Dos módulos de bomba, misma longitud de onda central de 976 nm en la etiqueta, misma potencia nominal, mismo diámetro de núcleo de fibra. Uno mantiene la salida plana en todo el rango de operación. El otro pierde eficiencia cada vez que el circuito de refrigeración se calienta.
Nada en la primera página de ninguna hoja de datos explica la diferencia. Está en la página tres, en la especificación de ancho de línea y en la curva de longitud de onda versus temperatura, y es el número más decisivo en la selección de bombas para láser de fibra.
Esta guía cubre las decisiones que realmente determinan si una elección de bomba funciona: qué longitud de onda, qué arquitectura de diodo, cómo el brillo limita tus opciones y qué te aporta la estabilización.
Tabla de Contenidos
Un diodo de bomba para láser de fibra es un láser semiconductor de alta potencia, generalmente acoplado a fibra, que excita el dopante de tierras raras en una fibra activa. Los sistemas de iterbio y erbio-iterbio se bombean cerca de 915 nm o 976 nm. Los módulos van desde emisores individuales de alrededor de 15 W hasta apilamientos de barras que entregan kilovatios.
La primera decisión: 915 nm o 976 nm
Para fibra de iterbio y erbio-iterbio, esta elección se extiende a todo lo demás. Vale la pena hacerla deliberadamente en lugar de recurrir por defecto a lo que usó el último diseño.
976 nm Se sitúa en el pico de absorción fuerte y estrecho del iterbio. Debido a que la absorción es alta, necesitas menos fibra activa, y como el fotón de bomba está más cerca en energía de la salida de ~1064–1080 nm, el defecto cuántico es menor y se genera menos calor por fotón absorbido. La eficiencia óptica a óptica es la mejor disponible.
El inconveniente es el ancho de ese pico. La característica de absorción del iterbio a 976 nm es estrecha, al menos un orden de magnitud más estrecha que la del erbio a la misma longitud de onda. Un diodo Fabry-Pérot simple emite con un ancho de línea de alrededor de 6 nm y se desplaza aproximadamente 0,3 nm por °C de temperatura de unión. Junta todo eso y una bomba de 976 nm no estabilizada se aleja del pico de absorción a medida que se calienta. La potencia absorbida cae, la bomba no absorbida se propaga por la fibra donde no debería, y la salida disminuye de una manera que parece un problema de fibra.
Así que un diseño de 976 nm esencialmente requiere estabilización de longitud de onda, control térmico estricto, o ambos.
915 nm Se sitúa en una característica de absorción más amplia y débil. La sección transversal es más pequeña y más ancha, y eso cambia la ingeniería en varias direcciones a la vez. Debido a que la absorción es más débil, necesitas una fibra activa más larga y aceptas una menor eficiencia óptica a óptica. Debido a que la característica es amplia, el desplazamiento del diodo por temperatura apenas importa: la bomba permanece útilmente absorbida en un amplio rango de longitudes de onda, lo que hace que la estabilización y el enfriamiento de precisión sean opcionales.
Hay un segundo beneficio menos obvio. La absorción más débil distribuye la deposición de la bomba sobre una longitud de fibra más larga, lo que reduce la carga térmica local y eleva el umbral de inestabilidad de modo transversal. Eso importa enormemente a alta potencia. Un oscilador bombeado a 915 nm de clase kilovatio de 2024 reportó una eficiencia óptica a óptica del 75,4&37; — por debajo de lo que 976 nm puede alcanzar, pero logrado con requisitos térmicos y de estabilización notablemente relajados.
| 915 nm | 976 nm | |
|---|---|---|
| Absorción | Amplia, débil | Estrecha, fuerte |
| Eficiencia | Menor | La más alta disponible |
| Defecto cuántico / calor | Mayor por fotón | Menor por fotón |
| Longitud de fibra necesaria | Más larga | Más corta |
| Estabilización | Generalmente innecesaria | Efectivamente requerida |
| Control térmico | Relajado | Estricto |
| Umbral de inestabilidad de modo | Mayor (calor distribuido) | Menor (calor concentrado) |
| Mejor para | Robusto, alta potencia, despliegue en campo | Crítico en eficiencia y tamaño |
La versión corta: 976 nm compra eficiencia y compacidad a costa de complejidad de control. 915 nm compra robustez y margen térmico a costa de eficiencia. Si tu sistema vive en un gabinete con control de temperatura, 976 nm suele ser lo correcto. Si vive en un vehículo, un barco o un piso de fábrica sin acondicionar, 915 nm merece una consideración seria.
