Weird MAF table...
01-27-2007, 10:42 PM
#51
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
As for the filtering, I wonder how many samples is really needed? Definatly need something. That noise in the signal could be felt at high RPMs. That was the source of my mystery WOT vibration. Could even see it in the tach. 15-20 looks ok over that .3 second frame, but there is some response lag in the second one. Definatly dont want the spikes on the conditioned signal if I can avoid it. Those are whats causing the issues.
01-27-2007, 11:57 PM
#52
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
As for the filtering, I wonder how many samples is really needed? Definatly need something. That noise in the signal could be felt at high RPMs. That was the source of my mystery WOT vibration. Could even see it in the tach. 15-20 looks ok over that .3 second frame, but there is some response lag in the second one. Definatly dont want the spikes on the conditioned signal if I can avoid it. Those are whats causing the issues.
Its kind of funny that you mention the Heavy throttle Vibration.
I have one as well. Changing the bad U-Joint should help, but I am sure that it has something else going on as well.
It runs very well up top, but still feels like a loaded up blender that was just turned on. It is rough.
01-28-2007, 01:17 AM
#53
Like fast said, it goes right into the PCM, no converters. The resolution is actually quite good. A 'good chunk' was bad wording on my part. Me not thinking before typing. I think the main problem is the fast response of the MAF sensor. There is a small ammount (<2%) error due to uncertainty in the counters as well as ~2% from the sensor itself, from what Ive read. This doesnt do much by itself, but it probably does factor in with all the other airflow noise in the intake. At light throttle and idle, the noise is just a percent or two. It was actually faster with the throttle partially closed at high loads without filtering. Probably because the restriction acted to filter out the intake pulses by itself. The 'noise' does seem to be actual airflow, so the average of all that should yeild the true airflow rate over time.
Fast, it shouldnt be, and never was, an issue with an auto as youll never actually get to low RPM, high load operation. At low RPMs, the instantanious airflow varies by +/- 20%. This causes the motor to run real erratic. When Im going around a corner in 2nd, without the filtering, the engine would plain crap out when I got back on the gas. Works ok with the filtering, but still goes somewhat lean at 800 RPM or so. Even that is more then tolerable, but Id like to see if using a rolling average filter on the frequency can help smooth things out even more. Ill have to tinker with it in excel and see what looks good. GM has one in these PCMs for filtering the derivative of the TIS input. That uses a full 50 samples as well as a convoluted method for weighting the start and end values in the sample set. Im going to take that over for the MAF and tweak it a bit. Maybe have the # of samples used be variable based on RPMs.
Fast, it shouldnt be, and never was, an issue with an auto as youll never actually get to low RPM, high load operation. At low RPMs, the instantanious airflow varies by +/- 20%. This causes the motor to run real erratic. When Im going around a corner in 2nd, without the filtering, the engine would plain crap out when I got back on the gas. Works ok with the filtering, but still goes somewhat lean at 800 RPM or so. Even that is more then tolerable, but Id like to see if using a rolling average filter on the frequency can help smooth things out even more. Ill have to tinker with it in excel and see what looks good. GM has one in these PCMs for filtering the derivative of the TIS input. That uses a full 50 samples as well as a convoluted method for weighting the start and end values in the sample set. Im going to take that over for the MAF and tweak it a bit. Maybe have the # of samples used be variable based on RPMs.
01-28-2007, 01:28 AM
#54
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
O.K., so that is data the PCM is seeing. I wasn't sure if you were logging PCM data or using an external datalogger with a F/V converter as a crude oscilloscope. Now I am curious as to how it looks on an O-scope. Maybe it warrants rigging up an LM2907 F/V converter and taking an O-scope mobile. I suspect the wires in the sensor and/or the sensor itself are vibrating causing your fluctuations. When the sensor vibrates towards the airflow, the frequency and therefore the reported airflow go up and when it vibrates towards the engine, the opposite happens. Are you running the sensor as far away from the engine as possible and using some kind of vibration dampener like a corrugated rubber piece of tubing or one with a bellows? The MAF sensor should be well isolated from engine vibrations. As you have discovered, they are very sensitive. Just a thought.
I found that my MAF sensor did not like the Cone style filter.
01-28-2007, 01:42 AM
#55
In my experience lately with OBDII stuff, it is not so much the cone filter but the cones proximity to the MAF sensor. The sensor needs to be at least 6" away or more from the cone to report accurately, especially at higher flow rates since the cone filter induces the venturi effect. This is just my working theory, then again, I could be FOS.
01-28-2007, 12:51 PM
#56
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
O.K., so that is data the PCM is seeing. I wasn't sure if you were logging PCM data or using an external datalogger with a F/V converter as a crude oscilloscope. Now I am curious as to how it looks on an O-scope. Maybe it warrants rigging up an LM2907 F/V converter and taking an O-scope mobile. I suspect the wires in the sensor and/or the sensor itself are vibrating causing your fluctuations. When the sensor vibrates towards the airflow, the frequency and therefore the reported airflow go up and when it vibrates towards the engine, the opposite happens. Are you running the sensor as far away from the engine as possible and using some kind of vibration dampener like a corrugated rubber piece of tubing or one with a bellows? The MAF sensor should be well isolated from engine vibrations. As you have discovered, they are very sensitive. Just a thought.
IMO, I dont think its an issue with the sensor vibrating as much as intake resonance and sudden pulses in the airflow through the MAF as the engine takes in air. The MAF is hooked up using an elastomer coupling, so it should be relatively isolated. On the datalog, I observed a peak step response of 3500+ gms/sec*sec on the PCMs side, and thats with the filtering thats inherent to using accumulation to read in the pulses, so this is a really fast sensor. Partially closing the throttle at high load does seem to sheild the MAF from alot of this, so Id guess its an issue of toning it down a bit. The MAF is pretty much in direct control of the fuel injectors, so the fueling would directly track along that output without filtering. With heavy filtering, it does run nicely and doesnt lean out, but it does seem like there is a transient response as itll dip to 14:1 in PE on sudden tip-in and then quickly come up to the desired AFR of around 12.5 or so.
