L
Admin
[Technical article, rfforum Editorial Team]
Analog RoF and high-speed PD receivers often freeze an RF match from a dark or low-current S-parameter, then watch match, compression, and even apparent responsivity move as photocurrent rises. The DC bias / return network is usually the quiet culprit.
1. What changes with photocurrent
- PD junction capacitance and series resistance shift with bias and optical power; a match designed at one bias point drifts
- Bias-T / inductive return and decoupling that looked invisible at small signal can resonate or limit video bandwidth under large DC photocurrent
- IP3 / compression: the RF match and the bias network share copper — a "perfect" 50 Ω dark match can still be the wrong place for linearity under illumination (bias-drift cousins: /t/316)
2. Freeze list for a PD / RoF front-end claim
- Optical power / photocurrent next to every S22, NF, and IP3 number
- Bias network schematic (or at least L/C values and grounding) on the same traveler as the RF match
- Whether match was optimised dark, at nominal photocurrent, or under the worst-case optical overload you allow
- Monitor PD / tap path: if you use a monitor diode for bias lock, its loading belongs in the RF budget too
3. Habit
Do not publish a PD S22 without the photocurrent and bias voltage. A dark Smith chart and an illuminated Smith chart are different devices for bring-up purposes.
Discussion: on your last RoF or high-current PD bring-up, what moved first as optical power rose — S22, IP3, or a bias-loop spur — and do you require illuminated S-params in the acceptance pack?
— rfforum 编辑部 / Editorial Team
Analog RoF and high-speed PD receivers often freeze an RF match from a dark or low-current S-parameter, then watch match, compression, and even apparent responsivity move as photocurrent rises. The DC bias / return network is usually the quiet culprit.
1. What changes with photocurrent
- PD junction capacitance and series resistance shift with bias and optical power; a match designed at one bias point drifts
- Bias-T / inductive return and decoupling that looked invisible at small signal can resonate or limit video bandwidth under large DC photocurrent
- IP3 / compression: the RF match and the bias network share copper — a "perfect" 50 Ω dark match can still be the wrong place for linearity under illumination (bias-drift cousins: /t/316)
2. Freeze list for a PD / RoF front-end claim
- Optical power / photocurrent next to every S22, NF, and IP3 number
- Bias network schematic (or at least L/C values and grounding) on the same traveler as the RF match
- Whether match was optimised dark, at nominal photocurrent, or under the worst-case optical overload you allow
- Monitor PD / tap path: if you use a monitor diode for bias lock, its loading belongs in the RF budget too
3. Habit
Do not publish a PD S22 without the photocurrent and bias voltage. A dark Smith chart and an illuminated Smith chart are different devices for bring-up purposes.
Discussion: on your last RoF or high-current PD bring-up, what moved first as optical power rose — S22, IP3, or a bias-loop spur — and do you require illuminated S-params in the acceptance pack?
— rfforum 编辑部 / Editorial Team
