Wideband short-comings

[We're on Boost]

I've been reading about widebands lately and finding that the more I read, the more questions I have, and there are places where I can't make the dots connect. So in this thread I would like to bring some of these things out for discussion. I imagine there are at least a few people in here who have access to technical information that is not readily available. If you have proprietary information that you can't give out, I understand.
I'm not going to ask all my questions at once. Let's go.

 

[jrohner]

Then go ahead and ask your questions.

 

[No_Skillz]

The answer is yes.

 

[We're on Boost]

First of all I wonder if an inexpensive wideband is really that much more accurate than reading the voltage from a good, new, OEM narrowband sensor. By inexpensive I mean like the AEM units that go for under $250 with the gage. I've read all kinds of stuff about the LSU 4.2 sensor itself, that the accuracy is thrown off by temperature, and by pressure. I've seen multiple people complain that the AEM analog output to logger is different by about 0.5 afr from what the gage says. One of our knowledgeable guys says the AEM white paper on how to convert the AEM output voltage to AFR is off by quite a bit. As far as I know, the end users (like me) can't get real technical support from AEM, so the questions pile up.

So I called tech support at Innovate. Yup they have real tech support, on the phone no less. They claim their LC-1 is accurate down to 7.35 afr, and that the AEM is less accurate at any afr, and especially less accurate at afr's below 11 or so. They use the same LSU 4.2 sensor that AEM does. So I said "where's the difference". They said it's in the controller. Temperature compensation for one thing, but just a better controller overall.
The AEM units don't give much of a spec. The stand-alone ones that cost about $400 just for the controller say "accurate to 0.1 afr" and "reads as rich as .55 lambda" which is 8.1 gasoline afr. The cheaper AEM ones just say "accurate to 0.1 afr".
Then yesterday I found an interesting article called "Why EGT and Narrowband, not Wideband for Turbos".

So I'm wondering a lot of things here.
I'm wondering why the AEM stand alone controller is so much more expensive than the gage controller combos - is it a lot better - more accurate?
I'm wondering if the inexpensive AEM units are really that much better than just reading the voltage from your OEM narrow band.
I'm thinking that even the more expensive and presumably more accurate wideband units have some short comings that the user should be aware of.

I've been logging my OEM narrow band voltage with my ecu+ and I am amazed at how consistant the logged values seem to be, in relation to where knock starts and so forth. The ecu+ reads the voltages in increments of .01 volts. I suppose most loggers read to .01 volts but the accuracy of the voltage readings must vary from logger to logger. After all, my Fluke 87 cost about $300 and I have pretty good confidence that it reads voltage with good accuracy. But the voltage accuracy of these loggers, I don't know. I guess that's another question!!

 

[knochgoon24]

I haven't had any problems with my LC-1 other than the o2 sensor going bad. If I were running my car on the ragged edge on a race setup, then I might be more concerned about accuracy, but since I DD my car, I keep it conservative.

 

[We're on Boost]

I haven't had any problems with my LC-1 other than the o2 sensor going bad. If I were running my car on the ragged edge on a race setup, then I might be more concerned about accuracy, but since I DD my car, I keep it conservative.

How many miles would you say your sensor was good for? Were all those miles on unleaded gas, or was some of it with leaded gas? Where is your sensor mounted? Like if it is in the downpipe, how many inches about is it downstream from the O2-to-downpipe flange?

 

[knochgoon24]

It's just before the flex section. I bought the kit and the sensor used, so I don't have a clue how many miles were on the sensor. I have a spare sensor, I just need to get around to installing it.

Otherwise, it has performed flawlessly (from what I can tell). I log using Evoscan, which can log both the ECU and the WBo2 at the same time (very nice).

I only use unleaded gas. I was stupid and didn't remove the sensor when I seafoam'd my car. I think that may have been what killed it.

 

[We're on Boost]

OK thanks! I'm interested in the LC-1 and LM-2 and might get back to you with some questions specific to the LC-1 at some point.

Reminds me though, when I asked the Innovate tech person if I could put an extension cable from the plug on the end of the sensor wire (which is only 20" long) to the input of the LC1 unit, he said absolutely not! He couldn't really explain why not, except to say that it would add noise to the signal. But he was adamant that I should not do it, that it would not work. 20" from sensor to LC1 period! No matter how good the extension cable is or who made it or what kind of shielding it has or what the impedance is! So if anybody has any insight on why that might be such a no-no, I'd be interested in that in this thread. One reason I'm interested in the LM2 is because it comes with an 8 foot cable to the sensor right from the get-go. Otherwise, I think my main questions are in post 4, at least for starters.

 

[knochgoon24]

I have my LC-1 (the actual computer part) strapped right to the power steering reservoir. If I took more time, I might be able to have it so it's inside the car instead of under the hood. It doesn't get hot where it's at, so I'm not too worried.

 

[Hubbard 0]

I've logged my AEM wideband through both EvoScan and Mitsulogger and have yet to find a discrepancy between the the log and the gauge.

I've actually heard quite a few stories about how AEM's are less accurate or that the way the controller is made, it inherently wears out. However, I've never actually seen a case anywhere written down where someone had an actual problem with their AEM.

 

[We're on Boost]

Here are some interesting claims from MoTeC on their Lambda meter. Link for this is here: http://www.westminsterspeed.com/mrt_tech/lambdabrochure.pdf

 

[We're on Boost]

Here is another interesting thing in the Motec link. They give some specs on their analog output.

