Patch request discussion: rev-down maf clamp (stalling issue)
Jun 3, 2008, 09:36 AM
#1
Patch request discussion: rev-down maf clamp (stalling issue)
tephra, mrfred, and all other patch writers:
I would like to start this thread to discuss a possibility of writing a patch that will create a type of maf clamp on closed throttle rev-down.
This is to address all of the people that have the clutch-in or free rev stalling issues...the scenario where the RPM will drop right past the target idle and dip well below and stumble, and sometimes even stall.
I don't have this issue, since I am still on mostly stock parts, but there are several people that are or have experienced this that I think would be eager to test if something was coded up.
This isn't a new problem to me as I had it on my DSM years ago and there has been a discussion on the DSMLink boards about this going back to 2004 or so. I remember reading that thread and going through tons of logs as well as my own logs and listening to what the DSMLink creators were saying and trying to code up themselves.
The reason why I decided to make this thread now is because lately I have been helping a couple of people over PM with this exact problem. Their logs show the exact same issue that I used to see years ago. Basically, the MAF overreads during the rev down after the throttle plate is closed. This causes the Hz readings to stay high and the load readings to climb (mass airflow/RPM). Sometime the ECU tries to correct this situation with the ISC and the idle will dip and rise a few times, but since the airflow reading is abnormally high, the oscillation continues, stalls, or finally fixes itself after the ISC opens and holds it open for an extended period of time.
The basic stalling and stumbling is because the misreported Hz and airflow values are causing the ECU to dump too much fuel. All logs that I have seen of this have wideband readings that get pegged rich, IPW readings that rise, etc.
So, to try to make a long story short, this is what I was thinking and what I would like this discussion to be about. I'm not sure if the DSMLink guys have this done or not, but rather than bothering them, I would like to work this out amongst ourselves and see if we can resovle this, so that we can continue to use the stock MAF reliably much longer into our modding path.
What I was thinking was this:
1. Define a set of criteria that has to be met in order for the maf clamp to kick in:
--To describe it in English, what we want the criteria to match is anytime the throttle plate is closed and the revs are dropping, preferably while out of gear. Basically, we want to capture the event of the throttle closing and then clamp a certain Hz value/load value (I forgot how the ECU calculates the Hz and load...I remember Bez explaining it, but we would have to clamp whatever is used to calculate the other).
2. Have this maf clamp of Hz stay active until the Hz reading comes back to the true Hz reading or until the throttle is pressed again. It would re-activate if the throttle is released back to 0 throttle (about 14% in our cars, I think).
--The true Hz reading would have to be something that would be adjusted per car. This should be an adjustable value. To find what it should be, simply log your car at idle. Let's say that your Hz reading at idle is 35Hz. Then, your idle or MAF clamp value should be maybe slightly above 35 Hz, like 40 Hz maybe. Once the measured Hz comes back below this value, the MAF clamp is turned off. (Basically, what this would do is eliminate that few seconds of time where the MAF is reporting abnormally high readings and feed the ECU with our MAF clamp reading...this way the correct fueling would be maintained and the overly rich stumble wouldn't occur...or at least that's the plan.)
We may or may not have to mess with other stuff like coolant temp, ac on/off, ISC steps, etc, but it may be as simple as clamping the Hz to a known good value and feeding that to the ECU, while monitoring what the MAF is telling us, and then lettin the MAF regain control once the Hz comes back to where it should be.
What do you guys think?
Eric
I would like to start this thread to discuss a possibility of writing a patch that will create a type of maf clamp on closed throttle rev-down.
This is to address all of the people that have the clutch-in or free rev stalling issues...the scenario where the RPM will drop right past the target idle and dip well below and stumble, and sometimes even stall.
I don't have this issue, since I am still on mostly stock parts, but there are several people that are or have experienced this that I think would be eager to test if something was coded up.
This isn't a new problem to me as I had it on my DSM years ago and there has been a discussion on the DSMLink boards about this going back to 2004 or so. I remember reading that thread and going through tons of logs as well as my own logs and listening to what the DSMLink creators were saying and trying to code up themselves.