01-29-2007, 07:40 AM
#57
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
If I were to design a software to use a frequency based MAF, I would first count alllll MAF pulses between cylinders, and use that for my "flow." Then filter it a little. How does the filtering work on what'er you're using? Is it simply a digital frequency to instantaneous V (or digital count) conversion that is not synced with the engine? Or is it synced to X crank degrees (to match high and low pulse between cylinder strokes) and averaged, and filtered? Or what? If I wanted to do it with a slower processor, I'd probably to frequency to voltage, and also sync it to the high and low between cylinders (and average), and then filter as necessary.
01-29-2007, 08:01 AM
#58
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
All the pulses are there as theyre accumulated, so that acts as a sort of filter on its own, but the engines fueling tracks along the MAF output with no other filtering. Without filtering, the engine and MAF form a sort of feedback loop with eachother and itll buck and surge at low RPMs as the fueling goes all over the place and the engine misfires. On the wideband it shows large ammounts of O2 when this is happening. It did this with the F/V converter as well. The filtering Im using is a slide filter on the airflow, but thats sort of sluggish. Goes slightly lean while the filtered airflow is catching up and way rich when the throttle closes.
Im thinking of using a rolling average on the frequency instead and having adaptive filtering. Basically, the # of samples used for filtering will be reduced some while the RPMs are surging up real fast or the throttle is opening/closing. From the datalogs, it looks like a rolling average filter will work well for cleaning up all the noise.
Im thinking of using a rolling average on the frequency instead and having adaptive filtering. Basically, the # of samples used for filtering will be reduced some while the RPMs are surging up real fast or the throttle is opening/closing. From the datalogs, it looks like a rolling average filter will work well for cleaning up all the noise.
01-29-2007, 08:45 AM
#59
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
I'd guess that a "synced" filter would work, where the pulse count is measured at precice degree intervals (both high and low of each wave), and then averaged between intake strokes to get a midpoint. Otherwise you fight aliasing.
01-29-2007, 09:47 AM
#60
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I could put the filtering routine into the DRP routine and filter it there, but that would also take additional processing time. Hmm...
I can see what look like intake pulses. There is also a vicious ringing in the intake when the throttle opens suddenly. Hard to see that in what I posted, but it shows up on the overall plot of the MAFs output. Large, exponentially decaying sine wave after the throttle opens up. Dampening that out also seems to reduce some of the issues as well. Messing with the averaging filter in excel, it looks like it can handle the signal ok without aliasing. the signal cant really change any faster then 80Hz, anyway as the accumulator filters out anything faster then that as it averages the pulses on each read, so Im not sure if aliasing will be an issue or not with the filter routine.
Somehow I have a feeling that Im beating the same dead horse that GM was when they developed the system, though. Maybe dipping into SD at high load/fast transients is the answer. Ill have to see how DSP handles it first, though.
I can see what look like intake pulses. There is also a vicious ringing in the intake when the throttle opens suddenly. Hard to see that in what I posted, but it shows up on the overall plot of the MAFs output. Large, exponentially decaying sine wave after the throttle opens up. Dampening that out also seems to reduce some of the issues as well. Messing with the averaging filter in excel, it looks like it can handle the signal ok without aliasing. the signal cant really change any faster then 80Hz, anyway as the accumulator filters out anything faster then that as it averages the pulses on each read, so Im not sure if aliasing will be an issue or not with the filter routine.
Somehow I have a feeling that Im beating the same dead horse that GM was when they developed the system, though. Maybe dipping into SD at high load/fast transients is the answer. Ill have to see how DSP handles it first, though.
01-29-2007, 10:43 AM
#61
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
Even with a TMAP (MAP with IAT that is mounted directly on the manifold), we sync our sampling. Without it, it can go completely nuts, like as much as 15 kPa, even when filtered a little. With the sync, it's less than the resolution of the A2D. Filtering via software really hurts transients. Anything above 1/2 of ambient pressure presumably can generate upstream waves. WOT is worst. The pulses are from intake openings, and thusly are RPM dependant, although it's not a constant crank angle, due to exhaust tuning, sensor lag, and whatnot. An o-scope on a TMAP and the crank 60-2 makes it ever so obvious. I'm not trying to tell you how to do it - just some advice on what works on my system with MAP.
01-29-2007, 11:44 AM
#62
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I hear ya... The input is much appreciated. Also, interestingly, the stock SD code in these operates the same way as you stated above. The MAP reads are syncronous with the DRPs/cam sensor pulses. The sync MAP reads definatly make sense as you want to see what teh pressure is while the cylinder is intaking. With MAF, to me it seems a little more muddy as even in between the cylinder intake cycles theres still airflow into the manifold that needs to be accounted for.
Ive thought of moving it into the DRP routine, which uses the DRP as its interrupt. One thing about the MAF is that I use a pulse accumulator to read in the signal. This thing basically sits there and just gathers pulses in between reads. This is the only way to get good (+/- .05% at idle) resolution with the hardware thats there presently in the PCM. I could read it in syncronously, but all the information (good or bad) will still be there in the pulse counter, regardless of when its read as its constantly gathering the pulses. The frequency is the total additional # of pulses recieved/time between reads. It serves as its own filter in a way and does induce some lag in the signal. I cant read the true instantanous frequency due to low frequency timers giving poor resolution.
Im thinking the adaptive filtering may be the way to go. During transients, ease the filtering as to not totally kill the response or induce overhang and then heavily filter the signal during steady state operation.
Maybe that and flip/flop between speed density and MAF. Probably during transients it would be good to duck into SD for a period of time at high throttle transients and then exit out back into MAF operation. This is starting to get complicated, kinda like the LS1.
