SharkTuner Mk 2
07-26-2010, 08:14 PM
#286
Nordschleife Master
Here's another question about LH, perhaps therefore appropriate for the ST2 thread.
What are the triggers that cause the LH to move from closed-loop state to open-loop state?
So far, in my ST2 data logs, I have only found one event that causes a transition from closed-loop to open-loop: The WOT switch. The WOT switch closing doesn't have an immediate effect, however. Instead, LH steps the O2 adjustment towards zero at rate of 3 adjustment units per second. A similar transition happens when the WOT switch opens again.
Is there a field in ST2 that controls the rate at which LH transitions from closed-loop to open loop? Two seconds to get down from 6 units of adjustment is a really long time when I floor it the on second gear.
Does anyone have any evidence of a transition from closed-loop to open-loop mode that is not triggered by the WOT switch closing?
[EDIT 1: I am seeing the some cases at startup where LH is open loop without WOT. That makes sense, since the O2 sensor needs to be heated.]
[EDIT 2: I am also seeing some reasonably high MAF voltage and reasonably high RPM cases but without WOT that are also in open loop. If select records with WOT switch open (part throttle) and regress my estimate of open loop condition on RPM and MAF signal, I get a statistically reliable positive coefficient on both. However, the R2 is low, about 1%.]
[EDIT 3: I defined open loop condition as a variable that takes value 1 if the O2 adjustment is exactly zero now and also was exactly zero in the previous data point from the same run. I then regressed this variable on some variables. High RPMs, high MAF signal values, and WOT predict open loop. The R2 is about 16%. The linear model doesn't fit very well, but this is some indication of what to look for.
Variable Coefficients t Stat
Intercept -0.1143 -13.94
RPM 0.0000 9.58
MAF signal 0.0012 17.17
MAF signal [%] -0.0059 -6.41
1 if WOT, else o 0.5597 49.94
]
What are the triggers that cause the LH to move from closed-loop state to open-loop state?
So far, in my ST2 data logs, I have only found one event that causes a transition from closed-loop to open-loop: The WOT switch. The WOT switch closing doesn't have an immediate effect, however. Instead, LH steps the O2 adjustment towards zero at rate of 3 adjustment units per second. A similar transition happens when the WOT switch opens again.
Is there a field in ST2 that controls the rate at which LH transitions from closed-loop to open loop? Two seconds to get down from 6 units of adjustment is a really long time when I floor it the on second gear.
Does anyone have any evidence of a transition from closed-loop to open-loop mode that is not triggered by the WOT switch closing?
[EDIT 1: I am seeing the some cases at startup where LH is open loop without WOT. That makes sense, since the O2 sensor needs to be heated.]
[EDIT 2: I am also seeing some reasonably high MAF voltage and reasonably high RPM cases but without WOT that are also in open loop. If select records with WOT switch open (part throttle) and regress my estimate of open loop condition on RPM and MAF signal, I get a statistically reliable positive coefficient on both. However, the R2 is low, about 1%.]
[EDIT 3: I defined open loop condition as a variable that takes value 1 if the O2 adjustment is exactly zero now and also was exactly zero in the previous data point from the same run. I then regressed this variable on some variables. High RPMs, high MAF signal values, and WOT predict open loop. The R2 is about 16%. The linear model doesn't fit very well, but this is some indication of what to look for.
Variable Coefficients t Stat
Intercept -0.1143 -13.94
RPM 0.0000 9.58
MAF signal 0.0012 17.17
MAF signal [%] -0.0059 -6.41
1 if WOT, else o 0.5597 49.94
]
Last edited by ptuomov; 07-26-2010 at 09:51 PM.
07-27-2010, 07:57 PM
#287
Nordschleife Master
Here are some more LH and EZ-K questions that I can't find answers to anywhere. I bought two books on Bosch fuel injection, and they were total vaporware.
(1) Is there a spec document to LH and EZ-K that describes how they work given the parameters?
(2) How do the ECUs come up with values that are between cell mid points? Do they just step discretely to the next cell? Or do they use a three-point linear interpolation? Or do they use four-point quadratic (aka bilinear) interpolation?
(3) ST2 logs show EZ-K column "Ign adv [°]". Does this value reflect any possible knock retard?
(4) With SuperMAF, what's the relation between LH "MAF signal" and EZ-K "Load [%]"? Why does it seem that the EZ-K map has half the range in the top 100% load cell compared to the LH map?
(5) Jim Morton sent me the Bosch spec documents for the stock 928 S4 MAF. (Thanks Jim!) Jim's spec document gives the mapping between MAF voltage and the mass air flow rate. Does a similar spec document or information exist for the super MAF?