The reason why I decided to make this thread now is because lately I have been helping a couple of people over PM with this exact problem. Their logs show the exact same issue that I used to see years ago. Basically, the MAF overreads during the rev down after the throttle plate is closed. This causes the Hz readings to stay high and the load readings to climb (mass airflow/RPM). Sometime the ECU tries to correct this situation with the ISC and the idle will dip and rise a few times, but since the airflow reading is abnormally high, the oscillation continues, stalls, or finally fixes itself after the ISC opens and holds it open for an extended period of time.
The basic stalling and stumbling is because the misreported Hz and airflow values are causing the ECU to dump too much fuel. All logs that I have seen of this have wideband readings that get pegged rich, IPW readings that rise, etc.
So, to try to make a long story short, this is what I was thinking and what I would like this discussion to be about. I'm not sure if the DSMLink guys have this done or not, but rather than bothering them, I would like to work this out amongst ourselves and see if we can resovle this, so that we can continue to use the stock MAF reliably much longer into our modding path.
What I was thinking was this:
1. Define a set of criteria that has to be met in order for the maf clamp to kick in:
--To describe it in English, what we want the criteria to match is anytime the throttle plate is closed and the revs are dropping, preferably while out of gear. Basically, we want to capture the event of the throttle closing and then clamp a certain Hz value/load value (I forgot how the ECU calculates the Hz and load...I remember Bez explaining it, but we would have to clamp whatever is used to calculate the other).
2. Have this maf clamp of Hz stay active until the Hz reading comes back to the true Hz reading or until the throttle is pressed again. It would re-activate if the throttle is released back to 0 throttle (about 14% in our cars, I think).
--The true Hz reading would have to be something that would be adjusted per car. This should be an adjustable value. To find what it should be, simply log your car at idle. Let's say that your Hz reading at idle is 35Hz. Then, your idle or MAF clamp value should be maybe slightly above 35 Hz, like 40 Hz maybe. Once the measured Hz comes back below this value, the MAF clamp is turned off. (Basically, what this would do is eliminate that few seconds of time where the MAF is reporting abnormally high readings and feed the ECU with our MAF clamp reading...this way the correct fueling would be maintained and the overly rich stumble wouldn't occur...or at least that's the plan.)
We may or may not have to mess with other stuff like coolant temp, ac on/off, ISC steps, etc, but it may be as simple as clamping the Hz to a known good value and feeding that to the ECU, while monitoring what the MAF is telling us, and then lettin the MAF regain control once the Hz comes back to where it should be.
What do you guys think?
Eric
Jun 3, 2008, 09:41 AM
#2
Here is some of the discussion that I saved from the DSMLink forums years ago:
Me:
Tom:
Dave:
Me:
Tom/Dave - If this truly is an airflow problem, which is seems to be, isn't there anything that can be done in DSMLink to combat the problem? Maybe some code to induce a 'Fake Maf' reading when the throttle plate is closed and speed is >'some number' or rpm > 'some number'. This way, you can set the MAF reading what it should be, somewhere around .20 g/rev or less, as Dave mentioned.
Dave is already testing code that does just that, yes.
It's actually not a trivial problem to solve. You essentially have to guess an upper limit for the airflow. You have to be rather aggressive in guessing a low enough airflow limit, otherwise it does no good, but not be so aggressive that the limit is too low. The limit is also affected by a number of variables such as engine temperature as well as power steering, alternator, A/C and automatic transmission load. It also depends on things the ECU has no way of knowing, such as whether or not there is engine-braking going on (although this may normally produce fuel cut, so it's not a big issue).
You might think that using a MAP sensor and momentarily switching to speed-density might help, but so far I haven't found that it responds quickly enough. Maybe it has something to do with where I'm picking up the pressure signal. It's also affected by EGR operation.
So far I have the limit activating only when the coolant temperature is above 170 degrees F and the idle switch is active. I use two limit tables, one for A/C off and one for A/C on. This seems to work well enough that it does not seem to limit when it shouldn't (causing lean running), but, particularly with the A/C on, doesn't help quite as much as I would like. Still, it's looking like this is going in the right direction and should be something we can include in a V3 release, maybe with some adjustability.
You might think that using a MAP sensor and momentarily switching to speed-density might help, but so far I haven't found that it responds quickly enough. Maybe it has something to do with where I'm picking up the pressure signal. It's also affected by EGR operation.
