Closed loop idle

[clint78z]

My accel dfi has been running better and better. I tried do closed loop idle and it is not working out so well. I get a surging effect, I am not too sure if it is my programming or mechanical/electrical issues. I made all cells around the idle fuel and timing equal, to avoid hunting. A weird thing is that the battery voltage is up and down from 13.5 to 14.7v. The O2 sensor is heated but some times it's voltage drops right off to .023V for a brief second.

 

Any suggestions of where to start snooping ??

[jgkurz]

What's your initial timing set at? I had a similar problem until I retarded my initial timing to around 15btdc. Your car may be different, but once I did this I could run in closed loop without hunting.

[pparaska]

Does the cam have alot of overlap? If it lopes, then it'll always run lean at idle in closed loop. That's what alpha-N is for (down low in the TPS range). Does the DFI allow that?

[Guest greimann]

Ditto what Pete said about a lumpy cam (even slightly lumpy) at idle. They like to idle way richer what an O2 sensor is capable of maintaining. One undocumented trick that the old Holley 4Di systems have that may or may not apply to your DFI, is to set the O2 correction to zero for the idle range in the map. It just goes open loop and runs off what ever you progam in the fuel curve. If you have Alpha-n mode or an open loop option, even better.

[clint78z]

Good suggestions from every one. Yes I have read that a high overlap cam can hinder idle quality. However mine is a stock L28et, the turbo cams are very mild as far as overlap. It pulls a solid 15"hg at idle, that should be more than enough. I can run open loop at idle, it runs decent like this a few little burbles. As soon as I start to lean out the base fuel map so the 02 sensor has zero trim, it surges. It runs decent for about 3 seconds then almost dies then back up again. One note is that to run stoich at idle the injectors require 1.2ms pulse on the 370cc injectors. Mabey the injectors are becoming unstable at this low pulse width, and are not spraying consistently??

 

I know a decent amount about datalogging but still learning, is the a set of variables I shoud be looking at ??

 

It has base pw, gross pw, voltage, 02 V, rpm, map ....

[pparaska]

One note is that to run stoich at idle the injectors require 1.2ms pulse on the 370cc injectors. Mabey the injectors are becoming unstable at this low pulse width, and are not spraying consistently??

 

You may be on to something there. I've heard that symptom with larger injectors / idle problems discussed hear and on the Megasquirt mail list/yahoo group.

[Guest jt240z]

For what it's worth, the L28ET engine with the stock cam does not like to idle at 14.7 AFR. The L28 is just not as effiecient as more modern engines are at low intake velocities.

[clint78z]

I got it to idle quite nicely at 1.4ms, not sure others get there cars to idle with big ole 550cc injectors. The bar says 13:1, but I take that with a grain of salt.

 

The car went on it's first road trip, it shocked me with 37miles/imp gallon, which =30mpg for US. On the way back I discover that the O2 sensor and Coolant temp sensor were giving intermittant readings. I think something is wrong with the wiring harness, might be why the previos owner sold it for cheap.

 

Most of my tuning is has been done on the highway double lane to be safe, I had to do quite a bit of smoothing on the map I got. It had several holes in it, this got rid of most of the bucking and chucking. Is this right the way I did it

 

In working with the lower half of the base fuel map I set closed loop and adjusted everything besides idle for zero trim by the O2 sensor feedback. IE if closed loop was turned off it should still run 14.7. Then I started to adjust the TPS enrichment to get rid of the momentary lean condition. Then I adjusted the map enrichment to take care of the hills when turbo spools up without throttle input.

 

At 7 psi on the highway it pulls like a frieght train, tuning is quite fun when everything is installed properly.

[TimZ]

One note is that to run stoich at idle the injectors require 1.2ms pulse on the 370cc injectors. Mabey the injectors are becoming unstable at this low pulse width' date=' and are not spraying consistently??[/quote']

 

You may be on to something there. I've heard that symptom with larger injectors / idle problems discussed hear and on the Megasquirt mail list/yahoo group.

 

Well, this is not quite right - 1.2ms is approximately the minimum pulsewidth that a low impedance injector can tolerate without becoming unstable. This means that you will need to have at least 1.2mS of pulsewidth at idle to have any hope of getting a stable idle.

 

Note that this dictates a minimum amount of fuel that you can inject into the cylinders at any time, and if this is still too rich, you will have a difficult time reducing the fuel delivery more. This is where jgkurz's suggestion of backing the idle timing off comes into play. All else equal, less timing = lower idle speed. So, if you reduce the base timing, the IAC will have to open more to compensate. The result is more air for your minimum amount of fuel (i.e., it gets leaner).

