EEC-IV Adaptive Control according to Mike Wesly

Anything technical worth keeping around related to EEC Tuning or just general engine tech info.

Moderators: cgrey8, EDS50, Jon 94GT, 2Shaker, tmoss

Locked
User avatar
tmoss
Gear Head
Posts: 46
Joined: Mon Feb 02, 2009 7:15 pm
Location: St Louis, MO

EEC-IV Adaptive Control according to Mike Wesly

Post by tmoss » Mon Aug 09, 2010 9:44 am

Here is a nice article on how the EEC adaptively tunes fuel for changing conditions.
Attachments
EEC Adaptive Control.doc
(34 KiB) Downloaded 601 times

User avatar
cgrey8
Administrator
Posts: 10348
Joined: Fri Jun 24, 2005 5:54 am
Location: Acworth, Ga (Metro Atlanta)
Contact:

Re: Mike Wesly's EEC-IV ADAPTIVE CONTROL

Post by cgrey8 » Mon Aug 09, 2010 1:06 pm

The only criticism I can give to this article is that it gave me the impression that an EEC adjusts Adaptive Control in a matter of 1 Injector Fire and that the HEGOs can tell it how rich or lean the mix is. Here's a the passage I'm refering to:
...The EEC will calculate the difference from what it wanted and what it got and update the Adaptive table with a 'correction factor' and use this correction factor to reduce the injector pulse width the next time the injector fires...
For the purposes of concept, I guess that works. But for the DIY tuner, it's important to know what's actually going on. It takes a bit of time for Adaptive Control to learn.

First off, a HEGO can't know the AFR of the burned fuel in the engine until the engine expels the exhaust out of the engine. And then, it may take 2-3 engine fires for the reading of all cylinders using that HEGO to stabilize. To adjust this stabilization delay, there's a table in the tune referred to as HEGO delay.

Once the HEGO reading is considered stable, the EEC only knows that the burn is rich or lean. Note, there is no "perfect" state. The exhaust is reported as either rich or lean. The EEC doesn't know by how rich or lean the burn is. Eventually, the EEC will need to figure that out but it won't know that just yet. For this discussion, lets assume the HEGO is reporting rich. In response, the EEC adjusts the Short Term Fuel Trim (STFT) known as the LAMBSE in the direction opposite of the HEGO feedback state. In this case, the EEC adjusts the LAMBSE lean (raising the value). The amount it raises depends on the RPM/Load, time since last HEGO switch, the HEGO Bias at the current RPM/Load, and a number of other parameters in the EEC that are beyond the scope of this discussion.

The EEC will continue to increment the LAMBSE leaner until the HEGO reports that the burn is lean. Upon a lean report, the EEC trims the LAMBSE back rich which repeats the whole process. The EEC wants to keep the HEGO switching between rich and lean all the time. If the EEC can keep the HEGO switching, then the AFR is being maintained at or near 1.0 lambda (aka stoic) or 14.64 AFR for gasoline. At this point, the EEC is maintaining a stabilized Closed Loop running condition. For this discussion, lets say the EEC stabilized with a LAMBSE value around 1.10 lambda or 16.1 AFR assuming gasoline. A LAMBSE value of 1.10 represents the condition the EEC expected from the amount of fuel currently being being injected. Since the HEGOs are confirming that actual lambda is 1.0, this means the EEC needs to learn that the engine doesn't need as much fuel for this RPM/Load condition. Any combination of the following could explain the error between the LAMBSE and the actual condition being confirmed by the HEGO:
  • The MAF curve is mis-representing the MAF's behavior and thus has the EEC believing more air than actual is entering the engine, which is resulting in too much fuel being injected
  • The Injector Parameters are mis-representing the Injector's actual capabilities and the injectors are delivering more fuel than the EEC expected
  • Some other physical problem causing more air to be reported or more fuel to be delivered than should be expected.
Whatever the reason, this error can be learned assuming it is consistent. If the error isn't consistent, then all the learning in the world isn't going to correct for it. For the purposes of this discussion, lets assume the error is consistent and thus learnable.