Ive thought of moving it into the DRP routine, which uses the DRP as its interrupt. One thing about the MAF is that I use a pulse accumulator to read in the signal. This thing basically sits there and just gathers pulses in between reads. This is the only way to get good (+/- .05% at idle) resolution with the hardware thats there presently in the PCM. I could read it in syncronously, but all the information (good or bad) will still be there in the pulse counter, regardless of when its read as its constantly gathering the pulses. The frequency is the total additional # of pulses recieved/time between reads. It serves as its own filter in a way and does induce some lag in the signal. I cant read the true instantanous frequency due to low frequency timers giving poor resolution.
Im thinking the adaptive filtering may be the way to go. During transients, ease the filtering as to not totally kill the response or induce overhang and then heavily filter the signal during steady state operation.
Maybe that and flip/flop between speed density and MAF. Probably during transients it would be good to duck into SD for a period of time at high throttle transients and then exit out back into MAF operation. This is starting to get complicated, kinda like the LS1.
01-29-2007, 06:54 PM
#63
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Ive decided on just ducking into SD during transients. This way I wont need to have to worry about any crazy filtering or lag in the filtering. Basically go into SD just long enough during heavy transitions or high load/very low RPM operation to give the MAF some time to stablize. The MAF is quite precise during all weather conditions under steady state, but during transitions at low RPMs its damn near useless.
01-29-2007, 07:27 PM
#65
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Lol, tell me about it
I knew this early on from my first setup way back when. The MAF does see it, but it also sees everything else, like standing waves, intake cycles, etc. I added TPS AE just for that purpose when I developed my current code to provide momentary compensation, but its difficult to get the correct fueling as Im completly reliant upon it during throttle openings while the filtered flow stablizes. Hence, Ill be using speed density intermittently so I have some other, more concrete form of fueling during transition.
I knew this early on from my first setup way back when. The MAF does see it, but it also sees everything else, like standing waves, intake cycles, etc. I added TPS AE just for that purpose when I developed my current code to provide momentary compensation, but its difficult to get the correct fueling as Im completly reliant upon it during throttle openings while the filtered flow stablizes. Hence, Ill be using speed density intermittently so I have some other, more concrete form of fueling during transition.
01-29-2007, 09:30 PM
#66
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
Yeah, transients are almost not much about airflow, and more about wall wetting. If the throttle is snapped, there is some airflow overshoot, that presumably a MAF would catch, but a MAP would miss. Otherwise you just need to build and evaporate (or drain) a puddle of wall film. That is more dependant on temperature and pressure. You could use the MAF to calculate "load" and then replicate a SD transient. Are you using both MAF and MAP sensors? Spark is also one of those things that MAP is better for, IMO.
If i were fighting this, I'd try sampling during the peak and valley of each pulse, and average the 2 together. That's how we do it, but with the pressure pulses. Then take that average sample and filter it only a little with the next average sample.
I think I agree with your chosen method of switching to an alternate method during transients, as MAF alone doesn't have much meaning unless you back out load from it. Transients are where I spend a lot of time (days for warm engine, weeks for cold) trying to get it right (or as right as the software can get, and then think of changes to the code to improve it / make calibration less of a maze). I happen to be near Russia right now doing cold transient (and starting, and whatever else to do with cold) calibration (.9L, 1L 3cyls, 1.3L, 1.5L 4cyls), and it gets to be a bit of fun finding the balance.
If i were fighting this, I'd try sampling during the peak and valley of each pulse, and average the 2 together. That's how we do it, but with the pressure pulses. Then take that average sample and filter it only a little with the next average sample.
I think I agree with your chosen method of switching to an alternate method during transients, as MAF alone doesn't have much meaning unless you back out load from it. Transients are where I spend a lot of time (days for warm engine, weeks for cold) trying to get it right (or as right as the software can get, and then think of changes to the code to improve it / make calibration less of a maze). I happen to be near Russia right now doing cold transient (and starting, and whatever else to do with cold) calibration (.9L, 1L 3cyls, 1.3L, 1.5L 4cyls), and it gets to be a bit of fun finding the balance.
Last edited by RednGold86Z; 01-29-2007 at 09:34 PM. Reason: added the averaging paragraph
01-29-2007, 11:51 PM
#67
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
From what I can tell, the issue I was having was the standing waves in the intake. There was a large standing wave that would form immediatly after the throttle snapped open and it would decay out. This obviously isnt actual airflow as the air is just ocillating in the intake tract. Not an issue with filtering, but the filtering itself then causes issues as the gms/sec really lags behind the airflow. And then when the throttle closed, the gms/sec way overshot the actual airflow and caused the engine to go real RICH. I could sit there and watch the airflow decay down on the WB. It would start at <9:1 and approach my desired AFR as an exponential decay. Maybe a better filtering stratagy would sove it, but like you said, theres always going to be some lag, so it would be a balancing act.
01-30-2007, 12:11 AM
#68
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
As for the TPS AE, its short acting enrichment based on throttle opening rate. The actual AE is done through the MAP AE, which comensates for the wall wetting. That part is relatively ok. The car responds well currently, but god only knows how itll do in the summer like that since the AE is covering for the MAF as well as the fuel dynamics. Could be way rich with warmer air. Lately its been in the teens here.
01-30-2007, 12:19 AM
#69
Do you mean that a standing wave(s) are forming from the steady flow before the throttle crack and are then getting sucked in past the MAF sensor causing the false reporting because a standing wave(s) implies stationary wave(s) that build from a steady frequency of pulses. A standing wave as I understand it cannot form from a transient but can cause a transitional state when absorbed. This could explain some of the issues but standing waves would quickly dissipate with a change in flow. I don't see that causing a decay in the transient flow. Resonance, however, can a ringing out of flow rate. Just some food for thought.
