Technical notes:

PiLas costs ~$10k, LED lasers cost about $100-1500 (they have large difference in power from pulse to pulse),

the second expensive part is reference counter, which measures the laser stability.

The Giessen contribution to the GlueX DIRC will be at the order of 50k$



Here is the www address from where we bought that laser:

Fiber, light connectors, splitters  and diffusers(if we found we need them) we buy usually from Edmund Optics.( )

For the optical fiber feedthrough, I have this one,

Concerning fibers, this one might be good for us:

NA(Numerical Acceptance) is the sin(theta) of the opening angle after fiber, so they(thorlab) have 5.7 degree......29.85 degree


Belle II talks:

Belle II CM talk about laser calibration (R.Stroili)

Belle II CM talk about laser calibration (R.Mussa)

Timing distributions for 3, 2, and 1 fibers pro optical box


1. option with 1 laser per optical box is possible. Laser position in this case is next to the 3-segmented mirror in the middle of the box. The laser is oriented as Avetik has suggested.
2. the aperture of the optical fiber can be really large (~gaussian sigma of 50 degrees), photons reflecting off the mirrors do not introduce distortions in the timing distributions. This is consistent with the SuperB timing calibration
4. simulation does not include reflections off the photocathode right now. Probably laser orientation from here is more efficient.
5. simulation includes cookie between PMTs and window. The material is silicon (const n, no absorption).
Comments from Avetik:
1. Two fibers per optical box is a better option, as it requires narrower angular distribution of photons.
For the fibers we buy NA(Numerical Acceptance) corresponds to sin(theta) of the opening angle after fiber, so they (thorlab) have 5.7 degree - 29.85 degree
2. We can monitor the DIRC radiators by shining the laser light throuhgh them. If you keep open the time window to look more than 25ns and find the photons travelling fort and back through bar then the amount of photons we register will be sensitive to optical quality of quartz bars too, this I mean when typed  “monitor the whole DIRC”. Over time bars might get radiation and change color(transparency) or mirror at the end, or they change the surface one has statistic how quartz bars change over 30 years in experiments.
Further studies:
Timing distributions and occupancy for two fibers pro optical box, sigma of the angular distribution = 11 degrees (PD plane is withing 2 sigma interval)
For both of these configurations the PD occupancy is not uniform due to reflections off the 3-segmented mirror.
Since it is important to have a uniform occupancy, Avetik suggested to incline the laser beam more towards PMT#19.
Weekly Meeting (03/25)
The inclination of the optical fiber inside the box should not be a problem from the technical point of view.
For Belle II TOP they are splitting the laser in 40 fibers, so for our design 5-10 fibers per bar box providing uniform distribution could be an option.
The time delays between optical fibers are calibratable.
The fibers can be mounted in a gap between the flat and the 3-segmented mirrors. For this the flat mirror can be made shorter. This gap can be as wide as 1-2 cm. Such design does not limit the number of fibers in the optical box.
Comments from Avetik:
See, iTOP has 9 fibers per box and now they cannot define the relation between fiber and time spectrum peak. We have to find a compromise between hitmap homogenity and simplicity in time spectra(I mean that we can identify a peak and relate it with fixed photon path).

For 1 fiber per box we have this simplicity, for big N fibers we might reach some sort of hitmap equality, but then the simplicity will be gone.
With two fibers the difference between the most and the least occupied areas is very large. Try 3 fibers:
The occupancy is too non-homogenious, try larger sigma:


Some pics of the simulation for the point-like laser source located on the bottom of the optical box

The opening angle is 0.28 rad = 16.1 degrees is chosen so that all photons travel from the origin directly to PD (photo detectors) without reflections inside the optical box. Right now the photon distribution is uniform in cos theta.

The origin of the laser is located closer to the radiators so that the photons have larger incidence at the PD (and so they have lower probability to get reflected off PD).

On the second pic the light spot from such a laser on the photodetector surface is marked with a circle. There are more than 3 such spots fitting into the width of the optical box. Right now I located the laser in the middle of the bar box. 

Comments from Avetik:

1. It would be hard to have diffuser+collimator in circulating water. The diffuser has to be glued to the fiber and the collimator has to be well aligned. Therefore just use simple fiber with gaussian angular distribution

2. The requirement is to hit every pixel, so unavoidably there will be photons hitting sides and mirrors. As long as they have similar time (with respect to the PMT resolution of about 1 ns), it is ok

3. Put the origin of photons inside the fiber material (n=1.5)

Another location for the laser. The PD surface is in 2sigma interval. All photons are supposed to have 1 reflection off the flat mirror. In this case the photon incidence is close to perpendicular, which makes the probability to get reflected on the PD boundaries lower. On the other hand, some fraction of photons is lost on the mirror (reflectivity ~0.9).



Location of the laser suggested by Avetik. PD surface is covered by 2sigma interval. The laser is oriented perpendicular to the radiators. In this case the photon incidence at the PD is quite shallow, which makes the probabilty to get reflected before getting detected higher (currectly the photocathode is made of fused silica. I have to put in the photocathode index of refraction).



I'm not sure that unambiguous paths are needed for the laser calibration, given that our target accuracy is not so ambitious (much less stringent than for PANDA or Belle II). The MC simulation should tell us if the uncertainty in the peak position is noticeable.

I would agree with Avetik, getting all pixels covered with the laser cone is more important than a clean single-path setup. I am attaching the slide from Doug Richard's SuperB FRDIRC talk at the RICH2013 which shows their proposed solution for the laser fiber. Of course, their situation was different, they could couple the diffuser to the fused silica camera without the need to worry about water.

Your scenarios 2 and 3 both seem reasonable to me. One thing to remember is that you may want to add long delays between adjacent laser fibers (like the delay vernier we used for the BaBar DIRC) to avoid ambiguities due to different fibers reaching the same pixel.




We can get the laser head with pulse power of ~30pJ, at wavelength of 405 nm, this is 3.0613eV photon energy. This makes ~61.2 Million photons per pulse. Let assume that we have 2 fibers per mirror box, so we have to split the laser beam to 5 fibers. Each insertion of the fiber has 4dB loss(typical for optical connections. I assume that we install laser close to FDIRC, which means our fibers length is ~10m a negligible number if we calculate the attenuation inside fiber, so we neglect it. Then we loose 2.5 times due to insertion and splitting makes 5 times. We get then at the end of each fiber ~4.9 Mio photons, so your 5M was nice shot. The fifth fiber goes into Reference counter to check Laser stability. The laser we can operate even with MhZ repetition, rate 20kHz I said because we have operated in that rate, for FDRIC we can operate till 100kHz(as high as DAQ can write the data, again 20kHz??).