L
Admin
An externally modulated RF-over-fiber link built around a lithium niobate Mach-Zehnder modulator (MZM) can be very linear, but only if the bias stays where you put it. Bias drift is one of the most common reasons a link that was fine on Monday shows poor second-order distortion on Friday.
1. Bias points and what they are for
The MZM transfer function is a raised cosine of the applied voltage. Three bias points are used:
- Quadrature (half power): the most linear point for a normal intensity-modulated link; even-order distortion is ideally zero here.
- Null (minimum transmission): carrier-suppressed operation, used for frequency doubling and some photonic mixing schemes.
- Peak (maximum transmission): also used for doubling and certain filtering schemes.
For a standard analog link, quadrature is the target.
2. Why the bias drifts
Lithium niobate modulators show DC bias drift from charge build-up, photorefractive effects at high optical power, and temperature changes. The drift can be slow (hours) but large compared with the tolerance for low second-order distortion. Note that the DC half-wave voltage (Vpi DC) and the RF half-wave voltage are usually different, so do not use the RF Vpi to compute the DC bias.
3. How bias controllers work
Most controllers add a small low-frequency dither tone to the bias, tap a small fraction of the optical output to a monitor photodiode, and detect the dither's first and second harmonics:
- At quadrature, the first harmonic of the dither is maximised and the second harmonic is near zero.
- At null or peak, the first harmonic goes to zero.
A feedback loop then holds the chosen point. Check that the dither frequency and amplitude do not produce spurs inside your RF band or in the receiver's baseband.
4. Verify with the RF second harmonic
The most direct check of quadrature is the RF second harmonic (or the second-order intermodulation product) at the link output. Drive a single tone, watch the 2f level, and adjust the bias manually: the minimum should coincide with the controller's lock point. If it does not, the controller is locking slightly off quadrature and your SFDR for multi-octave signals is worse than the datasheet suggests.
5. Other details that bite
- The MZM is polarisation sensitive. Use polarisation-maintaining fibre from the laser to the modulator, or a polariser and controller.
- High optical input power improves the link gain and noise figure but can accelerate photorefractive drift; follow the vendor's power limits.
- Thermal stability of the package and of the bias electronics reduces how hard the loop has to work.
What controller or dither scheme do you use for multi-octave links, and how much second-order improvement did locking give you?
1. Bias points and what they are for
The MZM transfer function is a raised cosine of the applied voltage. Three bias points are used:
- Quadrature (half power): the most linear point for a normal intensity-modulated link; even-order distortion is ideally zero here.
- Null (minimum transmission): carrier-suppressed operation, used for frequency doubling and some photonic mixing schemes.
- Peak (maximum transmission): also used for doubling and certain filtering schemes.
For a standard analog link, quadrature is the target.
2. Why the bias drifts
Lithium niobate modulators show DC bias drift from charge build-up, photorefractive effects at high optical power, and temperature changes. The drift can be slow (hours) but large compared with the tolerance for low second-order distortion. Note that the DC half-wave voltage (Vpi DC) and the RF half-wave voltage are usually different, so do not use the RF Vpi to compute the DC bias.
3. How bias controllers work
Most controllers add a small low-frequency dither tone to the bias, tap a small fraction of the optical output to a monitor photodiode, and detect the dither's first and second harmonics:
- At quadrature, the first harmonic of the dither is maximised and the second harmonic is near zero.
- At null or peak, the first harmonic goes to zero.
A feedback loop then holds the chosen point. Check that the dither frequency and amplitude do not produce spurs inside your RF band or in the receiver's baseband.
4. Verify with the RF second harmonic
The most direct check of quadrature is the RF second harmonic (or the second-order intermodulation product) at the link output. Drive a single tone, watch the 2f level, and adjust the bias manually: the minimum should coincide with the controller's lock point. If it does not, the controller is locking slightly off quadrature and your SFDR for multi-octave signals is worse than the datasheet suggests.
5. Other details that bite
- The MZM is polarisation sensitive. Use polarisation-maintaining fibre from the laser to the modulator, or a polariser and controller.
- High optical input power improves the link gain and noise figure but can accelerate photorefractive drift; follow the vendor's power limits.
- Thermal stability of the package and of the bias electronics reduces how hard the loop has to work.
What controller or dither scheme do you use for multi-octave links, and how much second-order improvement did locking give you?