When the Beam Goes Soft: Laser Output Decay and Power Meters

Nobody changed a setting. That's what makes this one hard to catch.

It shows up sideways. Your senior operator starts adding a pass, and doesn't mention it because it felt like judgement rather than a problem. The console reads what it read on day one.

Decay is measurable, and the kit to measure it costs less than one wasted lamp. This piece is the physics and the measurement. The money half of the same story, which part costs what and when it's worth replacing, sits in our breakdown of lamps, crystals and handpiece wear economics.

Fades Have a Shape, and the Shape Is Diagnostic

Start with what ages inside a flashlamp-pumped cabinet. Every discharge erodes the electrodes a little and throws a trace of that material onto the inner wall of the quartz envelope. The wall darkens. Transmission drops. The rod sees less pump light for the same stored charge, and delivered energy at your reference setting drifts down.

Gently, at first. That's the trap.

How steep the slope gets depends on how hard you drive the part, which is why two rooms running the identical lamp aren't on the same clock.

Slope versus step

Plot delivered energy over time and you get one of two pictures, and they mean opposite things.

A slope is ageing. Gradual, tracking the shot counter rather than the calendar. Nothing happened; things got older. You plan for a slope.

A step is an event. Output sat flat, then dropped and stayed down. Something happened between those two readings: a handpiece met a trolley, a window got cleaned with the wrong thing, an arm took a knock in transit, a service visit disturbed an alignment. Steps get investigated, not budgeted.

Without a log you can't tell these apart, and the difference decides whether you're phoning a supplier or opening a purchase order.

When only one wavelength goes quiet

Dual-line cabinets hand you a free diagnostic. A tired pump drags every output down together, since everything downstream feeds from the same place. So if superficial 532 nm work weakens while 1064 nm work still lands the way it always did, the fault sits downstream of the pump, in the doubling stage and its coatings. Different part, different invoice, different lead time. One reading taken on both lines splits that question in minutes.

Rank the Suspects Before You Order Anything

"Settings unchanged, results worse" has a cause list, and most clinics work it in roughly the wrong order, starting with the expensive part. Work it cheapest and most reversible first.

1. It isn't the machine. A new operator presses differently. Gel goes on thicker. Clients arrive with more sun on them in July than in February. Protocols drift when nobody re-reads them.
2. The last few centimetres. Film on the output window, an abraded sapphire tip, a gel layer doing what gel does. Our technical archive is blunt that cooling gel is an optical medium in its own right, reflecting and scattering and absorbing, so it always costs energy on the way in.
3. The spot. On a continuously adjustable head this one is brutal. Fluence is energy over area, and area goes as the square of diameter, so moving from a 3 mm spot to 4 mm cuts fluence by roughly 44% with the joules on the display unchanged. Our brochure archive shows that adjuster as a graduated ring the operator sets by hand. Anyone can nudge it. Check it before you suspect a lamp; the spot range for that cabinet sits on the QE-01 specification page.
4. The delivery path. Arm mirrors and their coatings, fibre end faces, connector seating.
5. The source. Lamp, rod, doubler, cavity reflector.
6. The display. It was never a measurement, and we'll come back to that.

Buying a Meter That Won't Lie to You, or Die Trying

Match the sensor to how your machine emits, because there isn't one detector that covers everything.

Thermal, pyroelectric, photodiode

Thermal sensors, meaning thermopiles, read power indirectly by measuring how much a disc heats up. Ophir's technical explainer puts their span at microwatts to tens of kilowatts, ultraviolet through the far infrared including 10.6 microns, with little angle or wavelength dependence and a response time from a fraction of a second to several seconds. Slow, wide, tough. That's your instrument for average power on a CO2 platform like the CF-01.

Pyroelectric sensors measure per-pulse energy by the temperature change they undergo when a pulse lands. That's the family you want on a nanosecond Q-switched cabinet, where the number that matters is joules in a single shot, not watts averaged over a minute. Photodiode sensors are the third option and the wrong one here: they saturate at low optical powers and belong to alignment beams and fibre work.

