When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
I'm trying to get a better understanding of these tables and their function in the EBL system.
INT delay vs airflow. Is this the transport delay between the cylinder and the O2 sensor? How does it come into play when tuning?
PRP duration vs airflow. Is this the "time" the computer allows while waiting for a x count or O2 reading within the O2 R/L limit tables? Again how does it come into play when tuning.
What I am experiencing during a steady state cruise (~2000RPM) is with the tables "stock" for a tuned port bin my INT will rise and then the BLM will rise but the O2 sensor (and wide band) is reporting rich. It will stay there a while then both move the other way. First the INT will dive then the BLM will follow, pulling fuel but the O2 (& WB) report lean. This cycle seems to repeat. I have reduced the Prop gains 10% at a time until they are now past half way and it still occurs.
Finally, I notice that in the TPI bins the prop gains vs O2 error are in the 2-4 range but the same table for a TBI bin has values in the 16 range.
I'm trying to get a better understanding of these tables and their function in the EBL system.
INT delay vs airflow. Is this the transport delay between the cylinder and the O2 sensor? How does it come into play when tuning?
That along with some delay for the proportional gain to have an affect.
PRP duration vs airflow. Is this the "time" the computer allows while waiting for a x count or O2 reading within the O2 R/L limit tables? Again how does it come into play when tuning.
It is the length of time the proportional gain is being used. Once that duration has timed out the gain is set to zero until the next cross count. The earlier versions of the EBL it was just a suggestion. The newer versions it is a true duration (V2.3+ for EBL Flash, V1.2+ for EBL P4 Flash, all versions of EBL SFI-6 Flash).
What I am experiencing during a steady state cruise (~2000RPM) is with the tables "stock" for a tuned port bin my INT will rise and then the BLM will rise but the O2 sensor (and wide band) is reporting rich. It will stay there a while then both move the other way. First the INT will dive then the BLM will follow, pulling fuel but the O2 (& WB) report lean. This cycle seems to repeat. I have reduced the Prop gains 10% at a time until they are now past half way and it still occurs.
With the slow oscillating INT need more proportional gain.
Finally, I notice that in the TPI bins the prop gains vs O2 error are in the 2-4 range but the same table for a TBI bin has values in the 16 range.
There is also the prop gain vs airflow multiplier that adjusts the O2 error base gain.
In regard to the proportional duration, would a smaller value in the table net less affect on fueling? I purchased EBL almost exactly 1 year ago, would this be the latest version or how can I identify it?
It appears I may need to try more prop gains. I kept lowering the values and it didn't improve. For some inexplicable reason I didn't try going the other way.
As far as the two different tunes having different values in the tables I do understand the multiplier changes the amount of prop gains from the first table, I am curious as to the reason one would have a large value in the gain table and a small multiplier in the airflow table. I suppose you could have pretty much the same affect either way but there has to be a method in the madness and I would like to understand why.
I know you probably don't have time to go too deep but anything you can add will be appreciated. I have read your PID paper many times and have been studying it and the stickies lately and making lots of notes. But in the end without being able to see the equation real time and how each table affects the PW it's difficult to figure this out. I'm trying. My car runs very well with EBL. After the season changed problems emerged. I made adjustments which seemed to help and it runs right 95% of the time but then runs up and down with the BLMS and targets lean or rich.
In regard to the proportional duration, would a smaller value in the table net less affect on fueling?
Yes, the proportional gain fuel doesn't last as long. As it is being added for a shorter period of time. And that at lower RPM (airflow) the injections are spaced further apart, so the duration is typically higher at lower air flows.
Note that a cross-count truncates any further duration and starts new to force the AFR in the other direction.
I purchased EBL almost exactly 1 year ago, would this be the latest version or how can I identify it?
With engine running it is displayed at the bottom of the main WUD screen. Can check our update page to see what the latest versions are along with what changes have been made.
It appears I may need to try more prop gains. I kept lowering the values and it didn't improve. For some inexplicable reason I didn't try going the other way.
As far as the two different tunes having different values in the tables I do understand the multiplier changes the amount of prop gains from the first table, I am curious as to the reason one would have a large value in the gain table and a small multiplier in the airflow table. I suppose you could have pretty much the same affect either way but there has to be a method in the madness and I would like to understand why.