01-30-2007, 12:32 AM
#70
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
Do you mean that a standing wave(s) are forming from the steady flow before the throttle crack and are then getting sucked in past the MAF sensor causing the false reporting because a standing wave(s) implies stationary wave(s) that build from a steady frequency of pulses. A standing wave as I understand it cannot form from a transient but can cause a transitional state when absorbed. This could explain some of the issues but standing waves would quickly dissipate with a change in flow. I don't see that causing a decay in the transient flow. Resonance, however, can a ringing out of flow rate. Just some food for thought.
Recorded a log a couple of days ago. Even ran it up over 100 mph. Ran great. Mostly stock 350 programming as well. B-car LT1 timing advance. Mostly stock 350 TBI truck calibration. 5.3 MAF calibration. etc. Held 125-133 blm all the way through the log. Transitions were smooth and quick. No complaints from me.
I did have a cone style filter attached to the MAF sensor. The cone style filter disrupted the proper flow through the MAF sensor and contributed to poor running. I now have a cut-down Astro Van air box up front. Runs better now than it ever did with the cone style filter.
01-30-2007, 08:01 AM
#71
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Do you mean that a standing wave(s) are forming from the steady flow before the throttle crack and are then getting sucked in past the MAF sensor causing the false reporting because a standing wave(s) implies stationary wave(s) that build from a steady frequency of pulses. A standing wave as I understand it cannot form from a transient but can cause a transitional state when absorbed. This could explain some of the issues but standing waves would quickly dissipate with a change in flow. I don't see that causing a decay in the transient flow. Resonance, however, can a ringing out of flow rate. Just some food for thought.
Ive been time crunched lately so half of what I type is probably wrong. In a nutshell, its a driven ocillitaion in the intake tract that gets much worse with a sudden throttle change. Getting rid of all the ducting solves it to an extent. Ive both run an air cleaner only as well as a whole ducted system and its only really a problem with a ducted system, but it still acts funky both ways down low when you jamb the throttle down. The decay is due to me having to filter the signal. I could have improved filtering, but its a balance between rejecting noise and response.when I do finally get some time, Im going to add in the SD portion of the code.
----------
Recorded a log a couple of days ago. Even ran it up over 100 mph. Ran great. Mostly stock 350 programming as well. B-car LT1 timing advance. Mostly stock 350 TBI truck calibration. 5.3 MAF calibration. etc. Held 125-133 blm all the way through the log. Transitions were smooth and quick. No complaints from me.
Last edited by dimented24x7; 01-30-2007 at 08:05 AM. Reason: Automerged Doublepost
01-30-2007, 10:42 AM
#72
Dont take my wording verbatem. Ive been time crunched lately so half of what I type is probably wrong. In a nutshell, its a driven ocillitaion in the intake tract that gets much worse with a sudden throttle change. Getting rid of all the ducting solves it to an extent. Ive both run an air cleaner only as well as a whole ducted system and its only really a problem with a ducted system, but it still acts funky both ways down low when you jamb the throttle down. The decay is due to me having to filter the signal. I could have improved filtering, but its a balance between rejecting noise and response.
when I do finally get some time, Im going to add in the SD portion of the code....
Ive been time crunched lately so half of what I type is probably wrong. In a nutshell, its a driven ocillitaion in the intake tract that gets much worse with a sudden throttle change. Getting rid of all the ducting solves it to an extent. Ive both run an air cleaner only as well as a whole ducted system and its only really a problem with a ducted system, but it still acts funky both ways down low when you jamb the throttle down. The decay is due to me having to filter the signal. I could have improved filtering, but its a balance between rejecting noise and response.when I do finally get some time, Im going to add in the SD portion of the code....
The ringing out is a telltale indicator though. Filtering is making it more pronounced but it is there in the logs you posted. It looks like you have found the tuned frequency of your intake tract when you crack the throttle and it is resonating up through the intake tract instead of back into the manifold. The Helmholtz (sp?) resonance that helps tuned port setups find torque peaks could be amplifying your reversion. Perhaps some sort of diverter box or whatever you call those boxes hanging off the intake tract would help. I don't know, I'm just throwing out ideas. While dealing with the issue with code is a good way to solve it, I always like to try the simple stuff first. Trial and error is not the most scientific, but hey, it works.
01-30-2007, 11:31 AM
#73
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
Again, though, almost all pulses are from the intake valve events (it's the driving force), so, sampling at the peak and valley between strokes (add and divide by 2) would instantly elimanate any ring, and just result in average (and practically instantaneous) airflow. Even if it resonates (which I actually doubt is going on in your tube), this method would absorb it.
01-30-2007, 02:41 PM
#74
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Reading it in when there is a DRP, or some time thereafter would be possible. Getting the instantanious value isnt possible, though. The MAF reading will always be some average due to accumulation. It would probably better coordinated, though, but the resolution will drop the faster the sensor is read. There will also still be the issue of the resonance in the intake. Its superimposed on all the rest of the noise. See my attachment in: https://www.thirdgen.org/forums/diy-...log-check.html
Im still thinking that just filtering the signal from the MAF with an averaging filter and going into speed density during transients is the path of least resistance. It would be nice to sync the MAF reads, but it may really not be feasible with the hardware I have. More overhead, and reduced resolution in some areas of operation.
Im still thinking that just filtering the signal from the MAF with an averaging filter and going into speed density during transients is the path of least resistance. It would be nice to sync the MAF reads, but it may really not be feasible with the hardware I have. More overhead, and reduced resolution in some areas of operation.
01-30-2007, 07:31 PM
#75
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I looked into how feasible it would be to fuel in the DRP loop rather then the main fuel loop. Not only would I be running the fueling at up to 400+ Hz on 10+ year old 2.1 MHz processor, but there would be around a 10% variance in the airflow reading the MAF this fast, and thats at higher loads. Probably wouldnt fly...
01-31-2007, 12:01 AM
#76
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I was tinkering around with this instead of sitting in front of the TV and working on a beer gut and the noise isnt actually that bad. The frequency variance is only +/- 5%, which is around 3% on the airflow side. Syncronous reads probably would help, but the cure would be worse then the disease as Id introduce about double the error from uncertainty.
