I would like to ask a basic question, on a standard watch, if you listen you can hear the watch running, on a correctly and accurately adjusted (IE keeping good time) watch are the time periods between the audible ticks the same for say all American pallet type escapements? or do they vary with different watch sizes and gearing ratios, being a beginner please forgive my basic questions, but if I dont ask I will never learn.

Chris Abell

No, they can be different. For example, older watches, such as an 1860 vintage, Waltham m1857, has a slow-train and a slower tick, compared with American rail-road watches of say 1920 which have a quick-train = faster ticking. Open a couple of these watches and you can see the difference.

There are other variations as well in sound as well. A good watchmaker can tell plenty buy listening to a watch in various positions on the ear prior to opening the watch for inspection.

Best regards,
Greg

“…can tell plenty by listening to a watch…”

At marts (I’ve never been to one) are most watches that run running so buyers can listen? If not are you generally welcome to wind and listen?

Stu

So if we took for example the quick trains, of the 1900 onwards would these be all the same audible timing ratio, would this be consistent pattern or would each maker having there own, basically using time intervals would it be possible set a watch from a audible source for a given make of watch?

Chris

Not counting "dollar watches," probably at least 98% of American pocket watches are geared to run at the same rate, which is 18,000 beats per hour; that's 300 ticks per minute, or 5 ticks per second. Some "dollar watches" and a few of the really old key-winds run at 14,400 beats per hour, which is 4 ticks per minute, or 16,200 beats per hour, which is 4.5 ticks per minute, but those are the exception rather than the rule.

Watchmakers "timing machines" or "rate recorders" utilize small microphones, which amplify the ticking of the watch, and compare it to a highly accurate time standard within the machine. The "beat" of a watch is determined by the difference in the interval between the "tick" and "tock" sounds of the escapement, which ideally, should be equal. If a watch produces a "tick-tock --- tick-tock --- tick-tock" sound, it's out of beat. The sound should be a perfectly rhythmic: "tick -- tock -- tick -- tock -- tick -- tock," with the ticks and tocks occurring at perfectly equal intervals.

Many (if not most) vendors at Marts have their watches wound, set and running, and I've never heard of anyone getting upset at anyone for wanting to wind one. Most people at mart events are "professionals," and they typically extend a certain degree of "professional courtesy" to each other, which might not be the norm at an antique shop or flea market.

==========================

Steve Maddox
Past President, NAWCC Chapter #62
North Little Rock, Arkansas
IHC Charter Member 49

along with listening to the tick-tock, one also listens for the sound of the hairspring and listening for the 'fall' of the balance as the endshake goes from side to side....

hold the watch to your ear, tilt head side to side..

yep, you get stares when doing this...

OK, I have a curious question about the sound of the tics and tocs as I listen to various watches in my collection. Some have an almost ringing sound as when a good quality anvil is struck with a hammer, and some just make the clicks and tick without the resonance. Any idea why the difference from watch to watch.

Also, is it the pallet stone stopping the escape wheel that makes the tics and tocs or is it the pallet striking the banking pin ...or both??

Thanks,

Bruce

A multitude of different sounds are made by the escapement in a watch. Obviously, the loudest sound is normally made by the escape wheel striking the pallet stones as they interrupt its advance, but other significant sounds are made by the pallet striking the banking pins, and the roller jewel interacting with the pallet fork.

In my experiences, the watches that tend to "ring" most are those with lively balance motion, and unusually stiff hairsprings. In such watches, the hairsprings resonate the escapement sounds, much like the striking of a miniature gong. A "rake" across such spring with a fine broach will easily reveal it's potential for "ringing," although that isn't something I'd recommend for anyone other than experienced watchmakers. I'd expect that the balance screws would largely deaden any potential "ringing" of a balance wheel itself, but I might be wrong.

====================

SM

Thanks for all the inforamtion above, where and how much would a rate receoreder cost.

Chris

Chris... I recently purchase an old Watchmaster G47 on Ebay with the top bid of about $30 and it came with a nice pickup that rotates to all positions, the recorder in working condition and all the schematics for the unit. Shipping was high...($30), but it's a really nice old unit. There's also a downloadable program called "WTM" that only requires a microphone plugged into your computer's sound card to operate and the software program seems pretty nice for timing your watch. It's a free trial download.
I've seen other timing units on Ebay going from $30 on up to several hundred depending on it's age, make and condition.

Here's the URL for Window Timing Machine Version 2.

I haven’t tried it – would appreciate comments from those who have.