Existen esquemas híbridos — bombear parcialmente en cada longitud de onda, o bombeo en tándem a 1018 nm para reducir aún más el defecto cuántico — pero esas son optimizaciones para alcanzar después de que la arquitectura base esté establecida.
Modo único o multimodo
La segunda bifurcación trata sobre calidad de haz versus potencia, y está mayormente decidida por lo que estás bombeando.
Bombas de modo único Se acoplan a una fibra de modo único con un núcleo de alrededor de 6–9 µm. La calidad del haz es excelente y la salida puede lanzarse directamente a un núcleo de señal. La potencia es limitada — típicamente muy por debajo de 1 W por dispositivo — y el costo por vatio es alto. Estas son para amplificadores bombeados por núcleo, etapas de semilla, aplicaciones de bajo ruido y amplificadores de banda de telecomunicaciones de erbio.
Bombas multimodo Se acoplan a fibra multimodo, comúnmente 105/125 µm, y escalan desde cientos de milivatios hasta varios cientos de vatios por módulo. La calidad del haz es pobre en comparación, que es precisamente por lo que existe el bombeo por revestimiento: lanza la bomba a un revestimiento interno multimodo grande en lugar del pequeño núcleo dopado, deja que cruce el núcleo repetidamente a lo largo de metros de fibra, y obtienes salida de modo único desde una bomba multimodo. Casi todos los láseres de fibra de alta potencia funcionan así.
Si estás construyendo un sistema bombeado por revestimiento — y para cualquier cosa por encima de unos pocos vatios lo estás — multimodo es la respuesta, y la pregunta real se convierte en qué arquitectura multimodo.
Cuatro arquitecturas multimodo
Los módulos de bomba se construyen de cuatro maneras, y las diferencias aparecen en potencia, brillo, costo y cómo fallan.
| Arquitectura | Potencia típica | Fibra de entrega | Carácter |
|---|---|---|---|
| Single emitter | ~10–25 W | 105/125 µm | Highest brightness per watt; one emitter, one failure point |
| Multi-emitter module | ~50–several hundred W | 105–200 µm | Several discrete emitters beam-combined into one fiber; graceful degradation |
| Single bar | ~50 W | 200–400 µm | Many emitters on one monolithic bar; lower brightness, low cost per watt |
| Multi-bar / stack | Hundreds of W to kW | 400 µm and larger | Bars stacked; highest power, lowest brightness, water cooling typical |
A few things worth understanding about this table.
Single emitters give the best brightness. One emitter’s output can be collimated and focused into a small fiber with high efficiency, so you get more watts per unit of étendue. Multi-emitter modules preserve much of that by combining several single emitters spatially and by polarization, which is why they dominate mid-power fiber laser pumping.
Bars trade brightness for cost. A bar packs many emitters side by side on one chip. Total power is high and cost per watt is low, but the emitters are spread across a wide slow axis, so coupling into a small fiber is inefficient. You end up needing a larger delivery fiber, which constrains the cladding diameter you can pump into.
Failure behavior differs. In a multi-emitter module, one dead emitter costs you a fraction of the power and the system usually keeps running. On a monolithic bar, a single emitter failure can propagate. This is a reliability argument that rarely appears on datasheets and matters a great deal for unattended or field-deployed systems.
Brillo, NA, y por qué el número de fibra importa
Two pump modules can both say “50 W” and be entirely different products. What separates them is brightness.
Brightness is power divided by the product of emitting area and solid angle. In fiber-coupled terms, that translates to power divided by the product of core diameter and numerical aperture. A 50 W module in a 105 µm / 0.22 NA fiber is a far more useful device than 50 W in a 400 µm / 0.22 NA fiber, even though the label power is identical.
Two consequences follow directly, and both bite late in a design if they are not budgeted early.
Brightness cannot be improved downstream. Optics can reshape a beam, but the product of size and divergence — étendue, expressed for beams as beam parameter product — cannot be reduced by passive optics. If your pump arrives with too much étendue for your fiber’s inner cladding, no lens fixes it. You lose the excess as heat at the launch.
