Wideband short-comings

[twdorris]

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

Strange. Can you post that log somewhere? I've attached a log here showing what we typically see from an AEM sensor. Those "stair steps" are each about 155ms in length. Here's a picture too, in case you can't open the log. Yellow is the AEMWB. Green is the factory front O2.

You must be logged in to view this image or video.

Thomas Dorris

 

[twdorris]

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

Seems to work as far as I can tell. We've had no particular issues with that kit I suppose.

Thomas Dorris

 

[Hubbard 0]

Strange. Can you post that log somewhere? I've attached a log here showing what we typically see from an AEM sensor. Those "stair steps" are each about 155ms in length. Here's a picture too, in case you can't open the log. Yellow is the AEMWB. Green is the factory front O2.

You must be logged in to view this image or video.

Thomas Dorris

I'll be sure to do that when I get home.

One thought did occur to me though. Is it possible the programs I use to log with recognize that the AEM updates less frequently and then interpolates the missing entries?

 

[twdorris]

One thought did occur to me though. Is it possible the programs I use to log with recognize that the AEM updates less frequently and then interpolates the missing entries?

Good question. I'm not sure. Are you logging the WB analog output voltage or are you logging directly from the AEM unit? I'm not sure if the AEM supports a direct serial logging interface, but if it does, it might just be that it updates that value more quickly?

Thomas Dorris

 

[Hubbard 0]

Good question. I'm not sure. Are you logging the WB analog output voltage or are you logging directly from the AEM unit? I'm not sure if the AEM supports a direct serial logging interface, but if it does, it might just be that it updates that value more quickly?

Thomas Dorris

I'm logging with the blue wire attached to a serial cable to my laptop, then through evoscan or Mitsulogger.

Also, for those asking questions about the AEM and its specifics, this link may help as it seems they've updated a lot of info in it from earlier versions I've seen.

http://www.aempower.com/images/products/Installation Instructions 30-4100.pdf

 

[->PrOjEcTGS<-]

I'm logging with the blue wire attached to a serial cable to my laptop, then through evoscan or Mitsulogger.

When you log via the serial connection, aren't you only able to log the WB reading? How then do you reference that against RPM, Boost, Timing, Knock, etc.?

That is why most of us run the analog output to the ECU and use one datalogger to retrieve all the information through one serial connection.

 

[We're on Boost]

One thought did occur to me though. Is it possible the programs I use to log with recognize that the AEM updates less frequently and then interpolates the missing entries?

I was wondering something like that too. Maybe the sample rate you see is the sample rate of the logger?

 

[Hubbard 0]

When you log via the serial connection, aren't you only able to log the WB reading? How then do you reference that against RPM, Boost, Timing, Knock, etc.?

That is why most of us run the analog output to the ECU and use one datalogger to retrieve all the information through one serial connection.

Both the programs I use are setup to take in another input for a wideband. I have my tactrix cable from the obd to usb and I have a serial to usb that go into my comptuer. The programs take care of the rest.

 

[We're on Boost]

Would like to get a handle on the "calibration resistor" which is the 6th wire on these 5 wire sensors.
According to WBO2.com: "The laser trimmed calibration resistor (Rcal) is unique to each sensor and adds a sixth pin to the connector. Each sensor's Rcal (shown as 30-300&#937 has a unique value that only works properly with that sensor. Rcal connects internally to the Pump Current (Ip) sensor lead."
AEM talks about this resistor in their WB manual, but it is a little confusing because they talk about it as if it was their idea or unique to them or something. Whereas WBO2 says it is just how all the LSU4.2 sensors are made. AEM considers their units to be "pre-calibrated" because of this resistor, and I don't believe they talk about any need to recalibrate their units periodically.
Innovate, as near as I can see, doesn't talk about this resistor. I think it must be there with the units they sell but I'm not sure. I'll call their tech support sometime to ask but want to do some homework first. With the Innovate units the first thing you do with it is calibrate it. Then you are supposed to recalibrate it every 6 months or some number of miles forever after.
So I got some dots that don't connect here. Do these sensors really need to be recalibrated periodically even though they have the Rcal? There are some OEM's that put these things on their new cars, German mostly I think. I'm sure the car owners are not bringing their cars in to service every 6 months to get their O2 sensors recalibrated! When Bosch designed this sensor it wasn't for us tweaks. It was for the OEMs. Does anybody know what the story is on this subject?

