Hey guys,

I can't really find a clear answer out there on the AEM calibration values in ECMLink for AEM PN 30-4110 (Classic Wideband). Yes, I've watched this video: ECMLink AEM WB Calibration, read the ECMLink writeup about how AEM widebands don't play well ECMLink. I've read pretty much every DSMTuners forums post on this issue. I'm kinda lost at this point.

My wideband has never really read accurately, no matter what values I put into it (Very possible it's user error). I have the wideband setup correctly in ECMLink, minus the calibration values. I just took it out and did my 2500-4500RPM pull for my tuner, and he mentioned that my AEM WB in the log was going extremely rich (9.8 AFR) however when I watched the gauge, it never went below 11AFR.

Here's the voltage values for my wideband (page 11): https://documents.aemelectronics.com/64fa707c21a65cf90802b55982b08abbdde8ee80.pdf
I did what the video said at the top of this post (I put the voltage and lambda values for 11AFR for min and 14.7AFR for max). I noticed that once it wanted to go less than 11AFR in the log, it just went straight to 9.8AFR. I also attached the log to this post.

Have you guys found any voltage and lambda values that worked for you for this part number? And do any of the preset settings in DSMLink work with this wideband?

Thanks guys,
Kyle

Any idea which brand sensor you have? I'm running a diffrrent model AEM but the FAE sensor would display fine on the gauge but was all over the board in ECMLink. I even made a filter as suggested from ECMTuning and was no better. Once I switched to the Bosch sensor, suggested by AEM support, everything worked perfectly without any fine tuning.

dwb said:

Any idea which brand sensor you have?

Click to expand...

I have the Bosch sensor as well. Before I went about calibrating it, it still read incorrectly even on this sensor.

I did have better results after putting in different settings yesterday. At least it wasn't going super rich. The sensor saw a low of 10.9 and the video I took of the sensor said 10.6 at that time. Here's the settings I put in.

for starters, volt range should be 0-5, not 0-4.

nice97civic said:

for starters, volt range should be 0-5, not 0-4.

Click to expand...

Gotcha, I just put that value in because the 1.23 lambda correlates with the 4V, if that makes any sense. This table from AEM doesn't go past 18.50AFR and ECMLink doesn't allow 3 decimal places in the volt or lambda spaces.

If you have your wideband gauge grounded to chassis ground instead of sensor / ECU ground you can have voltage offset as well which usually causes most of the offset issues.

Anfurnyy said:

If you have your wideband gauge grounded to chassis ground instead of sensor / ECU ground you can have voltage offset as well which usually causes most of the offset issues.

Click to expand...

There's not a dedicated sensor ground wire attached to this wideband I don't think... I'm pretty sure the X-Series does though. Do you mean instead of connecting the black chassis ground wire directly to the chassis, connect it to the sensor ground on the ECU?

kyler021 said:

There's not a dedicated sensor ground wire attached to this wideband I don't think... I'm pretty sure the X-Series does though. Do you mean instead of connecting the black chassis ground wire directly to the chassis, connect it to the sensor ground on the ECU?

Click to expand...

I'd have to read through some documents and old posts but IIRC you should be able to wire the ground of just the gauge to sensor ground on the ECU so that both the ECU and gauge have the same ground reference so there's not any offset between the 2.

Just make sure not to wire the actual wideband sensor circuit ground to the ECU ground cause that will most likely fry your ECU.

Pretty sure this is how I had mine setup in my 1g before switching to Link G4+ and a CANlambda and I didn't have any issues.

Also like someone else mentioned, make sure your voltage scaling is setup right in Link too.

Anfurnyy said:

Also like someone else mentioned, make sure your voltage scaling is setup right in Link too.

Click to expand...

I think this is the main problem I had... How should I go about doing this? Because a lot of the lambda numbers and voltage outputs for the sensor have 3 decimal places in the AEM paperwork, meanwhile ECMLink only allows 2 decimal places, so I guess it'll always be a little off? Or should I just estimate what I think it should be?