Basically it comes down to the range of possible results. This is also balanced against the duration. Can have a lower gain with more duration for the same volume of fuel for that period of time. The gains equation is simply:
gain_O2 * gain_airflow = gain_total
So with a gain_O2 of 16 and a gain_airflow of 80%, the result is 12.8 (rounded to 13). This '13' is added or subtracted from the main injector PW. With a resolution of approximately 15.26 usec per bit. So a gain of 13 changes the PW by approximately 200 usec.
I know you probably don't have time to go too deep but anything you can add will be appreciated. I have read your PID paper many times and have been studying it and the stickies lately and making lots of notes. But in the end without being able to see the equation real time and how each table affects the PW it's difficult to figure this out. I'm trying. My car runs very well with EBL. After the season changed problems emerged. I made adjustments which seemed to help and it runs right 95% of the time but then runs up and down with the BLMS and targets lean or rich.
Be aware that with seasonal changes the fuel composition also changes.
So the proportional gains event does nothing or adds/subtracts X amount of fuel (determined by the prop gains x prop mult) for a certain amount of time (determined by the duration table). In the mean time the INT is waiting (determined by the delay table) to "look" at the O2 sensor. Then, if the O2 reports lean/rich the INT will adjust. All of this is "zeroed" out by a O2 measurement cross-count as determined by the O2 R/L and boundary tables.
If the INT has to keep adding/subtracting fuel the BLM kicks in and raises/lowers the fueling by a large amount to keep the INT back at 128/neutral.
The entire Prop routine happens (or does nothing) every time the injector "bank" fires or once per revolution. So within that time there are 2 cylinders firing (per bank/O2 sensor) and the residual from 2 more and depending where the O2 sensor is positioned would get a lot of O2 input per cycle.
I think I have a fair understanding of this part. Please correct me where I'm wrong.
One area which is still unclear is the procedure to determine how much to change the prop gains. In a different thread Bob wrote;
"When tuning proportional gains I use the NB O2 signal and the INT for input. With the INT, see if it is oscillating. A rapid rise and fall is from too much prop gain or duration. This is where the INT is chasing the gains to correct the over/under fueling.
While a slow rise and fall is from not enough prop gain or duration. In this case the ECM is using the INT to create cross counts."
In this procedure would the BLM's be locked at 128? What is meant by "rapid" and "slow" rise or fall? In what amount of time? Also, is this during steady state conditions or after some other change like TPS/MAP etc.
Another problem I have is my engine has been modified since I began tuning so I really don't have a baseline nor know what is "normal" in regard to AFR/O2 fluctuations. I understand the system is attempting to raise and lower the AFR around 14.7. Maybe the AFR changes are within that range.
What happens is this. The engine has been brought to temp and run for 20 or so minutes at a 2000-2200 RPM/45-50MAP cruise. The BLM's are generally between 128 and about 134. The wide band will be reporting around 15-16:1 although the O2 R/L tables target 635mv. O2 signal reports good cross counts but favors rich. Then, for no apparent reason the INT will start to climb past 136 and the BLM will start to follow, the wide band will report 13-14. After the INT recovers it begins to drop and the BLM follows down to under 120 and WB reports lean 17-18. It will later level back off but will generally be a few counts rich or lean and stay there until the engine is restarted or driven in traffic. I just can't seem to figure out what may be causing the EBL to pull then flood fuel.
Other than Prop/Int fueling changes the only other issues I can imagine are related to my cam and the O2 sensor. Since I have ~70' of overlap maybe the sensor is cooling off and changing voltage even though it is heated. Or maybe "extra" oxygen in the exhaust if fooling everything and there's no way I can tune this to stay in closed loop. I don't think my cam is too wild at 236/242 112lsa in a 414 CI motor. I'm hoping for suggestions or ideas. Other than a bit of jerkiness at light map cruise under 2K the motor runs very well.
Last edited by antman89iroc; 10-01-2015 at 12:43 PM.
No one? I was wondering if mine was moving too slow also is there a right or wrong swing speed or is set to what the engine likes? or what the ECM wants?
