watch more susceptible to rate error due to the watch moving around.

So, back to a watch that is otherwise functioning correctly, and focusing on positional rate error, it must be noted there is no simple way to adjust the rate for vertical and horizontal orientations independently.  There are things that have been done to deliberately change the vertical rate, but these are considered bad practices, for example deliberately inducing extra friction on the balance pivots in the horizontal by flattening the pivot ends to slow the watch when dial up or dial down.

Adjustment to Position

A watch adjusted to positions means that the watch has been run in various orientations for 24 hour periods (or more) and that the start and end times are recorded.  The errors after the test period in each position can thus be noted.  Watches may be tested in some number of positions.  These would be dial up, dial down, pendant up, pendant down, pendant left, pendant right and more.  Higher grade watches are typically tested in more positions.  More basic grades, just two or three, or perhaps none.

When a person wears a watch, it will spend some certain time in different positions. So, once serious problems are corrected, "adjustment to positions" means setting the rate over all slightly faster or slightly slower, so that on average, at the end of a period of time, the watch will read fairly accurately, even though during the course of that period of time, it has run at slightly different rates moment to moment.

The challenge is that the amount of time a watch spends in different orientations can vary from person to person.

A couple of times lately I have fielded questions from people noticing that the rate of a vintage watch seemed off under certain conditions.  But even noticing this is a result of modern exposure to quartz watches that have a rock-solid, constant rate.  The best vintage watches have rates that vary.  Adjusting was always a process of setting it so the errors cancel out more or less over a period of time.

As complex as modern electronics are, a quartz movement, supplied with the correct power, is essentially on or off.  In some ways, the mechanical designs are more complicated.  There are many, many factors that affect the rate of a watch.  I have left a lot out.  Whole books have been written about hairsprings alone.  Not to mention the special issues of vintage pieces that have been altered by watchmakers over and over.  It's a never ending learning process...


Guaranteeing Watch Performance

From Horology magazine, March, 1939

Guaranteeing Watch Performance

The subject of watch guarantees and watch performance possibilities is a continual topic of discussion among jewelers and horologists, but in spite of all that has been said, nothing has been actually done. From the standpoint of value received, there is no doubt that the public receives an honest equivalent for the money it spends on new watches. However, the indiscriminate guarantees of service given by so many jewelers does react to the detriment of the entire jewelry industry. It is one thing to guarantee that a watch case is made of 14 or 18 karat gold and another to guarantee that a watch will run for a year without stopping or that it will run within 30 seconds a month.

True, a jeweler or horologist should be responsible for the satisfactory performance of the timepieces he sells or services, but the question arises, "What is satisfactory performance?" Here is where a great many watch salesmen and men at the repair counters of jewelry stores need a little education, for they are in the habit of promising their customers performance which is beyond the possibilities of attainment from ordinary watches.

Everyone will agree that very few watches with observatory certificates of rating are being sold, but for the sake of arriving at some standard of comparison let us see what the Swiss Official Controlment Offices for the Rating of Watches require of wrist watches before granting a certificate. They allow a difference between mean rates of the dial up and dial down positions of 30 seconds in 24 hours. Between the dial down position and either the pendant down, pendant right or left positions, a difference of 50 seconds in 24 hours is allowed.

Now, if a world famous institution considers a watch coming within such tolerances worthy of a certificate, what performance can be expected from the average wrist watch being sold? Obviously, one is only kidding himself and his customer if he tries to regulate a bracelet watch when it is only a minute or two off in a week. Some idea of what should be expected may be gained from the Horological Institute of America's requirements for the Certified Watchmaker examination. The Institute specifies that the horologist must adjust a wrist watch so that it will have a position error of not more than 90 seconds in 24 hours between the dial up or down and pendant down positions. In considering these figures it should be borne in mind that a Certified Watchmaker is better than average.

The motor car industry seems to have stabilized the matter of guarantees. An auto is guaranteed for several months against defects of workmanship but there is no guarantee of how many miles per gallon it will go or how infrequently it will need service. Why should a watch be guaranteed to run for a year without cleaning and within a minute a week, or some such figure, when the jeweler or horologist has absolutely no control over the conditions governing such performance?