The three specifications people skip

• Aperture. It has to be comfortably larger than the beam at the plane where you measure. Never focus a beam down to fit a sensor. You'll be measuring the destruction of your own instrument.
• Damage threshold. Quoted in J/cm² at a stated pulse duration. Do the arithmetic against your own worst case before the first shot: a several-hundred-millijoule pulse of a few nanoseconds concentrated into a small area is a genuine threat to a sensor face.
• Wavelength setting. Calibration is per wavelength. A meter set for 1064 nm and fired at with 532 nm reports a number that looks plausible and isn't.

Traceability, and the method behind the number

A meter is only a transfer standard, and it drifts too. What you want on the certificate is an unbroken chain to a national standard, plus a date. NIST's Sources and Detectors Group calibrates laser power and energy instruments from single photons up to the hundred-kilowatt range and develops SI-traceable optical power standards; commercial calibration labs hang off that chain. Put the recalibration due date in the same diary as your water changes.

There's a published method too. ISO 11554 covers test methods for laser beam power, energy and temporal characteristics, stability included, and it exists because "we measured it" means nothing without a stated procedure. Cite it in your acceptance protocol and both sides have something to point at. That protocol is also where your first reading belongs, and the rest of the sequence is in our guide to acceptance testing a new laser.

One caveat, and its answer. Calibration coefficients are quoted per wavelength, so a factor set for a single laser line and applied to a 530 to 1200 nm flash carries a systematic bias. That rules out the meter borrowed from the pigment room, not the measurement. Thermal sensors with a spectrally flat broadband absorber hold their absorption across widely differing wavelengths, and Ophir lists thermopile heads built specifically for air-coupled pulses from IPL dermatological sources, flat from 250 to 2200 nm, wide aperture, entrance window removed. Pick one whose clear aperture is larger than the light guide exit and whose damage threshold clears the worst single-flash fluence your head can produce. Then use the reading as a relative trend against your own baseline under one fixed recipe, filter set included because the filter sets the band, logged beside the flash counter. An E-Light platform runs the same protocol below as any laser cabinet.

A Baseline Anyone in the Room Can Repeat

The factory already works this way. Our brochure archive documents pre-shipment energy testing on the Q-switched line, done on a laboratory-grade energy meter at the far end of the articulated arm, quoting single-pulse energy where it leaves the machine rather than where it's generated. Copy that habit. Measure where the light leaves, not where it's born.

Fix the recipe and never vary it:

• One reference setting. One spot size. One repetition rate. Written down, taped inside the cabinet door.
• Same warm-up time before the first shot. Cold machines and hot machines don't read the same.
• Same sensor at the same distance and plane, ideally with a marked jig so it's not eyeballed.
• Discard the first few shots, then log ten. Record mean, minimum and maximum, because the spread is data too.
• Log the shot counter alongside every reading, and plot energy against shots rather than against dates. Wear is counted in discharges; a slow quarter and a busy one aren't comparable on a calendar axis.
• Repeat monthly, and after every service visit, transport, or water change.
• Eyewear on. You're firing an open beam into a sensor, and optical density is specified per wavelength.

Turn your own protocols into a replacement threshold

Suppliers rarely publish an end-of-life criterion you can check, so build one from arithmetic you control. Fluence equals energy divided by spot area, and area is pi times the radius squared. A 4 mm spot is about 0.126 cm², so a protocol calling for 6 J/cm² there needs roughly 0.75 J actually arriving at the tissue.

That's your floor. When measured delivered energy at your reference setting can no longer clear the fluence your protocols assume, the machine has stopped being able to do the work you sell, whatever the datasheet said when it was new. Write the number down.

Add an earlier trigger for investigation: a sustained fall from baseline across three consecutive monthly readings, at a percentage you choose and record. Try to get that percentage into the service contract while you still have bargaining room, which means before you've paid the balance.

What happens when the meter itself is out of tolerance

Our quality documentation handles this the way any regulated shop does. Work stops on that instrument, whatever was checked with it gets traced, earlier results are re-evaluated for validity, and somebody defines the scope of re-testing before the thing goes for repair.

Translate that to your clinic and it stings: a meter found out of tolerance casts doubt on your whole trend since its last valid certificate. So overlap. Take a reading immediately before the meter goes away and another as soon as it's back, identical reference setting, and keep both. Those two readings are what let you stitch a trend line across a calibration event.

