What this application demands
Optical communication components are judged by two numbers that do not appear on a general-purpose filter datasheet: how precisely the centre wavelength is placed on the ITU grid, and how far it moves between the temperature limits of the equipment. A filter that is correct at 25 °C and drifts outside the pass band at 70 °C is not usable, regardless of its room-temperature curve.
The substrate contributes directly to that stability. Fused silica has a low thermal expansion and a small thermo-optic coefficient, both of which reduce the shift the coating alone would produce. Where the component sits in a dense wavelength-division-multiplexed chain, the blocking specification is equally critical, because adjacent channels are close together and leakage is not tolerable.
Specifications we recommend
| Component | Typical specification | Why |
|---|---|---|
| DWDM band-pass filters | Narrow FWHM centred on the grid, high adjacent-channel isolation | Separates closely spaced channels without crosstalk |
| Low-loss coatings | AR and HR stacks engineered for minimum absorption and scatter | Insertion loss accumulates across a chain of components |
| Substrates | Fused silica and quartz with tight thickness and wedge control | Reduces thermal drift and avoids etalon effects in the transmitted path |
Failure modes we design against
Temperature-induced centre-wavelength shift is the defining failure mode. It is addressed through substrate choice, coating design and, where necessary, athermal packaging rather than through tighter room-temperature sorting. Wedge — a deliberate small angle between the two faces — is used to suppress the interference ripple that would otherwise appear in a thick parallel plate.
The second issue is angle sensitivity. In a collimated path the design angle is well defined, but in a converging or diverging beam the effective centre wavelength shifts. Specifying the beam geometry at the start avoids a filter that meets its curve on the bench and misses it in the module.
Frequently asked questions
Can you hold a centre wavelength to the ITU grid?
Yes, for the standard communication bands, with the achievable tolerance depending on bandwidth and temperature range. Send the channel plan and we will confirm.
Do you provide temperature-shift data?
Measured spectral curves at temperature can be supplied for qualification, so the drift can be verified rather than assumed.