Anfurnyy said:

I'd have to read through some documents and old posts but IIRC you should be able to wire the ground of just the gauge to sensor ground on the ECU so that both the ECU and gauge have the same ground reference so there's not any offset between the 2.

Click to expand...

I'll try and give this a look too. I'm not the one who wired it, so I'm not totally sure how things are wired.

Just set your min voltage to 0 and your max voltage to 5

You could also try not setting it up as linear. It's been a while since I used an AEM gauge so not sure if this is required or recommended but doing it non-linear removes the voltage requirement.

I had to get creative with my Series X. Couldnt get it to read correct. my min max V and Lam are min 0.85V 0.74Lam Max 3.12V 1.0Lam I set this based off the raw voltages. I tried 0-5V and it wasnt even close

Well I stand corrected then .

that was just me haha the older styles are different. I gave mine away and just got the X because it was completely unreliable. Im sure someone smarter than I could get it to work.
I think like stated above, he needs to verify he is on P0 and P1
should be something like 0V=0.68lam and 4.99V=1.37lam if you continue the scale

kyler021 said:

I'll try and give this a look too. I'm not the one who wired it, so I'm not totally sure how things are wired.

Click to expand...

The 30-4110 only has one ground wire and it is a chassis ground, a power ground. So you can't connect it to the ecu sensor ground input - it would put too much current through the sensor ground circuit in the ecu. It has to connect to the chassis.

kyler021 said:

I think this is the main problem I had... How should I go about doing this? Because a lot of the lambda numbers and voltage outputs for the sensor have 3 decimal places in the AEM paperwork, meanwhile ECMLink only allows 2 decimal places, so I guess it'll always be a little off? Or should I just estimate what I think it should be?

Click to expand...

2 decimal places is enough for this. You don't have to worry about not having the 3rd decimal place. That's the least of your worries.
The voltage and lambda numbers you show in post #3 are correct per the AEM chart, but a lot of people have found that the chart numbers don't actually work right.
That's why people sometimes use the lock in open loop method to find their own numbers for volts versus AFR.
That's what we did with OhhShmit's car, which has a "Classic" type AEM wideband. We first tried setting it up per the chart like you show in post #3 and that was wrong. Then we tried some of the other presets and they were wrong.
Then we used the lock in open loop method and got numbers that seemed ok, but really you can only verify it at around 14.7 in closed loop. In the extreme rich region there's nothing to really verify correctness. His car has a for real narrow band O2 sensor, which really helps in the stoic region. Better than just simulating O2 with the wideband.

So he ran his car for about a week with our newly found volts vs AFR numbers, things seemed ok, and then we noticed the logged AFRs were starting to drift a little from what they were, in closed loop where you can tell right or wrong (sort of). As if the wideband was not stable.
So I think the bottom line is, you gotta have a wideband that a decent number of people have found to be reliably accurate, and the 30-4110 is not one of those. Another thing that helps is as I said already, it's better to have a real narrow band to do the narrow band function, instead of simulating it from the wideband. Locate the wideband in the downpipe, and put the narrow band in the regular O2 housing location.

Anyway when we used the lock in open loop method to get our own numbers for OhhShmit, we had it locked in open loop for both pairs of numbers, both the rich and the lean pairs. We didn't want to run it so rich or so lean that it would be mis-firing or any other bad running behavior. At a fast idle we went rich to a gauge AFR of 11.6. Any richer than that and it didn't run clean enough. On the lean end we went up to 16.2 on the gauge with clean running. We got 3.82 raw volts at 16.2 AFR and 1.51 volts at 11.6 AFR. Converted AFR to Lambda using the AEM chart (you can interpolate). Put the numbers into the LinWideband scaling. And he is still going with those numbers, although as I said they have drifted a little, so probably will need to do it again. In the end I think he will buy a different wideband. He is using the 1g EGRtemp input which is a crummy input. Your 2g Rear O2 input is a better input so you might have better results.