 

Using this method, I was able to get stoich idle (well, ~14.4) using 72lb injectors and a mild turbo cam (Isky L475). My base timing was down to about 5 degress, btw.

[BLKMGK]

FWIW - I idle low impedance 72lbinjectors down to 1ms with no problems, one part of my map even goes to .75ms. Some ECUs cannot idle down this low, it will depend on the resolution of the ECU. You can fiddle fuel pressure to get idle PW up if you have headroom up top under boost.

 

Some cars also just hate to idle at stoich and will idle better richer. I can idle my 2JZ as lean as 15.5:1 but it begins to surge - this is with 20degrees of timing. Lowering timing certainly helps and mine is actually higher than stock. What are you using to measure A/F BTW? If it's not a UEGO aka W/B then don't trust it to be accurate except at Stoich.

[clint78z]

Well last thursday I was reading a ton of suggestions TimZ had in the archives of Vishnu forum for the Tec II I learned so much. After I lowered the timing and tried to lean it out. I found one problem I had was the closed loop hysterisis lower limit was 650 and high was 1800rpm DOH !!!. I set it to 750rpm on the top. I was able to get it to idle closed loop not too bad.

 

I took it back on the highway and noticed that in closed loop cruising it was too off and on like a water tap changing from 1.9ms to 2.7ms too jerky in my opinion. So I turned closed loop off and it ran smooth at 1.9ms and got better vacuum.

 

I have to track wiring issues since my voltage varies up and down almost one volt. I will let you all know how it goes.

[TimZ]

I took it back on the highway and noticed that in closed loop cruising it was too off and on like a water tap changing from 1.9ms to 2.7ms too jerky in my opinion. So I turned closed loop off and it ran smooth at 1.9ms and got better vacuum.

 

I have to track wiring issues since my voltage varies up and down almost one volt. I will let you all know how it goes.

 

You really don't want to rely on the EGO feedback to fix big holes in your fuel delivery curve (I assume the DFI has a VE table or something equivalent, right?). I always leave the feedback off until the VE table is in good shape. The EGO feedback simply cannot respond quickly enough to compensate for big swings as you go through incorrectly tuned cells in the table. The result is pretty much what you described - inappropriate swings in the injector pulsewidth and bucking and spitting.

 

On the voltage, you might want to check the connection at the regulator - I am running a 140A GM alternator, and found that the connection to the output post have come loose internally and was making intermittant contact, and this gave similar results to what you describe.

 

I agree with blkmgk that unless you have a wideband sensor, you shouldn't rely on the readings as they move away from stoich. BUT, I think it's pretty safe to say that if the narrowband sensor says it's rich it really is rich and if it says it's lean it really is lean. The inaccurate part is in the "how much" it says you are rich or lean.

[clint78z]

The DFI simply has a base fuel map, no VE Table. It is a 16x16 table load vs rpm and you punch in a pw injector value. It has TPS and Map rate on top of that which I set back to zero to tune the table. What is funny is my cells in the cruising range are fairly tight to the 14.7, the only number I can be sure of. The surronding cells are have no big holes in them. It goes from 1.8 to 2.0 ms around cruising range. Then when I turn on the closed loop correction it's going wild pulling it back to 1.4ms then compensating to 2.7ms This got me thinking that mabey it's a dirty O2 signal, poor ground possibly.

 

I did clean my battery terminal this weekend, and noticed the +ve cable was attached to the starter had a stripped nut. I swapped in a newer type starter and made sure my it had a good clean connection. Do you recomend ground the DFI to engine block or battery -ve post ?? I am also looking for a good corrosion cleaner mabey even acid to clean connections of my fuse box. It got nailed by soda from previous paint job, sand paper is not getting the job done.

 

I was about to check regulator, it has the non solid state type external. However the battery died in the multimeter.

 

I agree that wideband is the only way to go for tuning any ratio besides 14.7 to 1. That is why I am leaving the boost set at 7 psi until I have acess to a dyno. I want to get all the bugs I can see fixed before I turn up the wick.

 

PS TimZ, I am beginning to really like the Tec II the more I understand it. The instructions could be a 1000 times better, once you get your head around it has some really nice stuff. Alas I could not find one for $350 like my DFI, which is good value.

[TimZ]

The DFI simply has a base fuel map, no VE Table. It is a 16x16 table load vs rpm and you punch in a pw injector value. It has TPS and Map rate on top of that which I set back to zero to tune the table. What is funny is my cells in the cruising range are fairly tight to the 14.7, the only number I can be sure of. The surronding cells are have no big holes in them. It goes from 1.8 to 2.0 ms around cruising range. Then when I turn on the closed loop correction it's going wild pulling it back to 1.4ms then compensating to 2.7ms This got me thinking that mabey it's a dirty O2 signal, poor ground possibly.

