Some comments to the below:

>> That said, your situation with disconnecting the O2 sensor instead of the
>> gauge and the resulting response tells me that the gauge is a problem in
>> itself.
>
> Only if disconnecting the gauge makes the symptoms disappear...

That will be tested today. Yesterday's conditions were: O2 sensor
disconnected from ECU input and meter input. Meter input and ECU input
still connected together, meter sat steady at midscale. Unless, as was
mentioned, the signal line had a lot of high frequency noise on it that
the meter could not display.

<snip>

>
> True -- although Mike said that the meter stayed centered, suggesting
> that the ECU was back in open-loop mode?

I don't see how that indicates the ECU was in open loop, only that the
bias voltage at the ECU input did not change -- right? Pin 5 (O2 sensor
signal) is just a high-impedance input, yes?

Possibly useful would be to disconnect the sensor signal from the ECU
input and connect it to the meter to monitor what the engine is doing when
the ECU can't see the sensor. David said something about biasing the
sensor with a half-meg resistor?

--
Mike Elliott