Stu

The price of used equipment can and does vary widely. A friend in North Carolina recently purchased a nice L&R "Tic-O-Print" for $10.00, but that was just a fluke; "Tic-O-Print" machines in good working order typically sell in the $300 range, and they're reasonably good machines (actually produced by Bandelin, of West Germany).

The various "Watchmaster" machines are all at least 40 years old, and while many would argue the point, I consider them to be essentially obsolete by today's standards. I have three or four old "G-11" and "G-47" models, as well as a nice "G-57," but in my estimation, they now have more "historical" value, than practical.

The best timing machines in the industry for the last 40 years have been produced by Vibrograf, of Switzerland. The oldest ones with which I'm familiar are the B-100 models, which are gray, tube-type machines that are fairly practical even today. Nice, working B-100 models should run about $200 to $300.

The B-100 models were followed by the B-200, which are two-tone blue in color, and posses the advantage of all solid state components. They were produced for a number of years, and the electronic components progressed from transistors to IC chips over the years, but the features and external appearance remained essentially unchanged. Today, nice, working B-200 models can typically be found in the $400 to $800 range.

Vibrograf's latest and greatest is the B-300 model, which is the "stealth bomber" of timing machines. It will do things that no other timing machine has ever done before, and it combines the best qualities of both digital and analog machines in one unit. Unfortunately, the retail price is nearly $5k, and they're so rare that it's virtually impossible to find a used one.

Witschi is a relative latecomer to the timing machine business, and I really don't have any experience with them, but a lot of people really like them. They're MUCH cheaper than contemporary Vibrograf models, but that economy comes at the expense of features, and some would say a certain degree of quality.

I've written on the subject of timing machines before, but I can't remember where. Perhaps a "search" of the old topics here, or on the "green board," might reveal additional info, as well as alternative experiences and opinions.

====================

SM

I have been playing with the software and have managed to get it to record using my computer microphone enough to get a clear trace, but so far have found it very random. I have tried several watches known accurate ones, ones running fast and ones slow, the readings so far have been ramdom to say the least, however looks a great idea, anyone had any luck with this?. I may try building a hard cover to shield outside noise a bit more and see if that improves things, is there anymore software out there to try?.

Chris

I'm doubtful about the potential of computerized "timing machine" programs, because they rely upon the computer's internal time base as a standard. Depending on what I'm doing with my computer on any given day, its internal clock can be relatively accurate, or it can lose as much as five or ten minutes. While that might be unusually bad, everyone I know who's paid close attention to the subject, says that the clocks on their computers are among the worst timekeepers they own.

Older analog timing machines are difficult to read with accuracy much greater than one second per day, but that's essentially within the range of timekeeping potential for mechanical timepieces anyway. Newer digital machines, however, such as the Vibrograf B-300, are capable of accurately recording the rates of watches to less than 1/100 of a second per day, which is a rate most home computers can't come close to.

Needless to say, in order to get meaningful information from a timing machine, the machine itself has to be accurate, and any deficiency therein will cause varied results, such as those reported above by Chris.

===================

SM

My computer clock was horrible like Steve describes but I found an option in my preferences to hook it to a "network time" and its dead on. It has not varried more tham one second from an atomic clock since. Kind of cool.

BTW I wse a Mac and can help anyone who does to set this up. It is really easy.

Aaron
Casual junk collector

The clock built into a computer and a "time base" used on other software programs does not use the same control. While the cmos clock in your computer can drift and is also dependant on the state of charge on the motherboard's storage battery a program running certain clocking speeds should not vary. Even Microsoft admits computer clock problems with certain versions of Windows, but software programs with timing circuits seldom use the cmos clock as a timebase. I wouldn't have too many reservations about using timing software that wasn't tied to the computer's clock. Just my opinion though....

.For what we are doing sampling a relatively slow pulse over a very short period of time the loss of a few seconds/minutes per day would have such a minute effect on this software would be negligible.

