What is the difference between photoconductive and photovoltaic detectors? What is the main difference between photoconductive (PC) and photovoltaic (PV) detectors? I notice that PC detectors are typically AC coupled (requiring modulation of the light source to generate a signal) while PV detectors are typically DC coupled. I imagine this has to do with different 1/f noise properties, and if that is the case then why do PC detectors have worse 1/f noise?
EDIT: I understand that PC detectors are operated in a reverse bias configuration while PV detectors are typically operated at zero bias. Both types of detectors are typically amplified. However, the reverse biasing appears to lead to substantially worse noise at low frequencies in the PC case, and I am wondering what the physical mechanism is that causes that.
References are much appreciated.
 A: I think you are quite confused. In short:


*

*Photoconductive (PC) = is connected to a power supply.

*Photovoltaic (PV) = is NOT connected to any power supply.


PV means connecting the sensor directly to the meter. For example, a photodiode directly connected to the amperimeter, nothing else. Usually we change the amperimeter for a resistance, in which we measure the tension drop (it is equivalent).
Hoever, in PC, there IS a power supply in the circuit. For example, a source connected to the negative of the photodiode, and then the positive plug of the diode connected to the resistence in which we measure the voltage drop.
We do this because PC is linear and PV is not, and besides, PC is much more responsive than PV (in fact PV's response is usually needed to be amplified).
A: It's not clear what you think photoconductive means.  But, it seems you are referring to sensors like photoresistors and photodiodes.
The main difference is that in photovoltaic sensors, the output signal is a direct conversion of the incoming light.  In other types of light sensors, the light modifies something, and that something is then measured.
Neither type of sensor is inherently better or worse, because there are different criteria for good and bad in different applications.
For example, old photographic light meters (before light measurement was built into cameras) often worked by coupling a photocell directly to a meter, with a calibrated resistance adjusted by a dial.  These had the advantage that they always worked, and never needed batteries.  The meter was powered directly by the received light that was being measured.
A photoresistor, by contrast, is a light-variable resistor.  More light usually means lower resistance.  However, it takes a current or voltage to measure a resistance.  The power for that has to come from somewhere else.
Photodiodes effectively "leak" current proportional to light.  It takes external power to get a signal out of a photodiode.  Since the output is then a current, it can be fed into a transconductance amplifier.  Since the input impedance of a transconductance amplifier is ideally 0, there is no issue of charging and discharging the inevitable parasitic capacitance slowing down the signal.
So, both types of sensors have their place and are routinely used in different applications.  Properties like noise, dynamic range, and response speed differ.  Some come directly from the sensor.  Others are more attributes of the circuit that handles the sensor signal and converts it to a signal that is actually usable directly by the remainder of the system.
A: As you point out Photoconductive (PC) photodiodes are reverse biased, while Photovoltaic (PV) diodes are not. The reverse bias increases the size of the depletion region (in a semiconductor junction) and thus decreases the junction capacitance. This means that PC photodiodes have a larger bandwidth (i.e. a faster response time). But the reverse bias causes more dark current to flow and thus sets a higher noise floor. For precision DC sensing photovoltaic (non-biased) detectors are preferred: see precision photodiodes. A more general reference, covering all photo-diode basics (including PV vs PC) can be found from TU Delft (pdf link). Another, shorter, tutorial can be found from everyone's favorite optics store.
To appreciate the difference it may help to start with an IV characteristic (from the TU delft link):

A reverse bias means operating the device to the left of the x-axis. Note that changing optical power is represented by moving between the three curves and not along them. We can see that when reverse biasing the photodiode current is non-zero even for zero optical power. This is the dark current. But we can also see that reverse biasing extends the linear range of the diode (improving the "dynamic range"). 
We cannot see the change in bandwidth directly from the IV curve, but as noted above it is another beneficial consequence of the reverse biasing. A simple intuition for the change in bandwidth is as follows: Applying reverse bias moves the charge carriers away from the P-N junction, creating a higher barrier and widening the depletion zone. A wider depletion zone has a smaller capacitance (recall that for a parallel plate capacitor $C\propto 1/d$). And, finally, having a smaller capacitance implies a higher LC resonant frequency ($\omega_0=1/\sqrt{L C}$) and thus a faster time response.
Both PC and PV diodes have a 1/f noise spectrum and can benefit from a chopper and lock-in measurement (see perkin elmer application note, pdf link). The chopper makes an turns a continuous light source into a periodic one and then the measurement "locks-in" with a homodyne measurement at the (known) chopping frequency. This lets you move down the 1/f noise curve to get better precision.
You also seem to be a little mistaken about AC vs DC coupling. AC coupling refers to a photodiode arrangement with a capacitor that acts as a low-pass filter in order to remove the DC offset (e.g. from reverse biasing). See AC vs DC Coupling.