After a certain amount of time in stabilized Closed Loop, the EEC evaluates the LAMBSE value to determine if it can learn. In this case, it can. Being off by more than ~.05 lambda (tune adjustable) is enough for the EEC to attempt learning. But it doesn't just take the difference of 1.10 and 1.0 and apply that immediately. It is much more gradual than that. Assuming the engine is in an RPM/Load area allowed for learning, the EEC begins adjusting the Long Term Fuel Trim (LTFT) lean by lowering the value. The LTFT is known as KAMRF, which is sometimes shortened to KAM (Keep Alive Memory). The KAMRF value IS the EEC learned value. Most EECs can associate a different KAMRF value for the different RPM/Load conditions an engine can be in. So to get the LAMBSE from 1.10 down closer to 1.0, the EEC begins adjusting the KAMRF. The amount and rate it adjusts the KAMRF is again, details of the tune that are beyond the scope of this discussion. On each KAMRF adjustment, the intention is that the EEC runs the engine slightly lean. The lean condition causes the normal Closed Loop behavior to trim the LAMBSE from 1.10 down towards 1.0 by the same amount the KAMRF was adjusted. Once the LAMBSE is within an acceptable range (tune adjustable) from 1.0, the EEC quits trimming the KAMRF any further and calls it good.

From this point on, the hope is that any time the engine re-enters this RPM/Load, the EEC will not be over-fueling and instead will be delivering the correct amount of fuel while keeping the LAMBSE value near 1.0. For the average stock vehicle, the EEC has learned the error in the system and at this point should be performing the way the factory intended.

Although on a modified engine that has the benefit of a DIY tuner, this KAMRF info is important because it gives the tuner insight into whether the tune is inherently rich or lean and thus gives guidance as to how to adjust the tune so the EEC isn't reliant on KAMRF learning to get fueling right.

Again, I'll reiterate that the effectiveness of Adaptive Control is based on how consistent the error is. If the LAMBSE adjusts rich sometimes and lean other times while in the same RPM/Load, then that's not consistent and is usually indicative of some other problem. Common causes of this type problem are bends immediately before or after a MAF or failing Fuel Pressure Regulator that isn't able to keep fuel pressures stable. Some problems are not easy or feasible to correct. For these cases, you sometimes find that Adaptive Control just gets in the way and makes things worse and that you are better off relying on normal Closed Loop LAMBSE adjustment to chase the current condition without Adaptive Control trying to help.
...Always Somethin'

89 Ranger Supercab, 331 w/GT40p heads, ported Explorer lower, Crane Powermax 2020 cam, 1.6RRs, FMS Explorer (GT40p) headers, Slot Style MAF, aftermarket T5 'Z-Spec', 8.8" rear w/3.27s, Powertrax Locker, Innovate LC-1, GUFB, Moates QuarterHorse tuned using BE&EA

Member V8-Ranger.com

User avatar
tmoss
Gear Head
Posts: 46
Joined: Mon Feb 02, 2009 7:15 pm
Location: St Louis, MO

Re: EEC-IV Adaptive Control according to Mike Wesly

Post by tmoss » Tue Aug 10, 2010 10:13 am

Nice synopsis cgrey8. Mike wrote this quite a while ago and while informative, it does not go into a great deal of depth - but any good information is better than none. I never got the impression that it said adaptive would be done in a very short period of time.

I would try to keep the MAF at least 2 MAF diameters away from any bend on either side - more if you can.

User avatar
cgrey8
Administrator
Posts: 10348
Joined: Fri Jun 24, 2005 5:54 am
Location: Acworth, Ga (Metro Atlanta)
Contact:

Re: EEC-IV Adaptive Control according to Mike Wesly

Post by cgrey8 » Tue Aug 10, 2010 11:37 am

The document is perfect for the intended audience, which I'm assuming are people that were not DIY tuners with dataloggers. I think it is an excellent writeup for anybody new coming into DIY tuning to read over, learn from, and be aware of. Thanks for the contribution.

BTW, I edited my original post, quoting the specific passage that I was referring to. I made some other cleanup mods as well...nothing major.
...Always Somethin'

89 Ranger Supercab, 331 w/GT40p heads, ported Explorer lower, Crane Powermax 2020 cam, 1.6RRs, FMS Explorer (GT40p) headers, Slot Style MAF, aftermarket T5 'Z-Spec', 8.8" rear w/3.27s, Powertrax Locker, Innovate LC-1, GUFB, Moates QuarterHorse tuned using BE&EA

Member V8-Ranger.com

Locked

Who is online

Users browsing this forum: No registered users and 3 guests