Part of the problem with the transitions is that stupid 2 1/2 foot long section of pipe I have. It actually is standing waves at the ends of the pipe that are the problem. Its singing like a fat lady at the opera. With nothing else changed, Im going to try the cone air filter alone again and see what happens. If it reads ok and the stumble is reduced, then its time to hit the yard and find a stock air box that can be made to work. Wonder if a TPI air cleaner or an LT1/LS1 airbox could be made to work?
Im still torn between going with adaptive filtering, or just going into SD during transitions.
Part of the problem with the transitions is that stupid 2 1/2 foot long section of pipe I have. It actually is standing waves at the ends of the pipe that are the problem. Its singing like a fat lady at the opera. With nothing else changed, Im going to try the cone air filter alone again and see what happens. If it reads ok and the stumble is reduced, then its time to hit the yard and find a stock air box that can be made to work. Wonder if a TPI air cleaner or an LT1/LS1 airbox could be made to work?
Im still torn between going with adaptive filtering, or just going into SD during transitions.
01-31-2007, 12:15 PM
#77
Member
Join Date: Oct 2005
Location: Dyer, In
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
Car: 91 Camaro RS
Engine: 355 Vortec
Transmission: 700r4
Axle/Gears: 3.42
Dimented,
I have not read this entire thread and do not mean to butt in....but I think your problem may be much more basic in how you are reading the MAF frequency. It sounds as if you are reading MAF pulses over a time period to determine frequency, therefore in order to get any accuracy in the readings you can not read the accumulated pulses at a very high rate. This leaves you with an inherent lag in the MAF reading which would be far to slow for transient close loop fuel control. Adding filtering just further aggrevates the situation do to futher lag in the system and it will end up causing resonance and oscillation in your control loop dependent on sample rate and filter tuning.
I would think the only way to get MAF reading to be fast enough to be usable for transient fuel control would be to measure frequency by measuring the period between MAF pulses and then the reciprical to determine frequency. In this way you could sample frequency as fast as every pulse if needed and and then be able to add some filtering to get rid of noise. This is a common method of frequency measurement for high speed closed loop control systems.
Now, I realize this is probably not possible with the PCM hardware you are using. But you could consider building an external F/V converter that used this method of frequency sampling and then bring in the analog votage for sampling by the PCM.
Hope this makes some sense.
Al
I have not read this entire thread and do not mean to butt in....but I think your problem may be much more basic in how you are reading the MAF frequency. It sounds as if you are reading MAF pulses over a time period to determine frequency, therefore in order to get any accuracy in the readings you can not read the accumulated pulses at a very high rate. This leaves you with an inherent lag in the MAF reading which would be far to slow for transient close loop fuel control. Adding filtering just further aggrevates the situation do to futher lag in the system and it will end up causing resonance and oscillation in your control loop dependent on sample rate and filter tuning.
I would think the only way to get MAF reading to be fast enough to be usable for transient fuel control would be to measure frequency by measuring the period between MAF pulses and then the reciprical to determine frequency. In this way you could sample frequency as fast as every pulse if needed and and then be able to add some filtering to get rid of noise. This is a common method of frequency measurement for high speed closed loop control systems.
Now, I realize this is probably not possible with the PCM hardware you are using. But you could consider building an external F/V converter that used this method of frequency sampling and then bring in the analog votage for sampling by the PCM.
Hope this makes some sense.
Al
01-31-2007, 12:37 PM
#78
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Currenty, the MAF is read at 80 Hz. Even at 160 Hz, the problem persisted. This was also an issue with the old F/V converter, as well as the inherent resolution issues with using an 8-bit A/D. It seems like its an issue with the MAF and engine forming a feedback loop with eachother. With filtering, theres a lag in response, but it recovers quickly and the AFR tracks along the desired AFR in the cal. Without it, its lean at low RPMs and takes some time for it to recover and start reporting the correct AFR, regardless of the intake arrangment in use. This only happens at low RPMs and seems to be RPM dependant. Maybe its noise, or resonance in the manifold? Im not sure. There doesnt seem to be enough noise in the steady state signal to account for it (only around 4% or so), but I havnt datalogged the actual unfiltered MAF frequency without the filtering in place on the airflow side yet.
01-31-2007, 02:09 PM
#79
Member
Join Date: Oct 2005
Location: Dyer, In
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
Car: 91 Camaro RS
Engine: 355 Vortec
Transmission: 700r4
Axle/Gears: 3.42
That is my exact point. It is a trade off between sampling rate and accuracy. If I calulated correct with your current sampling rate at 80 hz and I believe a max MAF freq of about 12khz, you only have the equivalent of 150 counts of resolution. If you sample at 160 hz you are down to 75 counts of resolution. Much less than even the crappy 8bit A/D which gives you 256 counts of resolution.
You may actually be better of going to a 20 or 40 hz sampling rate to see if the better resolution helps, but then of coarse you may start to see other problems at high rpms due to the slower sampling rate.
You may actually be better of going to a 20 or 40 hz sampling rate to see if the better resolution helps, but then of coarse you may start to see other problems at high rpms due to the slower sampling rate.
Last edited by alvanwie; 01-31-2007 at 03:29 PM. Reason: changed "bits of resolution" to "counts of resolution"
01-31-2007, 02:33 PM
#80
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Alvan,
The MAF is computed from the # of accumulated pulses/time between reads in seconds x 65536. The # of accumulated pulses has an uncertainty of +/- 1 while the time has an uncertainty of +/- 15.26 usecs. The resolution is actually fairly good at 80 Hz, +/- 2% or so. This is the preferred method from delphi's SAE article if using a frequency based MAF. Using the voltage gives ok resolution at high load, but it falls to around 7-8% uncertainty at idle with an 8-bit A/D, which is one of the worst places to have it. It worked well during cruise, but at idle it would surge a bit.