Use the INT (integrator) value to see if the proportional gains are correct. If the O2 sensor has been moved further down stream it can also help to increase the INT update delay a little. Here are three screen grabs showing the O2 response. This is on a mostly stock engine with an _old_ O2 sensor (it should be replaced).
One is at idle, another at steady state speed, with the third showing what occurs when the throttle position is changed.
A different vehicle, this has the O2 windows increased to run richer then stoich. Can see how the O2 end points (high & low) are higher then normal. Not for a vehicle with cat-cons. The heated NB O2 sensor is about 6" past the collector of long tubes.
The slow O2 signal image is from not enough proportional gains. The INT is slowly rising and falling to create the cross-counts.
I am struggling to get good cross counts like graphs above. I am running VAFPR at 22psi like the 3001 file, so I copied much of the file changes relative to fueling. I mult .58 by the prop gains and durations (also AEs). I then get less cross counts than before. So, I looked at the 3001 files values and copied them. Again, less cross counts. Held high/low for approximately 5s at steady drive. Above it recommends to increase the prop gains to get more cross counts. I ended up trying mult by 104/61=1.7 instead and started getting more cross counts. They aren't as crisp as the graphs above. This seems to be conflicting to the recommended approach, anyone have some guidance. Part of the reason on stumbled on this is that I tuned with NB and then WB. When I went back to NB my BLMs went back up and INT drifts up. Realized I missed correcting PRPs.
The multiplying of the prop gains is to rough it in. Needs to be further tuned from there.
I use the INT and the sound of the engine.
RBob.
What engine "sound" are you listening to?
My NB signal looks more like "M's" mostly high then dips low briefly. Currently my Prop tables are stock.
I have my INT R/L mean table (& hi/low tables) fairly high. 680 at 40-50 map and slightly leaner at higher loads. I remember with the stock MAF computer it would subtract 100 when in "divert" so I guessed EBL was also subtracting 100 from these tables. Otherwise I should be running pig rich huh?
My INT doesn't oscillate up and down much, at least not quickly. What generally happens after a long drive is the INT will drop some and the BLM will follow to about 124ish. The NB will be (and is almost always) high but even then my WB will report lean. Looking at plugs is difficult to say as with the HSR manifold the fronts look about right and the rears look rich. Runs pretty good and I've sort of accepted it but I can't always get predictable results from changes.
The tone of the engine. Is it smooth? Does it have a bad u-joint sound (too much prop gain). Or does tone of the engine increase and then decrease in a steady pattern? The biggest to me is to have the engine sound and feel smooth. That is a happy engine.
Note that with performance engines open loop is usually a lot smoother then closed loop. This is a tradeoff of what is acceptable to the driver.
the stock MAF computer it would subtract 100 when in "divert" so I guessed EBL was also subtracting 100 from these tables.
The EBL Flash doesn't support AIR, the O2 tables are used as they are shown (P4 & SFI-6 do support AIR).
Bob, moderator or anyone who may know, how do I get the screen shots of the performance window in EBL? I am trying to post some from my log to show what I am experiencing.
When you say, "The INT is slowly rising and falling to create the cross-counts." and "I use the INT and the sound of the engine." are you referring to the INT in the INT/BLM columns?
Bob, moderator or anyone who may know, how do I get the screen shots of the performance window in EBL?
With the window that you want to capture as the active one, ALT+PrintScreen will copy it to the clipboard. Then paste it into something such as Paint or IrFanView and save it as a jpg image. Upload the jpg image here via the Quote/Advanced reply and Attach Files dialog.
When you say, "The INT is slowly rising and falling to create the cross-counts." and "I use the INT and the sound of the engine." are you referring to the INT in the INT/BLM columns?
Yes, the INTegrator (short term fuel trim). Above I posted a picture of what not enough proportional gain looks like. This is where the INT is being used by the ECM to create cross counts. Works like this, the O2 is reporting rich and the prop gain can't create a cross count. This is due to the volume of fuel being delivered not being lowered enough.
The prop gain works directly on the injector PW.
So without a cross count the INT delay times out and the INT is decremented by one.
