Mastiff

I just finished converting from carb to TBI using a kit from CutomEFIs.com (factory GM TBI). Well, everything is installed, now I'm trying to get it to run properly. The guy is trying to help me out, but it looks like he doesn't work weekends or something, so maybe you guys can help.

I have a mild 350 engine, with headers and a performer intake. The TBI kit came with an adapter to the 4-barrel manifold. I have no EGR anymore.

The engine fires right up and idles decent in open loop mode, then when it goes closed loop the RPM oscillates up and down within a 200 RPM range for only about 30 seconds, then throws code 44 (lean O2) and goes open loop. Even after it throws the code, the idle still goes up and down, just not as much.

The thing has never successfully run in closed loop, so I have no BLM adjustments, I'm just raw PROM which may not be very close for all I know. Something seems to be up with the O2 sensor, since the WinALDL shows it going between 0.000 and 0.004 volts all the time. I think that's pretty much way low? It's a heated O2, and we traced the wires so the center is getting +12, pin C traces all the way back to the right wire at the ECM and pin A is grounded (this is a brand new Bosch 3-wire heated O2). I don't know for sure what the wire definitions are for the O2 sensor, but the guy who sold me the kit had me trace this stuff out and thinks it's correct.

Oh, and if I disconnect the O2 altogether, WinALDL says something like 0.45 volts.

Any ideas what might be going on? It seems like I'm very close to getting this thing working.

dimented24x7

iTrader: (2)

Check for exaust leaks around or before teh O2. If you have AIR, try disconnecting it to see if the problem improves. AIR itself can really push the O2s down while its routing air to the manifolds. Very low O2 voltage can also be caused by a leak to ground (i think) so check the wiring. As a check for this, the O2 should be around .45 volts untill the sensor warms up to around 600 deg F . If its near zero volts as soon as you turn the key with everything completly cold then there is a problem with the circuit.

Mastiff

Thanks for the input. At this point, I believe that the O2 is working fine. I pulled it out and tested it with a propane torch and it went right up to 0.8 or something like that. Also, when it's in the vehicle, I can pull the vacuum line to the MAP and see the O2 readings go right up to very rich, like 0.7-8.

So, for some reason, the computer is deciding just not to give it enough gas in closed loop to make the O2 happy. I was told that the ECM won't deviate more than 20% from its baseline programming? If I have an exhaust leak or fuel delivery problem, it may be asking the computer for insane amounts of gas, so it just gives up and flags a bad sensor?

The guy I bought the TBI kit from suspects that I'm not getting enough fuel pressure because either the pump is too far from the tank, or there is a restriction someplace. Maybe this is possible, but this is the factory original fuel lines and I never had any trouble before, how do you just develop a restirction in the tank?

I'm also skeptical about the pump mounting. I have it on the lower frame rail. I could move it back about 2 feet and still be on the lower part, to go further back I'd have to go up above the tank, which I think is bad.

I have one theory, that maybe my tank venting is poor. I'm venting up into the engine compartment (where there used to be a charcoal cannister). Maybe I'm making a vacuum in the tank that the pump is having to fight. I'll test this idea by pulling the filler cap.

Unfortunately, I'm waiting on a new injector pod since I cracked mine over torquing the gas outlet feed.

Dewey316

since you seem to be getting fuel with the MAP unplugged, that would leave me to beleive that your pump is ok (at least for idle).

what are you other sensor readings while in open-loop? specifily your MAP.

Mastiff

That's what I thought. I was told that the ECM won't deviate more than 20% from the nominal fuel delivery based pulse width, etc. If the fuel pressure is low, the ECM would have to increase the pulse width by a lot to satisfy the O2 sensor, but it doesn't know the fuel pressure is low, so it thinks it's dumping in tons of gas. The 20% thing would be a protection from pouring gas into the engine when the O2 sensor is busted.

I'll have to look when I get home. I can't ever remember the MAP reading because my mind doesn't work in kpa and WinALDL won't convert to inches for me. I think maybe 40?

Dewey316

40 Kpa isn't bad for idle, but it might be off enough to cuase idle problems. on a 'Stock' engine, that is partail load, idle is more like 20-30 Kpa.

1981TTA

Mastiff,

You should also keep an eye on the commanded injector pulse time/% and commanded AFR as the ECM tries to go to closed loop. It might be that the ECM is simply commanding a richer AFR during the first 30 seconds of operation when it idles well. Then, as the AFR moves to 14.7, there's not enough fuel due to poor VE information? (I.e. when the ECM commands 12:1 at first, you're really getting the equivalent of 15:1. When it goes to 14.7:1, you're really getting 16:1.) Do you see "better" O2 readings during this first 30 seconds of operation before closed loop?