So far I have the limit activating only when the coolant temperature is above 170 degrees F and the idle switch is active. I use two limit tables, one for A/C off and one for A/C on. This seems to work well enough that it does not seem to limit when it shouldn't (causing lean running), but, particularly with the A/C on, doesn't help quite as much as I would like. Still, it's looking like this is going in the right direction and should be something we can include in a V3 release, maybe with some adjustability.
Jun 3, 2008, 10:11 AM
#3
I think the criteria are too complex. The standard cars have a little issue like this with a brief MAF over-read on partial throttle lift off and it would be nice to get rid of it.
I think the best method is to convert to speed density.
I think the best method is to convert to speed density.
Jun 3, 2008, 10:46 AM
#4
I agree speed density is the ultimate solution. As a matter of act, when I had this problem with my DSM, I compeltely solved it by installing a GM MAF in blow-though.
However, I think we should be able to code this relatively simply. One of the devices that I was mentioning that does this sort of thing is the HKS EIDS. All it has is an input for the MAF Hz and TPS. So, I'm sure it's doing something along the lines of what I was mentioning above.
I think it would be good to include coolant temp and AC on/off just so cold start and AC conditions wouldn't be affected.
Eric
However, I think we should be able to code this relatively simply. One of the devices that I was mentioning that does this sort of thing is the HKS EIDS. All it has is an input for the MAF Hz and TPS. So, I'm sure it's doing something along the lines of what I was mentioning above.
I think it would be good to include coolant temp and AC on/off just so cold start and AC conditions wouldn't be affected.
Eric
Jun 3, 2008, 10:51 AM
#5
This is how I would explain it a little simpler:
There would be two user inputs: an RPM value and a Hz value
There would be two things being monitored by the code: the Hz reading from the MAF and the throttle % or TPS volts (if that's easier)
I'm thinking activation would be the following:
1. Coolant above certain temp
2. RPM above user specified value (should be maybe 100-200 above idle RPM)
3. Throttle plate is closed (whatever TPS % or TPS volts that may be)
For deactivation:
1. Maf Hz reading reaches user specified Hz reading (whatever your idle Hz is) for 1-2 seconds
or
2. Throttle plate is opened again (gas pedal is pressed).
Under these conditions the Maf Hz would clamp and feed the clamp value to the ECU until the MAF settles down (1-2 seconds of good reading). Then the code allows the MAF Hz to be fed back into the ECU. Or, if the user steps on the throttle again, the clamp deactivates.
If this seems to work, then later something with the AC on can be added, where the hz and load would be higher.
Eric
There would be two user inputs: an RPM value and a Hz value
There would be two things being monitored by the code: the Hz reading from the MAF and the throttle % or TPS volts (if that's easier)
I'm thinking activation would be the following:
1. Coolant above certain temp
2. RPM above user specified value (should be maybe 100-200 above idle RPM)
3. Throttle plate is closed (whatever TPS % or TPS volts that may be)
For deactivation:
1. Maf Hz reading reaches user specified Hz reading (whatever your idle Hz is) for 1-2 seconds
or
2. Throttle plate is opened again (gas pedal is pressed).
Under these conditions the Maf Hz would clamp and feed the clamp value to the ECU until the MAF settles down (1-2 seconds of good reading). Then the code allows the MAF Hz to be fed back into the ECU. Or, if the user steps on the throttle again, the clamp deactivates.
If this seems to work, then later something with the AC on can be added, where the hz and load would be higher.
Eric
Last edited by l2r99gst; Jun 3, 2008 at 10:55 AM.
Jun 3, 2008, 01:31 PM
#6
Evolving Member
Join Date: Sep 2006
Location: Germany
Posts: 152
Likes: 0
Received 0 Likes
on
0 Posts
I think it is not very difficult to code this.
But how do you explain my problem:
I have replaced Stock injectors with PTE 780er
Use E85
Stock Airbox
Without AC off it idles fine, but with AC On it fell to deep when RPM drops from higher range (clutch hold pressed). It osscillate 3 or 4 times then idles fine, or in rare cases, it stall.
So I think "AC on/off" condition should be considered.
But how do you explain my problem:
I have replaced Stock injectors with PTE 780er
Use E85
Stock Airbox
Without AC off it idles fine, but with AC On it fell to deep when RPM drops from higher range (clutch hold pressed). It osscillate 3 or 4 times then idles fine, or in rare cases, it stall.
