O2 sensor voltage
A narrowband sensor switches; only a wideband measures. That one distinction decides which of these two tables your reading belongs in — and whether the number in front of you can be converted to an AFR at all.
| Reading | Mixture | What it does not tell you |
|---|---|---|
| 0.10 – 0.20 V | lean of stoichiometric | how lean — 0.15 V covers λ 1.02 and λ 1.40 alike |
| 0.30 – 0.40 V | slightly lean, crossing down | nothing usable for tuning |
| ~0.45 V | the switch point, λ 1.00 | a steady 0.45 V usually means a cold or dead sensor, not a perfect mixture |
| 0.50 – 0.70 V | slightly rich, crossing up | nothing usable for tuning |
| 0.75 – 0.90 V | rich of stoichiometric | how rich — 0.78 V is not an AFR and cannot be converted to one |
| not moving at all | sensor lazy, dead, unheated, or the loop is open | a healthy narrowband at cruise crosses 0.45 V roughly once or twice a second |
This is the answer to "what AFR is 780 mV": rich of stoichiometric, and nothing more precise than that. The zirconia cell's output collapses into a near-vertical step at λ 1.00, which makes it excellent at finding stoichiometric and useless at measuring distance from it. Tuning a power mixture on a narrowband is guessing with extra steps.
| Output | λ (AEM/PLX) | AFR petrol | λ (Innovate) | AFR petrol |
|---|---|---|---|---|
| 0.00 V | 0.680 | 10.0 | 0.500 | 7.3 |
| 0.50 V | 0.748 | 11.0 | 0.600 | 8.8 |
| 1.00 V | 0.816 | 12.0 | 0.700 | 10.3 |
| 1.50 V | 0.884 | 13.0 | 0.800 | 11.8 |
| 2.00 V | 0.952 | 14.0 | 0.900 | 13.2 |
| 2.50 V | 1.020 | 15.0 | 1.000 | 14.7 |
| 3.00 V | 1.088 | 16.0 | 1.100 | 16.2 |
| 3.50 V | 1.156 | 17.0 | 1.200 | 17.6 |
| 4.00 V | 1.224 | 18.0 | 1.300 | 19.1 |
| 4.50 V | 1.292 | 19.0 | 1.400 | 20.6 |
| 5.00 V | 1.360 | 20.0 | 1.500 | 22.0 |
The mapping belongs to the controller, not the sensor: the same 2.50 V is λ 1.02 on an AEM-convention output and λ 1.00 on an Innovate default. Check what yours is set to before you log anything — and check it again if you ever reconfigure it, because most controllers let you change the endpoints. Boxes that output 0.5–4.5 V use the same idea over a narrower swing, so the volts-per-lambda figure changes but the method does not.
| Type | Wires | Use it for |
|---|---|---|
| Narrowband zirconia | 1 – 4 | closed-loop trim, checking a catalyst is working, confirming stoichiometric |
| Narrowband titania | 4 | same job, different electrics — resistance-based, and the voltage sense is often inverted |
| Wideband (LSU 4.2 / 4.9, LSU-ADV) | 5 – 6 | every actual measurement: WOT mixture, fuelling tables, logging |
| Wideband raw, no controller | — | nothing — the cell reports a pumping current that needs a controller to become a number |
A wideband sensor is a consumable. Leaded fuel, silicone sealant and repeated cold-start condensation all shorten its life, and a tired one reads lean before it fails outright — which is the dangerous direction to be wrong in.