Chris

Well...once again I'll have to admit my stupidity....I'm having trouble trying to figure out what you folks are trying to accomplish....are you trying to vibrate a haispring...or...just attempting to check the accuracy of a watch...all ready assembled.... in a short period of time???....here's what I do...see if this makes any sense...after assembly...I set the regulater in the middle...fully wind the watch...and dial up....I compare the seconds hand activity to an old cheap digital stop watch I bought for about seven bucks....this stop watch is frequently checked against my computer clock..which is automatically updated against an atomic clock...the stop watch is usually...right on....PW and stop watch run side by side for about 5-10 minutes...by comparing the seconds hands to the split second timer....I can quickly get a good idea as to whether my repaired watch is going to run fast or slow...and...I can calulate how fast or slow it will be over a 24 hour period....am I on the right track here???....or....do I need to start shopping for a $500.00 TIMING MACHINE?,,,Tony C

What you are doing I'm sure is just fine Tony...I've talked with others who do it in a similar way. I think we're just trying to play with new toys! I've recorded two wav files of a 992B and an old Elgin and uploaded them to my site. If anyone wants to play with the WTM program this would let you do it and show you how it works.
I used the mike from the G47 and also used the amplifier in it. I set the WTM program mike level at 91 and it works fine...

Hi Bill, your recordings are cooool!, very clear and interesting how different the Elgin an 992B sounds are. Thanks for sharing.

[This message was edited by Carlos Flores on December 21, 2003 at 11:52.]

I like the Elgin better.

Aaron
Casual junk collector

I usually refrain from writing about topics I don't know much about, and I probably should do that in this instance, but from my limited knowledge of the subject, it seems to me that no computer, nor any computer program, could be its own timekeeping standard.

Anything that keeps time accurately has to rely upon some sort of known, stable standard. With mechanical watches, the standard is created by the oscillations of the balance wheel; with pendulum clocks it's created by the cycling of the pendulum; with quartz clocks and watches, it's created by the regular frequency at which the quartz crystal oscillates, etc.

It doesn't seem to me that a computer could keep time without either an internal, or an external time standard. Computers could access an external standard via a network, or through the Internet, but short of that, one would be left to its own devices, and I strongly suspect that at least one quartz oscillator "clock" is built into every computer specifically for that purpose. The computer's processor would not be able to maintain a constant rate by itself because its load at any given instant affects all its processes.

It's true that an approximation of rate can be determined by comparing a timepiece of unknown accuracy with one of known accuracy, but the degree of precision that will result is directly linked to the duration of the trial. During an interval of 10 minutes, ordinary 18,000 bph watches are supposed to tick exactly 3,000 times. Since the smallest increment that can be measured either audibly or visually without mechanical assistance is 1/5 of a second, the rate can only be judged accurately to 1 part in 3,000. While that might seem highly accurate, it really isn't; a timekeeping error of 1 part in 3,000 is 6 ticks per hour, or 144 ticks per day. Since ordinary watches tick 5 times per second, the limit to which the rate of a watch can be estimated over the course of a 10 minute comparison, is 28.8 seconds per day, and that's assuming the person making the comparison is capable of accurately discerning the necessary 0.2 second increments.

Those who've never experimented with a chronograph or a stopwatch should really spend some time with one, and see just how difficult it is to stop one with accuracy of 1/5 of a second. Start the watch at any given time, and while directly observing the sweep second hand, try to stop it exactly on the 10 second mark without any deviation. It's possible, but it isn't very easy, and for every 1/5 second error, the degree of error over a 10 second time trial is increased by nearly half a minute per day.

The idea that the brief duration of time required for a rate check on a watch somehow allows for an added measure of "tolerance," is exactly opposite to the true nature of the subject. In order to measure a rate accurately in a brief period of time, the tolerances are highly critical. A watch that runs with 99.9% accuracy, could gain or lose 86.4 seconds per day, which is nearly 1.5 minutes. To achieve accuracy within 10 seconds per day, a watch has to run with an accuracy of 99.988%.

To put that into more easily understandable terms, consider a person who's attempting to walk a distance of exactly 2 miles (10,560 feet). To accomplish that with the same degree of precision as a watch that runs accurately within 10 seconds per day, the 2-mile walk would have to be accurate to within 15.2 inches! For a watch that runs with an accuracy of 1 minute per month, the same math reveals that a 10-mile hike would have to be accomplished within an accuracy of 14.6 inches, or 99.9976852 %!!!


As for timing machines, in addition to merely showing rate errors instantly and accurately, they also visually display the beat of a watch, and the better quality modern machines also show the number of degrees of balance motion (in each direction from rest). The recordings below were made on my Vibrograf B-300, and they show the "before" and "after" records produced by a Rolex I repaired not long ago. "A" is the degrees of balance motion, "R" is the beat deviation in tenths of a millisecond (1/10,000 of a second), and "M" is the mean rate in seconds per day. Each of the recordings was produced during a trial of 1 minute.