Also, when I put your post #3 numbers into the scaling in your log, it looks kind of reasonable. It might actually be pretty close.
You could try to verify it with the lock in open loop method at 2 points.
For your lean point, don't just assume that closed loop running is at lambda 1 like that video says. Do an open loop point at a lean AFR.
We didn't use an SD cell to make it richer and leaner. We used Global deadtime. You could also use Global Fuel. Use "Save changes to ecu as they are made".

Don't worry that your 2 points are not at the very top and bottom of the scale. Remember that in math and geometry, 2 points define a line of infinite length. And that's how the right side of that scaling chart works. The Limits of the line are determined on the left side of that chart. There you can either leave the default numbers in it, or you can put in 10 and 18 like you did in post #3.

BTW I tried OhhShmit's scaling in your log from post #1 and it doesn't look right for yours.
So to verify your scaling, I think you would need to come up with your own numbers by the lock in open loop method.

We're on Boost said:

Anyway when we used the lock in open loop method to get our own numbers for OhhShmit, we had it locked in open loop for both pairs of numbers, both the rich and the lean pairs. We didn't want to run it so rich or so lean that it would be mis-firing or any other bad running behavior. At a fast idle we went rich to a gauge AFR of 11.6. Any richer than that and it didn't run clean enough. On the lean end we went up to 16.2 on the gauge with clean running. We got 3.82 raw volts at 16.2 AFR and 1.51 volts at 11.6 AFR. Converted AFR to Lambda using the AEM chart (you can interpolate). Put the numbers into the LinWideband scaling. And he is still going with those numbers, although as I said they have drifted a little, so probably will need to do it again. In the end I think he will buy a different wideband. He is using the 1g EGRtemp input which is a crummy input. Your 2g Rear O2 input is a better input so you might have better results.

Click to expand...

Okay... Let me see if I understand this...

So, I need to lock in open loop, find the points (rich and lean) where it will still run normally without any weird behaviors, log the wideband voltage at those points and plug that AFR and voltage into the boxes on the calibration for the wideband? And even then, the gauge can fall out of calibration eventually. Man, I love this car hahaha.

Also, when I had it calibrated to the 11AFR - 14.7AFR like the video shows, the wideband went extremely rich (9.8AFR) as soon as it wanted to pass the 11AFR threshold. Theoretically, if I were to find these lean and rich points, and let's say the rich point is 11.6 like in OhhShmit's case. If the car went past the 11.6 threshold (because it's in the tuning stage still, so I think he'll have to richen it up to be safe), wouldn't it just freak out again and display a really wrong reading? I'm sorry if I'm misunderstanding all of this. This is honestly making me happy that I didn't try tuning my car myself.

This seems like it may be way above my knowledge level. I'm not sure if I want to continue with this wideband at this point. Do you have any experience with the X-Series gauge? If so, was it less of a hassle than this one? I really don't want to take the easy way out but if the wideband will just constantly fall out of calibration then what even is the point?

Kyle

kyler021 said:

So, I need to lock in open loop, find the points (rich and lean) where it will still run normally without any weird behaviors, log the wideband voltage at those points and plug that AFR and voltage into the boxes on the calibration for the wideband? And even then, the gauge can fall out of calibration eventually. Man, I love this car hahaha.

Click to expand...

Right

kyler021 said:

Also, when I had it calibrated to the 11AFR - 14.7AFR like the video shows, the wideband went extremely rich (9.8AFR) as soon as it wanted to pass the 11AFR threshold. Theoretically, if I were to find these lean and rich points, and let's say the rich point is 11.6 like in OhhShmit's case. If the car went past the 11.6 threshold (because it's in the tuning stage still, so I think he'll have to richen it up to be safe), wouldn't it just freak out again and display a really wrong reading?

Click to expand...

