[Z06] PCM FT Adaption and Torque Management
Thread Starter
Racer
Joined: Mar 2002
Posts: 355
Likes: 0
From: Germantown WI
PCM FT Adaption and Torque Management
My apologies for not being clear in the "OBD2 Adaption" post earlier. I meant to note that I was reviewing the BMW system and trying to draw some parallels to the GM PCM system.. oops... I am trying to weed out the marketing hype from the realities of performing engine performance mods. On the BMW M coupe they were pretty useless and a waste of money, for some of the reasons I noted. I want to determine if the GM PCM system is different in a way that would make upgrades truly "UP" grades and not simply drafts on my bank statements. Watch any infomercial for the "Tornado" claiming a 10hp gain in a 200hp car, and one has to be suspicious of any claims of performance increases, wihtout understanding how the data is being generated and how it can be easily scewed and exploited by marketeers.
After a thorough goings over of the Service manual and a couple of discussions with a freind at a Dyno test/tune shop, I've come to the conclusion that the GM PCM system is MUCH simpler, and can be easily re-set with a power programmer or during software downloads.
It appears the only thing the GM system uses as "learned" data is the Fuel Trim (FT) setting. From what have gathered so far, it averages long term demand with short term demand to establish a trim factor based on the various inputs it recieves. The long term trim is stored in flash memory each time the car is shut down, and is used on start up for the next run cycle. The short term trim is not stored, thus the engine starts up on the settings stored in the long term trim memory. Because of this, the long term memory becomes a sort of extended mean moving average of demand made on the car. In this way it is adaptive to a given drivers historic demand.
Since the system's goal is to continually produce the lowest emmissions and fuel consumption, and maintain proper exhaust oxygen and heat levels (to support catalytic functions), by adjusting fuel delivery, it makes sense that the only thing it needs to store is the FT data. BMW takes all this further to cam setting, ignition settings, fuel, etc... but that may just be due to more sphincter restrictive German engineering methods than to anything else in reality as we know it.
It also appears the computer does store some other data, such as number of occurances that redline rpm is exceeded (and a few associated tidbits at that instant), and fault code occurances. Further, the system has a residual memory of something on the order of 5 seconds, of all activity, that can be harvested during an accident or warranty claim investigation.
Based on this, any back-to-back dyno tests of mods cannot be accurately evaluated without clearing the long term FT data from the system before the first test and between each test. Otherwise all that is being tested is how the mod creates a short term shift in power utilizing the original equipment's adapted fuel trim settings. As the car is driven, these changes will change the long term FT data, thus may substantially reduce or increase the impact of the mod.
Further, to get a really accurate picture of any gains made, the followign test procedure might be in order:
1. Clear all long term FT data
2. Drive the car a given number of laps on a closed course at a perscribed rate of speed, with acceleration and deceleration zones... say for 50 miles.
3. Test baseline or original configuration, using all pertinant atmospheric correction factors.
4. Clear long term FT data
5. Intall the mod
6. Repeat the closed course driving attempting to duplicate speeds and conditions used for original equipment.
7. Test new system using all appropriate correction factors.
Even with this, the fact is, that any mod that changes driving behavior is going to be factored into the long term FT data. However, it does appear that the simpler GM PCM system is going to be less likely to reduce the impact of a mod that increases airflow or decreases in sensed intake air temperatures, than the BMW OBD2 system is.
Does anyone have any information that indicates the system is using learned (adaption) data anywhere else in controlling engine performance?
Finally - I also noted that the "Torque Management" system appears to be focused on auto trans C5's. The only case where I note that it might have any reason to be activated in the M12 trans cars is the "under severe activity" component. In other words, reving the engine to 6,000rpm and side stepping the clutch, power shifting under full throttle acceleration, or rocking from first to reverse and back again to get unstuck, etc. It does not appear that under even the most hard run on a road race track, that the system would ever be activated, or if it is, noticed in any way. The setting of 349 ft. lbs does not limit the car to 349 ft. lbs (from 400), all the time, just when it is determined, through various sensors, that the car is in danger of being damaged. Does this sound accurate?
[Modified by kwillmorth, 3:56 PM 4/25/2002]
[Modified by kwillmorth, 3:58 PM 4/25/2002]
After a thorough goings over of the Service manual and a couple of discussions with a freind at a Dyno test/tune shop, I've come to the conclusion that the GM PCM system is MUCH simpler, and can be easily re-set with a power programmer or during software downloads.
It appears the only thing the GM system uses as "learned" data is the Fuel Trim (FT) setting. From what have gathered so far, it averages long term demand with short term demand to establish a trim factor based on the various inputs it recieves. The long term trim is stored in flash memory each time the car is shut down, and is used on start up for the next run cycle. The short term trim is not stored, thus the engine starts up on the settings stored in the long term trim memory. Because of this, the long term memory becomes a sort of extended mean moving average of demand made on the car. In this way it is adaptive to a given drivers historic demand.
