Of course, I understand that you tried to get -gh/c^2.

Excerpt from document at
  http://www.savefile.com/files/2045250
referred by Dono:

BEGIN

	2. Imagine that we could transport the Harvard tower to
	a location very far from any gravitational field. At the
	time an electromagnetic pulse is emitted from the top of
	the tower, the tower floor is accelerated away from the
	direction of the light front with the acceleration g in order
	to generate the equivalent redshift. The light front encounters
	the tower floor after the time t given by the equations:

	The relativistic Doppler effect is

	fo/fe = sqrt(1-V/c)/sqrt(1+V/c)), where

	V/c ~ gh/c^2

END

When V/c is very small, the denominator sqrt(1+V/c) can be
neglected, then one is left with

fo/fe = sqrt(1-V/c) or, approximately,

fo/fe ~ 1-V/2c
      ~ 1-gh/2c^2

The corresponding shift is thus -gh/2c^2, which is different
from -gh/c^2.

Btw, you see that one doesn't need 2 pulses.

Marcel Luttgens