Bench Degree·LASERSchapter

Chapter 16: Measuring a Beam

A photodiode will tell you a 5 mW laser and a 500 mW laser are the same brightness, confidently, and be wrong by a factor of a hundred. Knowing why is the difference between a measurement and a number.

Everything in this volume has depended on knowing what a beam is: its power, its wavelength, its divergence, whether it is pulsed. This chapter is about finding those out for yourself, and about the specific ways the easy methods lie.

Two things make it harder than measuring almost anything else. Beams are often invisible, so you cannot confirm you are measuring the thing you think you are. And detectors saturate, so a detector that is overwhelmed does not report an error, it reports a plausible number that happens to be wrong.

Section 1: The Photodiode, and the Trap

A photodiode is a semiconductor junction that produces current in proportion to the light falling on it. Cheap, fast, sensitive, and available in every drawer of salvaged parts.

And it is proportional only over a range. Above some level the junction cannot generate carriers any faster, and the output stops rising. The device is saturated.

Which is the trap, and it is worth stating baldly: a saturated photodiode gives a steady, believable reading that does not change when the light gets brighter. Put a 5 mW beam on it and read 4.9. Put 500 mW on it and read 5.0. Nothing warns you. There is no flag, no flicker, no error.

Three ways to know you are in trouble, and it is worth having all three as habits:

Change the light and see if the reading changes proportionally. Put a known attenuator in the path, or double the distance, which should quarter the reading. If the reading barely moves, you are saturated. This single check is the most valuable thing in the chapter.

Attenuate deliberately, always. Measure through a filter of known transmission and multiply back up. Working an order of magnitude below saturation is the normal way to use these.

And do not put the whole beam on it. Bounce the beam off a diffuse white surface and look at the scatter, or use a known small fraction of it.

A photodiode also does not measure power. It measures photons weighted by its own spectral response, which varies enormously with wavelength. Silicon peaks around 900 nm, falls away steeply past 1000, and is essentially blind past 1100. So the same silicon diode reading “1.0” for a red beam and “1.0” for a 1550 nm beam is telling you nothing comparable, and for 1550 nm it is telling you nothing at all.

IN PLAIN ENGLISH: A photodiode answers “how much light of the kind I happen to be good at seeing, up to the point where I stop counting”. To turn that into a power in watts you need to know its response at your wavelength and prove you are below saturation.

Section 2: The Thermal Power Meter, Which Actually Measures Power

The honest instrument works on a different principle entirely: absorb the beam and measure the heat.

A thermal sensor is a black absorbing disc with a temperature sensor behind it and a heatsink behind that. The beam heats the disc, the temperature rise is proportional to the power, and a calibration converts one to the other.

Three properties follow, and they are exactly the complement of the photodiode’s.

It is wavelength-flat. A black absorber absorbs, so 405 nm and 10.6 µm read the same for the same power. This is why a thermal meter is the only sensible instrument for an unknown beam, and why it is what you use to find out whether your green pointer is emitting infrared.

It does not saturate in the same way. Overload it and you damage the coating, which is visible and obvious, rather than silently reporting a wrong number.

And it is slow. Seconds to settle, because it is waiting for a thermal equilibrium. Useless for pulse shapes, fine for average power.

A usable thermal head costs a few hundred dollars, which is a great deal more than a photodiode and is the honest price of a trustworthy number.

For pulsed work the equivalent is a pyroelectric detector, which responds to the change in temperature rather than its level, and so gives energy per pulse rather than average power.

Section 3: What to Measure With What

Note the row about picosecond pulses, because it is a genuinely interesting limit. No oscilloscope and no photodiode is fast enough to see a femtosecond pulse; the electronics is thousands of times too slow. So a short pulse is measured against itself: split it in two, delay one half by a known amount, recombine them in a crystal that only responds when both are present, and scan the delay. The width of the resulting curve gives the pulse duration. You cannot measure it with a clock, so you measure it with a copy of itself, which is a good example of the sort of move this whole subject is full of.

Section 4: The Cheap Methods, Honestly Rated

Not everyone has a thermal head. Several improvised methods exist and it is worth being clear about what each one is actually worth.

A phone camera as an infrared detector. Chapter 2’s test. Verdict: excellent as a presence test, worthless as a measurement. It will tell you there is infrared coming out of your pointer, which is the important question, and it will not tell you whether it is 2 mW or 200 mW.

A photodiode from a salvaged part with a multimeter. Verdict: good for relative comparisons, poor for absolute ones, and only if you have proved you are below saturation.

Burning things. Paper, black tape, a match. Verdict: a rough intensity indicator and nothing more. It depends on the material’s absorption, the spot size, the exposure time and the ambient conditions, and none of those are controlled. Two lasers of identical power will burn quite differently at different wavelengths, per Chapter 10.

A solar cell as a power meter. Verdict: better than it sounds and still wavelength-dependent, with the same saturation problem, but the large area makes it convenient for catching a whole diverged beam.

Counting fringes to get wavelength. Verdict: genuinely good. Chapter 1’s grating method gives three significant figures with a ruler, and Chapter 11’s interferometer does better still.

And one that is not a method at all: the number printed on the device. Chapter 8 and Chapter 15 are both, in the end, about the gap between a label and a measurement. The label is a claim by someone with an interest in the answer.

SLOW DOWN. Check Your Understanding: You have a thermal power meter and a green pointer that you suspect is leaking infrared. You measure 5 mW total. You then hold a filter that blocks 532 nm in front of the meter and read 4 mW. What have you learned, and what have you not? Think before reading on.

You have learned that about 4 mW of the output is not green, which on a device sold as a 5 mW green pointer means roughly eighty percent of its power is somewhere you cannot see. That is a serious finding and it is exactly Chapter 8’s warning, measured. What you have not learned is the wavelength of that 4 mW. It is probably 1064 nm, because that is what a green pointer’s neodymium stage produces, but it could include 808 nm from the pump diode leaking through. The distinction matters for choosing eyewear, since a filter rated for 1064 nm may not block 808. You have measured how much and not what, and getting the second answer needs a spectrometer or a set of filters with known cut-offs. A thermal meter is wavelength-blind, which is its virtue for totals and its limitation for identification.

Section 5: The Habits

Six, and they are the whole chapter compressed.

Attenuate first, always, and multiply back up.

Prove linearity by changing the input, not by trusting the detector.

Use a wavelength-flat detector on any beam whose spectrum you do not know, which includes every beam you did not build.

Never measure a beam you cannot account for. Know where every reflection goes before switching on, which is Chapter 2 and it applies to metrology more than to anything else, because measuring is when you have your face near the apparatus.

Write down the conditions, not just the number. Distance, aperture, filter, wavelength assumed. A power reading without its conditions is not a measurement.

And treat the label as a hypothesis.

Next, and last: the ledger. What this volume promised in Chapter 1, whether it delivered, and what it did not cover.

Bench Degree

Get the degree without the diploma.

Learn the material, not how to pass the exam.