Frequency accuracy

[Robert Scott]

Brian Trout asks:

>...I got a chrome plated steel tuning fork.  It's not
>very loud, but it seems to be pretty stable.....
>At work amongst the tuners, we have 3 tuning forks, and 3 "machines",
>....My question is, how does one determine which A-440 is right?  Is there
>an easy way to do that without finding an electronic frequency counter?
>I can't think of anyone who has one.  I'm just trying to figure out if
>real tuning fork accuracy is possible in a real world.  To what do we
>pay homage as the "ultimate" standard measuring device?  Who's
>"ultimate" standard measuring device is right if they don't agree??? 
>If I find one person with an SAT and one with an RTC and the A-440's 
>aren't the same, who's right?

The ultimate frequency standard is the National Institute of Standards
and Technology, formerly called the National Bureau of Standards.
This government agency maintains a group of Cesium atomic clocks in 
Fort Collins, Colorado. These clocks are accurate to better than one part
in 100000000000, or, in piano tuning terms, 0.000000017 cents.
This standard is disseminated in a number of ways.  You can check their
web site at http://www.bldrdoc.gov/timefreq if you are interested.

For practical piano tuning purposes, it is usually safe to trust the
work of Sanderson, Reyburn, etc.  (Off-brand $200 devices, or something
in the guitar-tuner class may not be that accurate.) When a SAT is
manufactured,
I am sure that its internal quartz crystal frequency reference is calibrated
to some standard that ultimately is based on the NIST source.  With the
RCT, I believe the Mac version is calibrated by Dean when he sends out
a complete package.  And when he sells the software only (such as in
a Windows version) he supplies a little A440 quartz reference so that
you can calibrate your own particular laptop.  This step is necessary
because makers of sound cards for computers do not take the trouble to
ensure that their frequency references are all that accurate.  But if
they are not accurate, at least they are stable, so once the software
learns the error through calibration, that error will remain constant and
can be compensated for.

If you would like to set up a secondary frequency standard in your
shop without any cost at all, there is a program called CalSound that
is part of the TuneLab package, available for free on my web site at

     http://www.wwnet.net/~rscott

The CalSound program allows you to calibrate the frequency reference in
your computer's sound card using time-of-day.  Most local telephone
companies offer precise time-of-day ("At the tone, the time will be....")
Or, if you really want to go to the source, you can get precise
time-of-day from NIST's shortwave radio station, WWV, on 2.5, 5, 10, 15,
and 20 MHz.  By comparing time over an 8-hour period, CalSound can 
provide a frequency accuracy of about 0.03 cents.  The longer the
calibration period, the more precise the calibration.  Once you
have set up a computer with CalSound, you can use it to check
tuning forks, SATs, RCTs, oh, and by the way, you might also want
to use it to tune a piano :-)

Robert Scott
Ann Arbor, Michigan