Peter F suggested to check the bottom surface of the PD housings if there is any protrusion/interference/whatever.
And that was true! It was found that the front side of QPD1 (Left) was lifted by a machining burr.
It seems that this burr consistently exists as the other one also have it (see QPD2 picture (right)) although it is not too terrible compared to the one in QPD1.

Once these burrs were removed, the spots were found on the right position of each diode.
From the measurement of the power on each segment, the positions of the spots were estimated. (listed in the table)
They indicate that the spots are within 0.1mm from the center. This is good enough.
The quantum efficiency was measured from the incident power and the sum output. It seems that there are
some difference between the diodes. The numbers are consistent with the measurement the other day.
QPD# QPD1 QPD2
Diode# #43 #38
-------------------------------------
Power Incident 84.7 uW 86.2 uW
Sum Out 56 mV 61 mV
Vertical Out -6.8 mV 10 mV
Horizontal Out 4.2 mV 8.8 mV
SEG1 -17 mV -15 mV
SEG2 -14.5 mV -11 mV
SEG3 -11 mV -15 mV
SEG4 -13 mV -20 mV
-------------------------------------
Spot position X +25 um +46 um (positive = more power on SEG1 and SEG4)
Spot position Y -42 um +46 um (positive = more power on SEG3 and SEG4)
-------------------------------------
Responsivity[A/W] 0.66 0.71
Q.E. 0.77 0.82
-------------------------------------
Arrangement of the segments
View from the beam
/ 2 | 1 X
|---+---|
\ 3 | 4 /
---------------
I(w,x,y) = Exp[-2 (x^2 + y^2)/w^2]/(Pi w^2/2)
(SEG_A+SEG_B-SEG_C-SEG_D)/(SEG_A+SEG_B+SEG_C+SEG_D) = Erf[sqrt(2) d/w]
d: distance of the spot from the center
w: beam width
|