 

[knochgoon24]

The calibration for the LC-1 is done inside the LC-1 unit itself.

As sensors age, they deteriorate. The LC-1 tries to correct some of this.

I log my LC-1 through a serial input right along side all my date being logged from Evoscan through a usb input. I can pull up a log or graph of the log, and all the data is right in front of me, in one file.

EDIT:

You must be logged in to view this image or video.

My NBo2 sensor is going out, so it's not cycling quite properly, but you can see how the WB reading is right in evoscan.

 

[the_mork]

It is also nice that that is one less thing that those of us using evoscan have to worry about pulling through MUT logging so we can run it all the time regardless of what we want to be logging because there is no slowdown.

Twdorris and I had a short conversation about serial logging at one point where he expressed that he preferred using an ecu input but dsmlink guys don't have to worry about logging speed and I still prefer the idea that my wideband controller is directly outputting the data in digital form rather than converting it to a voltage, crossing two different gauge wires and one connection, then converting that voltage back to a value. He says its reliable and I believe him but I'll keep my serial port in my car.

 

[knochgoon24]

It is also nice that that is one less thing that those of us using evoscan have to worry about pulling through MUT logging so we can run it all the time regardless of what we want to be logging because there is no slowdown.

Twdorris and I had a short conversation about serial logging at one point where he expressed that he preferred using an ecu input but dsmlink guys don't have to worry about logging speed and I still prefer the idea that my wideband controller is directly outputting the data in digital form rather than converting it to a voltage, crossing two different gauge wires and one connection, then converting that voltage back to a value. He says its reliable and I believe him but I'll keep my serial port in my car.

And if us Evo 8 ecu guys really feel like it, there's some "logging superspeed" mods that can be done to get even more readings/second.

 

[the_mork]

I know DMA logging is very similar in concept to dsmlink logging though the implementation is less refined. I've looked at it quite a bit but I haven't tried it because at this point I'm really perfectly happy with the speeds I'm getting through normal MUT logging. Perhaps as I become a better tuner I'll feel the need for greater logging speed but at the moment just tuning boost, AFRs, and timing at my basic skill level I have all the resolution I need.

Livemapping is more interesting to me but I haven't had the chance to play with that yet.

 

[We're on Boost]

I log my LC-1 through a serial input right along side all my date being logged from Evoscan through a usb input. I can pull up a log or graph of the log, and all the data is right in front of me, in one file.

I looked at this log pic to see if I could guess what the sample rate of the logger is. It looks to me like all 5 lines are shown at only about 4 samples per second. Does that sound about right?
I don't know anything about evoscan. Is there a way to speed up the sample rate with just a software setting?

 

[knochgoon24]

I looked at this log pic to see if I could guess what the sample rate of the logger is. It looks to me like all 5 lines are shown at only about 4 samples per second. Does that sound about right?
I don't know anything about evoscan. Is there a way to speed up the sample rate with just a software setting?

I sample at over 115 samples per second. I was just logging a lot of things in that one log.
If only log 5 or 6 things, I can get a very high resolution (almost 20 readings a second for each data point).

 

[We're on Boost]

I've attached a log here showing what we typically see from an AEM sensor. Those "stair steps" are each about 155ms in length. Here's a picture too, in case you can't open the log. Yellow is the AEMWB. Green is the factory front O2.

Tom, I installed ECMLink on my computer and took a look at the log you have posted above. But I don't see the AEMWB plot in there at all. What I wanted to do is get some numbers on the vertical scales for the WB and the NB plots. I see the NB plot vertical scale has increasing voltage as you go up, like "normal". What about the WB plot you showed in the pic, does that also have increasing voltage as you go up on the vertical scale?

 

[twdorris]

But I don't see the AEMWB plot in there at all.

Go into Edit->Display Values to adjust what's available in the graph legend. You can also do a Ctrl-Q to hide/display all on the graph. With everything removed from the graph (but available in the legend), you can then add individual items back into the graph one at time by clicking on the item's "line box" in the legend.

Here are some pages with more details on the different things you can do and the different ways you can use the ECMLink datalog viewer.

