ONU optical power: what the dBm means and how to read it on every OLT
A practical guide to the one reading that predicts fibre trouble before subscribers do: what a healthy RX level is, what the patterns across a PON port tell you, the command to check it on each OLT, and the placeholder values that look like readings.
What the number means
RX power is the light arriving at the subscriber’s ONU, in dBm. It is negative, and more negative is weaker: −22 dBm is a comfortable signal and −29 dBm is one dirty connector away from an outage.
| ONU RX power | Shown as | What RadixOLT does |
|---|---|---|
| Above −25 dBm | Healthy | Nothing. The reading is kept, so a slow slide shows up on the ONU’s history. |
| −25 to −28 dBm | Weak | Drawn in the warning colour with the word beside it, never by colour alone. |
| −28 dBm or weaker | Critical | An alert, raised once for an online ONU and updated with each new reading; on Pro it goes to Telegram. |
| No reading | — | A blank, never a zero. A down ONU reports no optics on these platforms. |
The bands are RadixOLT’s own, and the alert uses the same −28 dBm the colour does, so a red reading and an alert never disagree. They are fixed across vendors on purpose: one plant, one rule.
How low an ONU can go and still work depends on its optics. GPON optics are specified by class in ITU-T G.984.2: a Class B+ receiver down to −27 dBm and a Class C+ receiver down to −30 dBm, both with an overload point of −8 dBm. EPON optics come as PX20 and PX20+, whose receivers differ by a few dB. The datasheet of the ONU’s own optics is the authority.
Reading the pattern, not the point
One weak reading says little. Where the weak readings are says where to send someone.
One ONU weak, its neighbours fine
The problem is past the splitter: the drop cable, a bend, or the connector at the premises.
Every ONU on one port weak together
The problem is shared: the feeder fibre, a splitter, or a dirty connector at the OLT or in the splice box. Measured on one live fleet: 5 of 128 online ONUs at or below −28 dBm, all five on the same PON port.
Every ONU on one port dark at once
A cut. The OLT reports a fibre-type reason for each ONU in the same sweep, and RadixOLT sends one summary naming the port instead of a message per subscriber.
A slow slide over days
Contamination or a stressed bend getting worse. Every sweep is kept as a point on the ONU’s RX history, so the slide is visible before it crosses −28.
Reading it on each OLT
The commands to check one ONU or one port by hand. RadixOLT reads the same values on a schedule, so a terminal is for the follow-up, not the search.
| Platform | Command | What it shows |
|---|---|---|
| Huawei | display ont optical-info <f/s> <port> <ont-id> | One ONT: RX and TX power, voltage, bias, temperature. |
| ZTE | show pon power onu-rx gpon_olt-<shelf/slot/port> | RX power for every ONU on one PON port. |
show pon power attenuation gpon_onu-<shelf/slot/port:id> | One ONU’s levels in both directions, and the loss between them. | |
| Genexis | show ont optical-info <slot/pon/ont> | One ONT’s optical readings. |
| TP-Link | show ont optical-info gpon 1/0/1-8 | Every ONU on all eight ports at once: what RadixOLT’s sweep reads. |
show ont optical-info gpon 1/0/<port> <ont-id> | One ONU. | |
| Syrotech EPON | show onu <onu-id> ctc opm_diag | Inside interface epon 0/<pon>. Values like 1.73 mW (2.39 dBm). |
| Syrotech GPON, DBC | SNMP only | show onu state has no optical column, so RadixOLT reads optics over SNMP, ONU by ONU. |
| Optilink OP-EOLT_11444 | show optical-diag <pon-port> | Every ONU on one PON port, in config mode. |
| Optilink generic EPON | show ont optical-info <pon> all | Every ONU on one PON port. |
Readings that are not readings
Each firmware has its own way of saying “no value”, and several of them look like numbers. Read naively, a dark subscriber shows a healthy level and no alert fires. RadixOLT drops every one of these and shows a blank.
| Platform | A dark ONU reports | Shown as a reading, it would say |
|---|---|---|
| Huawei | 2147483647 on every optical column | Two billion dBm |
| Syrotech GPON | 0.00 on all five columns | A perfect signal on a subscriber who is off the air |
| DBC | NULL | Nothing, unless a converter turns it into 0 |
| Syrotech EPON | An empty string on all five columns | A measured zero |
| Optilink OP-EOLT_11444 | -2147483648 | Minus two billion dBm |
| Genexis | N/A or -, as text | Nothing, unless a parser accepts a placeholder that starts with a digit |
Units are per firmware too
- Optilink OP-EOLT_11444 answers in hundredths:
-2376is −23.76 dBm. Checked against the chassis’s ownshow optical-diagfor the same ONU. - Syrotech EPON answers text:
1.73 mW (2.39 dBm). The dBm in brackets is the reading, not a conversion of the milliwatts. - Genexis answers text for every value, optical power included:
"-20.45".
Every platform here is swept every 15 minutes, and on Huawei, ZTE, Genexis, TP-Link, Syrotech and DBC an open ONU page re-reads that one ONU every 5 seconds over SNMP, without taking a CLI login.