Again without a cross count the INT delay times out and the INT is decremented by one.
Again without a cross count the INT delay times out and the INT is decremented by one.
And so on until a cross count finally occurs.
Now the reported O2 is lean, and since the INT was decremented to cause the cross count, it has undershot the correct value.
So without a cross count the INT delay times out and the INT is incremented by one.
Again without a cross count the INT delay times out and the INT is incremented by one.
And so on until a cross count finally occurs.
And we are back to where we started, rich, with the INT delay timing out and the INT now being decremented.
Note that this under and over shooting INT can also drag the BLM along with it.
Thanks for the clarification RBob. My INT doesn't creep up and down in a pattern so I don't think it is forcing the cross count as you are describing. I tried reducing my INT R/L mean, high and low tables to stock PT4 values and it's very lean. Strange think it continues to pull fuel even when NB is lean and WB is very lean. I will get some screen shots posted later tonight to show you what it's doing.
I know we were discussing the PROP gains and adjustment but it somehow seems like it's all related. When the INT R/L is high and stock PROP values I get a "M" shaped O2 graph, it doesn't just "move up" in the range as you've shown on your screen shots. I think it may be in the Prop gains and INT R/L area where I'm getting off.
Another factor somewhat related is the motor drive train combination. Obviously the motor is larger than stock with larger heads, injectors, intake, exhaust etc but also with the large CI motor, manual trans and 3.70 gear when I'm steady state cruising every conditions is different than stock. The RPM, throttle position and possibly more important the Map and Gms/Sec are different. So possibly some of the tables need to be reset that relate to Map and airflow. Cruise at say 50MPH is a light load compared to stock. Just not sure how to really dial it all in.
Last edited by antman89iroc; 12-22-2015 at 01:31 PM.
I ran across this post that has helped me out. (THANKS RBob!!!) https://www.thirdgen.org/forums/diy-...ml#post3808288
I first tried what was recommended by RBob, but I had some drifting in some of the lower map areas around idle. I tried a modification of adding +25 to the entire INT - Rich O2 (Upper) and subtracting 25 from INT - Lean O2 (Lower) and left the R/L the same. I did use the 2023 file tables.
# INT - Lean O2 (Lower)
# INT - Mean R/L
# INT - Rich O2 (Upper)
My prop gains and durations still need some work, but the above has helped reduce drift a great deal. My VE table is staying closer to my WB tune. I actually now have -VE points along with +VE on learns. Which I was only getting +VE.
INT delay vs airflow. Is this the transport delay between the cylinder and the O2 sensor? How does it come into play when tuning?
So when you change to full length headers, is there something you need to do with the the Closed Loop Fuel Integrator Delay vs Airflow table? The headers essentially doubles the distance from the cylinder head to the O2 sensor... so to the first order, would one want to double the values in the table?
So when you change to full length headers, is there something you need to do with the the Closed Loop Fuel Integrator Delay vs Airflow table? The headers essentially doubles the distance from the cylinder head to the O2 sensor... so to the first order, would one want to double the values in the table?
With the full length headers don't usually need to double the INT delay. The transport delay starts with the ECM making the PW calculation (with corrections). Then for the injectors to fire, next cylinder to fire, and so on.
Although, I've been known to double the INT delay(s) just to help settle/slow things down. Then speed it back up once the tune is closer.
Use the INT (integrator) value to see if the proportional gains are correct. If the O2 sensor has been moved further down stream it can also help to increase the INT update delay a little. Here are three screen grabs showing the O2 response. This is on a mostly stock engine with an _old_ O2 sensor (it should be replaced).
One is at idle, another at steady state speed, with the third showing what occurs when the throttle position is changed.
RBob.
I made the mistake of targeting crossing at the MPH shown, it should be at the RPM. Is this correct RBob? More air more switching.
So .5 cycle / sec at Idle and 2.5 cycles / sec at 2400RPM (.5Hz Idle, 2.5Hz at 2400RPM)
More by airflow, which is reported as gms/sec (grams per second) in the WUD. Both the duration and gain can be adjusted via the airflow (each has a table based on airflow).
