New wiki and manual has launched! Check it out at
Help with building your Speeduino, installing it, getting it to run etc.
By Black Knight
Narrow Band O2 sensor can tune nicely.

Just some back ground on what I have done over the years.

I have been tuning cars before there was an O2 sensor.

I am so glad that they were invented because they sure help with good tuning. I have been using NB O2 since the early 1990s.
I have tuned mechanical injection systems with them
(k jetronics) All types of carbs including propane. NB O2 has worked very well using a voltmeter.

Now I just got into ECU building ten months ago. So far I find them very fun and good to work with.

Then I kept reading that NB O2 cannot tune correctly yet I have been using NB O2 for nearly thirty years.
So I set out to find out what was the truth.
How can it work for 30 years and now they don’t work on ECUs?

I can get volt readings from 1.2v down to minus .7 volts with an NB.

Then I get to the point where I am calibrating my NB O2 on TS and it just fell off and gave no readings on TS below .4 v.
So I set out to find out why.

This lead me to need to learn about electronics. I took readings of both volts and amps of the NB O2. The volts were there but the amps drop off very quickly around .4v. So I concluded the O2 is just fine and it had something to do with the ECU.

It is the forward voltage or voltage drop of the ECU making the NB O2 not work. Mostly.

Now I was on a mission to beat this problem.

I got a good WB O2 and a good NB O2 and set up a test.
I created a controller that could work with the NB O2 and use a friends WB O2 for a comparison.
I like the WB. It works very well. But I was still on a mission to get the NB to work. So lots of learning of electronics and hard work I got the NB to start to give a wide range of volts/AFR to TS.
I finally got both NB and WB calibrated to work well and liked both.

But with all I had done on this test I should finish the NB controller and it’s calibration to see what it really could do.
In the end the WB can be more accurate in the extreme rich and lean areas but they are in AFRs that we don’t need.
Just to be clear. I like both WB and NB now.
My NB can show on TS and MLV as lean as 23 afr and as rich as 7 afr and will tune correctly with in about .5 afr of target AFR.

I set my NB on the controller to 11 afr and 20 afr because any more is not needed. On my NA economy car I set the controller to 12 afr and 18 afr. With that and setting up the calibration on TS to match tuning is much easier.

In conclusion I find WB just slightly better at the extremes but the NB is cheaper and less likely to have an internal heater to fail.
I have an NB that is 25 years old and it still works and is accurate.

I like both and can tune well with both so you will get no conflict from me.

I have now got both calibrated dead on and both now have dead on lambda delay set up with my new way of finding delay.

I am new to this and find all of it fascinating and fun when it is not being frustrating.

Just sharing.

Hope it helps someone.

Thanks to all here I can now install, configure, calibrate, base tune set up and road tune and hand tune.

There is much more to this and it would be hard to find the words to make the test work clear on a forum.

I will try to answer questions if I can, but understand I am new to ECUs, electronics and just started tuning 2 months ago, so I may not have all the lingo yet to say it right. I will do my best.


Black Knight
By noisymime
Whilst I won't ever knock someone's preferences if you like working with NBs, there are very good technical reasons why they shouldn't really be used for actual tuning. In (very) short terms, NB sensors are only accurate around their set point, typically 14.7:1. Either side of this point, not only are they extremely non-linear in their response, but the response curve varies greatly with temperature, sensor condition etc.
nb_sensor_output.jpg (17 KiB) Viewed 149 times
If you look at a typical NB response curve like this, you've got roughly 0.25v between 12:1 and 14.7:1, which is an area that is critical for tuning. 0.25v fed through a 10-bit ADC gives just 51 points of resolution, which is then scaled down in Speeduino by 2, so just 25 increments. IMHO 51 (let alone 25) is insufficient resolution points to tune a such critical area, and that is even before you account for noise, sensor variation etc.

For contrast, a (near) linear response WB controller might give a range of 10:1 to 20:1, which in the same 10-bit ADC gives you more than 5x as many points of resolution over that same 12:1 to 14.7:1 range.
By Black Knight
Thanks much for the response.

I have seen all this data and get the point behind it.

I am not trying to push one O2 over another. My main objective is to set an accurate base tune so auto tune is as good as it can be.
So far so good.

Just sharing that the NB may be more capable than the data on the Internet is claiming.

With my NB controller I am seeing stable AFRs way beyond that table above. Yes temp is important, very. That is true for all O2s.

The data on the TS O2 log file shows something very different than the Internet chart with my controller but shows somthing quite similar with out it.

Just to be clear. I am not in favor of one O2 over the other. I just had some fun making the NB work. Just a project.

I am a little shocked that it did work at all.

By all the info on the net it should not be working.

So perhaps the data is off, or perhaps I stumbled on to something that made it work better. Who knows.

I will enjoy testing both NB and WB until the truth proves out.
Just a hobby in the journey of learning ECU tuning.

Until I prove it out I will always confirm my tune by using a WB.

I have done road tunes with both and not seeing a great difference. WB is a little easier.

Before my controller was used NB was a pain in the ***.
Twitchy and unpredictable. Just like all the info says.
So I would agree that by itself NB is not so good.

It is more important to get things like the list below right.

Voltage correction
Dead time
O2 calibration dead on
Delay time
Sensor filters
All sensor calibration

And all the other things that make the tune work right, done first and as accurate as possible.

I really do want to thank you all for starting this Speeduino project as I am really enjoying what I can do with it.

I will always test the limits of what is possible so I can learn more.

I think the reply by noisymime is a good one and a good warning that until other wise proven go with the safe and proven way.

There are enough challenges with ECU install, calibration, and tuning with out going off the proven path.

Thanks to all again.

Listen to noisymime.

Black Knight
Hunting Idle: Caused by DFCO

Ahhh I remember this problem well. Took me way too[…]

I've been chasing a hard starting issue on my car […]

data logging

Will the engine continue to run? Are you using tu[…]

The teensy adaptor boards have supplementary pin i[…]

Still can't find what you're looking for?