ECMlink AEM wideband pulsing / erratic readings

[dwb]

My AEM X-Series Wideband, only about 4 months old, is giving me pulsing and erratic readings via ECMLink. It usually smooths out under heavy throttle or drastic change in RPM. The gauge numbers appear to read OK all the time. I'm using the EGRtemp input (1g) and I'm dealing with it's limitations, but in January when I first installed the WB it was reading much better. I hadn't logged since then but nothing changed. This is a daily driver so lots of power cycles.
I'm attaching a log from January for a baseline reading, then a log from yesterday. Screen snips also for ease of view (green line is the WB).

Let me know if you have any ideas why it's doing this and/or how to fix it.

May log:

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

January log:

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

 

[1990TSIAWDTALON]

Don you have your gauge and ECU both grounded in the same spot. On some WB's, it is important (my tuner had me move my gauge ground to the ECU ground behind the console. Just a thought.

 

[dwb]

Yes, the gauge get's it's ground from the ECU harness. Here's how I wired it: AEM Red/12v to pin #107, black/ground to #106, white/analog 0.5v to #15 (EGR Temp), and brown/analog sensor ground to #24 (ECU sensor ground). Followed the guide from ECMLink: https://www.ecmtuning.com/wiki/lc1install

 

[1990TSIAWDTALON]

I have never used a SENSOR ground for the wideband. My tuner said to ground it at the ECU MOUNTING GROUND (a body ground).
IDK if it is an issue for you but Mr Jewer had me change mine and then the gauge and Link were the same or super close.
I have never used power from the ECU either. Pull your power from a switched source to power the gauge not the ECU.
Also, I don't use one of the gauge outputs (the brown IIRC) and tapped into the white signal wire for the gauge to feed the ECU.
Excuse me if you have tried these and no luck.
Marty

 

[dwb]

My understanding is you don't want to combine a sensor ground (brown AEM wire) with a chassis ground for sake of accuracy.
Everything I've read says to attach the AEM brown sensor ground wire to the ECU sensor ground. That jives with both the AEM installation guide and ECMTuning's guide.

What stumps me is this worked properly a few months ago without any changes.

 

[1990TSIAWDTALON]

I am just repeating what my tuner had me do. ECU power and sensor grounds are easy to mess up. My LC2 is not hooked to ANYTHING on my ECU except the input at pin #105, I believe. It's been a while.

 

[dwb]

I've been scouring online resources and I haven't found much info. So I'm kind of at a loss as to where the issue might be. Since the gauge still reads correct, I think the sensor is OK. I have a few ideas in attempt to troubleshoot:

1. Inspect all electrical connections, grounds, and harness.
2. Inspect for exhaust leaks - although I believe this would affect both the log and the gauge.
3. As Marty suggested, use a common chassis ground instead of an ECU sensor ground for the analog ground (brown wire).
4. Pull the sensor and perform an open-air calibration.
5. Since I'm using the EGRTemp input, install a unity buffer per https://www.ecmtuning.com/wiki/ecuinputdata
6. Contact AEM for support (?)

 

[CrackedDSM]

I have an older/slower AEM UEGO in the same input in mine, without the unity buffer and mine doesn't do that at all. Do you have a multimeter? What does the voltage from rawlinwideband show when it does this? Wondering if it's a loose connection or a degrading connection somewhere causing it to freak.

 

[TJCTalon]

Yes, the gauge get's it's ground from the ECU harness. Here's how I wired it: AEM Red/12v to pin #107, black/ground to #106, white/analog 0.5v to #15 (EGR Temp), and brown/analog sensor ground to #24 (ECU sensor ground). Followed the guide from ECMLink: https://www.ecmtuning.com/wiki/lc1install

I agree I would not wire in a wideband that way either. I have my AEM wired to a switched 12v and grounded on the same bracket as the ECU's ground. I have zero issues with mine.

I had an innovate WB years ago that gave me nothing but trouble. Been with AEM ever since.