I know what you mean about wondering what happens when you go past either of the 2 points, because I've never found anything in the help or the wikis where it really says what happens, so I wondered about that too until I had a chance to check it out for myself. But the scaling does continue past the 2 points in both directions. It continues on the same line that you defined with your 2 points.

To check that out again right now with your log, I made a line in GeoGebra from 2 points picked out of your log.
Point A was x=1.57 (volts) y=12.0 (AFR)
Point C was x=2.88 (volts) y=16.5 (AFR)
Then I went in the log to where the AFR is super low, the log says 8.5 AFR at 0.53 raw volts.
Then I checked to see if that is still on the same line. (See the screenshot, Point H)
It's mighty close, a 0.53 volts value (x) on that line gives you 8.43 AFR (y). Log says 8.5 AFR. Good!
That little bit of being off is those missing decimal places at work
So your AFR logged values are not falling off a cliff after 11. The raw volts are going down fast too, and they are still scaled to each other the same way. The raw volts going down fast is not the fault of the scaling. It is either what is actually happening in the exhaust, or else it is a basic problem with the wideband.




Now you do want to have the 2 points as far apart as you can, because whatever error you have in reading your 2 points, that gets magnified as you go farther above or below them with the scaling. So read your 2 points carefully, with the engine running steady, don't be in too much of a hurry, be sure about what you are getting.
But only have them as far apart as you can without having the engine running rough and nasty. Because then the info you read might be just plain not valid.
By making your lean point at an AFR of about 16, that will make that point a lot farther away from the rich point than a lambda 1 point would be. So that helps, and 16 AFR should still run ok usually.

kyler021 said:

This seems like it may be way above my knowledge level. I'm not sure if I want to continue with this wideband at this point. Do you have any experience with the X-Series gauge? If so, was it less of a hassle than this one? I really don't want to take the easy way out but if the wideband will just constantly fall out of calibration then what even is the point?

Click to expand...

Right. Well unfortunately, I have a brand new X-series still in the box and I've been not wanting to have the car down to install it - I mean I always take a ton of time to do anything like that and I keep saying no, not now.
So I don't know what to recommend, except that ECMtuning still recommends the Innovate LC-2. The LC-2 is getting pretty old now. I don't think ECMtuning has ever said anything one way or the other about the newer Innovate which is the MTX-L series. I don't think they have said anything about the AEM X-series.
Justin Dubois in here uses a Zeitronix ZT-3 Wideband on the 2g (1995) Rear O2 input and he seems to like it.
That's about all I can think of for wideband recs.

If you do this lock in open loop thing, you should get a 3rd point while you are at it. A middle point. One where the gauge is reading somewhere around 14 AFR. Still in open loop. Then we can see how close that middle point is to being on the line made by the lean and rich points. If it is very close, that would give us more confidence that the lean and rich points are good data. I mean it's not always so easy if the gauge is jumping around a little bit when you are trying to get a steady reading.
I can graph it up so we can see it.

We're on Boost said:

If you do this lock in open loop thing, you should get a 3rd point while you are at it. A middle point. One where the gauge is reading somewhere around 14 AFR. Still in open loop. Then we can see how close that middle point is to being on the line made by the lean and rich points. If it is very close, that would give us more confidence that the lean and rich points are good data. I mean it's not always so easy if the gauge is jumping around a little bit when you are trying to get a steady reading.
I can graph it up so we can see it.

Click to expand...

I will try to do that today and let you know! Preparing for a bad storm later.

kyler021 said:

I will try to do that today and let you know! Preparing for a bad storm later.

Click to expand...

Sounds like fun

We're on Boost said:

Sounds like fun

Click to expand...

Tornados in MD... In my 20 years of living, I've never experienced a worse summer when it comes to bad weather.

Anyways, I grabbed 2 logs. First one should be the lean log and second should be the middle point... I wasn't really sure what to change to get to the rich point though... I tried tracking the datalog and changing the SD field, and lowering the global fuel but wasn't able to get it to idle less than 13. And I probably shouldn't be messing with deadtime (mostly because I'm not sure what it does and don't want to break something).

