If you are referring to liquid crystal spatial light modulators, the three leading suppliers are Boulder Nonlinear Systems (BNS), Hamamatsu, and Holoeye. Among these, the Holoeye is the cheapest, and also has the larger number of phase pixels, by a wide margin: up to 1920x1080. Unfortunately, Holoeye SLMs flicker badly, with the phase pattern returning all the way to zero for a brief interval 50 times per second. This can be a (big) problem for some applications.

The BNS and Hamamatsu SLMs are both excellent-quality devices. They have a choice of computer interfaces, including DVI interfaces for connecting to video cards, and dedicated PCI interfaces for higher-speed operation. Both have excellent optical properties. Both feature 512x512 arrays of phase pixels. The Hamamatsu device is more compact. The BNS device comes with integrated tilt adjustments for easy mounting and alignment.

If you are referring to acousto-optic modulators (AOMs), acousto-optic deflectors (AODs), electro-optic modulators (EOMs) or similar devices, other people are likely to be more knowledgeable. One thing to be aware of is that such devices tend to distort the wavefronts of light passing through them. This can create problems, for example, if you want to bring the modulated beam to a diffraction-limited focus.

Centellax 40G Optical Modulators are being used widely in the Fiber Optics Coomunications,

@Ghanshyam what technique is used in that?

Are you talking beam modulators such as Pockels cell or bragg cell, or spatial light modulator such as liquid crystal display or DMD? What kind of modulation frequency are you looking at Hz, kHz, GHz? Knowing the wavelength would also help.

actually when i have submitted this question , I was meaning optical modulator used in optical fiber channels.

Fiber modulators are very convenient for coarse applications: You can find very broadband devices (extending from DC up to 10 GHz range) and are plug'n play. Moreover, standard HF generators with 50 Ohms output impedence are well suited to drive them, avoiding possible electromagnetic parasites issues.

They are however not well adapted for precision/metrology experiments: They usually show high optical loss rates (> 6dB), and have a high level of Residual Amplitude Modulation (RAM, on the order of 1% in my experience). If you are interested in accurate frequency detection/stabilization applications (typically with quantum limited imprecision), you may prefer free space EOM devices, which have a much lower optical loss and RAM rates (in the 1% and 10^-4 ranges respectively). The bandwidth of these devices is however severely restricted as compared to fibered modulators, but it can be tuned by the supplier over a large range.

The last type of modulator you may find (and even in fibered version recently) are Acousto-Optic Modulators (AOM). They are limited to work below 100 MHz under typical conditions, and suppose relatively high optical loss rates in the optical mode of use, by construction. They also work only with RF amplifiers, and it may be mandatory to set appropriate electromagnetic shields in order to avoid measurement artifacts. Last, they require a careful alignement in order to optimize the sought modulation depth. I would say they have a pretty flat response from DC up to 100 MHz, and are therefore excellent testing devices.

@Pierre Verlot, I hadn't heard about fiber modulators so if you bring up a reference i would appreciate.

actually 100Mhz bandwidth is pretty old fashion in todays multi 10 giga bit tranmission! i have some background in EOM in the past, however i wonder if it is still the best choice for extremely high data rate wired transmission ? what technology in EOM is being used and why its bandwidth is lower than fiber modulator?

EOM can be used in GHz range by use of laser controlled modulators and as we know in EOM the modulation of phase, freq, amplitude, polarized beam is possible

@Anji, what kind of receiver ( i.e. down-converter/ phase demodulator or frequency demodulator) can be used in receiver?