 

[Dusty Landrum]

Yes, the gauge get's it's ground from the ECU harness. Here's how I wired it: AEM Red/12v to pin #107, black/ground to #106, white/analog 0.5v to #15 (EGR Temp), and brown/analog sensor ground to #24 (ECU sensor ground). Followed the guide from ECMLink: https://www.ecmtuning.com/wiki/lc1install

Maybe I am wrong, but I also use the x-series aem wideband and the directions on the manual say nothing about connecting power and ground to the ecu, nor would I think that would be a good location to get them anyway. Also, I am with you that the brown should go to ecu sensor ground, and even AEM specifically states this should be used unless there is not one, THEN ecu ground should be used (page10 of install manual).

Double check all those CRIMPED connections I know you love (AEM does say proper crimp/solder).... hahaha. Just joshin, but yeah maybe a pin is slightly pushed at the gauge or something odd. Hope you get it figured out man.

 

[dwb]

It's not like I just arbitrarily found +12v switchable source. It's all listed out on https://www.ecmtuning.com/wiki/lc1install for their LC1 guide. AEM isn't any different, other than wire colors. I trust that ECMTuning knows what they are talking about. But again, this all worked fine a few months ago.

Here are some snips from that page:

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

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

And a couple snips from the AEM installation guide:

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

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

 

[curt-s]

If it did work and no longer does, you won't be first to have issues with the analog out on an aem x series gauge. A friend with a 1gb just had his output straight up die on him out of nowhere. Traced the problem all the way back to the PCB itself. I gave him some tips to check that I got from a different platform's forum thread about another failed analog output.

I recently picked up a classic 4110, I wonder if it will pull the same crap or if there's something inherently wrong with the X. Maybe DSM voltage ripple is killing it.

 

[dwb]

I've tried some things but unfortunately no solution yet. So far here's what I've tested:
-Verified all connectors are tight and clean.
-All splices are tight and clean.
-Wiggled harnesses looking for changes in datalog.
-Moved sensor ground to chassis ground
-Moved ground to new chassis ground
-Moved +12v directly to battery
-Hooked up a dc volt meter between sensor ground (connected to ecu sensor ground) and loose 5v analog. Got fairly stable readings (see video).

The multimeter test leads me to believe something is up with the egrtemp input so I'm thinking about trying the unity buffer.

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

 

[CrackedDSM]

If it did work and no longer does, you won't be first to have issues with the analog out on an aem x series gauge. A friend with a 1gb just had his output straight up die on him out of nowhere. Traced the problem all the way back to the PCB itself. I gave him some tips to check that I got from a different platform's forum thread about another failed analog output.

I recently picked up a classic 4110, I wonder if it will pull the same crap or if there's something inherently wrong with the X. Maybe DSM voltage ripple is killing it.

I have an ancient AEM UEGO wideband, and it outputs fine. The voltages are just extremely weird, so basically you need to set a window where the logged values are "accurate" and sacrifice the other areas. For me it's from 11.0-14.7. Anything above or below is BS. I adjust it constantly but pretty much that's what I always come back to. I am super duper tempted to go with an LC2 for this reason just to cut out the issues. Only reason I haven't so far is pretty ricer-riffic, and that's because I want the gauges to match.

I was gonna replace this UEGO with a Series-X wideband, but I've been hearing a boatload of people having failures just like this, so it's pretty discouraging.

 

[curt-s]

Me too. I haven't even installed this uego so I could probably get back what I paid and just go innovate (again)

 

[dwb]

I couldn't get this to upload on Friday, so here is another video showing the analog 5v via a multimeter. This time with the car running/blipping the throttle, and a repost of the voltage to AFR chart. The voltage of my sensor does look to be reading a little high based on the chart, but no obvious oscillation like I see with ECMLink.

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

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

 

[curt-s]

Agreed, it should be below 3.25 consistently on idle but you're showing a full 100mV or even more. Of course, we don't know the actual delay from gauge to analog out voltage signalling but it can't be that long.

I'm wondering what the latent DC ripple voltage is which may be leaking in but I would think AEM has at least rudementary filtering since it's an automotive application.