The MAF is computed from the # of accumulated pulses/time between reads in seconds x 65536. The # of accumulated pulses has an uncertainty of +/- 1 while the time has an uncertainty of +/- 15.26 usecs. The resolution is actually fairly good at 80 Hz, +/- 2% or so. This is the preferred method from delphi's SAE article if using a frequency based MAF. Using the voltage gives ok resolution at high load, but it falls to around 7-8% uncertainty at idle with an 8-bit A/D, which is one of the worst places to have it. It worked well during cruise, but at idle it would surge a bit.
Last edited by dimented24x7; 01-31-2007 at 02:39 PM.
01-31-2007, 02:56 PM
#81
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Ive also been thinking of the sync'd reads. With some semi-sophisticated logic, I could devise a way to control the rate that the sensor is read. At high DRPs, I could skip reads to ensure it doesnt exceed 80-100Hz or so. Only issue with putting it in the DRP interrupt is that itll displace the MPFI code, which makes this even more TBI specific. Although on the flip side of the coin, unlike TBI, theres no reason not to run SD with MPFI as it wont suffer from the fuel dynamics issues.
01-31-2007, 03:41 PM
#82
Member
Join Date: Oct 2005
Location: Dyer, In
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
Car: 91 Camaro RS
Engine: 355 Vortec
Transmission: 700r4
Axle/Gears: 3.42
I think there is some confusion here between resolution, accuracy as a percent of full scale, and accuracy as a percent of the current reading.
01-31-2007, 04:06 PM
#83
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
The method above is the most precise way to read it in, resulting in very little precision error, up to around 160 Hz or so, resulting in at most a few % error in the indicated reading. Above that the error goes high enough to potentially cause issues. If you have time, DL the SAE article from Delphi on MAF interfacing. Its quite informative.
Accuracy, however, is another matter. I think RedNGold is on to something in that spacing the reads to coorespond with the intake cycles will result in the most accuracy as itll account for the total airflow per intake cycle (in theory). I just have to find a practical way to make it a reality, balancing resolution with read time.
Accuracy, however, is another matter. I think RedNGold is on to something in that spacing the reads to coorespond with the intake cycles will result in the most accuracy as itll account for the total airflow per intake cycle (in theory). I just have to find a practical way to make it a reality, balancing resolution with read time.
01-31-2007, 09:53 PM
#84
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Forgot that the MAF read-in is less than 250 cycles, so it should have minimal impact. Added to the DRP interrupt routine for sync reads and request added to the main fuel loop to keep the read intervals reasonable. Ill have to try it out when I get a chance.
02-03-2007, 04:12 PM
#85
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
RednGold, I tried it with the sync reads and not surprisingly, it really does work as advertised. No longer leans out at lower RPMs. Having the extra code in the DRP interrupt is the worst place to have it in a heavily loaded computer, but its well worth it. Provides steady fueling with no filtering needed.
The thing that really helped was ditching the cold air intake. I put it in the round file and I now have my hot air intake back in place. Its night and day. Throttle response is smooth and crisp without all the resonance. The steady state is much better, also. Even at 60 mph, it doesnt even feel like the car is moving. Runs real smooooth... I have the feeling that I also dont really have enough injector to meet the AE demands at very cold temps. Definatly confirms that cold air is a terrible idea with TBI.
The thing that really helped was ditching the cold air intake. I put it in the round file and I now have my hot air intake back in place. Its night and day. Throttle response is smooth and crisp without all the resonance. The steady state is much better, also. Even at 60 mph, it doesnt even feel like the car is moving. Runs real smooooth... I have the feeling that I also dont really have enough injector to meet the AE demands at very cold temps. Definatly confirms that cold air is a terrible idea with TBI.
02-03-2007, 05:07 PM
#86
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
RednGold, I tried it with the sync reads and not surprisingly, it really does work as advertised. No longer leans out at lower RPMs. Having the extra code in the DRP interrupt is the worst place to have it in a heavily loaded computer, but its well worth it. Provides steady fueling with no filtering needed.
The thing that really helped was ditching the cold air intake. I put it in the round file and I now have my hot air intake back in place. Its night and day. Throttle response is smooth and crisp without all the resonance. The steady state is much better, also. Even at 60 mph, it doesnt even feel like the car is moving. Runs real smooooth... I have the feeling that I also dont really have enough injector to meet the AE demands at very cold temps. Definatly confirms that cold air is a terrible idea with TBI.
The thing that really helped was ditching the cold air intake. I put it in the round file and I now have my hot air intake back in place. Its night and day. Throttle response is smooth and crisp without all the resonance. The steady state is much better, also. Even at 60 mph, it doesnt even feel like the car is moving. Runs real smooooth... I have the feeling that I also dont really have enough injector to meet the AE demands at very cold temps. Definatly confirms that cold air is a terrible idea with TBI.
02-03-2007, 06:29 PM
#87
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I will soon. Just have to decide what I want to do with the filtering. Eliminate it altogether, or just keep it for when the motor is idling. I dont know how itll react to a larger cam with overlap. Then again, overlap might not be a good idea with MAF.
02-03-2007, 07:30 PM
#88
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
Hey glad to hear it's working better! Wish I had a MAF input (hmmm.... maybe... would be nice to study... and it's not my money... or time...)
About cold AE (ramblings) - TBI has to be about the worst. Even with SD and sequential mpi, I only get it so-so with our algorithm (which I'm trying to re-design/improve, but only in the single mind brainstorm mode now). Like I said though, it's almost all about wall film, which has less to do with airflow (except at higher airflow, it's less, scale accordingly), and more to do with the several walls - like tb and manifold (for you TBI guys), runners and ports, and the cylinder walls itself. Each are different temperatures, and have different airflows across them. And fuel itself is made of light and heavy stuff... A model gets terribly complicated, and large ad hoc tables get almost impossible for a human to tune (only a few moments at each temperature range during a cold start). I'm thinking of making ours X-Tau ish with enough modifiers to melt it down into something tunable at almost all conditions. I think the wideband just makes me too critical, sometimes, though, hehe. Should just go by A) drivability B) emissions C) wideband - in that order of importance, but I'm c a b now.