What Each Reading Actually Tells You to Do

Readings are only worth having if they change a decision.

• Reading at baseline, results poor. Not a parts problem. Look at protocol, technique, patient selection and endpoint discipline.
• Gradual fall on every output, tracking the counter. The pump source is ageing. Schedule the replacement instead of reacting to it, and price it against your per-treatment wear cost.
• One wavelength down, the other steady. Doubling stage and its coatings. Get it inspected before ordering a lamp you don't need.
• Step drop with a flat line either side. An event, not ageing. Reconstruct what changed between those two dates and treat it as a service call.
• Healthy at the source, weak at the tip. The delivery path: window, tip, fibre, arm.
• Normal mean, wide spread. Charging, simmer or thermal behaviour. Your minimum and maximum columns just earned their keep.

One rule sits above all of them. A fading source gets replaced, not compensated. Raising energy to chase a result the machine used to give for free is the field practice that quietly walks a room toward blistering and post-inflammatory hyperpigmentation, particularly on darker phototypes, and it hides the fault while it does it.

Treat the display with the scepticism it has earned. Girasol and colleagues measured 24 therapeutic laser devices in clinical use and reported in PLoS ONE in 2022 that analysed power ranged from 2% to 134% of the values declared by manufacturers, with beam diameters between 38% and 543% of nominal. Those were low-level therapy units, not aesthetic Class 4 cabinets, so don't transplant the numbers. Transplant the lesson: what a device says it delivers and what it delivers are two quantities, and only one of them is measured.

This is engineering and procurement background for people specifying and running equipment, not clinical guidance; operators need training and patients need screening. Want a reference-setting protocol drafted against a specific configuration? Our service team can put one together, or send us the model details and we'll spell out where to measure and what to log.

Frequently asked questions

How often should we measure output?

Monthly is a sensible default for a machine that treats most days, plus an extra reading after any service visit, relocation or cooling-system work. The absolute value matters less than the trend, and a trend needs enough points to be one. Two readings a year tell you almost nothing; twelve will show you a slope while there's still time to plan around it.

Can I use a cheap handheld meter from a laser-pointer supplier?

No. Those are photodiode instruments built for milliwatts, and they'll saturate, mislead or fail outright in front of a treatment beam. Pick the detector family for your emission mode, a thermal sensor for average power and a pyroelectric one for per-pulse energy, then check aperture size and damage threshold against your own worst-case setting before the first shot.

The console shows joules. Isn't that already the measurement?

It's a request, not a result. The display reports what the machine was asked to produce, with everything downstream of that point unaccounted for: envelope darkening, coating ageing, a dirty window, a gel layer. Published measurements of light-based devices in clinical use have found real gaps between declared and delivered output, so treat the display as an input and the meter as the output.

We never took a baseline at installation. Is it too late?

Not at all. A late baseline can't tell you how far you've already fallen, but it starts the slope from today, and the slope is what drives decisions. Take your first reading this week, log the shot counter with it, and keep the recipe fixed from then on. Three months of honest data beats a year of remembering.

References

1. ISO 11554:2017 - Optics and photonics. Lasers and laser-related equipment. Test methods for laser beam power, energy and temporal characteristics
2. NIST Sources and Detectors Group - laser power and energy calibration services and SI-traceable optical power standards
3. Ophir Photonics - How It Works: Measuring Laser Power with a Thermal Sensor (thermopile range, response time, wavelength dependence)
4. Ophir Photonics - How It Works: Measuring Laser Power with a Pyroelectric Sensor
5. Ophir Photonics - L50(300)A-LP2-65 intense pulsed light sensor (spectrally flat LP2 broadband absorber, 250-2200 nm, Ø65 mm aperture, window removed for air-coupled IPL sources)
6. Girasol CE, Braz GA, Bachmann L, Celli J, Guirro RRJ. Laser light sources for photobiomodulation: the role of power and beam characterization in treatment accuracy and reliability. PLoS ONE, 2022
7. IEC 60825-1:2014 - Safety of laser products, Part 1: Equipment classification and requirements