ahickman

Starting to fine tune my 3.0 I recently rebuilt. After installing a new muffler, I noticed a faint smell of fuel and some black soot building up on the inside of the new tail pipe. Checked the O2 sensor, TPS and VAF sensor with a voltmeter. Nothing out of whack. After scratching my head a while, I hooked up the oscilloscope to the O2 and Vs leads going into the ECU. O2 sensor reacts as expected, but there is an significant ripple voltage superimposed on the carrier voltage. Two graps below, where the time base is the only difference. Any thoughts as to what would cause the ripple? Bad ground or a capacitive voltage somewhere along the O2 sensor wires running back to the ECU is the only thing I can think of.

motoriderbc

Turn on the "filter" function on the labscope if it has it. Should clean the signal up. The PCM won't be effected by that dirty signal it just looks at the low /high / low swing.

scope103

Nice scope trace! (What box is it?)

Your ripple is about 15hz, or 900ppm. Since your idle is probably close to 900 rpm (spec is 800), I'm going to guess that you're just looking at plug noise.

abecedarian

I see what you're talking about... and I think you're seeing things...
You're first plot is 250ms/Div and the second is 1s/Div. You've got finer resolution on the first plot than you have on the second so you can see more subtle variations in the amplitude of the signal. Zoomed out / less resolution, the average becomes the predominate observation and 'dilutes' the superimposition you see on the first plot.

Out of curiosity- are you sampling the O2 sensor at the same rate the ECU is?

Soot in the tailpipe IS normal. If you can find me a car with 1000 miles or so on it and no soot, I'd be surprised. I'd become worried if the plots were showing flattened peaks and valleys meaning that the fuel mixture ratio extended beyond the range of the O2 sensor.

ahickman

1. I am using a very inexpensive USB based scope from Parallax ($149). Small, portable and accurate.

2. Not sure what rate the ECU samples at, but figure it has be at least three times faster than the greatest oscillation rate (3 x 0.8 = 2.4 Hz) in order to detect the high/low transitions.

3. Below is a trace of the O2 output on another 94 (22re) I have. Noticeably smoother.

I can't help but think that the rich condition and low power is related to this ripple voltage. Not sure, however that it is the cause for richness or is just an effect of richness.

Other the following conditions what will cause an engine to run rich?

1. bad engine coolant sensor
2. timing
3. TPS sensor
4. VAF sensor
5. stuck cold start switch

ewong

be aware that for "narrow range" (aka OEM) O2 sensors - the sensor can NOT read the actual Air/Fuel ratio.

All the sensor can read is a BINARY value
-- RICH
-- LEAN

the intermediate voltages don' mean much

So the ECU use the "duty cycle" of rich and lean mapped back to internal maps that are based on LAB TESTING of the engine with a wideband ego back at the factor design office.

Use the Vf signal to tell you how much the ECU is compensating (um... = what year is the engine?)

Id suspect a TPS that has a worn spot on the trace...

thook

You guys blow my mind. Are everyone of you electricians or something?

JamesD

You have a normal pattern for the O2. That ripple is a small interference of the wiring, nothing you can really do about it. Like someone else said, you have it magnified. Your vane air flow could be out of wack but you said you tested that. Did you test it under a load or did you just use your finger? Also check for the computer ground it make get rid of the ripple. I have seen many cars like the pattern you shown and most of them run perfectly fine and small percentage do not.

James

JamesD

I have a degree in automotive technology that is where I learned. Some of these guys may have the same or some electrical engineering under their belt. Do not feel bad if you do not know, this isn't something everyone would know even if you build your own truck or work on many vehicles.

I use a PGM, power graphing meter by Snap On called the Vantage. Very good tool but the sample rate is just a little slow but not that slow. Best tool I ever bought.

