How do you richen up SPFI? HELP!

[Vertigo]

I think I am running a bit lean and I was curious if there is a way to determine how good I'm running. I know the ideal ratio is 14.7 so how do I find that out with a SPFI non turbo EA82. And knowing that... if it is lean or rich how do I adjust it? is it involved with replacing sensors or turning a screw somewhere?

[EmmCeeBee]

This doesn't answer your specific question (cuz I don't know about manual adjustment :-\ ).

 

But if your ECU is working and your oxygen sensor is OK, then I think it's out of your hands. The O2 sensor keeps the air/fuel ratio at 14.7 by adjusting fuel delivery many times a second (through signal to the ECU).

 

If you suspect the ratio is out-of-bounds, you might want to test your O2 sensor. There's a good article at:

http://www.wps.com/LPG/o2sensor.html

 

Other possibilities are in the fuel system, like low fuel pressure or a bad regulator. But if it's so far off that the O2 sensor couldn't compensate, I'm assuming it would turn up an error code in your ECU.

 

-- Mark

[calebz]

One possible way to check if you are running lean is by pulling your spark plugs and checking their condition.. all white and crispy indicates a lean condition. One of the most common causes is an intake vacuum leak.

 

The absolute best way to tell is by hooking up an A/F meter.

[asavage]

14.7 = atmospheric pressure at sea level

14.6 = stoichiometric ratio of gasoline to air

 

I have been giving this considerable thought recently.

 

In closed-loop mode, the ECU will generally attempt to keep the mixture close to its calibration. Changing the fuel pressure would be at least partly compensated by the O2 feedback. I've confirmed this empirically on my own SPFI system.

 

However, if you fool the ECU into thinking the engine is not up to temp yet, but keep it above the open-loop/closed-loop breakover point, I think that you could convince the ECU to fatten up the mixture -- leaving aside whether this is desireable.

 

This is all speculation on my part, of course.

 

The coolant temp sensor operates over a range of about 6k to 100 ohms. The only stated reference points I'm aware of are two:

[edited to reflect new info from Chilton's}

 

14°F = 7-11.5k

68°F = 2-3k

122°F = 700-1000

 

Want richer? Wire a linear-taper potentiometer in series with the coolant temp sensor, mount it on the dash, dial in (from zero to n) more resistance, and the ECU thinks the engine is colder than it actually is.

 

Of course, this may affect the IAC setting the idle speed as well, but I'm thinking about doing it for diagnostic purposes only, so I don't really care about the idle speed.

 

(I did this as a teenager to my Mom's early FI Volvo 145 -- Bosch K-Jetronic, I think it was. Worked pretty well. Idle speed was not under computer control on that system, though.)

 

Want leaner? Wire a pot in parallel with the coolant temp sensor, dial out (from max toward zero) to lean. The ECU thinks the engine is real warm, and the air will be less dense, so it'll reduce the amount of fuel to compensate the mixture. I think.

 

On measuring A/F: you have to measure before the cat(s). Hard to do. As I mentioned in another post, the O2 sensors that we have are not designed to output a relative A/F ratio: they monitor the oxygen level in the exhaust stream, and toggle their output to an OFF/ON state when the level of oxygen attains or falls below a certain threshold. The ECU takes that ON or OFF and (combined with other inputs) tailors the ignition timing and injector ON pulse width to try to keep the O2 sensor "crossing" that point a number of times per some unit of time. Think of a old-time points dwell meter (which is a tool that was actually used for working with early feedback carbs in systems that used the first automotive O2 sensors).

 

Also, the stoichiometric ratio is a theoretical number only. At least on older, carbed rigs, they never ran that lean: couldn't get efficient enough atomization->vaporization of the liquid fuel to burn reliably at that ratio. Typical mix ratios in, say, the 60's were closer to 12.5:1, on up to maybe a bit over 13:1 (IIRC; it's been a long time since I needed to remember those numbers). Until about 1970, nobody was seriously trying to get 14.6 (stoichiometric for gasoline) because of technical problems getting the existing fuel delivery systems to meter that finely over the entire range of operating conditions. Stationary plants, that's a different story.)