I also meant to ask, how do you pick your points? Like I have it idling at 17 in the first log, should I use the 17 value, 17.1? 17.2? Does it even matter as long as the voltage is corresponding to those points? Sorry if these are stupid questions or if they've already been answered.

Thanks!
Kyle

kyler021 said:

I wasn't really sure what to change to get to the rich point though... I tried tracking the datalog and changing the SD field, and lowering the global fuel but wasn't able to get it to idle less than 13. And I probably shouldn't be messing with deadtime (mostly because I'm not sure what it does and don't want to break something).

Click to expand...

Don't be afraid to use the Global deadtime for this. Just change it back to 220 when you are done.
The trouble with using an SD cell is that it is actually using at least 2 cells somewhat, even if it is only showing it in one cell in tracking. So as the engine hunts around a little bit, it is probably going to be less steady when you change just one cell. If you do it with Global deadtime, that changes the fuel everywhere, doesn't matter if it's looking at multiple cells.

That number in Global deadtime is microseconds. So 1000 in there is 1 millisecond of Injector Ontime that it adds to what you would have otherwise. It will go over 1000 if you want it to, and it will even go negative. I haven't tried negative to see if it actually works but I think it does.

So you should be able to get your Rich condition by upping the Global deadtime, and your Lean condition by lowering it.

kyler021 said:

Does it even matter as long as the voltage is corresponding to those points?

Click to expand...

Well this is it, ideally you need to be able to read the raw voltage in the log and the number showing on the gauge at the same time! And since both things are wandering around a little bit all the time, it's hard to do.
The best way is if you can somehow have the laptop screen and the gauge both in the frame of your camera and take video of the whole thing. Then you would pick out your numbers from the video!
Because that way you can pause the video wherever you want and get whatever numbers are showing at the time.
You probably can't realistically set up your laptop like that.

So a 2nd way that is probably more doable on your end is, you take video of the gauge after you have found the deadtime that gives you a solid running rich AFR.
At some point in the middle of the video you swing the camera over to the laptop screen for about 10 seconds to get a good look at the numbers in the log, being sure to get clearly the raw WB volts, and the Seconds (secs) in the lower right corner. Then you swing the cam back to the gauge so we are watching the gauge again for another few seconds without changing anything.
Then run the Global deadtime downward to get the Lean point, doing the same type of camera views.
Then close the log and stop the video.

Here's the video that OhhShmit took, which we used to get his Rich and Lean point. We took this to be 11.6 AFR at 1.51 volts and 16.2 AFR at 3.82 volts.

At about 1:15 in the video, he has the gauge AFR wagging back and forth between 11.6 and 11.7, pretty good. He first tried a little richer, didn't like it, dialed it back to 11.6 - 11.7. Kept the cam on that for about 10 seconds, then over to the laptop screen which he only showed for about 5 seconds which was barely enough to get a clear read of the raw volts and log seconds, but we got it.
Then back over to the gauge for a few secs to make sure it was still showing around 11.6 - 11.7 AFR.

Then he started lowering the deadtime to get the lean point.

It doesn't have to be at night. @OhhShmit also took some video during the day and those looked ok too. I guess it wouldn't have to all be in one video either, it could be in 2 videos if that's easier. It seems like the log clock (secs) gets interrupted when you change settings, which makes syncing up log seconds with video seconds harder.

If taking video is too much distraction and you just want to go by what you see in the moment when you are in the car, well, that's another way. But having video that other people can look at can help with deciding on the numbers and can help with knowing what you did.

Notice also that @OhhShmit 's idle was 1000 rpm, which probably helps the car to run a little more strong and steady while you are throwing these odd-ball AFRs at it. Yours is set at 750. Whichever gives you a steadier result!

A minor change I think I would make for higher accuracy would be changing your displayed stoich to 14.1. Our fuel in the northeast is e10, so stoich should be 14.1 vs normal 14.7.