Did you get the TBI injectors from the company? Or, did you supply them yourself? Either way, have you confirmed the injectors match the ECM calibration? If the ECM is calibrated thinking the injectors flow more than they actually do (i.e. smaller injectors are installed), it would cause a super-lean condition like you're experiencing. An abnormally low fuel pressure could cause this as well. That's probably why customer support is checking this out. I've heard of a neat trick you can try to look at the injector spray patterns. Hook up a timing light and point it inside the TBI. You should see a cone-shaped spray pattern. If things are clogged or at a low pressure for some other reason, the shape shouldn't be as pretty.

Does the engine exhibit the same speed oscillation if you step on the throttle a little? I'm thinking just enough (1-2%) to get out of any idle speed control algorithms would be good. During idle, the IAC will usually react to engine speed changes in an effort to maintain a setpoint speed. If possible, look to see if IAC counts are oscillating in time with the engine speed changes. There are a number of calibration parameters that will affect this. Oh, and I probably should have mentioned this first. But, have you reset the IAC to make sure it's really where the ECM thinks it is?

Mastiff

I do see better O2 readings at first, but it gets worse before it tries closed loop. It fails closed loop quickly and remains at about 0 on the O2 readings (0.000 or 0.004 volts).

If you're up to looking at it, here's some data I recorded from WinALDL. The motor was warm, but I cleared all the codes and started fresh:

http://www.taildragger.info/K5/TBI/Lean_O2.txt

I am trusting the manufacturer to have matched my injectors and the PROM and all that. It all came as a big kit. I'll try your timing light idea when the throttle body comes back. The pattern looked kind of sloppy to me, but this is the first time I've ever seen one, so I don't really know what I'm looking for.

I haven't tried this, but I will.

I followed the instructions. What I was told to do was to short ALDL A and B and turn the key on, but not start the engine for 2 minutes. Then pull the IAC connector, start the motor and set the idle screw to a good idle speed (no specific number given, I picked 750 or so in park). Then shut off the motor and reconnect the IAC connector. Sound right?

Do you think it's important to use an exact idle number that corresponds to how the chip is programmed?

Thanks for the help.

dimented24x7

iTrader: (2)

As for the IAC. No need to perform reset procedures or any other sort of stuff as the ecm automatically sweeps teh IAC to reset it to parking the IAC every time it shuts down.

dimented24x7

iTrader: (2)

It looks like it might be a very bad vacuum leak causing a lean misfire on one or more of the odd number cylinders (or all of them for that matter). The ecm is commanding the IAC to close all the way in an attempt to obtain the desired engine speed. This would be a good indicator of a vacuum leak, or the min throttle position being too great. A misfire will send lots of O2 into the exaust that will show up as low O2 volts. It could also be just going really lean but check for vacuum leaks. Another thing is that your knock counter is just free running. Are the module and knock sensor hooked up properly?

1981TTA

I didn't know the 7747 had this capability. (And, I ASSumed that's the ECM Mastiff is using...) I know I had a 7747 that didn't do this in *one* application I'd seen. But, that could have been a calibration/application issue, too. I know "later" ECMs like the 730 and 749 will do this at shutdown.

Either way, the procedure Mastiff followed is correct. So, the I's are dotted and T's are crossed.....

I don't know if this is enough to determine a vacuum leak given the unknown of what the desired idle speed is calibrated within the ECM. The procedure to set the idle speed is done with the IAC completely closed via the ALDL pin shorting. So, it would stand to reason (again ASSuming the ECM's idle RPM = set idle RPM) that the conditions would be replicated. So, when conditions are identical to what was present when you set the idle speed, you should see the same actuator commands (i.e. IAC @ 0 steps). From the log file, it appears engine speed is staying right around 750 RPM. (IAC actually blips up to 1 step when RPM = 675.) That being said, I do agree that IAC moving toward 0 *can* indicate a vacuum leak. I just don't think we can use this test in this case conclusively.

Given the way the IAC behaves relative to engine speed, I'd guess the ECM's desired Idle speed is close to 750. The ECM will change Idle speed under different coolant conditions, for example. So, it's impossible to mechanically adjust the TBI for all the speeds that might be desired. Effectively, you adjust the TBI for the lowest expected idle speed. From there, the ECM can always use the IAC to add airflow and raise RPM. However, it has a tough time using the IAC to "remove" airflow!!

I didn't think to ask this before. But, I'm guessing the engine ran better when you disconnected the MAP and richened up the mixture? I'm also ASSuming there aren't/weren't any backfires/popping/spark timing-type problems that would make you believe you're getting misfires when the O2 is reading higher than 0V? If this isn't the case, some of what I'm going to talk about below might not be the first thing to try......

