legacycentral bbs

I'm still working on this no-start issue. I checked the coil. Tried replacing the ignitor. It cranks but isn't firing.

The check engine light is on so I decided to check the codes...

Code 12
Abnormal signal emitted from starter switch
Code 21
Abnormal signal from water temperature sensor
Code 23
Abnormal voltage input entered from air flow sensor
Code 24
Air control valve inoperative
Code 33
Abnormal voltage input entered from vehicle speed sensor
Code 49
Use of improper air flow sensor

I can't believe it would be all of these at once! Improper engine ground? loose ECU connector? Main relay not powering a bunch of sensors?

Any ideas?

-Michael

You need a new MAF, to begin with, and probably a new IAC, as well. How old is the car, and how long may it have been sitting idle/not driven?

The car is a 1991 and it was daily driven last week. This is why I find it hard to believe that all this shit just decided to quit. I may be able to fit the MAF from my STi on it but there is no way I'm going to Subaru and paying $400 for a new one. At that price I'll pull a standalone from the shelf and install it.

-Michael

make sure everything is plugged in correctly, connections clean. fuses in good order? reset the codes (not the ECU damnit, just the codes) and see if they come up again.

Few questions about this... Where did they hide the ECU? On my Impreza it's under the pax kick plate. I looked quickly to verify the connections but on this car it doesn't seem to be there. Before I start digging under the dash I thought I'd ask.

The car had the battery removed overnight so I assume that would have cleared the codes. 2 buttons were installed on the dash. One is for checking the codes but I'm not sure what the other is for. The car was modded before I got it. FMIC, full exhaust, intake, 4.44 tranny, brake upgrade etc.

Ultimately I may build up the 2.2l engine, sleeves, pistons rods, STi heads, STi turbo and stick one of my development ECUs in it for some testing. Right now I just need a car to get from point A to point B.

-Michael

He can't reset the ECU without clearing its memory completely, since he can't get the car to start. You can clear the memory by disconnecting the negative battery terminal and pressing on the brake.

Of all those codes, only 12, 21, and 23 could actually prevent the engine from firing when it cranks.

The recommended procedure is to attempt to fix the codes in numerical order, since any code could be a result of a problem that throws a lower-numbered code.

I dunno, though... it's quite possible that none of those are really indicative of the problem. They could all be spurious codes. Sometimes when you crank for an extended period of time, system voltage drops low enough that the ECU suddenly thinks a bunch of stuff has gone wrong even though there's no real problem.

Hmm.

Chances are that the cam and crank sensors are okay, since if one works and the other doesn't the ECU throws a code. But it is theoretically possible that both are bad.

Did you check that you're getting spark? The simple way to do it is to stick an old plug in the end of the wire and rest it on a ground (like the intake manifold or negative battery post). You can see the spark jump the gap of the plug.

And fuel? You can pull a plug and see if you smell fuel.

I suppose it could be a grounding problem of some type, but that'd be kind of odd... Usually that causes intermittent problems rather than just non-start conditions.

The airflow meter and temperature sensor both have wires going all the way to the ECU for everything, so they don't rely on external grounds or power supplies.

There aren't any big intake leaks or anything, right?

What happened between the time the car last ran and the time the car didn't start?

The ECU is under the driver side dash. Search for other posts about it and I think you'll find pictures.

Leaving the battery disconnected may or may not clear the codes. It helps to press the brake pedal for like 30 seconds while the battery is disconnected.

Did you examine the entire length of the intercooler plumbing to make sure it was all solid and sealed up?

If it's a 91, I can try to sell/ship you the metal MAF that came standard, unfortunately, it's for an NA, and it sounds like you've gone hi-mods and forced induction... but if you'd like, the offer's there.

I think the intake tract is well enough sealed that it should start. While cranking the fuel pump does operate so I believe it's getting the crank & cam signals. I will verify that it's sparking.

I suspect there is no fuel being injected but the long list of codes seem to point to a wiring or ECU problem. The problem first showed itself as an intermitant no-start. Next day it was a complete no-start.

ECU plugs are the first thing I'll check. At the plugs I can also find the pins for the CLT and test it out. Probably not the easiest sensor to access. On my STi engine it's under the intake manifold.