Since the system's goal is to continually produce the lowest emmissions and fuel consumption, and maintain proper exhaust oxygen and heat levels (to support catalytic functions), by adjusting fuel delivery, it makes sense that the only thing it needs to store is the FT data. BMW takes all this further to cam setting, ignition settings, fuel, etc... but that may just be due to more sphincter restrictive German engineering methods than to anything else in reality as we know it.
It also appears the computer does store some other data, such as number of occurances that redline rpm is exceeded (and a few associated tidbits at that instant), and fault code occurances. Further, the system has a residual memory of something on the order of 5 seconds, of all activity, that can be harvested during an accident or warranty claim investigation.
Based on this, any back-to-back dyno tests of mods cannot be accurately evaluated without clearing the long term FT data from the system before the first test and between each test. Otherwise all that is being tested is how the mod creates a short term shift in power utilizing the original equipment's adapted fuel trim settings. As the car is driven, these changes will change the long term FT data, thus may substantially reduce or increase the impact of the mod.
Further, to get a really accurate picture of any gains made, the followign test procedure might be in order:
1. Clear all long term FT data
2. Drive the car a given number of laps on a closed course at a perscribed rate of speed, with acceleration and deceleration zones... say for 50 miles.
3. Test baseline or original configuration, using all pertinant atmospheric correction factors.
4. Clear long term FT data
5. Intall the mod
6. Repeat the closed course driving attempting to duplicate speeds and conditions used for original equipment.
7. Test new system using all appropriate correction factors.
Even with this, the fact is, that any mod that changes driving behavior is going to be factored into the long term FT data. However, it does appear that the simpler GM PCM system is going to be less likely to reduce the impact of a mod that increases airflow or decreases in sensed intake air temperatures, than the BMW OBD2 system is.
Does anyone have any information that indicates the system is using learned (adaption) data anywhere else in controlling engine performance?
Finally - I also noted that the "Torque Management" system appears to be focused on auto trans C5's. The only case where I note that it might have any reason to be activated in the M12 trans cars is the "under severe activity" component. In other words, reving the engine to 6,000rpm and side stepping the clutch, power shifting under full throttle acceleration, or rocking from first to reverse and back again to get unstuck, etc. It does not appear that under even the most hard run on a road race track, that the system would ever be activated, or if it is, noticed in any way. The setting of 349 ft. lbs does not limit the car to 349 ft. lbs (from 400), all the time, just when it is determined, through various sensors, that the car is in danger of being damaged. Does this sound accurate?
[Modified by kwillmorth, 3:56 PM 4/25/2002]
[Modified by kwillmorth, 3:58 PM 4/25/2002]
Le Mans Master
Joined: Jun 2001
Posts: 7,331
Likes: 518
From: Admit Nothing, Deny Everything, Make Counter Accusations.
Re: PCM FT Adaption and Torque Management (kwillmorth)
You definitely need AutoTap to see what is going on. Read some of these pages:
http://www.vetteguru.com/mods/howto/
http://www.zz0066.com/forums/showthr...threadid=14135
Change the zz0066 to z06 in the above URL.
This information should be helpful. :flag
[Modified by rbartick, 11:29 AM 4/25/2002]
http://www.vetteguru.com/mods/howto/
http://www.zz0066.com/forums/showthr...threadid=14135
Change the zz0066 to z06 in the above URL.
This information should be helpful. :flag
[Modified by rbartick, 11:29 AM 4/25/2002]
Safety Car
Joined: Oct 2000
Posts: 4,080
Likes: 565
From: Allen TX
Re: PCM FT Adaption and Torque Management (kwillmorth)
Kevin,
RE: LTFTs...this is why its very important to get them very close or at ZERO. If they are reading 10% lean or rich, for example (can be up to ~25% each way before a code is thrown) then making a mod and then testing it right away will NOT show accurate "long term" data.
But if you make a mod, then do scans (via EASE or Autotap software) and determine that the LTFT is close to zero (or manipulate it so that it will be ... via PCM tuning or something as simple as a MAF Translator) then you can test the mod to see how it is working.
Having the LTFT near zero will also allow you to do WOT tuning and not have any adaptive function come into play.
With regards to TQ management: I've heard in some cases it just reduces the timing dramatically if it kicks in. I will find out when the nitrous is on, I guess. ;)
RE: LTFTs...this is why its very important to get them very close or at ZERO. If they are reading 10% lean or rich, for example (can be up to ~25% each way before a code is thrown) then making a mod and then testing it right away will NOT show accurate "long term" data.
But if you make a mod, then do scans (via EASE or Autotap software) and determine that the LTFT is close to zero (or manipulate it so that it will be ... via PCM tuning or something as simple as a MAF Translator) then you can test the mod to see how it is working.
Having the LTFT near zero will also allow you to do WOT tuning and not have any adaptive function come into play.
With regards to TQ management: I've heard in some cases it just reduces the timing dramatically if it kicks in. I will find out when the nitrous is on, I guess. ;)