It would seem a desirable thing for the industry for watch manufacturers and importers to consider a uniform standard of performance for various grades and sizes of watches which the trade can actually guarantee to its customers. This would go a long way toward alleviating some of the ills of the jewelry business and would inspire greater confidence of the public. 

Watches Fail to Rate Properly

From Horology magazine, May, 1938

Watches Fail to Rate Properly
Editor Horology
Dear Sir:

As a subscriber to Horology I have a question I would appreciate if you could solve for me.
I am having trouble at times with high grade watches of the best American makes, 21-23 jewel grades, trying to make them run accurately. They seem to gain one day and lose the next.

These watches are not magnetized, the pivots are in perfect condition, the balance wheel properly poised and escapements match correctly. Sometimes these watches will show a variation of 30 seconds to 1 minute in 24 hours either way, and again they will run with no variation in 24 hours. At other times they run with a variation of only a few seconds.   I have had 15 years experience with watches and have been through a horological school.

On one occasion I sent one of these watches to the factory where it was made and had to send it back a second time before the trouble was overcome.

I would appreciate it if you could tell me what the process would be to overcome this kind of trouble.
R. T. C.

Answer: It is difficult to understand why watches of the grade you mention should have a daily variation of from 30 seconds to a minute. Such an amount is unreasonable, whether the watches vary this much from day to day while in the same position or whether they vary this amount from one position to another. 

The Bureau of Standards, in the Railroad Precision Test, allows a maximum difference of 12 seconds in daily rate between any 2 positions, although no tolerance is specified for a watch kept continuously in one position and temperature.

No matter how closely a watch may run under uniformly controlled conditions it will show some variation when kept in an ordinary room whose temperature is not constant.

However, in the case under consideration it would seem reasonable to assume that the watches are not in good enough mechanical order. If the variations in rate occur while the watch is in the same position, the trouble is very likely to be found in the transmission of power and not in the escapement. A poor mainspring, tight barrel arbor or center wheel are just a few of the defects which may contribute to an irregular rate. On the other hand, if you are troubled with position variations, the difficulty will probably be traced to the balance and hairspring.

You specifically mention that the pivots are in perfect condition, but the accuracy of this statement depends on the means of examination used. Mere observance with an ordinary loupe, of course, is not to be considered a test. Balance pivots cannot be adjudged straight or round unless tested in a true balloon chuck or in a Dorrington pivot polisher or similar tool.

It would be impossible to enumerate here all the possible faults which may cause a watch to take a variable rate.

For further enlightenment on the subject we refer you to "Practical Course in Adjusting" by Theo. Gribi, "Rules and Practice for Adjusting Watches" by Kleinlein and the section on adjusting in "Practical Benchwork for Horologists." 


Vintage Watch Accuracy Revisited

How accurate can a vintage watch be?  The Tick Tock blog has posted two enlightening pieces on the accuracy of watches, particularly vintage watches, since these two articles are from 1950.

"One of the things many watchmakers struggle with is customers who want modern accuracy (or even quartz accuracy) from vintage timepieces. It seems that over the years we have forgotten what was expected of wristwatches 50 years ago."

Indeed.
Take a look at the entire blog post here:

http://watchmakingblog.com/2011/04/06/vintage-accuracy/

Read the two pieces via Google Docs, one here and the other here.
My earlier comments on this subject are here...

How Accurate is a Vintage Watch?

There was a nice post over at the "I Already Have a Watch" blog, which is generally concerned with more "modern" vintage, regarding the accuracy of vintage timepieces. This post has been removed, but it was originally here:

http://ialreadyhaveawatch.com/2010/09/how-accurate-is-good-enough-for-vintage-watches/

Here's a quote:

"While many of these old watches can be repaired and adjusted to run very accurately there are simply some that are old, tired, worn out antiques. Very often mechanical watches were sorely neglected especially in the post-1970 era when inexpensive quartz replacements could be had and watchmakers began getting harder to come by. Some vintage watches are lucky to be running at all after the way they’ve been treated for three or more decades.

