Building a photonics laboratory in Bangladesh: the table, the mounts and the alignment habit

Why the optical table is the first purchase rather than the last, how to choose between a breadboard and an isolated table, what to specify in mounts, sources and detectors, and the alignment habits that decide whether the bench works.

A physics department orders the laser first. It is the exciting item, it has the longest lead time, and it takes the largest share of the grant. The optical table gets quoted separately, deferred to the next financial year, and for eighteen months a good laser sits in a store room because there is nothing stable to bolt a mirror to. On a photonics bench the order of purchase is close to the reverse of the order of excitement.

Start with the table, because everything else bolts to it

An optical table is a stiff, flat, dimensionally stable surface carrying a regular grid of tapped holes, built so that it damps its own vibration instead of ringing. A welded steel workbench with a granite slab on top has plenty of mass and almost no internal damping, so when the laboratory door closes it rings for a long time at frequencies that sit inside your measurement band. That is why a locally fabricated bench, which looks like a large saving against an imported table, usually is not one.

Four specification decisions are hard to reverse once the table is on site.

• Size. Take the longest beam path you expect, add turning space for the folding mirrors, then add room to stand and work at all four sides. Buying a table that is too small is the most common regret in this category, and extending it later means buying a second table.
• Thread standard. Imperial on a one inch grid, or metric on a 25 mm grid. Pick one for the whole department and write it into every subsequent order. ThorLabs supplies both, so the mistake is not choosing wrongly, it is choosing differently on two purchases and then owning every post, clamp and screw in two versions.
• Sealed holes. A honeycomb core with unsealed holes will take in a spilt solvent or index matching fluid and never give it back. Sealed cups under each hole cost little at the time of order and cannot be added afterwards.
• Top material and thickness. A ferromagnetic stainless top lets you use magnetic bases, which changes how quickly a teaching setup can be reconfigured. Thickness governs deflection under load and how many support points the table needs.

Isolators or a rigid frame

The tabletop damps itself. The legs decide whether floor motion reaches it at all, and these are two separate purchases that people routinely conflate.

A rigid frame is cheaper, needs no services and is often enough on a ground floor slab away from machinery. Pneumatic isolators earn their cost on an upper floor, near a lift shaft, above or beside a generator room, next to air handling plant, or in a building fronting a busy road. They also bring obligations: a clean dry air supply or a small compressor, self levelling valves, and somebody whose job it is to check the pressure occasionally.

Before you specify either, measure the floor. A day with a borrowed accelerometer or a geophone will tell you whether you are buying isolators or a frame, and it costs a great deal less than choosing wrongly. Include a measurement taken while the standby generator is running. A generator that starts during load shedding is a vibration source that appears only when the mains fails, which is exactly the moment people start blaming the experiment.

Optomechanics is a system, not a shopping list

The stability of a beam path is set by its worst joint. Every interface between the table and the optic behaves as a spring, so the number of parts in a mount stack matters more than the quality of any single item in it.

• Use the shortest post that reaches the beam height. A tall post in a small base is a tuning fork. For anything sensitive, use a pedestal or a clamping fork rather than building a tower.
• Fix one beam height for the whole laboratory and never change it. Any component can then move to any bench without re-shimming, and a target card cut to that height becomes the alignment tool everyone shares. Choose a height a seated student can reach that still leaves headroom for a periscope.
• Match adjuster resolution to the job. Fine pitch adjusters for interferometry and fibre coupling, standard adjusters for teaching, locking screws wherever a setup has to survive a whole term of student hands.
• Cage systems and lens tubes buy repeatability. They are pre-aligned, they exclude dust, and they are hard to nudge. On a teaching bench they are the difference between a demonstration that works in week one and one that still works in week twelve.
• Buy spares of the boring parts. Posts, post holders, bases, hex keys and cap screws in your table's thread size. Departments run out of screws, not lasers, and an order for screws takes as long to clear customs as an order for a laser.

Sources, and what really limits the experiment

For teaching, a helium neon laser or a stabilised visible diode module is usually right: low power, visible, and forgiving of poor alignment. Visibility is a genuine teaching specification, because a student who can see the beam learns alignment in an afternoon rather than a term.

