Sory to fog things up with that Regina stuff. Should have remembered that long thread on the LH. With all those splices, the sensor is a good suspect.
Here's a "canned" response I made up for testing O2/ECU responses:

---------------------->
The O2 sensor/ECU interaction can be tested by sending simulated Full RICH and Full LEAN O2 sensor signals to the ECU, and watching for appropriate responses from the O2 sensor. This is done "Open Loop", (i.e., with the O2 sensor signal unplugged), and with Voltmeter on the O2 sensor lead. The engine and O2 sensor should be fully warmed up for this test.

Fake Rich Test: Apply a + voltage potential to the ECU Green wire. Use either the + terminal of a grounded 1.5 volt (Flashlight) battery. Or, if you have Orangutan arms, hold the ECU (green wire) terminal with one hand and put the other hand on the Battery Positive terminal. The ECU should see this as a "Rich" O2 signal and set a very lean fuel mixture. The Idle should worsen (may even stall), and the O2 sensor should respond with a correspondingly low output (.1 to .2 volts).

Fake Lean Test: Ground the Green wire going to the ECU (I use the braided ground on the valve cover). This simulated "O2 Lean" signal should result in very rich setting from the ECU . The RPMs should increase, and the O2 sensor output should rise to nearly a full 1.0 volt.

These tests are described in the the Bosch FI & Engine Management manual (p 6-22, "Rich stop" and "Lean Stop"), for Continuous Injection (CI) K-jet, but are equally applicable to LH injection.
-------------------------->

The Ford Mustang 3-wire sensor is frequently mentioned as a good "universal" replacement. But it has only 6" leads, as I recall. So I've used it where the sensor was in the manifold, but would be hesitant to have the splices exposed under the car, if the '89 O2 is in that area.

I've also read that crimp splices are the way to go, because the sensor gets its reference air down thru the wire strands—and soldering would prevent this. Sounds a bit far-fetched, but the source was credible. And some true "universal" sensors do come with crimp connectors supplied.

Re: make-specific sensors, I believe it's all in the length of the wires and the types of connectors. AFAIK, the only real difference in sensors is between the Bosch type (Zirconia, voltage generator) and the Regina (Titania, variable resistance).

--
Bruce Young
'93 940-NA (current) — 240s (one V8) — 140s — 122s — since '63.