Out of spec o2 sensor voltages when connected, but fine when not!

[smokin]

Help requested from the forum before I trouble the factory (which is a long drive if I need to take it in) if I may

Background:

Car seriously in the red during MoT emissions test, even with Cat on. Gave me a pass with an advisory saying 'fluctuating emissions' - I love this garage. However, it was a similar story last year, but with a run round the block to get everything hot and some rev'ing in the bay the inspector managed to get me in the green. I replaced my lambda sensor under advisement from factory soon after (so it's 12 months old).

So, happy to have the pass but annoyed I didn't pay close attention to the readout during the test - I don't have the results printout you get with a fail. I do recall CO was red around 2.5% (0.2% pass) and Lambda was in the red but didn't see by how much.

So I purchased a resonably cheap CO gas analyser and it reads 2.5 at idle and drops to 1-1.5 at 2k rpm. Still a long was off and reads roughly same as garage. But when I disconnect the O2 sensor there's no change in CO reading or engine note.

So I decide to take voltage readings from the sensor. When connected to the car wiring, the green (ground apparently) wire reads over 2v AND so does the white! There are two other wires, both black and these have 12v (so guessing the heater). Googled and it's not usual for the sensors to fail high.

Then decided to disconnect sensor wires and test voltage at the sensor, unplugged but still installed. Voltage across white/green was in the 0.2 to 0.9v range but not fluctuating and actually reading low 0.4v, when I'd expected a high reading as I was expecting it to be running rich.

I've had the engine out over the winter but I'm reasonably sure I connected everything back up correctly (I took lots of pictures) and I believe I've had this problem since last years MoT. The car seems to run ok but I don't have a frame of reference.

Does anyone know which type of sensor is used in the Atom II? Apparently the Zirconia (most common) ones can't generate a voltage above 1v and supplying a voltage can break them. The rarer Titania based ones vary the supplied voltage.

With ignition on and sensor disconnected, I measured the voltage at each of the four pins of the female, loom connector. I have +ve voltage at ALL pins. 12v on one and 1.85, 2.76 and 2.15v at others.

I suspect a wiring problem but before I do anything more I wanted to consult you guys.

Thanks

[smokin]

Don't want to get too excited too soon, but I may have cracked it.

Today I discovered there are narrow band o2 sensors and wide band o2 sensors. The output voltages of these different types are shown below.

I think I have a narrow band sensor and the car is setup for wide (wide is better). The voltages I've got align to those in the graphic, too closely to be coincidence (I hope). 

The wide band o2 sensor Honda part number is 36531-PRB-A01 and posts suggest Denso 234-9005 is a suitable alternative.

Sensor will be removed and inspected in the morning and a call to the factory Monday is planned

[benyeats]

Don't the outputs from lambda sensors oscillate so they need to be checked with a 'scope rather than a meter to get a true picture of what it is doing?

[smokin]

From what I've read, you're correct for narrow band. These are basic and only tell the ECU if the mix is rich or lean, not how rich or how lean. If the sensor reports rich then the ECU leans it out and if lean the ECU richens the mix. This happens a couple of times a second so the voltage from the signal wire on a narrow band fluctuates from 0.2v to 0.9v. Narrow band are crude indicators and not useful in tuning.

Wide band sensors alter the current draw depending on the mix they detect. The ECU then knows the exact ratio and doesn't have to flip-flop (oscillate) from rich to lean as with narrow band.

My problem stems from not knowing about wide band and using a Sealey lambda diagnostic tool which caters for Zirconia and Titania sensors but not wide band. The Sealey unit was giving really odd data - I think I now understand why.

I found this link on hondata very useful http://www.hondata.com/techk20-ctr-o2wiring.html
and this one too  http://www.enginebasics.com/EFI%20Tuning/AF%20Ratio%20Basics.html

It's been a frustrating but informative discovery so far. I have an OBD II kit coming from the factory so once fitted I'll have a better understanding of what's going on.

Cheers

[benyeats]

[quote="smokin"]
From what I've read, you're correct for narrow band. These are basic and only tell the ECU if the mix is rich or lean, not how rich or how lean. If the sensor reports rich then the ECU leans it out and if lean the ECU richens the mix. This happens a couple of times a second so the voltage from the signal wire on a narrow band fluctuates from 0.2v to 0.9v. Narrow band are crude indicators and not useful in tuning.

Wide band sensors alter the current draw depending on the mix they detect. The ECU then knows the exact ratio and doesn't have to flip-flop (oscillate) from rich to lean as with narrow band.

My problem stems from not knowing about wide band and using a Sealey lambda diagnostic tool which caters for Zirconia and Titania sensors but not wide band. The Sealey unit was giving really odd data - I think I now understand why.

I found this link on hondata very useful http://www.hondata.com/techk20-ctr-o2wiring.html
and this one too  http://www.enginebasics.com/EFI%20Tuning/AF%20Ratio%20Basics.html

It's been a frustrating but informative discovery so far. I have an OBD II kit coming from the factory so once fitted I'll have a better understanding of what's going on.

Cheers
[/quote]

OBD II is very good at diagnosing emissions problems, it should be as that was why it was invented!  I should think it will make all clear

Ben