92silver964

I have looked and cannot find much on this subject. What is the feeling/facts/experience on the extent to which the stock ECU “learns” on its own after you make a relevant change to the engine? I made a significant change this spring to my air thru-put with a Fabspeed “cup” airbox, Fabspeed 100 cell cat, and Maxflo dual muffler, but my immediate dyno results were very disappointing when compared with the previous configuration (primary bypass – “cup pipe”). On the exact same dyno setup, I lost about 10% on my NA 3.6L due to “improving” the engine’s ability to take in air and get rid of it! Some told me that the ECU just needed time to adjust to the changes. So I ran it for a month but didn’t feel much difference. I didn’t feel like paying for another dyno run if I couldn’t feel any improvement, which is why I didn’t retest. I switched over to a more restrictive exhaust upgrade (does that make sense?) where I went back to the stock cat and secondary, with a FVD Dansk dual outlet (actually a modified single outlet, it seems to me). The effect was that the car got back pretty close to my starting configuration in terms of power, as tested on the same dyno again. I talked to Steve Wong during this time and he seemed adamant that “learning” by the ECU was a myth, only saying that there might be a brief settling period for the idle. BTW, the experience seemed to make a case for the primary bypass setup as a pretty good bang for the buck. But, I wanted more sound and duals, so I kept at it.

I have had the second configuration in place for almost two months and it does seem to have picked up a bit since installation. By the way, I really like the sound of the Dansk setup. The Fabspeed cat and Maxflo were too loud in the wrong way for me, especially at idle, and added resonance at 3-4K and during downshifting (I know Fabspeed says there isn’t any). The Dansk is relatively quiet at idle, gives me the sound volume/note I was looking for while pushing the car hard (and without any resonance - I think the secondary takes it out), and allows the free-flowing airbox to now contribute its tone (kind of a sucking roar – remember when you took the air filter our of your mom’s car?)

I am planning to re-dyno the car at the end of August since I want to know where I really am, and because I am thinking of trying the SCART 200 cell cat to see if I can squeeze out a few more hp (and dB) without upsetting the backpressure balance. Then, once I am hopefully where I want with hardware, I was planning to get a Wong chip in hopes of dragging some more power and low end performance out of the motor.

Sorry, I blabbed away so much once I got started, but it has been a frustrating experience so I guess I’m venting. At least Fabspeed took back the cat and Maxflo for a full refund. So, if anyone has any words of wisdom regarding ECU “learning” in response to airbox / exhaust changes I would appreciate it. Thanks for helping out!!!

I also attached a pic of the Dansk tips – they match the bumper contour better than the Fabspeed tips did, but what looks good to you is a matter of opinion.

springer3

I think Steve knows what he is talking about. Idle adaptation centers the idle control at a happy place that smooths the idle and minimizes stalling, but I have never heard or read anything about other "learning" in the 964 DME.

Don't put too much stock in small changes on a dyno. The machines are large and rugged, but not terribly accurate or repeatable due to a large number of variables. Tire design and tire pressure will affect what you get, for example. Changes in the weather affect engine power more than changes in exhaust configuration.

Engine power depends on the mass of air you can pump through. At any given speed, a piston engine moves a constant air volume plus or minus some variation for dynamic effects. Cold air has more mass per unit volume than hot air. Dry air has higher percent oxygen than humid air, since the water vapor occupies some of the volume. Therefore, do the "before" dyno on a hot humid day, and the "after" dyno on a cold dry day if you want to see maximum HP gains from your mod.

If you want valid results, bring a second car that is not modified to each dyno run. Control tire pressure and other variables as tightly as practical, and use the change observed in the "mule" to correct your dyno results for weather change and other variables including drift in the dyno calibration.

jimq

My TIP computer learns

elbeee964

My (car's) CPU is a neural net processor; a learning computer...

Steve Weiner-Rennsport Systems

Hi,

Steve is right,...these ECU's don't "learn" or adapt in the same way that later (OBD-2) cars do such as the water-cooled ones, aside from some idle adaptation.

I will tell you that these cars are sensitive to exhaust tuning and many changes can result in LESS performance than stock.