Anyway, congrats! - but are you sure it's working as well as you think? - never be satisfied (until you retire)! Show us some data, too, eh?
About cold AE (ramblings) - TBI has to be about the worst. Even with SD and sequential mpi, I only get it so-so with our algorithm (which I'm trying to re-design/improve, but only in the single mind brainstorm mode now). Like I said though, it's almost all about wall film, which has less to do with airflow (except at higher airflow, it's less, scale accordingly), and more to do with the several walls - like tb and manifold (for you TBI guys), runners and ports, and the cylinder walls itself. Each are different temperatures, and have different airflows across them. And fuel itself is made of light and heavy stuff... A model gets terribly complicated, and large ad hoc tables get almost impossible for a human to tune (only a few moments at each temperature range during a cold start). I'm thinking of making ours X-Tau ish with enough modifiers to melt it down into something tunable at almost all conditions. I think the wideband just makes me too critical, sometimes, though, hehe. Should just go by A) drivability B) emissions C) wideband - in that order of importance, but I'm c a b now.
Anyway, congrats! - but are you sure it's working as well as you think? - never be satisfied (until you retire)! Show us some data, too, eh?
02-03-2007, 09:22 PM
#89
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
I had the WB hooked up to it. Wasnt datalogging, so no data to show, but its not going lean at low RPMs. In fact, its running richer then target if anything under load. With the old intake and new code, it was still going wicked lean, but only for a moment. It would recover and get going, but it still wasnt running as evenly as it could. The AFRs where on target, but it still was slightly off. I think the air fuel dist. was just way off without warm air to heat the manifold and intake charge. At any rate, cold air was a really bad idea. The manifold was never heating up. It was always cool to the touch.
You are right about one thing. TBI is very hard to tuneif you dont set everything up right. It runs fairly well even cold now. Much better then it used to run with the stock system. It ran like dog **** with that. Id almost say it runs like it came right off the factory floor. On the highway, its like flying in a plane.
If I can, Ill get some MAF data posted. Id imagine it probably wouldnt look much different, but it seems like the fueling is well coordinated with the motor.
You are right about one thing. TBI is very hard to tuneif you dont set everything up right. It runs fairly well even cold now. Much better then it used to run with the stock system. It ran like dog **** with that. Id almost say it runs like it came right off the factory floor. On the highway, its like flying in a plane.
If I can, Ill get some MAF data posted. Id imagine it probably wouldnt look much different, but it seems like the fueling is well coordinated with the motor.
02-03-2007, 11:23 PM
#90
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Since its a nice balmy 18 degrees F out, I took the car out for a minute to see how it is in the cold. It does show some leanness at very low RPMs. Worst is around 450 RPM under load. Its around 15.5:1-16:1. It does quickly richen up as the RPMs build and is fine by 1600 RPM. With an auto, youd never notice, but with a stick and highway gears, its real easy to push its buttons. There is still some lag in the response as the MAF cant be sampled any faster then around 80-100 Hz and is also out of sync with the fuel routine, which also only runs at 80 HZ. Doing the math, during sudden WOT, the airflow rises at a rate of around 400 gms/sec^2, which is more then enough to cause some leaning out as the airflow being used in the fueling will be less then the actual airflow. This and Im sure theres some lag in the transient response of circuitry of the MAF as well. Its not enough to be really noticed unless you force it.
02-04-2007, 04:36 PM
#91
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Its hard to say if it really is the MAF now. I took it out an hour after I had shut it down and it mustve heat soaked real bad. On tip in the AFRs where pegging at 9:1 and the motor was bucking at low RPMs. It clears up and works fine once driven a bit, but evedently, it would take a whole lot more code to properly do the AE. Probably would also need to have a dedicated temp sender just for the manifold.
Fast, Ill send teh latest via email. For some reason, its getting corrupted when I upload it to here.
Fast, Ill send teh latest via email. For some reason, its getting corrupted when I upload it to here.
Last edited by dimented24x7; 02-04-2007 at 04:48 PM.
02-04-2007, 09:01 PM
#92
Supreme Member
iTrader: (1)
Join Date: Mar 2000
Location: All over China, Iowa, and San Luis Obispo, CA
Posts: 1,692
Likes: 0
Received 1 Like
on
1 Post
Car: 92 Form, 91 Z28, 89 GTA, 86 Z28
Engine: 5.7 TPI, LG4
Transmission: 700R4, 700R4
Axle/Gears: 3.27, 2.73
TBI, oi! The algorithm would get ugly! But a combination Manifold temp and Coolant (head / valve) temp would probably get it closer. Maybe a man-temp global AE multiplier (on top of the existing code) would be ok. You don't need IAT, due to using a MAF, so relocate it if you got it! Man-T Should be used to calc/mod cranking and startup too!
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
02-04-2007, 09:23 PM
#93
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
TBI, oi! The algorithm would get ugly! But a combination Manifold temp and Coolant (head / valve) temp would probably get it closer. Maybe a man-temp global AE multiplier (on top of the existing code) would be ok. You don't need IAT, due to using a MAF, so relocate it if you got it! Man-T Should be used to calc/mod cranking and startup too!
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
Its funny that you mention this. With the EBL, my IAT was screwed directly into the Intake manifold. I had the closed style MAT sensor from a TPI car. Screwed directly into the manifold vacuum take-off on the edelbrock intake. Saw the intake temps drop 20-40* at times when I opened the throttle and went from idle to cruise. I saw intake temperatures of 40-50* on 100* days at times.
You can see how far it dipped at times in these logs. Notice how far the IAT would drop under heavy load.
Last edited by Fast355; 02-04-2007 at 09:34 PM.
02-04-2007, 09:52 PM
#94
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Lets see if this attachment works on this computer...
Edit: Nope... Something up with the boards.
Edit: Nope... Something up with the boards.
