The telemetry a transceiver reports about itself — temperature, voltage, bias current, and transmit and receive optical power.
What DDM Reports
Every module of any consequence measures five things about itself and publishes them over the management interface: case temperature, supply voltage, laser bias current, transmit optical power and receive optical power. The switch reads them over the same two-wire bus it uses for identity.
These are not laboratory-grade measurements and they are not meant to be. SFF-8472 asks only for ±3 dB on optical power across the operating temperature and voltage range — and ±3 °C, ±3 per cent and ±10 per cent on the other three — so a single reading is a coarse indicator. The difference between two readings on the same module is far more accurate than either absolute value, which is why trending beats spot-checking.
Reading Receive Power
Receive power is the number that earns DDM its place. Take the far end's transmit power, subtract your receive power, and you have measured the loss of the plant between you without touching a meter. Compare it against what the design predicted — the link budget calculator gives the expected figure — and a discrepancy of more than a decibel or two points straight at the plant.
The habit worth building is recording that number the day the link is commissioned. A receive power of −9.4 dBm means nothing on its own; −9.4 dBm where it read −6.1 dBm at handover means 3.3 dB has appeared somewhere, and that is a fault report. The dB loss converter turns it into a fraction of the light lost.
What the Other Four Tell You
Transmit power drifting down while bias current climbs is a laser at the end of its life: the control loop is pushing harder to hold the same output. Bias climbing with transmit power flat is the same story earlier on — worth a planned replacement rather than an outage.
Temperature sitting within a few degrees of the high alarm is an airflow problem, not an optics problem — check filters and blanking panels before blaming the module. Supply voltage outside a few per cent of 3.3 V points at the host board or a badly seated cage, and makes every other reading untrustworthy.
What Goes Wrong
The commonest misreading is treating a low receive power as proof of a fibre fault when the far end is simply transmitting low. Always read both ends. The second commonest is trusting the absolute value: two modules on the same link can disagree by several decibels and both be within specification.
Externally calibrated modules are a genuine trap. The raw counts mean nothing until the host applies the calibration constants stored in the module, and a host that ignores them prints numbers that look plausible and are wrong. CMIS modules are internally calibrated, which removes the problem.
Where You Meet It
First step of nearly every optical fault and the basis of every sensible monitoring system. Poll receive power on every link, alarm on change rather than absolute value, and most fibre problems announce themselves days before an outage. The link-down diagnostic guide walks the sequence; the optical power converter handles the dBm arithmetic.
DDM reference
| Live values | SFP: address A2h bytes 96–105. CMIS: Page 11h, reported per lane |
|---|---|
| Thresholds | SFP: A2h bytes 0–39. CMIS: Page 02h |
| Temperature | signed 16-bit, 1/256 °C per count |
| Supply voltage | unsigned 16-bit, 100 µV per count |
| Laser bias | unsigned 16-bit, 2 µA per count |
| Optical power | unsigned 16-bit, 0.1 µW per count |
| SFF-8472 accuracy | ±3 dB optical power, ±3 °C, ±3% on Vcc, ±10% on bias |
| Calibration | internal, or external with constants the host must apply itself |
| Quantisation floor | 0.1 µW steps are about 0.4 dB apart at −30 dBm, negligible above −20 dBm |