(6) Does anyone sell a device for 928 that would have an adjustable Hobbs switch connected in parallel to the WOT switch, always tricking the ECU off the adaptation mode when boost exceeds some threshold?
(7) Here's a question to the supercharger crowd: Do you ever see knocks in the LH "adaptation mode"? One way to identify this is to look at your data logs and see if there are any records with the O2 adjustment not equal to zero and a knock event recorded.
More to come...
(1) Is there a spec document to LH and EZ-K that describes how they work given the parameters?
(2) How do the ECUs come up with values that are between cell mid points? Do they just step discretely to the next cell? Or do they use a three-point linear interpolation? Or do they use four-point quadratic (aka bilinear) interpolation?
(3) ST2 logs show EZ-K column "Ign adv [°]". Does this value reflect any possible knock retard?
(4) With SuperMAF, what's the relation between LH "MAF signal" and EZ-K "Load [%]"? Why does it seem that the EZ-K map has half the range in the top 100% load cell compared to the LH map?
(5) Jim Morton sent me the Bosch spec documents for the stock 928 S4 MAF. (Thanks Jim!) Jim's spec document gives the mapping between MAF voltage and the mass air flow rate. Does a similar spec document or information exist for the super MAF?
(6) Does anyone sell a device for 928 that would have an adjustable Hobbs switch connected in parallel to the WOT switch, always tricking the ECU off the adaptation mode when boost exceeds some threshold?
(7) Here's a question to the supercharger crowd: Do you ever see knocks in the LH "adaptation mode"? One way to identify this is to look at your data logs and see if there are any records with the O2 adjustment not equal to zero and a knock event recorded.
More to come...
Last edited by ptuomov; 07-27-2010 at 11:54 PM.
07-28-2010, 01:56 AM
#288
Supercharged
Rennlist Member
Rennlist Member
Join Date: May 2002
Location: Back in Michigan - Full time!
Posts: 18,925
Likes: 0
Received 61 Likes
on
35 Posts
Originally Posted by ptuomov
So far, in my ST2 data logs, I have only found one event that causes a transition from closed-loop to open-loop: The WOT switch. The WOT switch closing doesn't have an immediate effect, however. Instead, LH steps the O2 adjustment towards zero at rate of 3 adjustment units per second. A similar transition happens when the WOT switch opens again.
Is there a field in ST2 that controls the rate at which LH transitions from closed-loop to open loop? Two seconds to get down from 6 units of adjustment is a really long time when I floor it the on second gear.
Is there a field in ST2 that controls the rate at which LH transitions from closed-loop to open loop? Two seconds to get down from 6 units of adjustment is a really long time when I floor it the on second gear.
Originally Posted by ptuomov
Does anyone have any evidence of a transition from closed-loop to open-loop mode that is not triggered by the WOT switch closing?
Originally Posted by ptuomov
[EDIT 1: I am seeing the some cases at startup where LH is open loop without WOT. That makes sense, since the O2 sensor needs to be heated.]
Originally Posted by ptuomov
[EDIT 2: I am also seeing some reasonably high MAF voltage and reasonably high RPM cases but without WOT that are also in open loop. If select records with WOT switch open (part throttle) and regress my estimate of open loop condition on RPM and MAF signal, I get a statistically reliable positive coefficient on both. However, the R2 is low, about 1%.]
Originally Posted by ptuomov
[EDIT 3: I defined open loop condition as a variable that takes value 1 if the O2 adjustment is exactly zero now and also was exactly zero in the previous data point from the same run. I then regressed this variable on some variables. High RPMs, high MAF signal values, and WOT predict open loop. The R2 is about 16%. The linear model doesn't fit very well, but this is some indication of what to look for.
Variable Coefficients t Stat
Intercept -0.1143 -13.94
RPM 0.0000 9.58
MAF signal 0.0012 17.17
MAF signal [%] -0.0059 -6.41
1 if WOT, else o 0.5597 49.94
]
Variable Coefficients t Stat
Intercept -0.1143 -13.94
RPM 0.0000 9.58
MAF signal 0.0012 17.17
MAF signal [%] -0.0059 -6.41
1 if WOT, else o 0.5597 49.94
]
Originally Posted by ptuomov
(2) How do the ECUs come up with values that are between cell mid points? Do they just step discretely to the next cell? Or do they use a three-point linear interpolation? Or do they use four-point quadratic (aka bilinear) interpolation?
Originally Posted by ptuomov
(3) ST2 logs show EZ-K column "Ign adv [°]". Does this value reflect any possible knock retard?