Lastly, dyno testing can be a bit fickle, depending on the dyno, operator, whether the facility has sufficient cooling for these cars and of course, atmospherics. Thats one reason why we prefer our engine dyno for evaluating and testing bolt-on components.

boxsey911

This confirms what I was thinking myself. In the UK our standard petrol is 95 RON and the super is 97+ RON. The owners manual for my boxster states that the ignition will adapt to the different fuel put in (but advises that 98 RON is the preferred octane rating). Sure enough I can feel the difference if I switch between grades. The 964 manual has no such information as 95 RON was the only widely available grade when the car was introduced in the UK. Personally I can't feel any difference in performance when I swap between fuel grades in the 964. Nb. Mine has a standard DME chip.

Tom W

Your fuel issue isn't learning, it's the knock sensors changing the timing (and providing less power).

I posted my dyno results a few years ago when the Turtle was mostly stock, they should still be found with a search. When I 'upgraded' to a Fabspeed dual outlet exhaust and chip programmed for it, I lost 10 hp.

Lorenfb

"The owners manual for my boxster states that the ignition will adapt to the different fuel put in (but advises that 98 RON is the preferred octane rating). Sure enough I can feel the difference if I switch between grades."

Absolutely correct! So in affect, the closed-loop engine management system
has "learned" to compensate for the lesser quality/octane fuel. The system
also corrects ("learns") when the mixture is too rich, e.g. fuel pressure, AFM
setting, via the O2 sensor. This aspect is further augmented via the OBDII
TRA/FRA of the '96 and later Porsche engine management systems.
The "learning" process always reverts to its default state once the error,
e.g. excessive knocks, no longer occurs or the system is reset, i.e. key off.

"When I 'upgraded' to a Fabspeed dual outlet exhaust and chip programmed for it,
I lost 10 hp."

So, when one does an engine mod, e.g. "performance" chip, the engine
management systems corrects ("learns") for either the excessive "pushed"
timing or an excessive rich mixture, with the 993 and later Porsches having
greater correction ("learning") capability than the 964 for mixture correction
("learning").

92silver964

Thanks for all the helpful responses. Definitely cleared up the ECU learning myth for a 964. I continue to be amazed at how much you seasoned P-nuts know. "I am not worthy!" Cheers.

Cupcar#12

on the gain/loss subject - a few years ago i went through this and found results simlar to Tom's
full exhaust flow (off the shelf "cup" chip/weltmeister/others) a B7B header single outlet system netted 2hp with a stock airbox or the cup.

best bang is the either the Primary Delete with the cup pipe or the Primary with a secondary delete & Cat delete (fabspeed) is my best config welt chip with anti-stall fix + cup air box, yeilded by best whp of 274 with the critical mid range being strong, admitially i also have a Cartridge oil pressure relieve valve which is tuned to increases postive pressure in the case of the 964 and is good for a few HP (do a search if your curious)
my engine is an "RS" # and was built as such, as well so you results will vary.

i am running stock headers as well (no delta from delivered)
the 964 engine likes some back pressure to produce Hp - eliminate it and you see a loss normally, chip tuning can help within limits though. you have to change the ECU to go beyond its desinged limits.

no learning as has been said (use to think that as well) idle and ignition timing is all (it will pull it from the programed baseline is all).

use a single Dyno to try an eliminate variables with the same conditions (temp, humidity, etc)

(i'm a test manager for a software Co so that stuff is "built in" for me )

G-luck its a trial sometimes. but as long as your expecations are in line it can be done, however the huge gains are normally vaporware.

Lorenfb

"Definitely cleared up the ECU learning myth for a 964."

Really? It's hardly a so called myth as was explained in detail in two previous
posts. Obvious a number of posters lack a full understanding of the 964
(and later Porsche systems) engine management system and in particular
as explained in the 964 manual WKD 495 121. The later Porsche engine management
systems, as mentioned also previously, just "learn" to a greater extent.

As usual the hyperbole begins to "pile on" when the discussion includes the
topic of HP gains over stock and the so-called "tuners" usual exhibit of their
"learned it on the internet" knowledge, i.e. lack of it.

Bottom line: Again, when one mods one's engine with so-called "performance"
enhancements, generally expect the 964 and later engine management systems
to correct/compensate/"learn" a new optimum running mode. This is usually
why most, as exemplified by the thread starter, experence poor subsequent
performance results. The real myth part is the claimed performance improvement
of most/all mods which usually result in less performance.

Cupcar#12

"learning" can be many things for an ECU - the 964 DME is not that smart though and operates on a relitively narrow set of rules.

Loren's right it does adapt or "learn" but it cannot do any more than its realitively limited boundrys

you can free the engine it just requires a huge investment in tuning and a different ECU.

i liken it to trying to expect an IBM XT to run Office 2007 on windows XP. (just an anology...)