I thought this was good. You can guess the gist of the previous 'advice' on calibrating a stopwatch....

---------snippo---------

Date: Thu, 4 Feb 1999 09:00:00 -0500
From: "Klouse Kenneth P." KlouseKP MSHA.gov
To: Greg Gogates
Subject: RE: Re1: Stop Watch Cal (fwd)

Unless I'm missing something, anyone who spends more than 15 minutes calibrating a stop watch, or claims accuracies of .001%, better smell what their peddling, get a life, or start measuring the reflexes of the stop watch operators. I'm not saying that stop watches so not need to be calibrated, I'm saying that one should look at the accuracy needed for the measurement, and calibrate the instrument accordingly. As Jeff Gust so eloquently put it, " As you would note, the random error from the person dominates the uncertainty budget, and in fact is 41 times greater than that of the unit under test tolerances. " If that person hasn't had his coffee that morning or he didn't get enough sleep, his reaction time could be all over the map.

As an assessor of product certification schemes, it is enlightening to read responses from different people on these varied subjects. But sometimes I think we have to step back and see the practicality of what we are doing or requiring.

I sorta figured this also could fit here. Another reality check. What if there is no 'standard'?

----------snippo---------

Date: Sat, 20 Feb 1999 05:54:03 EST
From: Rcveale
To: iso25
Subject: Re: Linear Displacement Potentiometer

Pierre,

There are no American nor ISO standards for linear displacement potentiometers. There might me a military standard that I am not aware of.

A group was formed several yars ago to write a standard for linear transducers, and a draft was written; however, it was never finished nor published as a standard. A couple of years ago there was an attempt to revive the Group, but there appeared to be little interest or need for such a standard.

If a need can be shown to exist and someone is willing to put together a group to do the work, there is a procedure in place for producing the standard.

Ralph Veale, Chairman
ASME B89.1 Subcommittee for Length

Ah yes...stopwatches and timers:notme:

We have 8 timers and 4 stopwatches whose calibrations are due shortly. In the past we have just tossed them into the garbage and replaced them. During our last audit, the auditor asked why we don't just use the internet to calibrate them. OK...I'll bite:rolleyes:

So, I'm looking at the calibration certificates that came with them. It states:

The maximum error of this Digital Timer at the time of calibration did not exceed 0.01%.

As measured: +/- 0.367 seconds per 24hrs.

Expanded measurement uncertainty at K=2: +/- 0.13 seconds per 24hrs.

So then I went to thise site: http://www.time.gov/timezone.cgi?Central/d/-6/java

I assume this is the site the auditor meant. So now what? Check each one against "The Clock"? For how long? How would one document the results? How could one prove this was actually performed without a certificate?

;)

Probably there won't be a certificate. But you could accept World Time as a "Naturally Accepted Standard".

Verifying stopwatches is an interesting concept. One thing I have seen is a test rig. This "starts" the timers and the standard timer at the same time. It does minimize some of the human error.

BTW... Has anyone ever seen a stopwatch be wrong ? :D

Attached is NIST's procedure calibrating stopwatches/timers - it's is freely available from them. I picked up a copy at the Measurement Museum during my last visit there. I doubt anyone could reasonably question the validity of this. Hope this helps!

Thank you for the NIST document! I'll try to read it in the next couple of days.

If you try and calibrate a timer using the internet wouldn't you have to consider the speed of the internet connection as well as the computer speed?
How can you verify what you are seeing on the internet is really accurate?

You are welcome Brad.
Most of the cal labs I have worked in over the years had plenty of homegrown cal procedures for these devices, some of which were pretty good and others that were pretty hokey. If I remember correctly, our uncertainty for this measurement was like 0.3 or 0.4 seconds (K=2), mainly due to the human reflex component. Some procedures had electronic pick-ups attached to the stopwatch circuitry - gating electronic counters. I never failed an electronic stopwatch. There was an old mechanical timer I failed once..."back in the day".
When I heard that NIST published this one, I encouraged the large commercial calibration company I worked for at the time to adopt it. All the calibration techs really liked it.

If you try and calibrate a timer using the internet wouldn't you have to consider the speed of the internet connection as well as the computer speed?
How can you verify what you are seeing on the internet is really accurate?

The NIST procedure I attached earlier addresses your question nicely...

The NIST guide provided has a lot of good information, and is written by well known Metrology professionals.

I have assessed timer calibrations using GPS-sync'd time bases and that often results in MU under 0.1 seconds.

Hershal

Time study departments not using calibrated stop watches should have their rates be declared invalid! If you ever dealt with them, this could be very funny!

Ok, after reading through the Guide I tried the Direct Comparison Method on a couple of the stop watches. As per the NIST recommendation I used the 10800 second interval. This was selected to to keep the uncertainty of human reaction time (me) insignificant and to exceed the 0.02% requirement recommended.