Originally Posted by ptuomov
(4) With SuperMAF, what's the relation between LH "MAF signal" and EZ-K "Load [%]"? Why does it seem that the EZ-K map has half the range in the top 100% load cell compared to the LH map?
Originally Posted by ptuomov
(5) Jim Morton sent me the Bosch spec documents for the stock 928 S4 MAF. (Thanks Jim!) Jim's spec document gives the mapping between MAF voltage and the mass air flow rate. Does a similar spec document or information exist for the super MAF?
Originally Posted by ptuomov
(6) Does anyone sell a device for 928 that would have an adjustable Hobbs switch connected in parallel to the WOT switch, always tricking the ECU off the adaptation mode when boost exceeds some threshold?
Originally Posted by ptuomov
(7) Here's a question to the supercharger crowd: Do you ever see knocks in the LH "adaptation mode"? One way to identify this is to look at your data logs and see if there are any records with the O2 adjustment not equal to zero and a knock event recorded.
Originally Posted by ptuomov
More to come...
j/k
07-28-2010, 04:59 AM
#289
Nordschleife Master
Here's another question about LH, perhaps therefore appropriate for the ST2 thread.
What are the triggers that cause the LH to move from closed-loop state to open-loop state?
So far, in my ST2 data logs, I have only found one event that causes a transition from closed-loop to open-loop: The WOT switch. The WOT switch closing doesn't have an immediate effect, however. Instead, LH steps the O2 adjustment towards zero at rate of 3 adjustment units per second. A similar transition happens when the WOT switch opens again.
What are the triggers that cause the LH to move from closed-loop state to open-loop state?
So far, in my ST2 data logs, I have only found one event that causes a transition from closed-loop to open-loop: The WOT switch. The WOT switch closing doesn't have an immediate effect, however. Instead, LH steps the O2 adjustment towards zero at rate of 3 adjustment units per second. A similar transition happens when the WOT switch opens again.
I'm not sure whether you can change the coding plug dynamically - but assuming its possible, you may be able to rig something to ground the relevant pin above certain boost levels, using your boost controller?
Of course, unless you're super worried about the emissions, it's easier to fit a non-cat coding plug, the O2 signal ground terminator and the calibration potentiometer, and run open-loop the whole time so you don't have to worry about the transition.
07-28-2010, 06:27 AM
#290
Rennlist Member
Thread Starter
See below...
Here are some more LH and EZ-K questions that I can't find answers to anywhere. I bought two books on Bosch fuel injection, and they were total vaporware.
(1) Is there a spec document to LH and EZ-K that describes how they work given the parameters?
You must be joking ! Bosch never give away any information.
(2) How do the ECUs come up with values that are between cell mid points? Do they just step discretely to the next cell? Or do they use a three-point linear interpolation? Or do they use four-point quadratic (aka bilinear) interpolation?
3 point
(3) ST2 logs show EZ-K column "Ign adv [°]". Does this value reflect any possible knock retard?
Yes
(4) With SuperMAF, what's the relation between LH "MAF signal" and EZ-K "Load [%]"? Why does it seem that the EZ-K map has half the range in the top 100% load cell compared to the LH map?
Too complicated to explain here.
(5) Jim Morton sent me the Bosch spec documents for the stock 928 S4 MAF. (Thanks Jim!) Jim's spec document gives the mapping between MAF voltage and the mass air flow rate. Does a similar spec document or information exist for the super MAF?
Yes, we have constructed such a graph.
(6) Does anyone sell a device for 928 that would have an adjustable Hobbs switch connected in parallel to the WOT switch, always tricking the ECU off the adaptation mode when boost exceeds some threshold?
The LH switches to open loop every time you hit the throttle, otherwise the O2 loop would prevent the transient enrichment necessary to stop the engine hesitating. The LH also switches to open loop at quite moderate steady loads and rpm. The algorthm is complicated and changes the time the LH goes open loop depending on sequences of event.
(7) Here's a question to the supercharger crowd: Do you ever see knocks in the LH "adaptation mode"? One way to identify this is to look at your data logs and see if there are any records with the O2 adjustment not equal to zero and a knock event recorded.
???
More to come...
(1) Is there a spec document to LH and EZ-K that describes how they work given the parameters?
You must be joking ! Bosch never give away any information.
(2) How do the ECUs come up with values that are between cell mid points? Do they just step discretely to the next cell? Or do they use a three-point linear interpolation? Or do they use four-point quadratic (aka bilinear) interpolation?
3 point
(3) ST2 logs show EZ-K column "Ign adv [°]". Does this value reflect any possible knock retard?
Yes
(4) With SuperMAF, what's the relation between LH "MAF signal" and EZ-K "Load [%]"? Why does it seem that the EZ-K map has half the range in the top 100% load cell compared to the LH map?