Note that the NB O2 sensor also slows things down as it is a slow responding sensor. Probably can't get x-counts faster then 2.5 to 3 Hz.
More by airflow, which is reported as gms/sec (grams per second) in the WUD. Both the duration and gain can be adjusted via the airflow (each has a table based on airflow).
Note that the NB O2 sensor also slows things down as it is a slow responding sensor. Probably can't get x-counts faster then 2.5 to 3 Hz.
RBob.
I want to make sure that I am understanding things correctly. The frequency show in the attached is 2.5 to 3Hz of x-counts @ 2400 RPM?
I am trying to get the INT - Delay vs Airflow narrowed in on and was looking at the example bins. The 2023 file (94 -95 truck) is the type of exhaust where the 3 Wire O2 sensor is at the end of the Y-pipe just before the cat. The stock ECM is a BJYK and the delay was less than the 2023. I also looked at a 2011 file (90 truck) which is I think a 1 Wire O2 just a foot or so down from exhaust manifold and the delays were the same as the 2023. So, I was a little perplexed. What would be your recommendation to go with? My assumption is the 2023 values or should I go even more? Maybe something went wrong when looking at the BJYK values. I did switch my XDF file when view it.
On another note, my NB is still slowing creeping richer than WB. I have noticed on the 3001 file and reading the sticky that you have the R/L at 655, which I assume drives things leaner based on the smelly gas part of the sticky. Curious as to why you set this flat apposed to the curve on the base 2023 file for example. One of Grumpy last post noted that further down stream he said it resulting in about 30mv less, so if I were to bump my curve up, copied from 2023 file, about 30mv it should cause BLMs to learn and go leaner and match my WB. Am I thinking right?
Last edited by dbesko1; 01-01-2016 at 05:23 PM.
Reason: Add picture
So, I was a little perplexed. What would be your recommendation to go with? My assumption is the 2023 values or should I go even more? Maybe something went wrong when looking at the BJYK values. I did switch my XDF file when view it.
There is also the INT delay multiplier table to take into account. As for the values, give hte engine what it wants.
On another note, my NB is still slowing creeping richer than WB. I have noticed on the 3001 file and reading the sticky that you have the R/L at 655, which I assume drives things leaner based on the smelly gas part of the sticky. Curious as to why you set this flat apposed to the curve
IIRC, the 3001 BIN is mostly an open loop calibration.
Here are some recent logs and screen shots (thx Rbob) Take a look and let me know what you see. I included a sec of data for each screen shot. INT is stable for the screen shot.
The one at 2100 RPM looks to be on the slow side, while the one at 2200 RPM looks OK.
How does the engine run & sound?
RBob.
Overall it runs and sounds good. At 2000+ it doesn't surge but between 1600 to around 1900 it surges mostly at steady state cruise ~45map. Remember it's a stick shift car so at the lower rpm/speed range it is very lightly loaded. At any load above 50map it seems fine and as the rpm increases over 2000 it gets much better. I'm starting to conclude that it's mostly because of the cam overlap and that it probably about as good as it's going to get.
A different vehicle, this has the O2 windows increased to run richer then stoich. Can see how the O2 end points (high & low) are higher then normal. Not for a vehicle with cat-cons. The heated NB O2 sensor is about 6" past the collector of long tubes.
The slow O2 signal image is from not enough proportional gains. The INT is slowly rising and falling to create the cross-counts.
The other image has correct proportional gain.
RBob.
I think my prp gains are right, but I am still struggling to get the engine to run at 14.7AFR. My WB (Innovates, sorry I bought it before switching to EBL) is showing about 14.0AFR (likely right based on E-te$t$ with high HCs). My learns have been narrowing in on 128, but still some areas of increasing BLMs maxing out. I do notice that the BLM difference isn't always what the VE learn table reflects. Any insight into this: note I am running a VPFPR. I even noticed no VE learn change but BLM would read 138. In quote you mention shifting windows, exactly what factors. INT - Mean R/L, INT - Rich O2 and INT - Lean O2??? any others? Any recommendation on how much to shift?
A side question, will a WB learn likely burn up my cat-con since it will not stay rich and not be cross count switching?
09-24-2015, 12:28 PM