On the first stop watch the time after completion was 10800.02 s. So for a
24hr period I can assume the time would be 86400.16 or +0.16 s / 24hrs which is well within tolerance.

My question is how many 3hr cycles are needed or recommended to assure the item is calibrated? Would one cycle be enough or would you average multiple cycles?

Since you are gaining familiarization with this procedure, I would recommend that you repeat it a few times to gain confidence with it, just to verify that your uncertainty window is what you estimated. All your measurands should reside within your uncertainty value. Once you are convinced that you have a solid calibration method, then one cycle per DUT should suffice.
It has been my experience that one cycle is enough, unless there is a special requirement requested by the user (customer). I don't think the NIST procedure specifies more than once, unless I missed something.

[May]All your measurands should reside within your uncertainty value...

Pardon me for interrupting an interesting discussion, but I'd like to suggest that the snip above be accepted as the Official Metrology Prayer.:D

Thanks for the reply skappesser. I was thinking 3 per timer and reporting the average.

And as you said, just to verify that your uncertainty window is what you estimated.

Hershal - I agree about the NIST procedure. Have you ever been to their Measurement Museum? It is so cool. If you are ever in the Maryland area on an assessment, you should give them a call to see if you can gain access. Their security is tight - not sure if you still need a sponsor to get on the campus these days.

I like to suggest that the GPS frequency standard systems are a great money-saver for organizations that have a lot of basic time/frequency test and measuring equipment. A good used GPS system can be bought for like $5k, and can be used in-house to cal / verify almost anything with a time base.

All our electronic counters and synthesizers at our facility source a GPS system as the main time base. Saves a lot in outside calibration costs, at the same time maintaining low uncertainties as a high quality primary frequency standard. Sure is easier and more affordable than a Cesium or Rubidium system...

I love it Jim! How about an official metrology joke de jour?

Q: How many metrologists does it take to replace a light bulb?

A: None, they make a calibration technician do that...!

HEEEE-HAWWWWWWWWWW!
(Lord, I apologize for that...never met a cal tech I didn't like!)

My favorite is from my days as a bench tech (yes, I really was one.....), which went:

We do precision guesswork.

Course that was the good ol' days before MU.....

Hershal

We have 8 timers and 4 stopwatches whose calibrations are due shortly. In the past we have just tossed them into the garbage and replaced them. During our last audit, the auditor asked why we don't just use the internet to calibrate them. OK...I'll bite:rolleyes:


I have a customer who has found it cheaper to buy a new NIST traceable stopwatch rather than have the previous one calibrated at a lab. However by not having the previous stopwatch calibrated at a lab, he does not know if the stopwatch was previously out of cal at anytime since the original calibration.

Any comments on solving his dilemma?

Looks like it is better to review the NIST calibration procedure posted in this thread and calibrate it yourself. This is still probably more expensive though than just buying an inexpensive (aka cheap) new one.

I have a customer who has found it cheaper to buy a new NIST traceable stopwatch rather than have the previous one calibrated at a lab. However by not having the previous stopwatch calibrated at a lab, he does not know if the stopwatch was previously out of cal at anytime since the original calibration.

Any comments on solving his dilemma?

Looks like it is better to review the NIST calibration procedure posted in this thread and calibrate it yourself. This is still probably more expensive though than just buying an inexpensive (aka cheap) new one.

Good morning!

I don't understand how they can justify buying new ones, when all you need is a telephone to verify the existing timers. I can do five or six timers (not at the same time; I stagger them by 1 minute) in an afternoon.
Now, there is the caveat of the process and how accurate they need the timers to be. I usually check to a process requiring +/-3 seconds or more. If it is a tight tolerance, there may need to be a different procedure.

The cost of buying new ones, purchase orders, shipping charges, getting new ID numbers, etc. and then disposing, seems to not be a total cost not to be taken lightly at every interval.

Also, how good is the calibration they get with the new one? Does it give them true confidence?

Too, timers generally seem pretty reliable. You may can extend the interval a bit.

Aren't you supposed to calibrate any measurement equipment across its full measurement range? That could take some time! :lol:

Aren't you supposed to calibrate any measurement equipment across its full measurement range? That could take some time! :lol:

Well.. true. All of our timers/stopwatches are in perpetual calibration. The users get mad because the equipment is not available for use; but not my problem!! I let them know that the 5 year time verification is just about done!:tg::lmao:

Just a thought, but one could just open the stop watch, and make a measurement of the actual crystal timebase. 32,768 Hz if I remeber correctly in most cases. This method eliminates the repeatabilty of the Techs reaction time, and ups the actual measuremt resolution provided you have a decent counter, or sig gen and oscope. This shudl get you the MUC needed for just about any MFR spec. Now the UC of the stop watch in use is still goign to have issues, but the cal is solid.