Too complicated to explain here.
(5) Jim Morton sent me the Bosch spec documents for the stock 928 S4 MAF. (Thanks Jim!) Jim's spec document gives the mapping between MAF voltage and the mass air flow rate. Does a similar spec document or information exist for the super MAF?
Yes, we have constructed such a graph.
(6) Does anyone sell a device for 928 that would have an adjustable Hobbs switch connected in parallel to the WOT switch, always tricking the ECU off the adaptation mode when boost exceeds some threshold?
The LH switches to open loop every time you hit the throttle, otherwise the O2 loop would prevent the transient enrichment necessary to stop the engine hesitating. The LH also switches to open loop at quite moderate steady loads and rpm. The algorthm is complicated and changes the time the LH goes open loop depending on sequences of event.
(7) Here's a question to the supercharger crowd: Do you ever see knocks in the LH "adaptation mode"? One way to identify this is to look at your data logs and see if there are any records with the O2 adjustment not equal to zero and a knock event recorded.
???
More to come...
07-28-2010, 09:12 AM
#291
Nordschleife Master
Originally Posted by Andrew Olson
Oh joy!
What I want is a base map significantly rich in high load regions and slightly rich in medium and low load regions. Then, when the car is in closed-loop mode, I want the O2 adaptation to lean it to good fuel economy (and emissions). That should work, as long as I can be confident that the car doesn't go into 02 adaptation at high load regions and cause knock.
Originally Posted by Andrew Olson
Yes, I believe it is the actual adv with knock retard included. Speake can confirm.
07-28-2010, 10:00 AM
#292
Rennlist Member
Thread Starter
Adaptation is a relatively slow correction process, working only in closed loop. It serves two purposes
1) To keep the O2 loop in the middle of its range by "Adapting" the injector pulse widths.
2) To keep the ISV loop with sufficient margin to be able to keep the loop in control under all circumstances. This is also done by altering the Tinj.
So if you are at WOT or just high loads, then adaptation isn't suddenly going to do anything to weaken the mixture and cause detonation.
When you're less that WOT and in cat mode you can't change the steady state A/F to anything but stoich. To be able to graduate the A/F with load as you want, you have to go to open loop, as Hilton has said.
As you have a ST2 you don't need adaptation as it is concerned with long term slow changes i.e. MAF ageing. You can tune out any such changes. So you might be better going to open loop operation.
I will check again on the displayed igntition advance, I clearly was incorrect in my answer.
1) To keep the O2 loop in the middle of its range by "Adapting" the injector pulse widths.
2) To keep the ISV loop with sufficient margin to be able to keep the loop in control under all circumstances. This is also done by altering the Tinj.
So if you are at WOT or just high loads, then adaptation isn't suddenly going to do anything to weaken the mixture and cause detonation.
When you're less that WOT and in cat mode you can't change the steady state A/F to anything but stoich. To be able to graduate the A/F with load as you want, you have to go to open loop, as Hilton has said.
As you have a ST2 you don't need adaptation as it is concerned with long term slow changes i.e. MAF ageing. You can tune out any such changes. So you might be better going to open loop operation.
I will check again on the displayed igntition advance, I clearly was incorrect in my answer.
07-28-2010, 10:00 AM
#293
Supercharged
Rennlist Member
Rennlist Member
Join Date: May 2002
Location: Back in Michigan - Full time!
Posts: 18,925
Likes: 0
Received 61 Likes
on
35 Posts
I'd like to keep as many of the modern features enabled as possible. i think the 02 adaptation for low load states is clever idea. I think it's one of the reasons why the mileage on my car is still over 20 mpg on highway.
What I want is a base map significantly rich in high load regions and slightly rich in medium and low load regions. Then, when the car is in closed-loop mode, I want the O2 adaptation to lean it to good fuel economy (and emissions). That should work, as long as I can be confident that the car doesn't go into 02 adaptation at high load regions and cause knock.
What I want is a base map significantly rich in high load regions and slightly rich in medium and low load regions. Then, when the car is in closed-loop mode, I want the O2 adaptation to lean it to good fuel economy (and emissions). That should work, as long as I can be confident that the car doesn't go into 02 adaptation at high load regions and cause knock.
07-28-2010, 10:58 AM
#294
Race Car
Wait a minute... it seems the phrase "open loop" is being misused here. There are three modes of operation:
1) Closed-loop with adaptation
2) Closed-loop without adaptation - O2 sensor compensation still happens
3) Open-loop - no O2 sensor compensation at all
Lots of people using the ST/ST2 run on 2) because they find (including me) that adaptation gets in the way of tuning. But I am still running closed loop mode, just without adaptation.