So I think "AC on/off" condition should be considered.
Trending Topics
Jun 3, 2008, 02:35 PM
#8
I've always found a cure for the 'decel dip' or decel stalling to be in the idle timing and sometimes aided by making the low rpm/low load area's in open loop with more control on the AFR. I have helped 3 or 4 guys fix this problem and I have cured all of them with these practices. The hardest was with a ported head and revolver cams but we still fixed the issue in less than an hour of tweaking and driving. Granted, all of these damn intake pipes out now aren't helping things either.
Jun 3, 2008, 03:24 PM
#9
There is definitely more than one cause for decel dip and stalling, but the cases that I am referring to are because of disrupted MAF reading during the decel, causing the MAF to read too high and erratically for a few seconds after letting off the throttle.
Actually, in these cases, it causes the load to spike, which in turn raises the timing. In these particular cases, the root cause is the abnormal airflow readings. The only way to truly fix these cases is what I am describing above, go to speed density or move the MAF in to the UICP away from the disturbances (turbo, intake pipe, BOV, etc).
What I am suggesting is to fix these specific cases, while still retaining the stock MAF.
I definitely appreciate the input, but I think we are talking about slightly different issues. The issue I am referring to is the disrupted MAF readings. I can post a screenshot of a log or two of what I am talking about compared to a good log from my car. It may help in explaining it better.
Eric
Actually, in these cases, it causes the load to spike, which in turn raises the timing. In these particular cases, the root cause is the abnormal airflow readings. The only way to truly fix these cases is what I am describing above, go to speed density or move the MAF in to the UICP away from the disturbances (turbo, intake pipe, BOV, etc).
What I am suggesting is to fix these specific cases, while still retaining the stock MAF.
I definitely appreciate the input, but I think we are talking about slightly different issues. The issue I am referring to is the disrupted MAF readings. I can post a screenshot of a log or two of what I am talking about compared to a good log from my car. It may help in explaining it better.
Eric
Jun 3, 2008, 04:23 PM
#10
OK, here is a good screenshot of what I am referring to. Actually there are similar screenshots in the thread that redlanevo started entitled 'Buschur intake pipe' here. Redlanevo has the same exact problem as well, but in his case, his ISC wasn't responding like it should, so I think there is something awry there.
Anyway, onto these logs from someone else. This person is having this stalling problem and his logs show the same exact airflow disruption as redlanevo's log did. First I will post a screenshot of my log and an explanation, then I will post two screenshots of his log...two because the log gets too crowded with a lot of items shown.
My log:
This log shows me blipping the throttle a couple of times with the following logged (from top to bottom, if you look at the far left of the log):
ISC steps
load
RPM and target RPM (flat line)
Airflow Hz
The thing to note is the area immediately after the airflow spikes, like the next half of a second. The load dips very low and the Hz has a steep smooth drop back to it's original value. Since basic load is MAF/RPM, you can see my load gradually come back to it's normal idle load before the blips as the RPM approaches the target.
The main area I want to point out is the airflow reading, though, and the resulting load reading because of the airflow. Note the shape of the curve, the instant, steep drop, right after the throttle is released (peak of Hz in the log).
Now, here are the logs from the problem car. Look at the same area that I described above in my log.:
The first log has:
steps
load
RPM
Hz
The second log has:
AFR
RPM
Timing
Hz
In these logs, it appears the user did two rapid blips of the throttle. Look at the airflow Hz line at the bottom of the logs and compare to mine. Before the throttle blip it's fine, after the blip, it starts to drop like it should, then it just hovers at an artifically high reading for a while and gradually start to come down. It should look like a steep ski jump ramp, like mine. His looks more like a mountain.
Anyway, this is the airflow anomoly that I am referring to. This sets off a chain reaction of stumbling, surging, etc, until the ISC has a chance to settle everything down about 5-10 seconds later.
It is this airflow metering error that causes the load to spike (MAF/RPM) because the Hz aren't dropping to where they should be. Look at the load spike up during the decel whereas mine drops very low down, as it should, then comes back up gradually.
This is the exact area that I am referring to about the MAF clamp above. If we could clamp the Hz, in this case at 31Hz, during the decel, I am certain this problem would be completely solved. If your look at redlanevo's thread, his logs show the exact same phenomenon. And it's this phenomenon that was being discussed years ago on the DSMLink site. I don't know if they ever wrote their code or not to resolve it, since I don't have a DSM anymore.
