Page 1 of 1

AEM WB consitently lean on analog output

Posted: Mon Aug 17, 2015 3:36 pm
by csracing
Hello,

First post here so I'll try not to be too annoying.

I'm working on tuning a new engine for my 2007 Mustang. I have an AEM wideband setup that is using a LSU4.2 sensor. I am tuning with HPTuners and have the 0-5V analog output from the gauge installed on one of the analog inputs of the tuning interface. The gauge display is somewhat erratic and difficult to watch so I've obviously been using logged data from the analog output for my air/fuel tuning.

The ECU is forced into open loop for WB tuning. If I tune to the WB and then re-enable closed loop, the ECU is consistently pulling between 10 and 20% fuel out at pretty much every MAF voltage and load step per the short term fuel trims. I've tried a couple different formulas in the analog input of the tuning interface, but have not been able to come up with a WB curve that remotely aligns with what the ECU tries to do in closed loop. Essentially, the ECU says rich and the WB says lean.

To date, I've tried configuring as voltage to AFR as well as voltage to lambda. The gauge reads 10:1 at 0V and 20:1 at 5V per the manual for reference. I've replaced the O2 sensor thinking maybe I had a bad sensor but no delta in output was noticed.

With that said, I have a couple questions:
1) Is the AEM controller just not a good controller?
2) Is there a way to calibrate the sensor and gauge? The manual says no calibration is required.
3) I'd like to try an aftermarket WB controller that just has a known analog output, are there any available that are "inexpensive" that will produce acceptable steady state output?
4) Could I have some kind of grounding or other electrical issue that hasn't been obvious to me yet?

I had the car remote tuned by someone else with the original engine so I can't comment on whether or not the analog data was accurate before the engine swap.

Thanks in advance for any advice.

CSR

Re: AEM WB consitently lean on analog output

Posted: Tue Aug 18, 2015 8:38 am
by Fred
I can't remember if the AEM is good or bad, however this thread is a must read, read the entire thing, excuse any noisy patches, overall, it's pretty good: viewtopic.php?f=6&t=2267

If the unit says no calibration, then that's it. You could by lab gas, or run it in a tube with another of known quality, however it's likely not HUGELY far out, even if the AEM is crap.

The same thread will shed light on an inexpensive, accurate, and responsive unit made by Alan To, his products are all I plan to use for the foreseeable future. Really good stuff.

Grounding of widebands is critical, and could contribute to noisiness. Then again, a rough tune could contribute to percieved noisiness, too. Even a dialed in tune swings from lean to rich and back with throttle movements.

Is the ECU's target lambda in the same place as *your* target lambda? Is lambda just a target, or is it integral in other calcs? Understanding the system is important, and OEM stuff can be somewhat black-art like.

Good luck!

Fred.

Re: AEM WB consitently lean on analog output

Posted: Tue Aug 18, 2015 12:38 pm
by tpsretard
AEM is CRAP, actually no, that is an insult to crap......

It also has a problem with the ground reference voltage. so this is what you are probably seeing.

Re: AEM WB consitently lean on analog output

Posted: Tue Aug 18, 2015 2:07 pm
by csracing
Thanks for the replies. I read the WB thread, very informative. I've found plenty of posts online in my research now about AEM analog outputs being inaccurate so it seems I may be asking for something it cannot deliver. That said, I ordered a controller from Alan yesterday. Looking forward to getting it up and running so I can get back to tuning.

Re: AEM WB consitently lean on analog output

Posted: Mon Aug 31, 2015 4:50 am
by buzzard
The most common problem with AFR analyzer analog outputs it that they do not offer
ground isolated or virtual ground analog output grounds. You can "carefully ground"
all you like, however you will not resolve the issue.

I use Autronic analyzers almost exclusively. Some of the Tech Edge units have this feature.

I cannot comprehend why manufacturers who pitch their products for use with motorsport ecu's (everything that is not OEM) that typically have single ended analog inputs do not understand this
basic technical requirement.

Perhaps they seek to copy professional units such as ETAS,ECM-CO and others, not realising that these
analyzers are used with data loggers with differential inputs.

I recently contacted Ecotrons with regard to this matter (ALM board) and they have instituted a redesign.

My contact with Alan To of 14point7 in 2010 yielded this response

"There are 2 grounds for the unit, the electronics ground and the heater
ground. The heater ground carry upto 3 amps, the electronics ground
carry ~50ma. Just ground the electronics ground to where your
ecu/datalogger is grounded, 50ma is not going to cause detectable offsets."

Which is workable, if not technically elegant.

I welcome forum members responses.

Re: AEM WB consitently lean on analog output

Posted: Fri Sep 04, 2015 6:46 am
by Fred
Not sure why you'd say it's not technically elegant, removing voltage offsets by splitting current paths is standard practice across many types of circuits, including my beloved audio amps. It's the approach we use to ensure an accurate battery voltage reading with FreeEMS, too. Works very well, and is pretty much fool proof.

Re: AEM WB consitently lean on analog output

Posted: Fri Sep 04, 2015 6:58 am
by buzzard
Fred wrote:Not sure why you'd say it's not technically elegant, removing voltage offsets by splitting current paths is standard practice across many types of circuits, including my beloved audio amps. It's the approach we use to ensure an accurate battery voltage reading with FreeEMS, too. Works very well, and is pretty much fool proof.
Allow me to qualify that by saying it is not as technically elegant as a ground isolated/virtual ground for the analog output where the ground reference is to the ecu analog ground. It does not, and necessarily cannot fully resolve the voltage offset issues. My comments in the earlier post ought not to be considered a criticism of Alan personally.

In my view there is a good market for a external ground isolation interface.

Re: AEM WB consitently lean on analog output

Posted: Fri Sep 11, 2015 10:03 am
by Fred
Differential inputs are great things, this is true. I wonder about the practicality of producing those IO interfaces with sufficient fidelity that you don't defeat the purpose, and in a small package, at a low cost, too. As for "necessarily cannot" whilst true, there is a practical limit in terms of your measuring equipment and the output resolution of the device in question, and provided that you're sufficiently below that range, you've effectively done the job as well as possible: immeasurable difference in numbers.

Aside from the debate on signal strategies, at least some of Alan's new stuff has something that I think he calls the "output sequencer" for calibrating any static offsets. And yes, I know what you'll say next: Many offsets are not static. True. With an ECU we're talking a few hundred milliamps at most for control circuitry, though, so over a decent piece of wire the drop will be pretty damn small.

I'm not really arguing against your point, you're right. I'm more trying to point out that in almost all well designed practical systems, it's a total non-issue.

Fred.