02-04-2007, 10:11 PM
#95
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
TBI, oi! The algorithm would get ugly! But a combination Manifold temp and Coolant (head / valve) temp would probably get it closer. Maybe a man-temp global AE multiplier (on top of the existing code) would be ok. You don't need IAT, due to using a MAF, so relocate it if you got it! Man-T Should be used to calc/mod cranking and startup too!
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
On shutdown, our intakes can get up to >75°C even in moderate weather (nevermind at hot weather, whew!). At idle (even in sub -10°C ambient) it goes up to >50°C - but that's a dry manifold. Your wet manifold probably cools from fueling in a relatively short manner.
Ill say one thing, the transient fueling with TBI is a nightmare. Getting smooth, crisp throttle response across the board was a pain. And the fun part is when summer comes around, I'll probably have to start all over again because itll be pig rich in the heat. Now I know why GM used that teeny tiny little restrictive fully heated manifold and thermac.
02-05-2007, 11:45 PM
#96
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Im going to go ahead and add a manifold plenum wall temp sender to the mix. Its really annoying to have the car hesitate on accel after a long drive at high speed. Works great in stop and go, but once theres alot of cold airflow, the manifold gets really cold in this weather and the engine falls flat on its face for a second or two when I stab the gas.
04-22-2007, 09:39 PM
#97
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Re: Weird MAF table...
This low speed, high throttle problem still persists. I did some testing and it runs really lean at low RPMs when you give it some throttle. If I hold it around 1500 RPM under load, itll actually start breaking up and goes into detonation
I was also datalogging, and I got some interesting results. The MAF readings are what the should be. The interesting part was graphing the grams of air that each cylinder was ingesting. At low speeds, each cylinder appears to move alot more air. The VE at 1400 RPM is nearly 100%, but falls to around 85% around 2200 RPM, about where it should be. The MAF is seeing the extra flow, and adding the right ammount of fuel, but it looks like its going out into the exhaust, or somewhere, whenever there is high manifold pressure at low speeds. The engine always seems to lean out at high manifold pressure. High throttle, letting the clutch out at low speeds, etc.
I was also datalogging, and I got some interesting results. The MAF readings are what the should be. The interesting part was graphing the grams of air that each cylinder was ingesting. At low speeds, each cylinder appears to move alot more air. The VE at 1400 RPM is nearly 100%, but falls to around 85% around 2200 RPM, about where it should be. The MAF is seeing the extra flow, and adding the right ammount of fuel, but it looks like its going out into the exhaust, or somewhere, whenever there is high manifold pressure at low speeds. The engine always seems to lean out at high manifold pressure. High throttle, letting the clutch out at low speeds, etc.
04-22-2007, 10:25 PM
#98
Supreme Member
iTrader: (2)
Join Date: Jan 2005
Location: Hurst, Texas
Posts: 10,081
Received 403 Likes
on
345 Posts
Car: 1983 G20 Chevy
Engine: 305 TPI
Transmission: 4L60
Axle/Gears: 14 bolt with 3.07 gears
Re: Weird MAF table...
This low speed, high throttle problem still persists. I did some testing and it runs really lean at low RPMs when you give it some throttle. If I hold it around 1500 RPM under load, itll actually start breaking up and goes into detonation
I was also datalogging, and I got some interesting results. The MAF readings are what the should be. The interesting part was graphing the grams of air that each cylinder was ingesting. At low speeds, each cylinder appears to move alot more air. The VE at 1400 RPM is nearly 100%, but falls to around 85% around 2200 RPM, about where it should be. The MAF is seeing the extra flow, and adding the right ammount of fuel, but it looks like its going out into the exhaust, or somewhere, whenever there is high manifold pressure at low speeds. The engine always seems to lean out at high manifold pressure. High throttle, letting the clutch out at low speeds, etc.
I was also datalogging, and I got some interesting results. The MAF readings are what the should be. The interesting part was graphing the grams of air that each cylinder was ingesting. At low speeds, each cylinder appears to move alot more air. The VE at 1400 RPM is nearly 100%, but falls to around 85% around 2200 RPM, about where it should be. The MAF is seeing the extra flow, and adding the right ammount of fuel, but it looks like its going out into the exhaust, or somewhere, whenever there is high manifold pressure at low speeds. The engine always seems to lean out at high manifold pressure. High throttle, letting the clutch out at low speeds, etc.
The intake event is probably very early on the "RV" cam you have in your engine. It is likely contributing to the problem.
04-23-2007, 07:34 AM
#99
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Re: Weird MAF table...
Its definatly a really annoying problem. If I let the clutch out at very low engine speed, the AFR goes up to like 18:1 when the MAP is high and theres alot of popping in the exhaust. I wonder if the cam timing is off or something?
04-23-2007, 08:25 AM
#100
Supreme Member
Thread Starter
iTrader: (2)
Join Date: Jan 2002
Location: Moorestown, NJ
Posts: 9,962
Likes: 0
Received 3 Likes
on
3 Posts
Car: 88 Camaro SC
Engine: SFI'd 350
Transmission: TKO 500
Axle/Gears: 9-bolt w/ 3.23's
Re: Weird MAF table...
Another thing is that without the full PID idle speed control enabled, the engine idles like crap. With it enabled, the engine idles well, but from looking at the spark and fuel derivative terms, the PID is adding in ALOT of corrections to maintain the desired idle speed. The cam is also misground. the #1 cylinder has a .442/.442 (and likely a 214/214) profile with all the rest having a .420/.442 204/214 profile. I wouldnt have believed it if I hadnt measured it with a dial gauge. The funny thing is that otherwise, the engine runs great, is very responsive, and has good power.
I have a feeling that if I was running a carb, I probably wouldve fixed it long ago.
I have a feeling that if I was running a carb, I probably wouldve fixed it long ago.
Thread
Thread Starter
Forum
Replies
Last Post
BumpaD82
Tech / General Engine
37
02-26-2016 02:57 PM