It's kind of hard to tell whether the O2 sensor is really seeing a higher voltage at any time during startup in the log file. I believe the O2 value is initialized at 0.450V in the ECM. (Probably why you saw that number when you disconnected the sensor in your first post.) Then a filter is applied to the actual O2 reading. Seeing the O2 reading move straight down would lead me to believe the O2 sensor is always reading ~0V.

The ECM will usually command a richer AFR at colder engine temperatures. Can you run a similar log file with the engine cooled off? (I.e. first start in the morning or right after work.) If you have a heated O2 sensor, try keying up without starting the engine for a minute. This should help the O2 begin to warm up in preparation for engine start.

The other potential thing to try would be getting another coolant sensor (from a yard or parts store). Leave the existing sensor in the warmed up engine and connect the "spare". It should read whatever the ambient air temperature is. While that might feel "warm" to us, in engine terms, that's cold. The ECM should ask for a richer fuel mixture. You should then be able to see if there's a response in engine behavior and/or O2 sensor. You can use an ice cube or propane torch to vary the temperature to see how (and if) the engine responds to the temperature change. This isn't a permanent fix. But, if you can tell the vendor that the O2 reads ~ 0.5 V when the coolant sensor = X degF, that might let them know how much adjustment needs to be done. In general (but not always and not always linearly) colder = richer. If the O2 still reads low with very low coolant temps, we'll have to come up with something else.....

I'm again ASSuming you aren't burning your own chips for this. If you are, there are some calibration values that can be changed to bump up the injector pulse width and/or AFR a little easier than what I describe above. Even if you're not doing this now, I would encourage you to start. There are bound to be other things you'll want to tune for your engine after you get these initial problems worked out. The oscillating idle comes immediately to mind.

Dewey316

i had to try very hard (emmisions) to get my o2 values below .1v, would guess you have somthing going on other than just a 'bad tune'

the IAC steps at 0 are of a concern. FWIW, i set mine to idle at about 450 with the IAC disconnected and in its park position. i let the IAC system control idle from there (stock chip idle is about 600.) haveing the o2 voltage that low is a real concern. have you tried another o2 sensor? also, spell the exaust. does it spell like it is full of unburt fuel? if you were 'really' running that lean, you would not be able to smell any gas smell in the exaust. if it is a false lean, you should smell the exaust.

i would also let it idle like that, then pull a plug, look at the condition of it. what really needs to be determined here, is if you are lean, or not.

Mastiff

I did do a lot of checking for vacuum leaks. I was told that vacuum leaks generally will not cause a lean condition because it will show up at the O2, and then the ECM will dump in more gas to compensate. So in the end, it just looks like the throttle is open.

I definitely see your point about the ECM trying to shut off the air with the IAC though.

Mastiff

Yes, the idle was pretty smooth with the MAP disconnected, and the O2 reading was high. I took this log file with the MAP unplugged (and the vacuum line plugged).

http://www.taildragger.info/K5/TBI/MAP_pulled.txt

Well, it was smooth for a little while. It started running a little rough at the end and then just stalled. It seems like it could be consistent with a super rich condition. In this case, I had the MAP off from the time I started the motor. If it's in one of those lean/osciallation cases and I pull the MAP, it smooths up (like it's happy to get more gas).

I could believe that.

I can when I get my TBI unit back. I had to replace the injector pod because I cracked it overtightening the fuel outlet fitting.

Isn't the unplugged MAP giving similar info? I think it's saying that the ECM can put a lot of gas in when the MAP is basically indicating wide open throttle.

I'm not currently burning my own chips. I'm sure I'll get assimilated sooner rather than later though...

1981TTA

You're absolutely right. I didn't know if the engine ran as well as you describe with the MAP disconnected. I thought it just dumped in a bunch of fuel long enough to register on the O2 sensor and died. My approach with the coolant sensor was to allow you to *slowly* enrich/enlean the mixture via temperature change and keep the engine running. However, from what you've described and what I can tell, you've conclusively determined you're running lean. I don't know if congratulations are in order or not.... You're definitely headed in the right direction looking at fuel delivery. While the TBI is out, are they also checking the pressure regulator? It might be that you have enough pressure from the pump. But, the regulator isn't working properly....

I think you'll be happy over "here" in the chip burning world. Lots of toys to play with!

1981TTA

Dewey,

You have to remember that Mastiff set the idle @0 IAC to be approximately equal to the desired idle speed in the chip. In his case, it's perfectly normal to have IAC move to 0 at idle. It's not an indication of a vacuum leak *in that case*.