 

I don't know if this is still true of today's systems. The state of production FI systems is quite a bit beyond the stuff I learned way back when. Gaseous fuels such as LPG and CNG vehicles, for example, can run at up to and over stoichiometric in controlled situations, and work fine, but there is no issue of vaporization with a gas, as opposed to a liquid, fuel.

[Skip]

I disagree with this statement

"toggle their output to an OFF/ON state when the level of oxygen attains or falls below a certain threshold. "

as I did in the previous post and will set about experimenting to prove the 02 sensor's output is varriable and not an on/off toggle as you purpose.

I do not think you have seen the output on a scope of an 02 sensor connected properly to a vehicle that does not have feedback control.

The reading I have done, does not support your on/off theory. I believe you are confussing the ECU's reaction to the high low voltage output as a digital signal from the 02 sensor.

[asavage]

Cool. Do that.

I've got a portable 'scope, but lack the ambition to do the testing. Do let me know what you find.

 

[later]

As someone else recently posted, the WPS site (we LPG enthusiasts call it the "old Rambler site") has a good collection of decade-old O2 sensor info, some of which is contradictory, but most of which does support what I've been saying: the O2 sensor that we get for automotive use can sense A/F in a very narrow range only, and the range is so narrow as to be a "rich/lean" indication rather than a set of gradations that can be mapped to an output that makes sense.

 

Apparently, several firms sell inexpensive things that purport to give A/F output using a <$100 O2 sensor. While I don't doubt that an LED bar graph can be driven using the output of an O2 sensor, and that some relative richness/leanness can be implied by using an O2 sensor -- after all, that's what they're designed to do -- it does not appear to me or others that the range of O2 sensor output is sufficient to imply the range of LED bar graph output that I've seen on the units being sold.

 

It's better than nothing, I suppose.

 

Anyway, when you do your testing, I'm interested in the results. Esp. if you can get stable numbers without signal conditioning -- you may have to load the line and use a current probe rather than a voltage probe, depending upon your particular environment. And the amount of current is going to be very, very small.

 

Hey, I may be all wet, but I'm not going to assume so

[Vertigo]

so... replace the o2 sensor is the vibe I'm getting. It is at 152 and I havn't replaced it since I got it at 145 miles. And I don't know if its ever been replaced.

[asavage]

O2 sensors wear out, that's a fact. They wear out less frequently when they are not overheated, contaminated with silicone (from some Room Temperature Vulcanizing sealants, or a leaky head gasket that causes coolant to go out the exhaust -- coolant which contains silicates), lead, phosphorus (from older oil formulations and snake-oil additives and lots of oil burning in a worn engine) or subjected to very rich mixtures over an extended length of time.

 

Also, they will not operate correctly if the exteriour jacket is coated with crud, as they operate by the differential of oxygen on the outside with respect to oxygen on the probe side.

 

This page has a test procedure for testing for a dead O2 sensor that is pretty easy, and some decent info on O2 construction and integration (though there are several misleading statements on the site, such as assuming "fuel" = "gasoline", kinda like assuming all PCs run Windoze).

 

We Subaru owners have it easy, because our ECUs have an O2 monitor LED which can tell us whether the rich/lean signal is "crossing" .45v -- which is should, in normal (closed-loop) operation. But that only covers one failure mode: failure to detect lean.

 

A mostly-dead or contaminated O2 sensor will falsely tell the ECU that there is an excess of oxygen in the exhaust stream (lean), and the ECU compensates by increasing the injector pulse time.

 

So, if you think you're running lean, I wouldn't go to the O2 sensor as my first choice, unless your O2 monitor LED does not light at all. I'd be looking for intake air leaks, dbl-check the fuel pressure regulator (and triple-check it on Turbo engines), and EGR operation (because when the EGR operates correctly, it replaces intake charge oxygen with non-oxygen, and the ECU reduces fuel to compensate, but if the EGR exhaust port is plugged, the net effect is too much air for the fuel being injected -- in theory, the ECU would compensate this.)

 

A worn but not worn-out O2 sensor gets "slow", and does not change output from rich/lean fast enough, which misleads the ECU into always trailing behind actuality. This failure mode is harder to test for, and is the main reason why so many mfgrs have a blanket mileage replacement recommendation. Like timing belts, they're going to fail, but the actual number of miles to failure varies considerably. Unlike timing belts, an O2 sensor's performance usually deteriorates gradually, unless contaminated.