Datalogging basics
Datalogging tips and tricks

I see the NB plot vertical scale has increasing voltage as you go up, like "normal". What about the WB plot you showed in the pic, does that also have increasing voltage as you go up on the vertical scale?

Yes. You can also change the scaling of anything in the graph too. Just double click on the item in the legend or right click on it and select Properties.

That particular log shows 38 "frames" per second with each frame containing a snapshot of 29 unique "samples" inside the ECU. Which basically means just over 1100 samples per second.

Thomas Dorris

 

[We're on Boost]

Tom, thanks, Excellent!
So, I'm looking at this log now to get an idea what the "lag time" of the WB is. What I mean is the NB and WB should both be getting richer at the same time, and they should both be getting leaner at the same time. Rich for the NB is up, Rich for the WB is down. So the NB plot should be going up whenever the WB plot is going down, and vice-versa.
What I see is they aren't doing that and the reason they aren't is because of the lag of the WB. But if you took the WB curve and slid it about a half second to the left, then the 2 curves would be properly "time aligned". So the lag of the WB is about half a second! That's more then I thought it would be. Do you think this is this right? I thought there would only be about a 0.2 second lag.

Gary

 

[the_mork]

That is why many people prefer to use the NBO2 for cruise tuning where it is accurate anyway and much faster than a wbo2.

 

[We're on Boost]

That is why many people prefer to use the NBO2 for cruise tuning where it is accurate anyway and much faster than a wbo2.

Yeah!
I've gotta wonder then, why are some of the OEM's using wideband sensors? I mean, they are actually controlling the engine with them, not just looking at pretty pictures! How can you control an engine in real-time with a sensor that is lagging by that much?

 

[toofast82]

Yeah!
I've gotta wonder then, why are some of the OEM's using wideband sensors? I mean, they are actually controlling the engine with them, not just looking at pretty pictures! How can you control an engine in real-time with a sensor that is lagging by that much?

I run my wideband as my only oxygen sensor on my 1g. It cycles just fine in closed loop and there are no drivability problems.

 

[the_mork]

Yeah!
I've gotta wonder then, why are some of the OEM's using wideband sensors? I mean, they are actually controlling the engine with them, not just looking at pretty pictures! How can you control an engine in real-time with a sensor that is lagging by that much?

How much variation are you really getting in closed loop? Mine is dialed in well enough between fuel trims and the fuel map that is hardly ever went more than a few points off of stoich and even then cruise is a very forgiving time for the motor. A car in good repair with an ecu tuned for its components doesn't even need closed loop. I for one have closed loop turned off for anything about 2500RPM so that I can cruise at 15.3:1 and get better economy. I keep closed loop at idle to help with my intensely varying weather conditions on cold startups and because I'm too lazy to tune something that I honestly don't care about and too cheap the buy the AFPR that my car really should have.

 

[twdorris]

So the lag of the WB is about half a second!

Any lag you see in a datalog from a WB sensor relative to a narrowband is more likely to be related to position of the WB in the exhaust stream than being related to the WB itself. If I run my WB way back in the rear O2 bung on my 2G exhaust, I can easily get a 0.5 second delay in response at idle (slow airflow stream). When I move that same WB sensor up into the front O2 bung, it responds just as quickly as a stock narrowband sensor would.

When it's located back in the rear O2 bung, I just have to slow the fuel trim update rate down to control the overshoot. But other than that, it can be made to work there just fine if you had to.

Thomas Dorris
ECMTuning, Inc.

 

[knochgoon24]

Ok. I JUST went out and checked the read rates I could get on my LC-1 and my narrowband.

Evoscan
NB - .04 sec
WB - .15 sec

Logworks
WB - .09 sec

You must be logged in to view this image or video.

With Evoscan, when the WB reads too fast, it just keeps returning 14.36 except when it has a new reading. So you get the spiky look. - There may be a setting for the LC-1 which will fix this, not sure.

The NB gets the stepped look.

When logging just those two, I was getting 270+ sample/sec. So about 135 each.

 

[twdorris]

When logging just those two, I was getting 270+ sample/sec. So about 135 each.

At that rate, you're outrunning the main loop in the ECU. You're basically sampling a value faster than the ECU is updating it, resulting in that stepped look to the narrowband signal.

Thomas Dorris