Dan
'91 928GT S/C
475hp/460lb.ft
1) Closed-loop with adaptation
2) Closed-loop without adaptation - O2 sensor compensation still happens
3) Open-loop - no O2 sensor compensation at all
Lots of people using the ST/ST2 run on 2) because they find (including me) that adaptation gets in the way of tuning. But I am still running closed loop mode, just without adaptation.
Dan
'91 928GT S/C
475hp/460lb.ft
07-28-2010, 12:17 PM
#296
Race Car
07-28-2010, 12:40 PM
#297
Nordschleife Master
Wait a minute... it seems the phrase "open loop" is being misused here. There are three modes of operation:
1) Closed-loop with adaptation
2) Closed-loop without adaptation - O2 sensor compensation still happens
3) Open-loop - no O2 sensor compensation at all
Lots of people using the ST/ST2 run on 2) because they find (including me) that adaptation gets in the way of tuning. But I am still running closed loop mode, just without adaptation.
1) Closed-loop with adaptation
2) Closed-loop without adaptation - O2 sensor compensation still happens
3) Open-loop - no O2 sensor compensation at all
Lots of people using the ST/ST2 run on 2) because they find (including me) that adaptation gets in the way of tuning. But I am still running closed loop mode, just without adaptation.
[EDITED: based Jim's below comment]
["Closed loop without adaptation" pulse width = tables + fast moving adjustment towards AFR 14.7
"Closed loop with adaptation" pulse width = tables + fast-moving adjustment towards AFR 14.7 + a slow moving average of past fast-moving adjustments in this cell
"Open loop without adaptation" pulse width = tables
"Open loop with adaptation" pulse width = tables + a slow moving average of past closed-loop fast-moving adjustments in this cell]
I understand that when tuning and altering cells, you probably want to run it in open loop. However, if you choose to run the car in closed loop mode after the tuning is done, wouldn't you always want to also allow adaptation? If I've understood correctly, adaptation just makes the fast-moving adjustments centered around the value that's worked in the past, thus making the closed loop operation less erratic. Or does enabling adaptation do something else as well? [EDIT: Jim says that adaptation stays even if the car is in open loop]
Last edited by ptuomov; 07-28-2010 at 02:10 PM.
07-28-2010, 01:29 PM
#298
Supercharged
Rennlist Member
Rennlist Member
Join Date: May 2002
Location: Back in Michigan - Full time!
Posts: 18,925
Likes: 0
Received 61 Likes
on
35 Posts
#2. Your statement is only true if your run excessively rich all the time. Open-loop itself will not cause the cats to fail.
07-28-2010, 01:40 PM
#299
Supercharged
Rennlist Member
Rennlist Member
Join Date: May 2002
Location: Back in Michigan - Full time!
Posts: 18,925
Likes: 0
Received 61 Likes
on
35 Posts
If I've understood correctly, adaptation just makes the fast-moving adjustments centered around the value that's worked in the past, thus making the closed loop operation less erratic. Or does enabling adaptation do something else as well?
O2 Adaptation - Slow moving agerage correction - not sure if this is cell based or otherwise - I don't believe the STer logs/displays this value, but you can reset it in the LH Paramters menu.
O2 Sensor Adjust - Fast moving correction - This is what you see on the STer. It's range is -20% correction to +20% correction.
Does this help?
07-28-2010, 01:47 PM
#300
Nordschleife Master
I think we need to get the terminology right here.
O2 Adaptation - Slow moving agerage correction - not sure if this is cell based or otherwise - I don't believe the STer logs/displays this value, but you can reset it in the LH Paramters menu.
O2 Sensor Adjust - Fast moving correction - This is what you see on the STer. It's range is -20% correction to +20% correction.
Does this help?
O2 Adaptation - Slow moving agerage correction - not sure if this is cell based or otherwise - I don't believe the STer logs/displays this value, but you can reset it in the LH Paramters menu.
O2 Sensor Adjust - Fast moving correction - This is what you see on the STer. It's range is -20% correction to +20% correction.
Does this help?
Does the ST2 O2 sensor adjustment value (as displayed on the screen and the data logs) only include the fast-moving correction? You seem to indicate that it's just that. If adaptation is enabled, I would suspect that the value displayed is the sum of the slow-moving average correction and the fast-moving correction, but this seems to be inconsistent with what you wrote above.
Also, is it the sum of slow-moving and fast-moving corrections that is bounded by +/20% or is it just the fast-moving correction that is bounded by +/-20%? Above, you seem to indicate the latter. If this is correct, then how is the slow-moving correction bounded?