For research the headline power is rarely the limiting parameter. What limits the measurement is wavelength stability, linewidth and coherence length for interferometry; beam quality and pointing stability for coupling into single mode fibre; pulse duration and repetition rate for time resolved work; and intensity noise for anything running near its detection limit. Higher power and materials processing sources sit in the Coherent range. Match the source to the measurement first, then to the safety case, then to the budget, in that order, because reversing the order produces a laser you cannot legally run in the room you have.

Safety drives the room, not the other way round. An interlocked door, curtains or a screened enclosure, eyewear with an optical density rated at your wavelength, signage and key control all follow from the laser class. Eyewear is where departments get caught: goggles rated for 532 nm do nothing at 1064 nm, and the pair in the drawer is almost always the pair bought for the previous laser.

One room detail is worth settling early: do not let an air conditioner blow across the beam path. A stream of alternately warm and cool air crossing a long path produces a slow wander that looks exactly like a mechanical drift, and is diagnosed by switching the unit off and watching the signal settle.

Detectors and the rest of the measurement chain

Match the detector to the wavelength and to the light level rather than to the price. Silicon photodiodes cover the visible and near infrared, InGaAs takes you further out, thermal sensors handle broadband and high power, and cooled detectors or photomultipliers are for counting photons. Ask three numbers of any detector before purchase: the wavelength range, the saturation level and the noise floor. If your signal sits outside the window between the last two, the detector will not see it whatever the datasheet says about sensitivity.

• A power meter is two purchases. The console and the sensor head are separate, and the calibration lives in the head. Buy a head that covers your wavelength and power range, and put it on the calibration schedule with the rest of the instruments.
• Plan the attenuation with the beam path. Neutral density filters, a beam sampler and a beam dump cost less than a damaged sensor head, and the damage threshold of the head is the number everyone forgets to check.
• A beam profiler and a spectrometer pay for themselves quickly. Most disagreements about alignment are settled in a minute by looking at the beam instead of arguing about it.
• Lock in detection with a chopper turns an experiment that only works at night into one that works during the day. In a building with fluorescent lighting and no blackout blinds, it is often the cheapest upgrade on the list.

What separates the undergraduate bench from the research bench

Element	Undergraduate teaching bench	Research bench
Surface	Breadboard on a rigid bench, or a small table on a rigid frame	Full table sized to the longest path, on pneumatic isolators
Mounts	Standard kinematic mounts, cage systems for repeatability	Fine pitch and locking mounts, translation and rotation stages, some motorised
Source	Visible low power helium neon or diode module	Chosen for linewidth, pointing stability or pulse duration
Detection	Photodiode and a power meter	Lock in amplifier, fast or cooled detector, spectrometer, camera
Environment	Normal classroom with curtains for stray light	Temperature stability, light tight enclosure, controlled access
Safety	Class limited by design so students are safe by default	Laser controlled area, door interlocks, eyewear held per wavelength
Records	Written exercise sheets	Alignment log, configuration photographs, calibration records
Spares	Screws, posts and post holders	The above, plus one of anything with a long lead time

The alignment discipline is part of the equipment

1. Fix the beam height and mark it. Cut a target card to that height and use it every time, including for the setups that look too simple to need it.
2. Lay the beam path along the rows and columns of the hole grid wherever the geometry allows. A path that follows the grid is reproducible; a diagonal one has to be rediscovered every time it is rebuilt.
3. Set two irises at the ends of the path and walk the beam with two mirrors, one working mainly on position and one mainly on angle. That is the whole of beam walking, and it is faster than any amount of intuition.
4. Align at the lowest power that gives a visible signal, then raise it. Most damaged optics and almost all eye incidents happen during alignment at full power.
5. Tighten in the order base, post, mount, then check the alignment again, because tightening moves things.
6. Photograph the working setup and keep the photograph with the log. When the next student rebuilds it, that photograph is worth a week.
7. Never adjust a mount you did not intend to adjust, and label the adjusters that must not be touched.

The habit matters more than the hardware. A disciplined group on a modest breadboard will produce cleaner data than a careless one on the best table in the building, and the discipline is easier to establish at the start than to install later.

Back to all Insights