Cladding diameter is a system-level constraint. A larger inner cladding accepts lower-brightness pump light more easily, but it also lowers cladding absorption per unit length, so you need more fiber. Smaller cladding absorbs faster and gives better mode control, but demands brighter pumps. Pump brightness and fiber design are one decision, not two.
The practical instruction: never accept a pump power figure without the core diameter and NA attached. Power alone is not a specification.
Estabilización de longitud de onda: qué cuesta realmente y qué compra
An unstabilized broad-area Fabry-Pérot diode has a linewidth of several nanometers and its center wavelength moves with both junction temperature (roughly 0.3 nm per °C) and drive current. For a 915 nm design that is fine. For 976 nm into ytterbium it is usually not.
Three approaches exist.
Fiber Bragg grating (FBG) stabilization writes a grating into the pigtail, reflecting a narrow band back into the diode and forcing it to lase there. This is the dominant method for fiber-coupled pump modules. It typically narrows emission below 0.5 nm and largely decouples wavelength from temperature and current.
Volume Bragg grating (VBG) stabilization places a bulk grating as the output coupler, usually paired with a reduced front-facet reflectivity so the grating wins the competition for gain. Common in free-space and module-level designs.
No stabilization, paired with either tight temperature control or a deliberately tolerant absorption band. This is the 915 nm route, and done knowingly it produces the simplest system.
Two details worth knowing before specifying a locked pump.
The grating’s design wavelength must sit close to the chip’s natural peak emission — within roughly ±5 nm — or the diode will not lock reliably across its operating range. This is a manufacturing-side constraint, but it determines how much temperature margin a locked module really has.
Locking is not instantaneous. FBG-based stabilization takes a finite time to establish after turn-on, because light must make a round trip to the grating and back. For continuous-wave pumping this is irrelevant. For pulsed or rapidly modulated pumping, the first portion of each pulse can emit unlocked and spectrally broad, which affects absorption and can surprise you in a QCW design.
Introduciendo la bomba en la fibra
Between the pump module and the active fiber sits a component that quietly determines a lot of system behavior: the pump combiner.
A pump-signal combiner takes several multimode pump fibers plus, usually, one signal fiber through the center — the (6+1)×1 and (18+1)×1 configurations are common — and tapers them into a single double-clad output fiber. Pump light lands in the inner cladding; the signal passes through the core untouched.
Three things about combiners are worth knowing at specification time.
Brightness conservation applies here too. The taper cannot concentrate pump light beyond what étendue permits. Feeding a combiner with pumps that are too low in brightness for the output cladding wastes power as heat right at the taper — a common cause of combiners running hot.
Pump count sets your redundancy granularity. A (6+1)×1 combiner with six pump modules degrades gracefully; losing one costs roughly a sixth of the pump power. Two very large modules feeding a 2+1 gives you a far more brittle system.
Backward-propagating light is a real hazard. Unabsorbed pump and back-reflected signal travel toward the pumps. Cladding light strippers and proper isolation are not optional above modest power levels, and pump diodes damaged by back-reflection are frequently misdiagnosed as random failures.
Empaquetado, unidad e interfaz térmica
The package is where the pump meets your mechanical and electrical design.
14-pin butterfly packages are the standard for lower-power and single-mode devices, typically integrating a TEC and a thermistor so junction temperature can be actively held. They cost board space and power, and the TEC becomes a component in your reliability budget.
Fiber-coupled multimode modules at tens to hundreds of watts are usually conduction-cooled to a baseplate, with a specified maximum case temperature and thermal resistance. Above a few hundred watts, water cooling is typical.
Electrically, expect a single emitter to want roughly 1.5 V at 10–20 A. Emitters wired in series raise voltage and keep current the same; bars run parallel arrays and pull 45–60 A at similar voltages. Multi-bar stacks in series climb in voltage accordingly. The reason to care is that your driver’s compliance voltage, current stability, and transient protection all follow from this, and pump diodes are unforgiving of current overshoot at turn-on.
One thermal point that gets underestimated: the specified case temperature is not a suggestion. Junction temperature drives wavelength, and wavelength drives absorbed power. In a 976 nm system, a thermal interface that degrades over time — dried compound, a loosened screw, a fouled cold plate — presents as slow output decline that looks like diode aging. It often is not.
Qué preguntar a un proveedor de bombas
The generic advice about reading datasheets skeptically applies. These are the pump-specific questions.