 

Your base fuel map is pretty much the same thing at the TECII's VE table. I wouldn't worry so much about discontinuities in the table as I would in the actual mixture. Contrary to popular belief, there is no law that states that your map has to look smooth. It's not uncommon to have weird resonances in the intake/fuel system that make a properly tuned map look pretty funky.

 

Log some data that shows a trouble spot. At a minimum, collect MAP (I'm assuming speed/density), RPM, measured AFR (don't worry too much about whether you have wideband or not), and EGO feedback corrections (if you have it). You are probably going to find that right before the feedback goes insane, you have a rich or lean spot. First step is to figure out why - could be bad base map, could be bad acceleration enrichments, could be something else.

 

One thing that I have found helpful is to pull your data into Excel, and sort by let's say RPM. You can then make a scatter plot of AFR vs MAP for each RPM range in your base fuel map (you could swap RPM and MAP in this example if you like). From this it's pretty easy to see where the fuel map needs to be adjusted. Again, I wouldn't worry too much about not having a wideband - you can still do quite a bit with a narrowband sensor. If it says you are rich, you most likely really are rich - it's just not very accurate about exactly how rich you are.

 

Do you recomend ground the DFI to engine block or battery -ve post ??

 

Well, you do need to have a good connection to battery ground, but I would recommend taking everything to the engine block for ground. Preferably to the same point on the engine block.

[clint78z]

Thanx so much for everyone helping me, support is quite poor for the DFI surpising since it has been on sooo many Chev 350's.

 

The datalogging on the DFI is something to be desired, it's only text based but I can deal with that. The main downfall I hear is poor resolution when datalogging over 4 variables at a time.

 

I do understand that the amount of pulsewidth is based on many factors. Intake and exhaust tuning effects, cam profile .... It should generally follow the torque output line. I smoothed everything out because there were too many holes that made trying to tune the cells impossible. I have tuned a bottom half of base fuel map running closed loop, which shows the AFR bar. Example I ran closed loop cruising on the highway and made the all the numbers in the cruising cells 14.7 AFR with zero O2 trim. Above that I was trying to set to run 13:1 (calculated by DFI) but still in closed loop for brief periods of time, the )2 sensor was trimming back my values.

 

Thank you for the excel suggestions, I have done a bit with this before but still a rookie. A quirk with Calmap is that it only displays AFR when in closed loop on a bar in the base fuel table. I can't datalog it what a pain in the butt.

 

Options for datalogging this problem I see

rpm, map, O2 correction, O2 voltage, additive pw, base pw, mult corr, gross pw.

 

Can I use the O2 correction for that scatter graph instead of AFR you mentioned ??

[pparaska]

I don't think you'd want to depend on any O2, narrow or WB for high boost turbo apps. I'd be running closed loop, datalogging, and looking at EGT, O2, spark timing, etc., starting rich and sneaking up on leaning it carefully.

 

I'd make sure all the grounds came to a common point, probably on the block, or at least tied with large gage wire to the block. Run as large a gage as you think the power circuit needs on the +ve side. Run a nice big ground to the exhaust pipe near the O2 sensor to the same place you're grounding the ECU to. That way you will stay away from different potentials for several grounds that might cause the O2 reading at the ECU to be off.

[clint78z]

All grounds for injectors, O2 sensor, ECU, Fuel pump collect from the DFI from factory DFI wires to a gold ring terminal to stop corrosion that and it's not flimsy, and bolt connects 12 gauge stereo wire directly to starter ground on block.

 

I have done what you suggested Pete start rich then carefully lean it out in small steps. The stock O2 sensor isn't fast acting, nor can it be trusted to tell any info besides stoich, rich, or lean. I still think it is enough to get a fair amount of info to tune my engine, if I take things carefulley.

[pparaska]

clint, sounds like you're more anal retentive than I am about electrical stuff! Gold ring terminals! and the setup sounds great.

 

I was really referring to the reaction speed of the O2s. Not nearly quick enough to have the ECU prevent a lean condition when rpm is rising at 1000 rpm/0.1second rates, as you alluded to. That's why I said datalogging and starting out rich.

 

I will be asking alot of questions once my EFI is up and running on my

pushrod NA V8. Right now the Pro-jection 4D is half-way installed, and the Megasquirt is built. I've NEVER done EFI tuning, so I'll have lots of questions for TimZ, clint, etc.!