"But in general it’s perhaps unreasonable to expect a decades-old watch to perform like new, and a +/- 1 minute a day rate for most any 30-60 year old watch with unknown or even no previous service history isn’t too bad in my opinion, especially for watches that may not have been capable of chronometer performance when new."

I agree, and the entire post is worth reading. It raises a few issues that I'd like to expand on here.

Most, almost all in fact, of the watches I work on are family heirlooms and of a wide variety of original quality levels ranging from budget 7 jewel grades, to the 21 and 23 jewel railroad models, and ranging from the mid 19th century through the 1950s. The accuracy of these watches after cleaning and repairing any serious faults (that is, faults that prevent a watch from running), also varies. The vast majority fall in the ballpark of +/- one to two minutes per 24 hours. There are outliers though, both more and less accurate, creating a wide range. Every vintage watch is a truly different, unique individual. The important thing is that the watch be functioning correctly, without fault. The actual rate of a watch can vary by several minutes per 24 hours in a completely properly functioning watch.

In my work, I strive to get the best mechanical state I can for each watch, while using as many original parts as possible - even if those parts are not in perfect condition. It may always be possible, in theory, to improve the accuracy of a watch by replacing slightly worn parts that are not ideal. But to what end? There are several issues to consider.

Most of the watches I deal with are American-made vintage pocketwatches. Even when they are slightly rare pieces, at least some replacement parts are at hand - and frequently factory new parts are available. But the more that is replaced, the closer we are to a different watch. If a part is marginal, but the watch runs, that is, within a minute (or two or three) a day, and does not stop in a certain position, for example, I would tend not do the replacement. It would likely be possible to marginally improve the watch by replacing more and more components with pristine parts, moving closer to an "ideal" or exemplar watch, but it would be far more practical to simply purchase a different watch that is in better condition, than to bear the cost of replacing each individual part until one watch is made over into another watch anyway. The cost of replacing parts, a lot of parts, in an antique watch approaches the cost of buying a different movement quite quickly for common pieces. For most people, since they are dealing with a family heirloom, replacing the movement is not what they are looking for, and rightly so I feel. And so accuracy is a secondary concern.

Even when a new, and presumably "better," part is not available, it can usually be made. In addition, factory parts in a watch might be tweaked and adjusted, perhaps using a modern method, tool or material, and again the accuracy of a watch improved. But the reason for putting the word "better" in quotes above is that, again, accuracy is not all that goes into the choices made. A watch should also be original. Boring a hole in a plate and adding a bushing to improve a worn, un-jeweled pivot, is a "repair," but it is also an irreversible alteration to an antique.

Further, is it really desirable to create a watch that is better than new?  In the early days of watchmaking, before railroad standards, accuracy of +/-10 minutes a day would not be unheard of.   Although 10 minutes is extreme, restoration of an antique to a level of performance typical of it's period seems reasonable and appropriate.

I should also make another point with regard to historic watchmaking practices. The considerations above would not have been taken into account by the old watchmakers that have previously serviced a piece in the 19th century. In those cases a watchmaker would in fact do whatever they needed to do to achieve the best timekeeping they could. Keeping in mind that watchmaking historically is a "by-hand" tradition, and was not all about standardized factory parts until much later, it is no surprise that hand-made parts and clever, jury-rigged repairs are pretty common in vintage pocketwatches. I see them all the time, especially in pocketwatches in use in the 1800s.

Sometimes these are awful hacks.  But these repairs are also part of the history of the watch and (hopefully) a tribute to the skill of the old watchmakers If they are not interfering, if they are working, it they are doing no harm, and if they don't require an alteration to another original part, I leave these fixes in place, even if a "proper" repair would improve the accuracy of the watch. In fact once or twice I have even "repaired" an old "repair".

Lastly it should be said that some watches should not be repaired. Occasionally I see something handmade, very old, and/or very rare or special in some way, such as having been owned by a significant historical figure. These watches are much more valuable in an "un-restored", as-found state. The best thing to do in these cases is stabilization, usually making the watch run (or at least tick), assuring general cleanliness and good lubrication, and replacing little if anything, depending on the circumstances. The original character of such items matters, not their accuracy.

