This is a rather advanced question for the experts. How capable is the F-bomb ECU for proportional and integral fueling in closed loop? Some background, and then the REAL question below.

Typically, the goal of any PI or PID loop is to reach a targeted stead state - such as idle speed for example. There is Proportional, Integral, and Derivative air to control idle speed. Simply put, the idea of proportional air is "I'm instantaneously off by 'x' RPM right now, so I need to change air by 'y' grams/second." The idea behind integral air is "I'm off by average 'x' RPM over 't' amount of time, so I need to change overall air by 'y' grams per second to move the average." The idea behind derivative air is "based on the slope of the idle speed I'm predicting I'll be off by 'x' RPM, so I'm preemptively going to change air by 'y' g/s to prevent over/undershoot." but the end goal is a steady idle speed.

For some reason, fueling is not approached this way. With NB oxygen sensors, for some reason, the goal is to oscillate back and forth and NEVER settle on the desired NB O2 switch point mV. In fact, if the NB readings aren't bouncing around, the assumption is something is wrong. There is proportional fueling, which adds and cuts fuel to keep the NB switching, which is analogous to proportional idle air. There is integral fueling to correct long-term errors over time, analogous to idle integral air. There is no derivative fueling control, but there are plenty of PI loops out there that maintain good process control without a derivative control.

So the REAL question is -- is it possible to set the proportional and idle fueling parameters to achieve and maintain a steady state 450-mV NB reading similar to what is almost always the goal for any PI or PID controller? Understanding of course there would be some amount of fluctuation, but not the constant bouncing from top to bottom.

I drew these two charts to illustrate the nature of the question, which is basically, can proportional and idle fueling be configured to result in the bottom chart instead of the typical top chart for NB O2 readings?

 

I don't know. It's a good question.

I know the faster they switch, the closer you are to your target fueling. Or at least it seems that way.

 

you can control fuel in that way via a wideband sensor, especially the newest ones

but my understanding is that narrowbands work that way on purpose as they are a switching type sensor

and even cruder still is my understanding of catalytic converters but i'm pretty sure they need that switching to operate too.

 

Yeah I have heard narrow bands have such a tight AFR that they would be bouncing back and forth anyways because they fall off .450 volts pretty quick. I also agree with Doug and remember at one point hearing the converters operate better from being rich then lean.

 

Very capable.

The short answer to that question is NO.

The longer answer is.... the whole thing is designed so that the O2 sensors oscillate between rich and lean. The PCM is expecting the fueling to go rich and then lean and then rich again. From the rich and lean readings it then calculates if the average fueling is too lean or too rich and makes the necessary adjustments (fuel trims). The fueling parameters are a PID controller (just like the airflow PID controller) but they control how much and how fast your fuel trims change in order to maintain stoich AFR.

 

Thanks I appreciate the answers. I was thinking to start chasing it down, but now I don't have to. It helps to know how the ECU expects it to behave rather than to try to force it to do something it isn't designed to do.

 

Truckdoug summed it up well imo.

 

Honestly most my standalone when setup properly with wide bands did this. They would bounce from 14.8 to 14.6 back and forth. Sometimes further just depending on how much map was off.

 

This is what I would expect and honestly I would consider it steady state

 

haha i was talking to an engineer at work today about this and he tried explaining the actual chemistry behind the cats needs to "see" both lean and rich conditions to function properly....he eventually let up when he saw my eyes glassing over.

you chemistry guys are smart.

 

So just knowing how it's supposed to work made a huge difference. Yyyuuuugggggeee difference lol!

My proportional fuel while cruising at steady state was wandering from AFR 13-16 like a sine wave with a 3-second period. So I kept working with the closed loop proportional fuel adders and also idle, and it seems more steady. It greatly reduced the amount of air and spark correction necessary to maintain idle. It also got my STFT bouncing between -5 and zero vs -20 and +15.

But my values are down around 30% of stock

 

You can simply turn off proportional switching at idle, but I don't see a way to do it across the board without disabling closed loop. This statement is not intended to say there isn't a way to do it, only that I haven't noticed it yet.

 

You'll see better fuel economy now too.

Enjoy!

 

What changes did you make there?

 

Basically I cut the error tables back so that the multipliers are lower. I cut proportional fueling way back. WAYYYYYY back. I think idle proportional is 0.3x stock. The closed loop prop is even more lower, but idk what percent. The integral timing is reduced for zones 4-8, and the integral multiplier is stock.

I might cut proportional even more and cut the error time also.

Trying to make smaller quicker changes vs longer slower changes.

 

I saw this as an explanation for the Proportional and Integral as well from HPT:

The proportional fuel moves the base BPW around to find the 14.7 crosspoint with the O2 sensor. The higher the number, the more fuel gets added OR subtracted.

The integral is how much time must pass if the 14.7 point is not found before it can make a fuel correction. If the time values in base table are exceeded, the VCM makes a short term fuel correction. While the table time can go up to seconds before making an adjustment, keep it in the millsecond range.

Both tables use an offset vs O2 error to speed things up if the error is greater.


So as long as your VE table is pretty spot on, you want the CL system to only make the bare-minimum adjustments... so I'm going to play with this to see if it helps my return to idle...

I'm also going to turn off the idle proportional to see if it does anything.

 

Here's my thought on that Jake. Lock in your IAC. Position and your spark timing in the scanner. If it starts surging really badly it's fuel, and these adjustments could help - especially if you're seeing the wideband look like a roller coaster. If you lock in your spark and air, and its steady, probably not the issue.

 

My wideband doesn't look too bad. It's wavy. But nothing horrible. Not 3 point swings. Maybe .2-.5 swings. 14.2-14.8 type of thing.

Of course, I've made adjustments there anyway. I locked in my switchpoints at 450 across the board and made some adjustments to the proportional based on some idle adjustments I found on HPT back in the day. Haven't changed anything since and it seems to be close.

One thing to note - I do have James Short looking at my airflow now. He's made a lot of adjustments and it's started to behave like a car again. Waiting on him to analyze my latest log and provide more feedback. So maybe I'll be close after that. Still bouncing off the idle and surging up as I come to a stop. But less so now.

 

I'm trying to figure out if everyone is discussing the same thing here. It is my understanding that proportional fueling is a bit of a sanity check to insure that the O2's are switching properly ( as well as something to do with cat converter function). On the other hand, we have proportional, integral, and derivative tables for adjusting rpm error at idle. These tables have nothing to do with proportional fueling.

 

Correct. I was asking if the P-I fueling could be set up to achieve a smooth mV reading vs the up and down switching, and I used PID idle as an example to illustrate what I was asking.

And the answer is effectively - No, that's not the way the proportional fueling controller is designed to work.