-Michael

The fuel pump being on during cranking indicates that the starter switch is working, not the cam/crank sensors. The cam and crank position signals trigger injection and spark.

Yes, the coolant temperature sensor is kind of hard to access. It's mounted under the driver side of the intake manifold, kind of under/ahead of the coolant filler tank.

This is not correct. The fuel pump relay is driven by the ECU. In the run position it primes the system by running the pump for a second or two then the pump shuts down. This is a safety feature since in a crash you don't want to pump fuel all over your crunched up front end if the key is still in the run position. The only way for the ECU to know your engine is running and thus the fuel pump should run is if it's receiving a crank and/or cam signal.

-Michael

Not on these cars. I lost my timing belt and the fuel pump could be heard working when I tried to start.

Steve

Michael, your reasoning is sound, but it's not correct that the ECU can't know the engine is turning without the cam and crank sensors. It has a "starter switch" input spliced into the starter solenoid's trigger wire. That's how it knows you're trying to start the engine. It runs the fuel pump during that time so that the ECU can meter fuel correctly even before the first revolution.

...which explains Steve's experience.

Okay you guys may be correct on this point. I know on my STi ecu it was triggered from the crank sensor. In playing with their factory code I synthesised a crank sensor and the fuel pump driver turned on. Same deal with the WRX ECU.

It runs the fuel pump during that time so that the ECU can meter fuel correctly even before the first revolution.

This doesn't make sense. The only concept of fuel metering the ECU has is the pulse width on the injectors. Even the priming pulse happens after the first cam sync. The initial prime when you turn the key to the run position is intended to pressurise the fuel system so the prime pressure is consistant.

The only time I've seen a factory ECU use a scheme like this is on a MAP based system. They look at the output from the MAP sensor on startup before cranking starts to determine the baro reading. I think the older hyundai's used this scheme. Didn't work well either!

From here I'll check the ECU connector. If it looks good I'll find the CLT pins and measure them with a DVM. If I pluck away at this thing eventually I'll figure out why it's not starting.

-Michael

Most older speed density systems use the MAP sensor to read barometric pressure between the time the ECU is powered up and the engine starts cranking. It works okay as long as the ECU doesn't reset for some reason while the engine is running. These days it's common to use a solenoid that switches the sensor between manifold and barometric pressure, or to just use two sensors.

Running the pump while cranking keeps fuel pressure correct so that the amount of fuel injected for any given pulse width is predictable.

Note that the ECU does also turn on the fuel pump if the engine starts turning. This allows the engine to start even if the starter switch signal is faulty or if you're starting it without the starter motor (like when you push-start).

Annnnyway... Yeah, checking the connectors is a good idea. But you might also want to look for the cam and crank sensor signals just to make sure they're there.

hackish wrote:The car is a 1991 and it was daily driven last week. This is why I find it hard to believe that all this shit just decided to quit. I may be able to fit the MAF from my STi on it but there is no way I'm going to Subaru and paying $400 for a new one. At that price I'll pull a standalone from the shelf and install it.

-Michael

[hijack]please do install a standalone and post results avaiblity ect ect... many of us are looking for standalone options.

oh and pics would be nice too[/hijack]

Cam and crank sensors on this engine are VR are they not? I don't have an easy way to test those.

In a couple of hours I'm going to crawl under there and yank the ECU... Now I'm thinking of pulling it apart to see how hard it may be to chip. Has anyone ever heard of these things being chipped?

I don't even know that people still mod these things. I have a 323GTX and came up with a chip for it. Only problem is with 800 around you can't sell enough to recoup the development cost.

-Michael

Yes, they're variable reluctance. You can do a really basic test -- just connect the sensor to a voltmeter and watch the voltage as you crank the engine. As long as it moves around a little you can be pretty sure the sensor's fine.

But first just check for spark. You won't get spark without a crankshaft position signal.

There aren't that many Legacy Turbos around either, so chipping efforts pretty much have to be labors of love .