But looking at your erratic log, that doesn't explain the severe shift. I can only assume the 1g egr temp input is not playing nice. It doesn't explain the 100mV+ offset you're seeing on the meter, but 100mV also doesn't translate to a reading of 4.25v apparently seen by the ECU.
Have you tried building the buffer circuit that ecmtuning talks about? https://www.ecmtuning.com/wiki/ecuinputdata

 

[dwb]

Have you tried building the buffer circuit that ecmtuning talks about? https://www.ecmtuning.com/wiki/ecuinputdata

That's my next step. I have both recommended chips (TLV2371IDBVR, TLV2371IP) in my DigiKey shopping cart.

 

[jed344]

This sounds like a gauge/controller issue to me. Myself and many cars i tune now run the x line with great results wired as you had it. Senor wiring at ecu ect, when wired this way and the chart used for settings set up as lin wideband they work perfect. The video looks like voltage on the 5v wire to ecu does not match the chart witch would indicate this issue. I had this issues in past with other wideband's myself. The oscillation you see in link i have not seen personally but the voltage does appear to be off. Do you have the gauge set to correct settings p00 ect, there is different modes as i recall but would have to go to my car to check witch is correct.

 

[dwb]

I've contacted AEM Electronics to see if they can/will offer any support.

 

[curt-s]

This sounds like a gauge/controller issue to me. Myself and many cars i tune now run the x line with great results wired as you had it. Senor wiring at ecu ect, when wired this way and the chart used for settings set up as lin wideband they work perfect. The video looks like voltage on the 5v wire to ecu does not match the chart witch would indicate this issue. I had this issues in past with other wideband's myself. The oscillation you see in link i have not seen personally but the voltage does appear to be off. Do you have the gauge set to correct settings p00 ect, there is different modes as i recall but would have to go to my car to check witch is correct.

Indeed, it does, and the constant extreme high voltage swings every ~40ms would shift mean voltage higher causing a reading on the meter to be higher than the gauge because RMS meters just don't present values as fast as link does. The buffer circuit still is recommended for that input, hence the suggestion to build one. Plus, who wouldn't want to build one if presented the opportunity

Get that gauge on an oscilloscope and we'll put this to rest
I've been considering one of those 10Mhz versions for just such occassions.

 

[dwb]

I've contacted AEM Electronics to see if they can/will offer any support.

AEM has been very responsive since I emailed them this morning. After some back-and-forth, I received this:

"We have had customers in the past use the factory ECU however, the problem that arises is that OEM/factory ECUs that aren't designed specifically for a wideband 0-5v input have different pull-up/pull-down resistors on the input circuit that directly affect the behavior on the incoming signal from the gauge. After market ECU’s, like our Infinity ECU do not have this issue since the dedicated input has a different pull-up spec. This issue can be corrected by using a Bosch 17025 sensor, which has shown to fix the issue for customers using factory ECU’s."

I followed up asking for some additional confirmation regarding the Bosch sensor:
"form [sic] my experience with customers using a factory ECU that has been modified for a wideband sensor, the Bosch 17025 solves the issue you are experiencing."

I also asked about the pull-up unity buffer that ECMLink recommends:
"I can say that some customers have done that and had success"

I guess I'll continue forward with the unity buffer. I really hate to have to buy another sensor already, but I'll keep that Bosch 17025 in mind.

 

[curt-s]

Give it a shot but I'm still curious as to why your MM reads 100-200mV higher than called for by the document at that afr, even taking into account the reading delay created by a RMS meter.
Shrug
See what happens with the buffer

 

[CrackedDSM]

I'm curious as to how and why the different sensor would change things and work better. I'm nowhere near an electrical engineer, so that genuinely has me curious.

And if it does fix everything...may be investing in a series x WBO2.

 

[dwb]

I built a unity buffer per the ECMTuning wiki. Good thing I ordered two chips. I accidentally broke a leg off the first one when color coding the wires with a sharpie. I also forgot to order the 100 ohm resister, but quickly found one in my kids old snap circuits. It's not exactly elegant. A small bare PCB would have been ideal but this will have to do.

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

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

Alas...it did absolutely nothing for the oscillation/erratic WB voltages I'm seeing in ECMLink. DOH!

I now have a Bosch 17025 sensor waiting anxiously on my workbench. $82...thanks AEM
I'll report back once I get the new sensor installed.