H Steve....I'm impressed with your tape graph. I would dearly love to be able to afford such a fine machine for my hobby, but cannot...I have to make do with what I can afford or build. It surely seems like a great diagnostic aid though.
As to a computer not having an accurate timebase.....poppycock! Every motherboard made has multiple crystals on board, each of which could and/or does provide an exceedingly accurate timebase which can be frequency divided or multiplied into any timebase needed for a specific application...and it'll be dead on. The one I'm holding now has three.....the most common being a 32.768 mhz. I believe this frequency is common to all motherboards. If you can trust the quartz in a watch then you can surely trust the same type crystal in a computer...the only difference being that the computer has a checks and balance system that will either alert you when it is off frequency or simply not function. A watch with an off frequency crystal will continue to run, but provide bad timing. At 32+ million cycles per second I'd say that the crystal frequency could be off quite a few cycles and you'd never see any difference at all! I'd trust the computer first any day of the week..
BTW...The atomic clock uses a crystal oscillator too!

[This message was edited by Bill Cobb on December 23, 2003 at 8:33.]

Hi Bill,

Watchmaking isn't just a hobby for me; I do it professionally, and as such, I've found that good tools are conducive to good work. I'm sure that's why most professional mechanics have chests full of Snap-On and Mac tools, rather than attempting to rely upon the sort of tools one would expect to find in the shop of a typical do-it-yourselfer.

As for the accuracy of computer clocks, if we know that a typical quartz oscillator in a mother board operates at 32.768 MHz, there's some math we can apply to that. I recall from my days in chemistry and physics that whenever a number is recorded scientifically, the last digit is always considered "questionable." In other words, there could be a fractional deviation in that number due to "rounding errors" in subsequent digits.

That being the case, the actual number of oscillations produced by a 32.768 MHz crystal could vary from as little as 32.7675 MHz, to as much as 32.7684 MHz , both of which still round to 32.768 MHz, but allow a "tolerance" of 900 Hz, which is 900 parts in 32,768,000, or a potential accuracy of only 99.9973%. A clock or watch that runs with 99.9973% accuracy, is allowed a tolerance of 2.3328 seconds per day, which is 69.984 seconds per month, or 14 minutes per year. Needless to say, even cheap quartz clocks and watches usually keep better time than that.

In any event, if computers have such highly accurate clocks with multiple checks and balances, why are the clocks in most of them such lousy timekeepers? I use "AtomTime" to adjust mine every couple of weeks, and after running it just now, it reported that the clock on my computer was 2 minutes and 46 seconds slow. I can't recall exactly when I adjusted it last, but it wasn't more than three weeks ago. The sad thing is, it might be another 2 minutes slow tomorrow, or it might take it three months to lose that much time again, depending on what sort of loads I've been running on the processor.

As for atomic clocks, they do NOT use quartz crystal oscillators as their time standards. They operate by measuring the stable frequencies of electromagnetic waves produced or absorbed by hydrogen or cesium atoms, which are accurate to about 1 second in a million years. In 1958, scientists throughout the world adopted the vibration rate of atomic particles as the universal standard for defining units of time.

====================

SM

Great thread! Would any of you find gents who possess the instructions for a G47 or G57 be willing to copy them for me? I see lots of the older model instructions floating around but virtually none for the 47/57 models. Name your price. Thanks.
B

Just why did I know there would be a response to this?lol
I never said that an atomic clock used a crystal oscillator as a "time standard". I said it used a crystal oscillator...that's all. Without it the atomic clock would not exist.
Here's a cut and paste.....perhaps they're wrong too...
Merry Christmas!

quote:

---

---

Cesium 133 is the element most commonly chosen for atomic clocks.

To turn the cesium atomic resonance into an atomic clock, it is necessary to measure one of its transition or resonant frequencies accurately. This is normally done by locking a crystal oscillator to the principal microwave resonance of the cesium atom. This signal is in the microwave range of the radio spectrum, and just happens to be at the same sort of frequency as direct broadcast satellite signals. Engineers understand how to build equipment in this area of the spectrum in great detail.

To create a clock, cesium is first heated so that atoms boil off and pass down a tube maintained at a high vacuum. First they pass through a magnetic field that selects atoms of the right energy state; then they pass through an intense microwave field. The frequency of the microwave energy sweeps backward and forward within a narrow range of frequencies, so that at some point in each cycle it crosses the frequency of exactly 9,192,631,770 Hertz (Hz, or cycles per second). The range of the microwave generator is already close to this exact frequency, as it comes from an accurate crystal oscillator. When a cesium atom receives microwave energy at exactly the right frequency, it changes its energy state.

