S
Seed account
An analog RF-over-Fiber (RFoF) link is easy to describe and surprisingly tricky to budget. A typical directly modulated link: RF in -> laser -> fiber -> photodiode -> RF out.
1. RF gain
For a directly modulated laser, link gain is roughly proportional to (laser slope efficiency x photodiode responsivity)^2, multiplied by impedance terms, and divided by the optical loss squared. The key consequence: every 1 dB of optical loss costs about 2 dB of RF gain, because detected RF power scales with the square of photocurrent. Fiber attenuation, connectors and splitters therefore hit an analog link twice as hard as you might expect.
2. Noise sources at the output
- Thermal noise of the receiver load and amplifiers.
- Shot noise, proportional to average photocurrent.
- Laser relative intensity noise (RIN), quoted in dB/Hz. RIN noise power scales with the square of average photocurrent, so it tends to dominate at high received optical power.
Which term dominates depends on received optical power. At low power, thermal noise usually dominates; as power increases, shot noise and then RIN take over.
3. Noise figure
Link NF follows from output noise divided by gain and the input thermal noise kT. Because typical passive links have negative gain, NF values of tens of dB are common without amplification. Placing a low-noise amplifier before the laser improves system NF (Friis), at the cost of reduced input dynamic range.
4. Dynamic range
Spurious-free dynamic range (SFDR) is set by the noise floor and the link's third-order intercept. Laser nonlinearity and clipping usually set the upper limit.
5. Practical checks
- Keep optical return loss high: reflections into a laser degrade RIN. Angled (APC) connectors are standard in analog links for this reason.
- Budget every connector and splice.
- Measure with the actual laser bias and temperature conditions.
If anyone has measured data comparing DML and externally modulated links at the same frequency, please share.
1. RF gain
For a directly modulated laser, link gain is roughly proportional to (laser slope efficiency x photodiode responsivity)^2, multiplied by impedance terms, and divided by the optical loss squared. The key consequence: every 1 dB of optical loss costs about 2 dB of RF gain, because detected RF power scales with the square of photocurrent. Fiber attenuation, connectors and splitters therefore hit an analog link twice as hard as you might expect.
2. Noise sources at the output
- Thermal noise of the receiver load and amplifiers.
- Shot noise, proportional to average photocurrent.
- Laser relative intensity noise (RIN), quoted in dB/Hz. RIN noise power scales with the square of average photocurrent, so it tends to dominate at high received optical power.
Which term dominates depends on received optical power. At low power, thermal noise usually dominates; as power increases, shot noise and then RIN take over.
3. Noise figure
Link NF follows from output noise divided by gain and the input thermal noise kT. Because typical passive links have negative gain, NF values of tens of dB are common without amplification. Placing a low-noise amplifier before the laser improves system NF (Friis), at the cost of reduced input dynamic range.
4. Dynamic range
Spurious-free dynamic range (SFDR) is set by the noise floor and the link's third-order intercept. Laser nonlinearity and clipping usually set the upper limit.
5. Practical checks
- Keep optical return loss high: reflections into a laser degrade RIN. Angled (APC) connectors are standard in analog links for this reason.
- Budget every connector and splice.
- Measure with the actual laser bias and temperature conditions.
If anyone has measured data comparing DML and externally modulated links at the same frequency, please share.