I've started a project to hack the ECU but have only been able to devote a small amount of time to it because I've been trying to finish school. They're definitely chippable -- there's a "reverse socket" in there for the purpose. They use a Hitachi chip that's related to the Motorola 6803 and use an industry-standard 27C256 EPROM.

The stock injectors and MAF sensor are both fairly limiting, though, and the fuel control is kind of primitive (very 80s-style), so a chip alone has limited utility. But with the appropriate other mods it could be cool.

Well I played with it a little this afternoon. There is some sparking going on because when it' spinning from cranking I occasionally get a few fires. The tach also indicates while cranking.

I checked the wiring on the ECU. It is clean and dry. There are about 3 black boxes hooked up there from alarms to autostarts. UGH!

I tried some quickstart but no change. Next time I have 2 people around I will double check that both coils are firing.

For standalones I design and build them for a living so I'm sure I could get it working. I am planning to do a plug and play for the WRX and if we can develop a decent throttle driver the STi as well.

A plug and play with this ECU would be difficult because it requires designing and etching custom boards as well as getting the proper female connectors.

If it's a 6803 and a 27c256 then I've got emulators for both and provided the checksum isn't too hard to crack I don't see a big problem with developing custom fuel and ignition maps. Like you say it's a limited market and I am starting to get busy.

-Michael

Designing custom PCB's: all who are interested go to

http://www.expresspcb.com

You can design circuitboards and order prototype or production runs online. If you stay within a certain size restriction you can get a prototype run of 3 boards for like $50. I use the service all the time, it's great...

Oh, and the software can be downloaded for free.

Most times I etch the boards myself since we have all the equipment for rapid prototyping of them. I just find it takes a significant time investment to do the design. The board itself takes less than an hour from the printer to the drilled product. It would just be hard to justify 10h of work plus any software and firmware to sell one standalone. I'll get the car running, get my DMS rebuilt, install them and fix up a few small things before embarking on that project. As soon as we are finished beta testing with the tester cars then I'll have an abundance of boards sitting in my shelf.

-Michael

Hmm, it's sparking but not consistently every revolution? Maybe it is a cam/crank sensor problem after all... But no cam or crank sensor code is stored? Hmm.

Yes, it's a 6803 and 27C256. See what you find and maybe we can compare notes.

If you can get your hands on the datasheet for the HD63140CP, that'd be very useful. That's the chip used for most of the I/O.

The 63140 is actually a rebadged chip made by someone else. I think we have a datasheet for it somewhere.

I had someone crank while I tested each coil outlet. Sparks were consistant. quite weak looking but this isn't always a good indication. I tried shooting a lot of quickstart in while it was being cranked. Still nothing. In my experience this excludes any fuel problems. Still, there is lots of pressure after the filter.

I wondered if the ground to the engine was weak so I tried jumper cabling it to the battery ground. No change. Later today I may try some GM coils but the resistance and impedance of these coils appears good. The ground side of the ignitor is also very low resistance to the battery.

While cranking with the throttle shut it does make vacuum (visible on boost gauge) so it should have enough compression to start. Plugs and cables are also new.

Where to go from here...???

-Michael

Dude, if you can find me a datasheet for the HD63140CP I would worship you.

Quickstart is some kind of starting fluid? Like ether or something?

With the coil disconnected, cranking ought to dump a lot of fuel into the cylinders, right? So you could make sure you're getting some fuel by unplugging the coil, cranking for a bit, and pulling the plugs to see if you smell fuel from having flooded the engine.

I wonder if it could be a faulty ECU somehow. I know they don't typically go bad like that, but...

If you have a spare turbo ECU you might try swapping it in. Or if you have a spare non-turbo ECU kicking around you can try it by switching the cam and crank sensor wires on your harness. It won't work well, of course, but it should start the car if everything else is okay.

Quickstart is a starting fluid. I used some on an STi swap before so I know it works with subie engines. Soon as I can get it back I'll try swapping out the coil with my STi one. I was going to do a 2.5RS coil with the built in ignitors but the wire sockets are different.

I'll have a look around for that datasheet. It's probably still packed from when we moved.

In your reverse engineering were you able to find if there is a checksum on the ROM? Were you able to located the fuel and ignition tables?

-Michael