TheSamba.com :: View topic - Electromagnetic interference with O2 sensor?

Sun Apr 22, 2018 4:32 pm

The common symptom of a bad o2 sensor/wiring is a rich running engine. Does anyone know what would cause the opposite?

My o2 sensor is causing a lean running condition meaning a higher voltage signal is going to the ECU from the o2 sensor. Electromagnetic interference? I have a wideband air/fuel ratio gauge, with its own sensor bung in the cat. Could it cause EMI?

I found no clues anywhere as to how an o2 sensor signal can be faulty on the higher voltage side. And I replaced the o2 sensor... twice.

Sun Apr 22, 2018 5:39 pm

poor engine grounds can raise the voltage potential between a well-grounded ECU and the ground return path of the O2 sensor. in other words, if your alternator is putting out 14.5 volts but a poor ground path drops .5 volts between the alternator and the body, that .5 volts can be sensed by the ECU on top of the voltage the sensor is producing and seeming like the engine is running rich and needs to be leaned out.

-dan

Mon Apr 23, 2018 6:10 am

Dan, you might be right on there. But heres the thing, I tried using the emulated o2 signal from my AEM air/fuel gauge, connected it to the ECU and it runs perfect. The gauge's signal has no issues. So does that make sense? Because the additional voltage was coming through the stock o2 wire?

Anyways, I actually had "add more grounds" on my to do list. Now i have a reason to get on that.

Mon Apr 23, 2018 8:04 am

Where did you connect the AEM to the ECU?

Mon Apr 23, 2018 8:23 am

I connected the AEM white wire (analog output) with setting P4: 0-1V to the ECU connector- pin 2 (oxygen sensor input). I was surprised how well the AEM gauge worked with the ECU. I'll be going back to using this AEM gauge output if I can't figure this out. It worked fine, I would just prefer the ECU to be communicating directly with the oxygen sensor, the way it was designed to work.

Mon Apr 23, 2018 9:09 am

so you connected the AEM right at the ECU? by piercing insulation/backprobing? or did you connect it to the pigtail for the O2 sensor in the engine bay?

if the AEM is grounded to the body, which the ECU is, then your observation makes sense. both the AEM and ECU are using the same ground reference and the AEM generated signal is accurate with respect to the ECU as the possible elevated ground potential between the engine and body does not come into play.

resistance in a ground connection is really a function of the current flowing thru that connection. you can measure zero ohms with a meter because there's nearly zero current flowing thru the meter to ground. picture a connection of one fine copper strand from a wire-- it'll measure zero ohms with no current flowing thru it but it won't handle any current.

but put 20-60 amps of alternator charging current thru a poor ground and you'll create a voltage drop across that connection. now your sensors are giving readings to the body grounded ECU with that same voltage drop between the engine and body superimposed on the signal.

-dan

Mon Apr 23, 2018 9:52 am

I connected it directly into the ECU connector harness. I've bought a bunch of oem style wire terminal ends from digikey.ca, even the small pin style for sensors and ECU connections. It's actually cheaper to buy proper terminals from digikey than going to a local store and grabbing a pack of wire terminals.

So I enhanced the grounds today, added some and cleaned some. Apparently I'm not there yet because it didn't fix the issue. Since the AEM wired into the ECU works, I'm going to try using that wire to connect to the stock o2 sensor, route it over the seats and through the engine lid. Its not a shielded wire, but I just need it to test. If it is a problem with the factory o2 wire I can order some shielded wire from digikey to replace it.

Mon Apr 23, 2018 10:08 am

I'd be inclined to just use the AEM sensor for input to the ECU and abandon the OE sensor. This way you are monitoring what the AEM sensor is monitoring and what the AEM is sending to the stock ECU. To me, the wide band sensor is an upgrade to the original.

Mon Apr 23, 2018 10:11 am

Keep in mind the OE single wire oxy sensor is not heated and takes a while to start working. Like 3 to 5 minutes at 2500RPM. The Nernst effect doesn't start until something like 480*F and above. Your AEM sensor works near instantaneously because it has a heater in it.

Tue Apr 24, 2018 8:15 am

OP hasn't mentioned year of van, heated O2 sensors were stock starting in '86 m.y.

The wideband warmup takes some time, from 5-6 sec. in a warm weather cold-start to 20 sec. or more in cold weather. If using the wideband to emulate a narrowband (which is really the simplest thing to do for the OP, rather than having it run in parallel to the narrowband) then the analog output should be set to be high-impedance and at ~0.45V while warming up (the Innovate systems I use have this as a selection in the setup programming). This is so it looks like what the ECU is programmed to see while waiting for the narrowband to heat up and produce signal.

Dan Has correctly identified the most probable glitch here, a ground voltage offset arising from the narrowband having its ground going thru the exhaust tract and engine case to ground via the engine grounding straps which are also carrying charge current from the alternator, and the high charge current is causing a the engine case/exh to be at a higher V than the van chassis where the WBO2 is grounded. The ECU has a dedicated ground wire connected at the engine case (it's ganged with the ign. dist. wire bundle) that's supposed to reference the ECU's O2 ground to the engine case, this connection may be missing or has a lot of resistance.

You can never go wrong by providing more direct high-current conductors for the alternator circuit (both sides) which I talked about recently here:

https://www.thesamba.com/vw/forum/viewtopic.php?p=8748591&highlight=#8748591

And don't overlook that battery ground!