And so regarding accuracy...   I am reminded of my Grandfather, who had forgotten more about watchmaking than I will ever learn in a lifetime, once expressed to me his puzzlement that anyone would want an old pocketwatch repaired. He asked me, "why would anyone want an old watch when the new ones are so much better and cheaper?"

Read more here:
http://elgintime.blogspot.com/search/label/Accuracy


How pocket watches are adjusted for temperature and timing

From American Horologist and Jeweler magazine, December 1959

How pocket watches are adjusted for temperature and timing
by W. H. Samelius

editor's note: Replying to a question by "W.A.", Mr. Information, Please" has written a techical article of practical value to every serious watch repairman.

TEMPERATURE adjusting is the first adjustment necessary in watches, and is confined entirely to the compensation balance. This particular adjustment is always done at the factory, and watchmakers should be careful not to change it. The principles are simple and have been determined by some millions of experiments. They are based on the fact that brass has greater expansion than steel, and if the steel is soft, the metal having the greater mass will completely overcome the other.


Speaking generally, a proportion of 2-fifths steel and 3-fifths brass have been found well adapted for this purpose.


The relative proportions between the width and the thickness of the rim is a matter of importance. The thinner the rim in proportion to its width, the more active it will be and the greater it its deflection due to changes in temperature.  Also, the deflection by centrifugal force will be greater as will be the danger of losing the true form by slight jars or by careless handling.  It is also evident that in order to produce a total given weight, a thin rim must have a greater weight in screws than a thicker one. A rim with a thickness of 36 per cent of the width has been found about right. There are four screws located about 90° apart (two at the ends of the arms, and two halfway between them on each rim) which are used to determine the practical working diameter of the balance. These are called "timing" screws and should never be changed unless to slow the balance or to increase its speed.  The other screws are simply to add weight, and their position in the holes drilled, has in every case been found by trial at the factory.


Moving these screws toward the cut ends of the balance changes the temperature compensation. The nearer a screw is to the cut end of the balance segment the farther in or out it will be carried with changes of temperature and consequently, the greater the effect will be.


A watch with a brass balance and tempered steel hairspring, if regulated in a temperature of 40° F, when run for 24 hours in a temperature of 90° F, will lose about 6 minutes, 15 seconds per day, or about 15 seconds per hour. Of this, 47 seconds is due to the increase of size of the balance, due to the heat, while the rest, 5 minutes, 28 seconds, is due mainly to the loss of elastic force of the hairspring.


In a compensation balance, the screws are adjusted so that these losses will be made up by what amounts to a practical reduction in the mean diameter of the balance. This is all there is to temperature adjustments, and, as stated, they are determined by trial for each balance by running them at 40° and 90° in the factory.


It can be seen, therefore, that it is best to let the balance alone unless one is provided with means of maintaining accurate temperatures over extended periods while a watch is being rated. The watch factories have these necessary facilities, yet they find it necessary to determine by actual trial the proper position of the screws on each balance.


It is, therefore, a good plan to let the screws alone, except to screw them in or out a trifle if the balance is found to be out of poise. Every watchmaker should by all means have a good poising tool and test each balance for poise as this will have an effect not only on the timing but also on the pendant positions in adjusting.


A balance heavier on one segment than on the other will not perform at the same rate in the pendant positions. Screwing the timing screws out will slow the rate; screwing them in will reduce the practical size of the balance by bringing that much weight nearer the center and hence the balance will run faster.


With a well-poised balance properly adjusted for temperature at the factory, and not monkeyed with afterward, it is possible to bring the watch to time, solely by means of the timing screws with the regulator in the center and this is the method followed in the factories.


Many fine watches and pocket and marine chronometers have no regulator.  The manufacturer brings the watch to time in the factories by timing screws, and leaves off the regulator deliberately, so that the purchaser cannot change the rate. If the buyer finds the watch is losing or gaining he must then take it to a jeweler who knows what he is doing . . . one who can change the rate with a slight touch to the timing screws.


When a watch is known to lose time in heat and to gain in cold, or vice versa, the first thing to do is to compare it with a new watch from the same factory and see whether some watchmaker has not changed the screws without knowing what he was doing. If the competent watchmaker handles many watches of the same kind, insensibly he becomes familiar with the position of the screws in the balance and will know at once whether they are where they should be. If not, a comparison with a new watch of similar make will soon put him right and save a lot of time. 