 

[We're on Boost]

The thing that got my attention in this spec above is the response time. These times (about a tenth of a second) agree with what I've seen for the response time of the Bosch wideband sensor itself. So I think this lag time is inherent to the sensor and is not the fault of the Motec electronics.
Anyway the point is, my logger samples the engine sensors at 25 times per second. A 0.1 second lag in the logged a/f values would be a lag of 2.5 samples on my logger. This I don't like. When I am looking at a log of a pull where my engine gets too lean, I like to put the curser on the exact sample point where a spike in my knock sensor voltage first appears. Then I can read the voltage of the narrow band O2 sensor at that same sample point. Pretty reliably I will find that on the first sample after my O2 sensor voltage drops below 0.83 I get the start of the spike in knock sensor voltage (This I'm remembering off the top of my head so it might be a little off).
So I guess what would help me here is if I knew what the response time of the OEM narrow band sensor is when running hot and heavy like this (WOT and RICH). I've never been able to find a spec for that. Probably because narrow bands were designed to do their thing when you are near stoic - not hot and heavy. I think under these WOT conditions the narrow band actually responds quicker than the wideband. Anybody know?

 

[knochgoon24]

My LC-1 reads really fast if it's the only thing I'm logging. The Logworks software has and amazing fast connection, and even Evoscan keeps the speeds up pretty high.

I don't see why you're so worried about this. Each of your cylinders will have a different AFR due to several variables. All you're getting is an average of the 4 cylinders anyway.

If you want more accurate than an LC-1 or AEM UEGO, then you should probably be running a WB in each exhaust runner.... which is overkill.

 

[We're on Boost]

Nah, it's not a matter of being worried about stuff. I'm going to get one of these things and make the best of it, whatever way that is. I'm just trying to learn about some things that I'm curious about. I'm an engineer you know? Can't you tell?

 

[Ceddy]

OK thanks! I'm interested in the LC-1 and LM-2 and might get back to you with some questions specific to the LC-1 at some point.

Reminds me though, when I asked the Innovate tech person if I could put an extension cable from the plug on the end of the sensor wire (which is only 20" long) to the input of the LC1 unit, he said absolutely not! He couldn't really explain why not, except to say that it would add noise to the signal. But he was adamant that I should not do it, that it would not work. 20" from sensor to LC1 period! No matter how good the extension cable is or who made it or what kind of shielding it has or what the impedance is! So if anybody has any insight on why that might be such a no-no, I'd be interested in that in this thread. One reason I'm interested in the LM2 is because it comes with an 8 foot cable to the sensor right from the get-go. Otherwise, I think my main questions are in posts 4,5,6,7, at least for starters.

I made my own 8' extension for for my LC-1 and it works perfect.

I'd like to know their reasoning on why it won't work.

 

[4g64fiero]

Those motec articles read like propaganda.

I have an ngk afx and it has never let me down after 2 years of driving with the sensor in at all times.

 

[We're on Boost]

I made my own 8' extension for for my LC-1 and it works perfect.

I'd like to know their reasoning on why it won't work.

Hey!
Now that's cool.
If you have time ever, I'd appreciate it if you could run a thread on how you made the cable, what materials and plugs you used, where you got the stuff from. You know? And if you do that please put a link to it in this thread.

 

[Ceddy]

Hey!
Now that's cool.
If you have time ever, I'd appreciate it if you could run a thread on how you made the cable, what materials and plugs you used, where you got the stuff from. You know? And if you do that please put a link to it in this thread.

Cut the WideBand sensor harness in half, and splice in 8' of wire.
8'x4 wires = 32' of wire, plus some spiral loom tubing to make it pretty.
Total cost $3.

If you are are looking to make a harness with plugs, I think the newer VWs use the same sensor. You could grab the male and female plugs from a junkyard.

 

[twdorris]

The problem with the AEM units is the white papers don't match the output voltages.

I'll second that. It's further complicated by the fact that they have different voltage mappings posted in different versions of their own install instructions! And none of them seem to match very well. I really despise the AEM units for this reason and have suggested for years that people buy anything BUT an AEM WB unit... That one device has been our largest after-market sensor support issue for years.

One of our users did post a manual remapping that seemed to work OK for him, if anyone is interested.

aemwb [ECMTuning - wiki]

Maybe that'll help someone get their AEM readings a little closer.

IMO, the LC-1 and AFX kits are very nice.

Thomas Dorris
ECMTuning, Inc.

 

[We're on Boost]

Cut the WideBand sensor harness in half, and splice in 8' of wire.
8'x4 wires = 32' of wire, plus some spiral loom tubing to make it pretty.
Total cost $3.

If you are are looking to make a harness with plugs, I think the newer VWs use the same sensor. You could grab the male and female plugs from a junkyard.

Cool! Is there anything unusual about the Bosch harness wire? I imagine each wire is a multi-strand type wire, but are the strands like interwoven with fiberglass or anything weird like that?
Aren't there 5 wires?

 

[We're on Boost]

hakcenter and twdorris: Thanks for those comments.
What you are saying is about what I figured.
When I look at the information (or lack of it) that AEM publishes in their advertising for their wideband products, it makes me suspicious right away - because they don't really publish a spec for it. About all they say is "accurate to 0.1 afr". Now that's not what I would call a spec. That's what I would call a Mother Goose Story.

 

[twdorris]

When I look at the information (or lack of it) that AEM publishes in their advertising for their wideband products, it makes me suspicious right away

Oh, something else I forgot to mention...the AEM WB output only updates 6-7 times a second (roughly 155ms between updates). This looks horrible in a log.

Thomas Dorris

 

[Hubbard 0]

Oh, something else I forgot to mention...the AEM WB output only updates 6-7 times a second (roughly 155ms between updates). This looks horrible in a log.

Thomas Dorris

I just looked at a log and I have 14-15 entries per second from my AEM.

 

[romeen]

Thomas, do you have any thoughts on the PLX wideband?