Tue Apr 24, 2018 8:19 am

Quote: If it is a problem with the factory o2 wire I can order some shielded wire from digikey to replace it.

The shielding is only needed where the O2 signal wire runs in a bundle to the ECU with the injector wires parallel, the current pulses in the injector wires can induce current in the O2 signal wire and easily overwhelm the low-V signal. If you're running an O2 signal by a different route that isn't bundled closely and parallel to other current-carrying wires then shielding can't hurt, but probably isn't necessary.

Tue Apr 24, 2018 10:36 am

tencentlife wrote:

The shielding is only needed where the O2 signal wire runs in a bundle to the ECU with the injector wires parallel, the current pulses in the injector wires can induce current in the O2 signal wire and easily overwhelm the low-V signal. ....

I wondered about that. Thanks for posting that info.

Neil.

Tue Apr 24, 2018 6:15 pm

Ok, so, now I'm lost. I re-connected the AEM gauge O2 output to my ECU and now it's behaving the same way as the stock narrowband sensor. Lean spikes, jumping between 14.0 and 17.0 air/fuel ratio. Every time it goes lean the engine shudders, so I know it's actually happening. And I know the AEM gauge was definitely working with the ECU before, because I drove it all winter like that. It was oscillating the air-fuel ratio above and below 14.7 as it should. Now its all over the place.

The van is an 86 Syncro. I just checked the ground paths from the ECU harness to the engine with my ohmmeter (Pin 19). Also checked resistance through Pin 13 to the engine. Everything is well grounded.

Tue Apr 24, 2018 6:26 pm

long shot question... how old is your battery? i ask because the battery is a big capacitor that filters out all sorts of voltage ripple and noise. when batteries age, and some do this less gracefully, their ability to absorp peaks and flucuations diminishes. all sorts of strange symptoms can come from a battery that, yes, will start an engine, but has a high impedance to changing voltages.

-dan

Tue Apr 24, 2018 6:33 pm

You could also try disconnecting both battery terminals and touch both terminals together (not connected to either side of the battery) and turn the key on. Leave it like that for 5 minutes. "Hard Reboot" is what a lot of guys call it. Long shot as I think the 86 ECU is too stupid to have a learning process. Turn key off before reconnecting battery cables.

While it's running, touch the O2 sensor output wire (Green on the ECU harness) connector (peel back the cover on the green wire to black wire from sensor) with your bare hand and touch ground with your other hand. Does the voltage go up? Does the A/F ratio go down and stabilize? If it's working OK it should do both these things. Note engine should be hot and run at 2500 RPM for about 2 minutes before trying this.

You might try a motovac or similar fuel injection cleaning. You may have fuel coming out in "balls" not as a fine mist which will confuse the crap out of any O2 sensor.

Just for sh*ts and giggles what does your ECT sensor read at say room temp and when warmed up. If you have a non contact (IR laser) thermometer you could compare temp to ohms and see if yours is good or not.

If none of the these work look at the injector signal to see what's going on if you have a scope. Crudely you could use a duty cycle or even dwell meter.

Thu Apr 26, 2018 5:05 am

You may have guessed it scuba. I just replaced a fuel injector that died a couple weeks ago. The replacement injector was from a local vendor, i dont think it's the right flow rate. My air/fuel ratio was slightly higher with the new injector. When i pulled the corresponding spark plug... FOULED. Maybe it was confusing the o2 sensor which couldnt keep up.

My original injectors were the gowesty's "best injector for your van". Well, after about 400 miles on them, one has died. And it was failing intermittently, giving me problems since day one. So i replaced it with one from Frank Condelli. It looks identical but part numbers dont match up so i ordered a gowesty replacement injector.

Fri Apr 27, 2018 4:14 am

Your symptoms and O2 sensor readings could also come from a clogged exhaust. Measure the back pressure on the exhaust upstream of the cat at around 3500 RPM. Should be about 2.5 PSI max. If your exhaust is clogged you will get the engine constantly trying to lean itself.

Fri Apr 27, 2018 5:41 am

I would prefer a "matched" set of injectors regardless of manufacture, flow, etc. The oxygen sensor assumes on cylinders are equal and adjusts mixture accordingly. One cylinder running richer or leaner than the others plays havoc with the sensor system.

Even new injectors off the shelf should be flowed. You'd be amazed what we find testing new injectors. Ultrasonic cleaning will restore new injectors to spec usually. I think it is related to how long they sat on the shelf and the residue from the bench test fluid used at the time of manufacture.

Fri Apr 27, 2018 8:21 am

I did a flow test and the new injector definitely was giving too much fuel. I should get the proper matching injector early next week.

About the exhaust flow, i actually have noticed the exhaust note has changed slightly over the last month (unless its my imagination). I replaced the cat a couple months ago but yesterday I removed the cat and checked it to be sure. It looked fine, still looks new. How common is it for a muffler to plug up?

Fri Apr 27, 2018 9:04 am

Quote: How common is it for a muffler to plug up?

Uncommon unless the cat element broke. Usually has a rattle when running or struck with a fist if one of the internal baffles has fallen or if part of the old cat is stuck in it. Best way to tell if you have an exhaust restriction is to measure the back pressure. A 0 to 10# fuel pressure gauge with a metal cone or metal pipe on the end of the hose works fine as long as you measure quickly and don't leave it hooked up too long. Silicone hose works even better. I would usually drill a small hole in the exhaust pipe upstream of the cat and then fill hole with weld when done measuring or you can take out the O2 sensor and measure back pressure there.