F
Member
The RF link budget for a fiber-fed remote radio unit (RRU) is usually computed in two disconnected halves: the RF engineer works out the air-interface budget, the optical engineer works out the CPRI/eCPRI budget, and nobody checks whether the two halves meet at the same supply-voltage and temperature corner. This post walks through one worked example and points out where the dB actually disappear.
1. The two budgets are not independent
RF side (downlink, per antenna port, typical macro cell):
PA output at antenna port +43 dBm
Feeder + connector loss 1.5 dB
Antenna gain 17 dBi
EIRP 58.5 dBm
Receiver sensitivity (UE) -100 dBm
Allowed path loss 158.5 dB
Optical side (CPRI option 2, 24.33 Gbps line rate, 10 km):
SFP+ 1310 nm DFB, Tx power -5 to 0 dBm
Rx sensitivity -14 dBm
Optical budget 9 to 14 dB
Fiber loss 0.35 dB/km x 10 km 3.5 dB
Two connector pairs 1.0 dB
Two splices 0.4 dB
Total 4.9 dB
Margin 4 to 9 dB
Both halves look fine in isolation. The failure mode appears when the RRU is mounted at the tower top with the optical module inside a sealed enclosure that reaches +75 C in summer. That is where the assumptions break.
2. Where the dB actually go: temperature
A commercial-grade SFP+ is specified 0 to +70 C case temperature. An industrial-grade part is -40 to +85 C. Two effects matter:
(a) Extinction ratio degradation. DFB lasers lose ER as temperature rises. A module with 9 dB ER at 25 C may deliver 7 dB at 75 C. For 10 km 10G links that is usually acceptable; for 40 km it is not.
(b) DDM calibration drift. If the module's digital diagnostics are not temperature-compensated across the full range, the reported Tx power can be off by 0.5 to 1.5 dB at the extremes. This is the dangerous one, because the NMS will show a healthy link while the actual optical power is 1 dB lower than believed. Never trust DDM above +70 C without verifying with an external power meter.
Practical rule: if the enclosure exceeds +70 C, use industrial-grade optics and add 1.5 dB of extra margin to the optical budget, purely for measurement uncertainty.
3. Where the dB actually go: the CPRI jitter budget
This is the part most RF engineers never see. CPRI carries IQ samples with a strict timing relationship. When the optical link introduces errors, the radio framer either drops a basic frame or, worse, re-syncs. Symptoms at the RF layer:
- EVM degradation of 1 to 2 percent with no change in RF path
- Sporadic "radio link failure" alarms that clear on their own
- Uplink noise floor rising by 2 to 3 dB in certain time windows
All three look like RF problems. All three were traced, in our cases, to optical BER between 1e-9 and 1e-7. CPRI has no FEC in the classic sense; it relies entirely on the physical layer being clean.
So the acceptance criterion for a fiber-fed RRU should be BER better than 1e-12 measured over at least 10 minutes with a PRBS31 pattern, not "link is up".
4. The power-over-fiber trap
Some deployments use hybrid cable (fiber plus copper for DC power). Two things go wrong:
- The copper pair carries the RRU's return current, which can induce a ground potential difference of several hundred millivolts between the tower-top unit and the baseband unit. If the optical module's metal shell is bonded to the local chassis ground, that potential difference drives a common-mode current through the module housing.
- The fix is a 1 nF / 2 kV capacitor between the optical cage and chassis ground: DC isolated, RF bonded.
We have seen this manifest as intermittent CRC errors that correlate with the RRU transmit duty cycle, which is a very confusing signature until you know what to look for.
5. A checklist for the next deployment
1. Measure the actual enclosure temperature at the hottest hour, not the ambient. Add 10 C for safety.
2. Specify industrial-grade optics if enclosure temperature exceeds +70 C, and add 1.5 dB for DDM uncertainty.
3. Compute the optical budget using worst-case per-channel insertion loss, not typical.
4. Require BER better than 1e-12 over 10 minutes as the acceptance test, not link-up status.
5. For hybrid cable, verify the ground potential difference between ends before installing optics.
6. Log DDM baselines at commissioning so that later drift is detectable.
The RF budget is the one everybody reviews. The optical budget is the one that fails in the field.