[asavage]

Originally posted by Skip

I disagree with this statement

"toggle their output to an OFF/ON state when the level of oxygen attains or falls below a certain threshold. "

 

I think you're thinking about the upcoming "Wideband" oxygen sensors.

 

How 'bout this:

 

"The newest O2 sensor technology from Bosch builds upon the planar design and adds the ability to actually measure the air/fuel ratio directly for the first time.

 

Instead of switching back and forth like all previous sensor designs, the new wide-band O2 sensor produces a signal that is directly proportional to the air/fuel ratio. . . .

 

The result is a sensor element that can precisely measure air/fuel ratios from very rich (10:1) to extremely lean (straight air). This allows the engine computer to use an entirely different operating strategy to control the air/fuel ratio. Instead of switching the air/fuel ratio back and forth from rich to lean to create an average balanced mixture, it can simply add or subtract fuel as needed to maintain a steady ratio of 14.7:1.

 

. . . But instead of switching abruptly, it produces a gradual change in the voltage that increases or decreases in proportion to the relative richness or leanness of the air/fuel ratio. So, at a perfectly balanced air/fuel ratio or 14.7:1, a wide-band O2 sensor will produce a steady 450 mv. If the mixture goes a little richer or a little leaner, the sensor's output voltage will only change a small amount instead of rising or dropping dramatically"

 

It goes on . . .

http://www.forparts.com/BoswidebandO2.htm

 

"The trouble is, conventional oxygen sensors give on a rich-lean indication. They can't tell the computer the exact air/fuel ratio. When the air/fuel ratio is perfectly balanced, a convention O2 sensor produces a signal of about 0.45 volts (450 millivolts). When the fuel mixture goes rich, even just a little bit, the O2 sensor's voltage output shoots up quickly to its maximum output of close to 0.9 volts. Conversely, when the fuel mixture goes lean, the sensor's output voltage drops to 0.1 volts.

 

Every time the oxygen sensor's output jumps or drops, the engine computer responds by decreasing or increasing the amount of fuel that is delivered. This rapid flip-flopping back and forth allows the feedback fuel control system to maintain a more-or-less balanced mixture, on average."

 

"The newest generation of oxygen sensors are being called "wideband" lambda sensors or "air/fuel ratio sensors" because that's exactly what they do. They provide a precise indication of the exact air/fuel ratio, and over a much broader range of mixtures - all the way from 0.7 lambda (11:1 air/fuel ratio) to straight air!"

 

That's supposedly Bosch writing those words, though with the number of typos, I hope not.

 

Anyway, they said what I said: the typical O2 sensor in use today has a very narrow balance point (deadband/hysteresis) and falls off that point very easily. The whole feedback system is designed around this feature. The ECU can get three possible values from an O2 sensor (after filtering):

 

[*]Signal > .5v = rich condition exists

[*]Signal < .45 = lean condition exists

[*]Signal = .45v = sensor failure or not up to temp.

[/list=1]

 

In practice, the midpoint is likely a lot wider, possibly as wide as 0.1v, but the sensor is not going to output, say, 0.35v to indicate "a little bit leaner than stoichiometric".

 

The upcoming wideband O2 sensors will do just that, but that's not what we have available today, for less than several hundreds or thousands of dollars, and if you read the details of this dual-elemen, oxygen-pump widget, you'll see that this wideband tech is exactly what Mark Stavropoulos posted about UEGOs in 1994 at http://www.wps.com/LPG/o2sensor.html .

 

What is being sold as an A/F meter is, I assume, a conventional O2 sensor and some time-based averaging electronics glue parts, so that if the O2 sensor tends to spend more time ON than OFF, the gadget lights up more of the "rich" LEDs. And vice-versa for lean. Is this a correct relationship? Well, sorta. Is it useful? Probably. Is it telling you the actual A/F ratio? No -- it's just guessing.

[Skip]

thanks Al,

I bow to your expertise , this shows me the A/F ratio meter I use is basically an average of the exhaust 02 content being a result of the on/off toggling of the 02 sensor output, as you have purposed since the first post.

 

I guess I now question, is it better to know your average 02 content than to have no clue (other than doing a plug cut) about the relative mixture in the combustion chamber?