Given the way you set your idle with IAC @ 0, you would be concerned with vacuum leakes with the IAC reaching 0. On your engine, if RPM = 450 when IAC = 0, it stands to reason the IAC has to be greater than zero to maintain 600 RPM. If it's considerably below that, the engine is getting extra air from somewhere else.

Again, Mastiff *could* be experiencing a vacuum leak (but I don't think so). It's just that we can't conclude that based on 0 IAC counts given the way his throttle blades were set at idle. I'm probably beating a dead horse here. But, I want to make sure nobody reads this, logs data from their car with 0 IAC steps at idle and immediately concludes they have a vacuum leak without considering how the idle was set....

1981TTA

You're absolutely right. I didn't know if the engine ran as well as you describe with the MAP disconnected. I thought it just dumped in a bunch of fuel long enough to register on the O2 sensor and died. My approach with the coolant sensor was to allow you to *slowly* enrich/enlean the mixture via temperature change and keep the engine running. However, from what you've described and what I can tell, you've conclusively determined you're running lean. I don't know if congratulations are in order or not.... You're definitely headed in the right direction looking at fuel delivery. While the TBI is out, are they also checking the pressure regulator? It might be that you have enough pressure from the pump. But, the regulator isn't working properly....

I think you'll be happy over "here" in the chip burning world. Lots of toys to play with!

Mastiff

Can you point me to some info on chip reprogramming? I'm not aware of what kinds of software and hardware tools are available. Is there software available that lets you modify the settings in convenient units and such, or do you have to hack the binary directly?

By the way, can someone explain why it's beneficial to have the fuel delivery table dialed in perfectly as opposed to just getting it close and letting the BLM pick up the slack? There must be some reason, but on the surface it seems like once the BLM is done learning, it should be the same thing.

Dewey316

It doesn't use BL corrections under PE modes. or before the engine warms up.

1981TTA

If you take a look over on the DIY-PROM forum, I think your current questions (and questions you don't even know you have yet) will be answered. Start with the "stickies" at the top of the page. There are a number of people making hardware and software that make changing calibrations a snap. As long as you're using a "common" ECM, you'll probably find everything you need already available. The nice thing with more people doing chip burning is that the list of "common" stuff grows!

This is true in closed loop as long as the computer is able to compensate. However, when you're not in closed loop, the "learnings" aren't always applied. At a minimum, they're applied differently in different ECMs and software sets. So, if your open-loop calibration (without correction) will cause a lean condition, you'll see actual air-fuel ratios less than what you calibrate. When you run into situations similar, but not as severe, as what you're seeing, you won't know exactly why you're not getting the AFR you commanded. Is it because the ECM is wrong? Or, is it because fuel pressure is wrong? Or, due to injectors clogging..... The list goes on and on. If you "know" your ECM is correct, the list of possible problems is reduced.

Consider the times when the ECM isn't in closed loop like during "cold" idle and WOT operation. Without reasonably accurate fueling, you'll probably have problems in that area. Like so many things, this depends on your tolerance for engine/vehicle behavior. If you can live with having the engine stall a few times or idle rough at startup when it's cold out. Or, keeping your foot on the throttle until the engine warms up under similar conditions. You may be able to live within the ECM's learning window. I'm sure there are a number of people that do this every day!

Mastiff

Thanks for the help. I may come back for more input when I get the system put back together. I've got some good ideas to go with now though.

dimented24x7

iTrader: (2)

Generally large vacuum leaks around the TBI have the same effect as opening the throttle. Vacuum leaks near an intake runner, like at the base of the intake, will allow lots of air to enter directly into the cylinder. Ive had it happen before, and in large enough ammounts itll cause a lean misfire. From your MAP readings, it looks unlikely, but still worthwhile looking at if youve exausted all other possibilities.

As for the computer... Did it come with a stock prom? If someones had their hands in the calibration there is alot of stuff that can go wrong. Another thing, whats the inline pump rated at? It has to be at least around 10 psi at all times in order to fuel the motor properly.

When you disconnect the MAP, the fueling is then done by creating a pseudo-MAP value from the tps and rpms. Not the greatest but good enough to get home if there is a problem.

Oh, and as dewey said, smell the exaust. Itll tell you alot about how the motor is running.

Mastiff

I got my throttle body and am back trying to debug. Unfortunately it's not just a tank vent problem like I hoped.

I did manage to get it into closed loop mode. I'm not sure why it worked this way, but I lowered my idle to around 600 and it has started going closed loop, although it still surges some and the data looks bad.

