Wideband Controller Calibration

[Owen Thomas]

G'day gents,

I have discovered that our oxygen sensor controller has been out-of-whack for quite some time, despite my attempts at calibration. We do not have easy access to a dyno, so finding this out an hour into a tuning session is a considerable waste of very precious time. We use this guy (http://www.aemelectronics.com/inline-wideband-uego-controller-1159) (AEM), and calibrate the ECU to accept the range of voltages shown in the instructions PDF.

So here's the question: What do other teams do to make sure your wideband is reading correctly? I'm all for sticking my schnoz in an exhaust stream and gauging that way, but the accuracy just isn't quite there. Cheap gauges? Second controller/sensor? I would also be interested in hearing peoples opinions on what could cause an off calibration on a "never needs to be calibrated" unit (assuming sensor is OK). Please discuss.

Second: We may need to acquire another controller, and while having two of the same type makes the most sense, I just don't really trust things with the AEM stamp (long story). So, I browsed around and the Innovate LC-1 (http://www.innovatemotorsports.com/products/lc1.php) seems to be a popular and easy solution, but it has mixed reviews. There are old topics discussing it, as well as some alternatives, but I'd like to hear some up to date testimonials regarding the different options.

Thanks,

[jlangholzj]

We've got an innovative unit on our dyno from before our days of running a MoTec. It works fine and and was our primary source of measurement at the time. The Motec has a built in air calibration feature, so thats taken care of now for us in the ECU.

Additionally, there are several types of widebands available. We use an LSU 4.9 sensor which should have a "calibration number" on the sensor itself. Again, the motec also has the option to put in this calibration number to set the baseline.

Depending on the options of the setup, I'd assume you'd be able to use one of the above options and if not, (again assuming) you should be able to measure the actual voltage output of the sensor once its nice and hot in open air to set a level for baseline.

[Mbirt]

Is your issue the output of the WBO2 controller itself or the ECU's transfer table that turns it into a value to use and show you? If it were the latter, I guess the premise of your questions wouldn't make sense. The AEM UEGO gague controller I've used never requires calibration input from the user because it reads the calibration resistor at the LSU 4.2's connector. Most WBO2 controllers (Innovate and Powerdex at least) use a free-air calibration process. If your AEM WBO2 controller is supposed to read the calibration resistor, I would suspect failure of the sensor or controller (helpful, right?). Do you have another LSU 4.2 to try?

I've never had a problem with the NTK Powerdex AFX: http://www.summitracing.com/parts/ngk-91101. Greg Banish claims in "Engine Management: Advanced Tuning" that it shares much with the OEM lab-grade ECM AFM1600. It has an incredibly easy free-air calibration routine and the typical 0-5v analog output. The downsides are that the display only shows gasoline AFR and it will only go to 1.09 lambda on the lean side of things.

That's cool that you guys use the 4.9, John. Do you know of any name-brand contorllers for it that aren't AiM or MoTeC? Ecotrons and Depo come to mind, but there isn't much feedback available about those lesser-known units.

[Owen Thomas]

Mbirt,

I am assuming the issue is that the signal from the UEGO controller to the ECU is (somehow) wrong. The transfer table in the ECU is a re-construction of the one provided in the AEM instructions PDF. All I know is that it was reading ~1.5 AFR rich compared to the tuners in-pipe wideband. I can of course modify the transfer table until I get correct values of AFR, but I have no way of reading said AFR except through the ECU! So any change I make to the table currently is really just tampering without being able to verify the results. We have different LSU 4.2's I can put in, but the one in there now appears to be in good condition, and is relatively new. Again, no way to verify if it is properly calibrated or not.

Unfortunately, the more I think about it, it seems the solution will be to buy another controller or a gauge to check against.

[jlangholzj]

Originally posted by Owen Thomas:
Mbirt,

I am assuming the issue is that the signal from the UEGO controller to the ECU is (somehow) wrong. The transfer table in the ECU is a re-construction of the one provided in the AEM instructions PDF. All I know is that it was reading ~1.5 AFR rich compared to the tuners in-pipe wideband. I can of course modify the transfer table until I get correct values of AFR, but I have no way of reading said AFR except through the ECU! So any change I make to the table currently is really just tampering without being able to verify the results. We have different LSU 4.2's I can put in, but the one in there now appears to be in good condition, and is relatively new. Again, no way to verify if it is properly calibrated or not.

Unfortunately, the more I think about it, it seems the solution will be to buy another controller or a gauge to check against.

Birt, I don't. I guess running the LC-1 all we had was the 4.2 for it but had a bunch of 4.9 sensors laying around the shop. Someone before me didn't take the time to research what was what I'm guessing. I can't recall if the innovative unit allows a 4.9 or not...hmm....anyway....


Owen, When you say the "in pipe" afr meter, are you talking about the meter that is inserted in the rear of the exhaust? If so, food for thought: Are you running a slip pipe (ie is the pipe held on by springs)? if so, does the pipe fit on the inside or the outside of the exhaust manifold. There's one configuration there that can potentially suck in outside air...I can never remember which.

I guess an AFR reading difference of 1.5 is only about 0.1 lambda (assuming you're running 91 ron) and Its not something I'd be terribly worried about especially if you're on the rich side. Of course, being close is never really good enough anyway now is it? http://fsae.com/groupee_common/emoticons/icon_razz.gif

Keep in mind that the whole reason behind the 4.9 sensor is that there was an inherit fault in the 4.2 series. The 4.9 compares the pumping current to the reference current to determine the amount of oxygen in the exhaust.Whereas the 4.2 has a reference air cell that can (and usually) will be contaminated by the exhaust or other pollutants. This will typically shift it low (if you've got more crap in the cell its less O2 right?) so you'd be reading rich. There's some more documentation out there on this but AFAIK, thats the reason why bosch switched to the 4.9.

[Owen Thomas]

Yes, it was that kind of AFR meter, but it was hella long. We had a 20" muffler on and it was pushed all the way into the secondary, so I don't think air leaking in was a problem. The slip on muffler goes around the outside of the secondary pipe and is secured by a worm drive exhaust clamp (like a badass hose clamp), just for reference.

Interesting point about the construction of the 4.2. Any idea if this reference cell can be visually inspected or cleaned? I just sort of assumed these things were near indestructible what with the abuse tuners put them through. We were running rich for a little while with that sensor, so it could be the problem, I will check it against a new sensor when we have our engine running again.

[Mbirt]

It would be perfectly logical that you're reading rich compared to a tailpipe sniffer, especially if there are air leaks or poor setup of the sniffer. The first and easiest test would be to try another sensor. If you get the same reading, see if anyone has a WBO2 controller you might be able to borrow and try on your existing sensor.

How about this--if it reads 1.5 afr rich of the tuner's setup, try mapping in some more fuel and then do a plug chop. I would think that you would be able to see a color difference between 11.5:1 and 13:1.

Definitely don't change the lookup table in the ecu to compensate.

[Charlie]

Narrowband sensors measure a lambda value of 1 very reliably. They don't have the range of a wideband, but they can be a cheap and valuable calibration tool for wideband sensors.

Anything between 300 and 600 mV is going to be very, very close to lambda = 1.

[Owen Thomas]

Originally posted by Charlie:
Narrowband sensors measure a lambda value of 1 very reliably. They don't have the range of a wideband, but they can be a cheap and valuable calibration tool for wideband sensors.
I completely forgot about narrow band sensors and thier intended purpose. Thanks Charlie! Probably able to rip one out of my Honda and rig something up fairly easily, too. Definitely cheaper than buying a gauge http://fsae.com/groupee_common/emoticons/icon_smile.gif