[TimZ]

I don't think you'd want to depend on any O2, narrow or WB for high boost turbo apps. I'd be running closed loop, datalogging, and looking at EGT, O2, spark timing, etc., starting rich and sneaking up on leaning it carefully.

 

...you meant open loop, right? I would not recommend tuning on boost with untuned closed loop feedback.

 

EGT is a good tool to use, but it's response is much slower than either type of O2 sensor, and you can't datalog it (at least not on the TEC - I'm assuming the DFI is the same way). So even if you see a funky reading, it will be very difficult to correlate that back to which cells need adjustment, especially for fuel delivery (timing does generally need to be smoothed, so you can make broader changes there) .

 

Just like everything else, you need to apply some common sense while trying to tune on boost. Find a safe place to tune (hopefully this is obvious), and you will need to be able to limit the rate of change for engine RPM and MAP. This pretty much means use higher gears (downside = illegal speeds), or in many cases you can drag your brakes against the throttle to go through cells that you are interested. If you are dragging the brakes, be sure to let them cool between runs. Also, try to always start the run at the same intake air temp, assuming you can see this. If you can't, then it is also a good idea to let the underhood temps stabilize between runs (part throttle cruise to keep air moving).

 

Like Pete said, start rich and sneak up on it, and always listen for anything unusual (misfires, knock, etc).

[BLKMGK]

Couple of things - #1) I DO use a W/B for high load regions and I even use it closed loop. However the AEM allows me to set how much correction I'll allow. I only allow it to remove a limited amount of fuel but I allow it to ADD a decent amount of fuel to try and prevent a lean condition. When you have this kind of flexability you can run the W/B under boost.

 

#2) I've seen the behaviour that was mentioned here - the oscillation. The AEM that I use allows me to setup it's reaction time and how much reaction it will make to try and adjust things. An oscillation like this is uaually caused by the O2 sensor being too far or too close to the exhaust valve. It actually takes a measurable time for the exhaust gas to travel to the sensor (remember how fast a computer can calculate). It's possible for the sensor to react too slowly or quickly and to overundershoot. This will setup a nasty oscillation - it's happened on my car the AEM board has the calcs to figure out how to fix this for that computer, IF the DFI allows you to make those adjustments then you shoudl be able to find out how or maybe use the AEM calcs.

 

#3) Remember that a miss will be seen as lean to the W/B. Since it measures oxygen a miss will show up as a spike in oxygen and look lean. This can cause the ECU to add fuel and it can get ugly. Additional fuel will possibly spike EGT or it might lower it depending on timing.

 

#4) EGT is indeed too slow IMO although I've used an RTD sensor that seemed to react pretty quickly. Unfortunatly it seems I've fried it An EGT can help you tune timing but for A/F I don't like it much - I prefer the W/B. I like EGT for trying to figure out if I'm retarded or too advanced. If you're retarded EGT will be high even if A/F looks okay - burning fuel exits the exhaust still burning. The oxygen can suposedly still burn before hitting the O2. I guess my point is that you need the EGT and W/B together to work best.

 

#5) Timing is bigtime important to make power. What you want is for cylinder pressure to remain constant and high in the cylinder. Too high and it will knock, too low and you don't make power. Timing map will look like wave I'm told. High timing that tapers off at TORQUE peak and then as that passes advances again. You fire earlier to raise cylinder pressure, you fire later to lower it. To do this "right" you need a cylinder pressure sensor, big bux. These can either be drilled into the cylinder or built into a sparkplug and is the best way to setup timing. A few degrees of timing on a friends car made BIG differences in his trap speeds. Always remember that timing changes A/F too. If you advance timing, don't see knock, and A/F goes leaner then I'm told that the ignition is closer to being optimized and that the burn is more effecient. Always watch the A/F as you advance timing becasue you can easily go lean! I've seen a few degrees of timing on my Supra under boost move A/F nearly a full point. If you push timing too far it'll knock, too late and it'll be sluggish and bump EGT. Retarded timing can speed turbo spool but it can be hard on the turbo...

 

Anyway, I hope that helps some. This is all stuff I've learned trying to better tune my car and I think it's pretty accurate unless I've misinterpeted it fromthe tuners who've tried to teach me

 

P.S. I've found that ignition energy can make a big difference on my car. I went from weak coils to strong ones and I can now idle a full point leaner before it will begin to get unstable and I can actually have the car pull some as lean as 16.5:1 where before it would pop and sputter! Iridium plugs DO make a difference BTW and they last awhile if you're willing to regap them after they've worn a little. 8)