Is the center wavelength specified with a tolerance and a reference case temperature? A bare “976 nm” is not usable. You need the number, the tolerance, and the temperature it was measured at.
What is the linewidth, and is the device locked? For 976 nm into ytterbium, the answer determines whether the design works.
What is the delivery fiber core and NA, and what is the power measured at the connector? Not chip power, not pre-coupling power. What comes out of the fiber you will actually connect.
Was every unit burn-in screened, or a sample? Diode lasers fail early or run for years. Burn-in is how you separate those populations before shipment, and screening every unit costs the supplier real yield — which is why some do not.
Is the package hermetic? Facet degradation from moisture accumulates over years. Two modules can look identical on paper and diverge sharply at year three.
Can you see data across the full operating window? Wavelength, power, and threshold at the temperature extremes, not just at 25 °C.
Our own flow for fiber-coupled diode pump sources runs incoming inspection, chip test, die bonding and chip-on-submount assembly, automated fiber coupling and alignment, hermetic sealing by parallel seam welding, then burn-in and high/low temperature cycling before final test. The reason to be vertically integrated is not marketing — it is that when a unit behaves oddly, the trace back to a bonding step or a coupling alignment actually exists. You can review the equipment and test capability behind that, or browse available laser source modules if you are scoping a specific wavelength and power.
Errores comunes
Comparing pumps on watts alone. Without core diameter and NA, the power figure tells you almost nothing about whether the module fits your fiber.
Choosing 976 nm and then discovering the thermal budget. The efficiency advantage is real, but it comes bundled with stabilization and cooling requirements. Decide with the full cost in view.
Assuming brightness can be recovered. It cannot. Étendue only grows through passive optics. Budget it at the architecture stage.
Ignoring back-reflection protection. Cladding light strippers and isolation are load-bearing, not accessories.
Calibrating and qualifying at ambient only. Wavelength and threshold both move with temperature. A 25 °C characterization tells you very little about a −20 °C cold start.
Treating the combiner as passive plumbing. Its brightness acceptance, pump count, and thermal handling shape system reliability as much as the diodes do.
Preguntas frecuentes
Should I use 915 nm or 976 nm to pump a ytterbium fiber laser?
976 nm sits on a strong narrow absorption peak — best efficiency and shortest fiber, but it needs wavelength stabilization and tight thermal control. 915 nm sits on a broad weak feature — lower efficiency and longer fiber, but tolerant of drift and gentler thermally. Controlled environments favor 976; rugged deployments favor 915.
What is a fiber Bragg grating stabilized pump?
A pump diode with a grating written into its delivery fiber that reflects a narrow band back into the diode, forcing it to lase at that wavelength. It typically narrows emission below 0.5 nm and holds the wavelength largely independent of temperature and drive current — essential for 976 nm ytterbium pumping.
What is the difference between a single emitter and a laser bar?
A single emitter is one emitting stripe, typically 10–25 W, with high brightness and easy coupling into a 105 µm fiber. A bar packs many emitters across one chip for around 50 W at lower cost per watt, but the wide emitting area lowers brightness and requires a larger delivery fiber.
Why does pump brightness matter if I only need total power?
Because brightness determines whether that power can be launched into your fiber’s inner cladding at all. Passive optics cannot reduce étendue, so pump light that is too divergent for the cladding is lost as heat at the launch, regardless of how many watts you started with.
What is cladding pumping?
Launching pump light into a large multimode inner cladding rather than the small doped core. The pump crosses the core repeatedly along meters of fiber and is gradually absorbed, which lets low-brightness multimode diodes produce a single-mode output. It is the basis of essentially all high-power fiber lasers.
Do pump diodes need a TEC?
It depends on wavelength. A stabilized or 915 nm pump often runs on passive conduction cooling with a specified case temperature. An unstabilized 976 nm pump into ytterbium generally needs active temperature control to stay on the absorption peak.
Especificando la bomba junto con la fibra
Pump wavelength, pump brightness, cladding diameter, and fiber length are one coupled decision. Settling them independently is how programs end up with a combiner that runs hot and an efficiency figure nobody can explain.
Send us your gain fiber, output requirement, and thermal envelope, and our engineers will work through the pump options with you — including whether stabilization earns its cost on your platform, and what the wavelength-versus-temperature data needs to show for your absorption budget to close.