Hi Yuriy,

             The only problem with using mirrors for reflection is that, after multiple reflection, though the delay will be generated BUT the intensity of the outcoming light will be greatly reduced. hence, it will be rather difficult to detect such extremely low intensity fluorescence signal by the detector. I basically need to keep obtain the delay, but without decreasing intensity too much. And you had already mentioned the problems with using optical fiber.

Could you describe the light-source you want to delay (bandwidth, pulselength) and the reason to do it? Will you mix this signal optically with something?

Yuriy has made some highly relevant suggestions, but here are some related questions which may help in selecting an acceptable solution.

Is the time delay fixed, or must it be adjustable?  If so, over what range.

What is the wavelength?  Are efficient high and low reflectivity dielectrc coatings available for the wavelengths and power densities of interest? What is the beam diameter and transverse intensity profile?  How much power loss can you tolerate - none, 1%, 10%, more ??

Do you require an optical delay so that you can make some kind of optical correlation or interference measurement?

If not, do you really need a pure optical delay?  Could you achieve your aims using an electrical delay to the laser trigger or in the electrical signal from the detector?

I have an optical system, in which there are 3 wavelengths of light (fluorescence emitted by cells), all emited at the same time. But, we want to detect this using 1 single detector. So, between each wavelength of light emitted, I want to introduce fixed delay. That is, I want the three emitted light to reach the detector at 3 different times, starting from the beginning. For eg. light 1 is going to get detected after 1 ns, lighjt 2 is going to get detected after 2 ns, and light 3 after 3 ns. So we have a delay of 1 ns betwen each light. The final intensity of light should be atleast 80 - 90 % of the original intensity. Also, we have to keep in mind that, the delay should be more than the deadtime of the photodetector (commercially avialble) that we may use. Hence, this is a kind of optimisation that we need to do, so that we can choose a suitable photodetector and have suitable delays, more than the deadtime of the photodetector.

I agree with Yuriy's suggestion about multiple detectors.  A solution with 3 independent detectors could well be easier to use, more efficient,  more compact and less expensive than an optical relay providing a 15 m optical path difference.

Have you identified a means to separate the wavelengths with less than 10% loss?

If you must use a single detector, dielectric filters are a promising option.  If the selected band is transmitted the rejected wavelengths are generally reflected with high efficiency.  You will necessarily separate the signal into multiple optical paths, which can be of different length, and you will need additional prisms, filters, lenses and/or mirrors to guide the separate wavelengths to a single detector.

Is the source volume sufficiently small for you to collect the emitted fluorescence into a well-collimated beam?  What is the beam diameter and divergence?  What is your detector?  Is the detector sensitive area sufficiently large for you to collect light from 3 independent beams or will you need a second set of dielectric filters or other means to re-combine the 3 delayed beams at the detector?

Have you selected the detector and characterized its dead time?  An optical solution with 2 ns differential delay will be much easier to realise than one with 50 ns of differential delay.

With a sufficiently small fluorescence volume, use of multimode fibre delay lines might be feasible.

For example, with a 0.9 NA objective lens collecting light from a 10 micron diameter volume, it should be possible to split the collimated beam using dichroic mirrors or interference filters and direct 140 micron diameter 0.07 NA illumination onto the ends of three 200 micron diameter 0.24 NA step index fibres.

The 50% coupling efficiency suggested in Yuriy's first reply is reasonable for coupling a single transverse mode laser to a single mode fibre.  Here we are coupling from a source with a modest mode volume to a fibre supporting a much larger modal volume, and higher efficiency is possible.

With a large area photomultiplier detector, It should then be possible to couple most of the power guided by three optical fibres onto a single photocathode.

What temporal resolution do you need?  With step index fibre, differential mode delay will broaden the output pulses.  If this is a concern, reduced launch numerical aperture and avoidance of excess fibre mode coupling by bends, or use of low dispersion graded index fibre may help.

How about using a grating? The input 3 wavelengths will be diffracted and a linear CCD is used to detected the 3 wavelengths.

