Wavelength-conversion stages are blamed for more faults than they cause. This page sets out how a conversion stage is actually tested, what the temperature behavior really tells you, and when the crystal genuinely is the problem.
Second-harmonic generation converts part of a fundamental beam into light at half the wavelength. It cannot produce more converted power than the fundamental delivered into it, and in a real system it produces a fraction of that fundamental, set by intensity, crystal length, phase matching and beam quality.
That single constraint settles most conversion-stage questions before any hardware is opened:
Our evaluation establishes the fundamental power actually arriving at the input of the conversion stage, measured on a defined path with the fundamental separated from the converted output.
That result decides where the work belongs. A healthy fundamental with disproportionately low converted output is a conversion-stage question, and the rest of this page applies. An absent or strongly suppressed fundamental moves the work upstream, into the seed, the drivers, the amplifier chain or the fiber path.
Where a converted reading and a fundamental reading are of the same order, the two numbers cannot both describe a working conversion process fed by that fundamental. There are three ways that resolves, and our evaluation tests each rather than assuming one:
Crystal ovens are one of the most consistently misread signals in scientific laser diagnostics, in both directions.
A temperature-stabilized oven exists to hold its setpoint. Holding it is success, not failure. When significant optical power passes through the crystal, the absorbed fraction creates a heat load that appears as a disturbance the controller has to work against, and you can sometimes see that disturbance when the drive current changes. When little or no fundamental arrives, there is no meaningful load, and the temperature sits flat.
So a crystal temperature that does not move when the current command changes is consistent with an absent fundamental. On its own it is weak evidence of a conversion-stage fault, and treating it as the primary symptom sends the investigation to the wrong end of the machine.
Those are TEC, driver, sensor, wiring and control-loop questions, and they are repairable at component level in most systems, including obsolete ones.
When the fundamental has been confirmed healthy and the converted output is still low, these are the causes worth working through, roughly in the order they are worth testing.
Crystal temperature offset from its correct operating point, or an angle that has shifted. Frequently recoverable without touching hardware, once the correct condition has been found experimentally.
Conversion is polarization dependent. A rotated or degraded polarization state upstream, including stress in a fiber path or a failed waveplate, reduces conversion without damaging anything.
Deposits on crystal faces or on the input and output optics of the stage, and degraded anti-reflection coatings. Common in systems that have run in unfiltered laboratory air for years.
Bulk or surface damage from a previous overdrive event, or from operation with a degraded beam. Genuine, but far less common than it is assumed to be.
Conversion depends on intensity. A degraded spatial mode or a shifted focus reduces efficiency while the average fundamental power looks unchanged.
Dichroics, filters and beam dumps that separate converted output from residual fundamental. A degraded separator can remove the converted light you are trying to measure.
Thermal control hardware faults as described above, including sensor drift that leaves the crystal at the wrong actual temperature while reporting the right one.
A crystal mount disturbed by shipping, thermal cycling or previous service work, changing the angle or the beam position on the crystal.
Spectral measurement is used throughout to confirm that the light being measured is the light you think it is. On a stage producing weak output with residual fundamental in the same path, that confirmation is not optional.
Optical parametric oscillators and amplifiers share the same first principle: they cannot produce more than the pump delivers, and their output depends on a phase-matching condition that has to be held. The same diagnostic order applies, with additional questions specific to these systems:
We evaluate OPO and OPA subsystems as part of a system evaluation. On systems where factory alignment data is unavailable, we state in the proposal what can and cannot be re-established, and we do not promise restoration of a tuning range we cannot verify against original acceptance data.
Class 4 work is planned and executed under a project-specific laser safety plan. The required engineering controls, personal protective equipment, beam management and facility conditions are defined in writing before any energized optical work begins, and the infrastructure that plan calls for is provided within the project scope. Where a laboratory's own controls are involved, final hazard analysis and approval stay with the institution's LSO or EHS function.
PhaseX Laser Services is an independent third-party service provider and is not affiliated with, authorized by, or endorsed by any manufacturer unless expressly stated in writing. Brand names and model names are used only to identify equipment for which customers may request evaluation. We do not guarantee a repair before evaluation, and we do not certify a system to original published specifications without the original acceptance data and the corresponding standards.
The upstream chain: seed, drivers, amplifiers and the four diagnostic traps.
Read the pageA worked example where the conversion stage was the obvious suspect and the wrong one.
Read the pageWhat remains possible after end-of-support, and where the real limits are.
Read the pageModel, configuration, symptoms and photographs in one submission.
Open the formTell us what you measured at each wavelength, with what instrument and how the beam reached it. On conversion-stage enquiries that information tells us whether the question is here or upstream, and it costs nothing to send.
Request service