-- F-tone Networks, Optical Interconnect Engineering
Technical resources: https://www.f-tone.com
1. The two budgets are not independent
RF side (downlink, per antenna port, typical macro cell):
PA output at antenna port +43 dBm
Feeder + connector loss 1.5 dB
Antenna gain 17 dBi
EIRP 58.5 dBm
Receiver sensitivity (UE) -100 dBm
Allowed path loss 158.5 dB
Optical side (CPRI option 2, 24.33 Gbps line rate, 10 km):
SFP+ 1310 nm DFB, Tx power -5 to 0 dBm
Rx sensitivity -14 dBm
Optical budget 9 to 14 dB
Fiber loss 0.35 dB/km x 10 km 3.5 dB
Two connector pairs 1.0 dB
Two splices 0.4 dB
Total 4.9 dB
Margin 4 to 9 dB
Both halves look fine in isolation. The failure mode appears when the RRU is mounted at the tower top with the optical module inside a sealed enclosure that reaches +75 C in summer. That is where the assumptions break.
2. Where the dB actually go: temperature
A commercial-grade SFP+ is specified 0 to +70 C case temperature. An industrial-grade part is -40 to +85 C. Two effects matter:
(a) Extinction ratio degradation. DFB lasers lose ER as temperature rises. A module with 9 dB ER at 25 C may deliver 7 dB at 75 C. For 10 km 10G links that is usually acceptable; for 40 km it is not.
(b) DDM calibration drift. If the module's digital diagnostics are not temperature-compensated across the full range, the reported Tx power can be off by 0.5 to 1.5 dB at the extremes. This is the dangerous one, because the NMS will show a healthy link while the actual optical power is 1 dB lower than believed. Never trust DDM above +70 C without verifying with an external power meter.
Practical rule: if the enclosure exceeds +70 C, use industrial-grade optics and add 1.5 dB of extra margin to the optical budget, purely for measurement uncertainty.
3. Where the dB actually go: the CPRI jitter budget
This is the part most RF engineers never see. CPRI carries IQ samples with a strict timing relationship. When the optical link introduces errors, the radio framer either drops a basic frame or, worse, re-syncs. Symptoms at the RF layer:
- EVM degradation of 1 to 2 percent with no change in RF path
- Sporadic "radio link failure" alarms that clear on their own
- Uplink noise floor rising by 2 to 3 dB in certain time windows
All three look like RF problems. All three were traced, in our cases, to optical BER between 1e-9 and 1e-7. CPRI has no FEC in the classic sense; it relies entirely on the physical layer being clean.
So the acceptance criterion for a fiber-fed RRU should be BER better than 1e-12 measured over at least 10 minutes with a PRBS31 pattern, not "link is up".
4. The power-over-fiber trap
Some deployments use hybrid cable (fiber plus copper for DC power). Two things go wrong:
- The copper pair carries the RRU's return current, which can induce a ground potential difference of several hundred millivolts between the tower-top unit and the baseband unit. If the optical module's metal shell is bonded to the local chassis ground, that potential difference drives a common-mode current through the module housing.
- The fix is a 1 nF / 2 kV capacitor between the optical cage and chassis ground: DC isolated, RF bonded.
We have seen this manifest as intermittent CRC errors that correlate with the RRU transmit duty cycle, which is a very confusing signature until you know what to look for.
5. A checklist for the next deployment
1. Measure the actual enclosure temperature at the hottest hour, not the ambient. Add 10 C for safety.
2. Specify industrial-grade optics if enclosure temperature exceeds +70 C, and add 1.5 dB for DDM uncertainty.
3. Compute the optical budget using worst-case per-channel insertion loss, not typical.
4. Require BER better than 1e-12 over 10 minutes as the acceptance test, not link-up status.
5. For hybrid cable, verify the ground potential difference between ends before installing optics.
6. Log DDM baselines at commissioning so that later drift is detectable.
The RF budget is the one everybody reviews. The optical budget is the one that fails in the field.
-- F-tone Networks, Optical Interconnect Engineering
Technical resources: https://www.f-tone.com