It is this area that units like the HKS EIDS controls. I'm pretty sure that some of the coders in here can write something to handle this nicely.
Hopefully, this explains more of what I am talking about.
Eric
Anyway, onto these logs from someone else. This person is having this stalling problem and his logs show the same exact airflow disruption as redlanevo's log did. First I will post a screenshot of my log and an explanation, then I will post two screenshots of his log...two because the log gets too crowded with a lot of items shown.
My log:
This log shows me blipping the throttle a couple of times with the following logged (from top to bottom, if you look at the far left of the log):
ISC steps
load
RPM and target RPM (flat line)
Airflow Hz
The thing to note is the area immediately after the airflow spikes, like the next half of a second. The load dips very low and the Hz has a steep smooth drop back to it's original value. Since basic load is MAF/RPM, you can see my load gradually come back to it's normal idle load before the blips as the RPM approaches the target.
The main area I want to point out is the airflow reading, though, and the resulting load reading because of the airflow. Note the shape of the curve, the instant, steep drop, right after the throttle is released (peak of Hz in the log).
Now, here are the logs from the problem car. Look at the same area that I described above in my log.:
The first log has:
steps
load
RPM
Hz
The second log has:
AFR
RPM
Timing
Hz
In these logs, it appears the user did two rapid blips of the throttle. Look at the airflow Hz line at the bottom of the logs and compare to mine. Before the throttle blip it's fine, after the blip, it starts to drop like it should, then it just hovers at an artifically high reading for a while and gradually start to come down. It should look like a steep ski jump ramp, like mine. His looks more like a mountain.
Anyway, this is the airflow anomoly that I am referring to. This sets off a chain reaction of stumbling, surging, etc, until the ISC has a chance to settle everything down about 5-10 seconds later.
It is this airflow metering error that causes the load to spike (MAF/RPM) because the Hz aren't dropping to where they should be. Look at the load spike up during the decel whereas mine drops very low down, as it should, then comes back up gradually.
This is the exact area that I am referring to about the MAF clamp above. If we could clamp the Hz, in this case at 31Hz, during the decel, I am certain this problem would be completely solved. If your look at redlanevo's thread, his logs show the exact same phenomenon. And it's this phenomenon that was being discussed years ago on the DSMLink site. I don't know if they ever wrote their code or not to resolve it, since I don't have a DSM anymore.
It is this area that units like the HKS EIDS controls. I'm pretty sure that some of the coders in here can write something to handle this nicely.
Hopefully, this explains more of what I am talking about.
Eric
Last edited by l2r99gst; Jun 3, 2008 at 05:53 PM.
Jun 4, 2008, 09:24 AM
#12
just as a simple interjection: From what I remember the EIDS doesn't clamp so much as it smooths and delays the MAS reading. On the by and by it doesn't work. I fiddled with it extensively and it did keep the car from stalling but overall the improvement was lackluster at best.
After switching from the HKS SSQV to the APS DV I've never had this issue again though.
After switching from the HKS SSQV to the APS DV I've never had this issue again though.
Jun 4, 2008, 01:33 PM
#14
Evolved Member
iTrader: (7)
Join Date: Nov 2006
Location: Pittsburgh
Posts: 772
Likes: 0
Received 0 Likes
on
0 Posts
l2r99gst, I think this is a path worth travelling. Admittedly some of the minor hiccups can be solved by other means. But its still a tool worth having in the arsenal on really pita cars. And we all know they exist...
Jan 20, 2010, 11:11 AM
#15
Evolving Member
Join Date: May 2008
Location: Dallas, TX
Posts: 346
Likes: 0
Received 0 Likes
on
0 Posts
bump.
I am having this stalling issue since moving from stock clutch to a Twin plate (much lighter weight) clutch. I have increase ISCV raised idle, added timing, still will drop very low rpm or flat line upon coasting to stop or pulling car out of gear at higher rpm.
I am having this stalling issue since moving from stock clutch to a Twin plate (much lighter weight) clutch. I have increase ISCV raised idle, added timing, still will drop very low rpm or flat line upon coasting to stop or pulling car out of gear at higher rpm.