It'd be great if some of you guys could look at this and tell me what you think. This data set is a warm engine from startup. At the end I hit the throttle a little to get some dat at different RPM points (it's an log file from WinALDL):

http://www.taildragger.info/K5/TBI/Closed.txt

The other thing I noticed is that the throttle is extremely sticky when the engine is running. When the engine is off, it feels totally smooth. Any idea on that?

Thanks for any help.

dimented24x7

iTrader: (2)

From the datalogs, it looks like its really short on fuel. What pressure are you running at? Also the knock circuit apears to be open. Do you have a module and knock sensor hooked up?

The rough action of the throttle is typical of old tbis. Its due to wear combined with the force from the pressure difference across the tbi when the motor is at idle.

Mastiff

Thanks for the help:

My pressure gauge reads 13-15 PSI. The chip was programmed assuming 17 PSI according to the manufacturer (customEFIs.com). I could see that the BLM was big, I think this means that the computer was having to put a lot of extra fuel in compared to the way it was programmed? What do you think explains the way the O2 readings are going up and down instead of leveling out though?

Do you think a vacuum leak at the manifold could explain this?

I paid good money for this system, I sure hope it's not showing wear already. Is it typical of new TBI's, or only worn out ones?

Dewey316

actualy, your o2 values look good. in closed loop, the o2 values 'swing' back and forth, it appears that yours is doing just that. the only time the o2 values will stay steady, is cold idle (open loop) or WOT (open loop).

dimented24x7

iTrader: (2)

If its one of the larger tbis it can happen as well. Its mainly due to friction from the force of atmospheric pressure. On a 2 in tbi there will be around 60 lbs of force on the throttle blades, which can mean that the shaft will really resist turning. Cant lube it, either, as the shaft is in contact with gas when the car is running.

The fact that it was set up for 17 psi will mean that the computer will think that the injectors will flow more then they really do at 15 psi. That will throw the fueling off even more then it already would be. Also look into the knock sensor stuff. You want to get that fixed before you use the car or itll run like complete crap and cause codes to be set. Best bet is to get a budget chip burner and start tuning. From the looks of it the tune they gave you is way off. If its a stock ecm then you could try a stock prom from maybe a truck, car, or whatever uses the same ecm you have that is close to your present setup. Might be a better starting point.

dimented24x7

iTrader: (2)

I took a quick look and it looks like you most likely got a 7747 truck ecm with your kit. You could get a prom from an L05 and try that.

Mastiff

Oh, I didn't know that. What causes them to swing? If it stays constant in open loop, it must be that the computer is fluctuating the fuel delivery? Why would it do that?

Mastiff

This is a GM TBI, but bored over. I really need to do something about it though because I would be lurching all over the place. A large part of why I (thought) I wanted to go away from carb was to get smoother throttle response. At idle, part of my problem might be the the idle is set low and the butterflies are "wedging" into the throat. It stuck on me even when I was holding 1200 RPM though...

Yeah. I'm using an external fuel pump. Do you think that the pump and fuel line/routing I installed could cause the lower PSI at idle, or is it more likely just variation in the fuel pressure regulator, or possibly my pressure gauge? It's hard for me to imagine that the pump and lines would prevent higher PSI at idle when very little fuel is flowing out the injectors. I can see a lot of fuel flowing through the clear fuel filter, so it seems like the regulator could neck down more to get more pressure if it wasn't what it wanted.

I keep forgetting to say that the knock sensor stuff is disabled. I don't know why it gives knock counts, but there is no connection to the knock computer and there is no knock sensor on the block.

The guy owes me a decently functional chip as part of the deal, but maybe I'll get sick of iterating through him and start doing it myself. It seems so incredibly far off that I'm worried about installation problems, like either vacuum leaks or fuel delivery problems.

Dewey316

It does that for emmisions, the cat needs the fuel/air/fuel/air swing like that to function most effiently. there are of course other reasons also, like the way the o2 sensor works, ect.

dimented24x7

iTrader: (2)

The butterflies shouldnt stick in the bores... Definatly sounds like he didnt do anything about the wear around the throttle shaft. Defintaly worth complaining about that since the sticking can get incredibly annoying. He most likely used a stock base and just punched it out.

IIRC, the ecm will still retard teh timing whenever knock counts are present. There is a way to stop it from free running like that. I think maybe a 3.9k resister from the input to ground? Dont know for sure. Post over on the DFI/ECM board. Im sure someone will know over there.

Itll be hard to get a good chip from him since he doesnt have access to the vehicle. Plus, there is no way of knowing how much experience he has. For now try sending him aldl logs and descriptipions of how it runs and take it from there.