In article , James Waldby says...

jim rozen wrote:
... says...
My target is to bring this spread below 50 KHz.......I want some help

OK, what thickness variation does this correspond to?

Ie, if you measure two samples at the extremes of the distribution,
200 khz apart, what's the size difference?

The http://www.sentrymfg.com/crystalclear.htm ref given earlier has a
formula, about 40% into the page: Thickness = 66.2 / Frequency
(Thickness in inches, frequency in KHZ) for AT cut crystals. That
would give: 66.2 /20000 - 66.2 /20050 ~ .0000083 inches difference,
ie about 8 millionths for 50 KHz, and about 33 millionths for 200 KHz.
Near the end the page says, "a film of water one molecule thick will
lower the frequency of a 10 MHz crystal about 10 parts per million."

Note, 50 KHz at 20 MHz is 2500 parts per million. About the
cheapest crystals you can buy are 100 ppm; for a few cents more,
20 ppm is common in cheap crystals. So y2kvickyindia (Pankaj Trivedi)
must have some serious material, measuring, or process control problems,
being out of the commercial ballpark by a factor of a hundred.

That's the question. If he's measuring thicknesses that are inconsistent
with the spread in his frequencies, there's some other uncontrolled
parameter in his process.

If he would list up the extremes of thickness that correspond to the
extremes in frequency, then it would be obvious if this was a
routine exercise in process control (goal to get thickness controlled)
or if it is a snipe hunt, where the goal is to find the mystery
parameter that's pulling the frequencies, in *spite* of good statistical
process control.

Jim


--
==================================================
please reply to:
JRR(zero) at pkmfgvm4 (dot) vnet (dot) ibm (dot) com
==================================================