Mirrors,Optical fibers, photonic crystals and dispersive elements (optically) or light source simulation (digitally). Note that each one may distort beam profile,wave-front or spectral response of light! 

Fluorescence decay is usually rather long, no?

The best would be to buy special optics, which reflect with very high efficiency (>99%) one wavelength-range and transmit the other two and then repeat this again for the other two to separate them. I would then put each signal on a separate photodiode, you can also buy balanced ones to make sure that they have the same response and noise performance (keeping in mind that the response can anyway be dependent on the wavelength). You could then have a 1ns cable length difference between each output and even use a single ADC channel for all the signals, provided that the signals are short enough in time not to interfere. You can buy photodiodes with 10's of ps response time, so this should not be the showstopper. You just need a focusing element to couple into them, as they have smaller area. This would be a very compact setup.

Mirrors is still  best way. If you are using laser a reflectivity can be as good as 0.999.

According to Prof. Parvin's comment, if you need to keep the shape of the wave-front , mirror-based methods are the best.

I have had the similar problem. I would need to increase the optical length up to 1000 m without any distortion, changing beam profile shape and especially any nonlinear effects for optical output signal.  The purpose was to investigate nonlinear effects of slow response time nonlinear medium. I found the electrical delay circuit as the best way in a similar manner described by Dr. Marta Divall. Otherwise, such mechanism can be also found in literature with the keyword "delayed feedback circuit" especially to investigate nonlinear effects in photorefractive materials. 

iyt depends how long delay is needed. simple way is to use mirrors to prolong the longteh delay by having longer path as mentioned by Dr. Ali Bavil. 

Bonjour,

- The only realistic way for realizing a 50 nm optical time delay is to use a pair of mirrors. In 1976 Professor White designed a special optical cell for that purpose.

• Journal of the Optical Society of America Vol. 66, Issue 5, pp. 411-416 (1976)

- I designed a simpler device of my own for the purpose of amplifying light pulses of which I join you a copy of the paper.

- More complicated: you could trap a light pulse in a laser cavity (Fabry-Pérot cavity) using a Pockels cell and a broadband polariser to trap in and dump out the pulse. This technology has been used for some time for regenerative laser amplifiers.

• See: "Femtosecond laser pulses" 2nd edition, 2005, Springer p. 181 and references herein.

- Last, you might find valuable information along the VIRGO or LIGO projects as they are dealing with kilometre long delay lines.

You should be a little bit more specific in your question. I suppose you are dealing with light pulses. If these are nanosecond pulses you will need narrow bandwidth high reflectance multi-dielectric mirrors (99.999%), if you are dealing with 100 fs pulses the bandwidth will be so large that you will not be able to preserve the full spectrum and the pulses will suffer duration broadening. If you are dealing with high power pulses you would have to enlarge the beams in order to decrease the risk of mirrors degradation.

Hope it helps

Perfect comment from Charles and useful comments from Dr Ali Bavali and Dr M.A.Gondal

In practical scenarios, optical delay is achieved by fiber optic ring resonators that are being used in optical communication and sensing systems. This device is very efficient in creating optical delays. You may see the following Refs.:

Large delay times without losses can be obtained by the Herriott cell Resonator. See

Herriott et al., Appl. Optics 3,523 (1964)

Herriott et al., Appl.Optics 4, 883 (1965)

Summary in:

Hodgson at al, "Laser Resonators and Beam Propagation", Springer , Optical Sciences p. 654 ,Vol.108, NY 2005

Large delays where produced by  Welling et al, Appl. Optics 11, 1986 (1972)

A very long laser resonator to reduce the spiking frequency of Nd-YAG Lasers with techical details was investigatet by:

W. Lobsiger, "Optische Verzögerungsleitungen in Nd-YAG Laser Kavitäten zur Simulation langer Resonatoren" (Optical Delay Lines to Simulate Long Nd-YAG Laser-Resonators), Diss. 1975, University of Berne (Switzerland), Institute for Applied Physics.