Watch Running Too Fast? Too Slow?

Does your watch run much too fast or much to slow?

Whether new or vintage, railroad grade or something more common, mechanical watches rely on the period of the balance wheel to allow the escapement to advance the mechanism at a regular pace. When a watch is designed, and then built, a hairspring is used that will have a period appropriate to the watch, given the mass of the balance wheel.

The most common rate of a vintage mechanical watch is 18,000 beats per hour, although there are other rates. The watch's rate is the ideal period of the balance wheel, which is essentially a pendulum whose period is dictated by the length of the hairspring, and the mass of the balance. In most vintage watches, the length of the spring can be controlled by a regulator. The regulator turns around the center of the balance and moves two regulator pins along the beginning stretch of the hair spring. The spring passes between the two pins. Moving these pins makes the hairspring effectively longer or shorter and thus changes its period.

In most watches, the full range of the regulator will not result in a change of more than two or three minutes a day, give or take.

Mechanical watches do not have perfect rates though. If everything else is functioning well, the intent of the regulator is to evenly distribute rate errors, created by things like the watches physical orientation throughout the day, or temperature changes, so that on average the watch reads well over a period of time, such as a whole day. But it is a relatively fine adjustment.

What if a watch runs very, very fast? Like gaining a minute or more in an hour?

A very fast rate is well beyond the range of the regulator to control. It is an indication of a serious problem. The most common cause is dirt, oil or even rust, on the hairspring. This can cause the spring's loops to touch as the spring coils move in and out. One can see this happening with careful observation of the running spring. If the coils touch, at all, this simply has the effect of making the spring shorter, and thus faster - much shorter and much faster. Magnetism can also cause this same problem.

A hairspring problem is the most common cause of a watch running much too fast. There are others. For example a mainspring that is too strong can drive the escapement too fast. The escapement may be damaged causing it to sometimes slip. Or a wheel may be missing a tooth causing it to jump ahead as the bad spot comes around.

What if a watch is very, very slow, such as losing minutes an hour?

Interestingly, under normal conditions where all the parts are correct, a watch can not physically run very, very slow. Again, the period of the balance dictates the frequency at which the escapement allows the mechanism to unwind. The speed of motion during each instant that power is released is a small factor in the watch's time keeping.

If a watch is behind by an hour at the end of a day, it is almost certainly because it is actually stopping completely at some point and restarting without being noticed. A watch might do this once for a long period, or several times during the day. A huge array of factors play into problems causing a watch to stop, some of them causing just slight pauses each hour, which may be hard to notice.

The long and short of it is that vintage mechanical watches have imperfect rates, rates which also vary slightly as they run. Small rate errors can be accounted for, through the fine adjustment of the regulator. But larger errors indicate a more serious trouble. A very fast watch probably has a hairspring problem. A very slow watch is probably stopping.

You can read more about the accuracy of vintage watches here. For other questions (and answers!) check out the Q and A posts!

Learn more about antique watches here at the ElginTime website!



Now here is a simply beautiful, all natural and handmade soap, at a great price!

You'll love this soap's fennel fragrance. It reminds me of bike rides by the river in the summer time, fennel plants lining the river bank in full bloom. It's a nice, clean, outdoorsy-licorice scent (also said to be an aphrodisiac...). Colored with dried safflower, a very healing plant A touch of honey is added to this ethereal bar for soft skin.

Order some today, it makes a wonderful surprise gift! 





The Beginning of the End?

From The American Horologist magazine, December 1938...

No Watch Necessary
"The electric clock, once its accuracy is proved, will make all of this anxious watchfulness unnecessary. The special generator by which the clock is actuated produces an alternating current of 100,000 cycles a second, which is "de-multiplied" electrically to provide a 500-cycle supply for the dial.

"The generator is controlled by an oscillating quartz crystal contained in a chamber with a temperature kept constant to one-hundredth of a degree. The probable resultant margin of error is one in 100,000,000.
"

Blog Archive