Will you argue that 90 % of our problems lie not in the fine line a wideband will give us, but in a general window of mixture in the combustion process.

In other words is the obsevation of an A/f ratio meter during daily driving worthless?

Thanks again for the research and the links, I sit corrected.

[asavage]

Originally posted by Skip

is it better to know your average 02 content than to have no clue.

A poor diagnostic tool is often better than no tool at all, if you know its limitations..

Will you argue that 90 % of our problems lie not in the fine line a wideband will give us, but in a general window of mixture in the combustion process.

The averaging that I think is going on implies that what you see is a guess from the electronics, and one which is a few heartbeats behind reality upon which those guesses are predicated. As long as you know this, and factor it into your tuning, you can compensate for the tool.

 

The problem arises when you think the tool is doing one thing, but it's not. In knowledge lies power.

 

We now have wide, fast worldwide communication, via the internet. This is great! But information and misinformation and downright lies can be broadcast equally easily. It has become very difficult to determine the wheat from the chaff, the signal from the noise, the truth from the BS. In a typed msg, everybody sounds like an expert or a fool.

 

Just as a pretty font and nice, even margins do not imply that what the words say are true, my or your writing or grammar skills do not imply the veracity of what we type. For this reason, just as in Real Life , I require a source to be vetted (or, put another way, to qualify themselves) before I accept the proffered knowledge.

 

I'm not an expert, but I play one at work We all have different skills, talents, abilities, resources, and passion to bring to bear against problems. The trick is to keep the dissembling to a minimum, and when misinformation can be refuted, to do so.

 

In other words, is the observation of an A/f ratio meter during daily driving worthless?

It likely has some value. I don't know, I've never used one of these sub-$200 units, being skeptical of the methodology used in their construction.

 

I steer clear of tools that I think were designed for the same purpose as the Sears Self-Sharpening Chainsaw (am I going back too far?): it wasn't designed to cut wood or self-sharpen, it was designed to do one thing well: sell like hotcakes, and it succeeded.

 

I think that a conventional O2-based A/F display is too close to that Sears POS, but that doesn't mean that you can't prevail and cut wood with it, only that you have to bear in mind its limitations. I hope that I've helped dispel the myth of the A/F meter's infallability -- if you take its output with a shaker of salt, it may be quite useful. Better than seat-of-the-pants, not as good as a chassis dyno or an onboard OBD-II realtime scantool.

 

There are a couple of reasons why SnapOn, Mac, OTC, et al charge so much for their scantools, and it's not all in the hardware or support. Keeping up with changing tech is so challenging, that most of us will change careers something like seven times over our lifetime, and that number is rising! If you're young, this doesn't sound too bad, but if your on your fifth career, its scary, indeed, let me tell you!

 

[preach]

You, Skip, are a cornerstone of this online Subaru Enthusiast community. As such, you have invested considerable time and effort in this community, and people who come for an specific answer, tend to stay to learn more, and to build on the social glue that you and others provide. Because you "qualify" yourself with every new posting, you've become an influencer to those here, and your words carry weight. It's doubly important that you type what you feel to be truth, because others will take it as gospel, and truth and fiction travel at the same speed.

 

I, Al S., will likely move on from this (excellent) Subaru community in a few weeks, when I have this particular project behind me; I prefer slow, quiet, and cushy to the qualities of this '93 Loyale, and the reasons I bought this one are that the woman needed to leave the area immediately, needed money bad, and I had a friend lined up whose wife wanted it. I'm just making sure that all parties are satisfied -- I'm not a long-term Subaru customer. But, I can't not fight the darkness; I have to try to light a candle.

 

But you have a responsibility to keep flow of information on the Truth side of the deadband, and when it swings over to the BS side, nail it as you see it.

[/preach]

 

I hope I've helped.

 

Keep On Truckin'

[archemitis]

my spfi ran fat all the time, especially when i floored it and the o2 was doing nothing. and even with advanced timing, exhaust, and a modified airbox, i couldnt get it to lean out much at all(it went down one half green bar on the autometer, and it was usually about 4 bars into the green). why do you think you are runing lean?

 

the o2 sensor only keeps it CLOSE to 14.7 when you are cruising at part throttle, so for performance driving (WOT) the 02 sensor is just another guage sensor.