James

abecedarian

I've got 11 years in cellular / microwave work, as a teen was a CB hobbyist and among other things have been a certified diesel mechanic, california lottery machine tech and RV eletrical system designer / installer.

ahickman

OK, now I am even more curious about how the ECU uses the O2 sensor. Ewong, you say that the ECU sees the voltage (e.g. 0.9 = ON and 0.2 = OFF) as discrete binary values. So is there a cut-off or threshold voltage for the ON/OFF conditions? As I understand it from the FSM, autoshop101.com and Thook, the ECU goes into closed loop and looks at the O2 sensor, cooland temp sensor, VAF and TPS to determine the spark timing and injector duration. I would have assumed that the ECU would be quick and smart enough to consider the rate of change (slope) of the voltage and not just the binary condition. If I understand you correctly, once the O2 voltage reaches say 0.7 volts, the ECU considers the exhaust to be rich. Not until the voltage drops to say 0.3 volts does it recognize a lean condition. If this is correct and it depends on a lookup table, then resetting the ECU does nothing in terms of re-learning the sensor as so many indicate in other threads?

abecedarian

The information I've learned is like this:
Let's say that the O2 sensor sends a signal. The amplitude of the signal is not used by the ECU to determine how much it needs to enrich or lean the mixture. Instead the ECU only determines whether the signal indicates rich or lean and then adjusts the mixture in steps until the O2 signal returns the opposite condition, at which point it begins altering the mixture in steps until the condition reverses again... ad infinitum.
The ECU does learn an offset with regards to how often it is lean compared to how often it is rich and compensates using that as well. This helps the ECU compensate for an aging / failing O2 sensor. Also, if it is adjusting the mixture more often in one direction than the other, and that period exceeds a predefined amount, it triggers the corresponding trouble code for too rich or too lean as appropriate. The offset can be 'unlearned' when an O2 is replaced, but resetting the ECU returns it to a neutral value and accelerates the ECU's learning of the O2 sensor.

Flash319

I read that "learning" stuff on a lot of threads but I think it is BS.
I would think that the rate of "decision making" would have to do with the load and RPM and not set to a specific frequency (monitoring fast, decision slow). You can only make changes to timing and duration of fuel injection at a very slow rate compared to what you can read O2 and such. I would think that making map changes would be controlled by load (TPS and AFM). The 02 would then "fine tune" that specific load map to get better combustion. Maybe there are 2 maps for every load, one for a lean condition and one for a rich. Then it flips between them as the mixture oscillates.

Maybe I am overdoing the smarts of an ECU. It is probably easier then I think. I am an industrial robotics/controls engineer and usually this type of logic is simple when thought through. I would think Toyota's is the simplest, thats what makes them robust. This is also why I think you may be trying to fix something that is not broken.

Nice to see some fellow electrical guys on here.

ahickman

Abe, I think you answered my next question about how the ECU deals with an O2 sensor that is shifted in the rich/lean direction. Thanks.

Flash39 also make a good point about the rate of change in timing. Although possible, I doubt Toyota would design the ECU to be that responsive.

JamesD

The O2 is supposed to toggle above and below .5 volts to give you a wave pattern. The computer reads it and takes the average and it adjusts it from there. The computer doesn't learn the O2 sensor since it is constantly changing and varying. Oxygen sensors are self-creating voltage sensor and do not generate a binary value. Their is no on/off, it is a variable voltage generator. When an engine is running constantly rich, the voltage will stay more on the upper end, .5v-1.0v and lean would be the opposite.

Here is a trick that can help. If you are wondering what would cause the truck to run rich, force the computer to go into open loop. Jump the Te-Te1 connector or whatever is the correct jumper for checking ignition timing(base timing) and then put your scope on it. This way you omit the other sensors that are needed for closed loop and it can help you determine if one of those sensors is faulty. The O2 sensor will tell you how the truck is running even in open loop. Look for the wave pattern and see if it differs.

James

abecedarian

Maybe what I said was confusing. The ECU does not learn the O2, but learns about the trend of the mixture. This is reflected in the ECU's Vf signal, which when checked in the proper mode, can show the learned adjustment to the fuel mixture.