At the far end of the tube, another magnetic field separates out the atoms that have changed their energy state if the microwave field was at exactly the correct frequency. A detector at the end of the tube gives an output proportional to the number of cesium atoms striking it, and therefore peaks in output when the microwave frequency is exactly correct. This peak is then used to make the slight correction necessary to bring the crystal oscillator and hence the microwave field exactly on frequency. This locked frequency is then divided by 9,192,631,770 to give the familiar one pulse per second required by the real world."

[This message was edited by Bill Cobb on December 23, 2003 at 20:42.]

Brian....All I have for my G47 is the schematic. Perhaps Steve can print them out for both of us...I need it too!

This is the type of crystal used in standard PC systems. It came out in 1998 so I feel comfortable in thinking that most PC systems have it incorporated.

DALLAS, TEXAS-October 12, 1998-Dallas Semiconductor today announced an oscillator that meets the precision timekeeping requirements of large computer networks, financial transaction processing applications, and timed-access communications. A temperature-compensated crystal oscillator (TCXO), the DS32KHz 32.768 KHz TCXO meets the 1998 European network server recommendation for real-time clock (RTC) accuracy.

The new device provides accuracy as great as ±1 minute per year (±2 parts per million) in operation from 0 to 40°C, making it the industry's most accurate 32.768-kHz oscillator. Economical quartz crystals provide the time references in real-time clock circuits for computers and many other electronic systems

[This message was edited by Bill Cobb on December 23, 2003 at 21:11.]

Now let's see....my Ticks are good guys.....my Tocks are bad guys....my Computer's clock can't be trusted....the watches I fix never keep accurate time...the experts can't agree on how time can be accurately maintained...Soooo....Oh well...tine is just an illusion anyway...what a way to end 2003....happy holiday.

Certainly no expert here...just someone who trusts circuits that I've been using for years. Just because a device has a rating printed on it of 32.768 doesn't mean that the number in reality stops there. It might have another 10 zeros after it, but that would be difficult to print on a crystal! Steves math theory holds no water. The accuracy of the oscillator is dependant on the manufacturing tolerance of the device itself. Many of them are now within 1.0ppm or less.
I'm outta this thread before it goes to contest again!..I've said my piece
Happy holidays everyone!

[This message was edited by Bill Cobb on December 23, 2003 at 22:51.]

Hey Bill,

We are watchin for do-do! Just kidding!


Big sister is watchin though!

Stephanie O'Neil
NAWCC Member 143979

Hi Steph! The boots are on...still cleaning them from yesterday!

Well, we've discussed a number of things in this thread, and explained a number of complex ideas, but the one thing no one has yet explained is why the clocks in most PCs (including mine) keep lousy time, and how a PC that can't keep accurate time on it's OWN clock, is going to help someone accurately judge the rate of anything else. Perhaps an explanation of that is still forthcoming, and if so, I'll be greatly interested.

As for Bill's suggestion that it "would be difficult to print on a crystal" a number longer than 5 digits, to use one of his previous expressions from above, I think that's "poppycock." My guess is that the part number probably has more digits than that, but that's just an educated guess.....

In any event, I do have an original copy of the "Operating Instructions for the Watch Master" "Model G-57 Type T Watch-Rate Recorder [and] Model 386 Watch Holder and Hairspring Vibrator." They're in the form of a booklet measuring 8.5 by 11 inches, which contains 22 printed pages, plus the outside covers. One of the pages is a schematic (dated 24 Sept. 1956), and there is included within the booklet a hand-drawn tube chart, which was presumably created by the original owner.

Unfortunately, I have neither the time, resources, nor inclination to copy the booklet myself, but I'll be more than happy to make it available to either Brian or Bill, whichever wants to accept responsibility for it's safe return to me, as well as for copying it and providing copies (at cost) to any of our fellow Chapter #185 members who may need them. If either Brian or Bill is interested, please e-mail me privately to provide your current mailing address.

----------------

For those interested in such things, about 14 hours prior to this writing, I allowed a program called "AtomTime" to connect to an international time standard, and synchronize the clock on my PC with it. A subsequent connection a few moments ago, revealed that the present time shown on my PC was 1 minute and 28 seconds slow. Another check tomorrow may reveal that it's 2 minutes fast, or any other number the processor feels inclined to provide at any particular moment.

If rates like that are acceptable "standards" to everyone else here, then I suppose they're fine with me, but I don't think the majority of MY customers would be happy if I adjusted their watches by such a standard. For all practical purposes, one might just as well use a "Big Ben" alarm clock.

================

SM