EJ22 random idle, pig to start - what else?

[silverbullet]

Finally go to fishing out the ECU from the o/s airbox, with the idea of relocating it somewhere safe/dry and rigging up a fault light. Turns out that it's got one built in - bonus! (even though there's a wire tail for the lamp, bit odd)
So, ignition on and oh *£%^ it's flashing. Code 49 = "improper air flow sensor fitted"
What?!
I guess that's another bit of Subaru-speak for "it's gone wrong, but in a different way, not specific"
It was going ok, apart from the random idle when warmed up (800-1500) and difficult starting (not always) and the hesitation under full throttle...
Anyone know where I can get the correct MAF for an (EJ22) 22611 AA670 / A11-000 R97 ecu?
Part number would be great as the label's long gone - Hitachi?
Where does everyone get their engine sensors from at the right money - this is my first Japanese engine!

Apologies if you've read all this before on the brickyard...

[jamesc76]

Has it got ODB2 diags or odb1???

[silverbullet]

I'm working with this:

ecu 22611 AA670 A11-000R97 (very early, integral fault code lamp)

maf 22680 AA160 A36-000R60 (suspect, as have code 49)

Hang on I'm repeating myself here...

Numbers seem to tally up as correct for Legacy 90-94 EJ22, according to most sources.

No diagnostic plug fitted. Inherited someone elses work...

[jamesc76]

same age as mine but dont have a fault light on the actual ecu!, do you have to plug the 2 blacks together to ge thte faults?

[silverbullet]

Same age Jim, but what part numbers have you got? Could be that I'be bought a car with a mismatched collection of bits. Unless it's US or Jap market and not Euro

RJES informs me that only the very earliest ones have the lamp in the ecu. They probably deleted it to save the cost of drilling one hole! Or to keep dust out of it.

What do you mean "join the black ones together" ? Don't follow.

Engine has running faults pretty consistently now - wondering if it is the maf or maybe another sensor "falsing" the ecu or even the ecu going up the wall. It goes ok though, just has a loopy idle. Doesn't smell ritch.
Have heard that knock sensors can mess up the idle, bu it's rumor, not concrete facts.

Might try unbolting it and going for a run to eliminate.

[jamesc76]

Dunno numbers on mine but I would try new leads and plugs they burn leads out and need high spec ones plus try dissconnecting the lambda sensor see if it runns any better ?

[silverbullet]

Aw, c'mon!
maf label is opposite side to the plug (green label?) and unless your ecu's buried somwhere bad like mine you should be able to read it straight off...
Don't have a lambda tapping so no sensor to unplug.
The more we share, the more we all know.

[silverbullet]

Well I swapped the MAF and its still the bleeding same.
Starting to think that the throttle pot is a mass of carbon dust inside and the idle switch doesn't always make it.
Anyone know if it's safe to open one up and how?

Still got a few wiring mysteries - found one pair over the back nearside which join the body loom but only one connected and the other's been off so long its full of oily cack...remembered that previous owner said that it had an exchange motor about a year back done by local garage - could explain a lot.

Will put up a pic when I have a chance and we can all play "What's this Wire?"

[toomanytoys]

I'd be tempted to check all connections, unplug and replug a couple of times... clear the code...
Also worth looking at any potential dodgy scotchloks/solderjoints etc etc...
Peps has an intermitant fault (at cruising speeds) that reckons its a faulty idle valve... yet it idles perfectly... normally

[silverbullet]

Ahh...clear the code...that was going to be my next question!

How is it done on a Subaru ecu without any fancy diagnostic gear (that no-one's got anyway)

Short together a couple of terminals for 5 seconds or something? Do tell!

Could be the idle stepper motor getting iffy, but how to tell other than fit a new (!) one?

[RJES]

I rarely get to see peoples airflow meters, so don't have much info on which ones go with which ECUs.

I have never heard of anyone having error code 49 before. Strange how there are two MAF error codes listed (23 and 49), but with no useful difference in the fault description. The error description seems pretty self explanatory to me.

An induction system leak somewhere between airflow meter and the valves could probably give an MAF error. It would also give a fast idle,and would be less noticeable at high revs. On an ECU with no lambda sensor it is also likely to result in a lean mixture.

There is a procedure to reset the ECU, but I find it far easier to just disconnect the battery over night.

If someones been messing with the throttle stop, the throttle position sensor position, or if the cable is too tight or the throttle motion is sticky, it sounds quite possible that maybe the idle switch isn't working every time the throttle closes. While you're at it check that the rises and falls smoothly as you move the throttle. As you have an ECU with a built in error code light, it's early, so may have no lambda sensor input, and will have a 4 terminal throttle position sensor (with a potentiometer and an idle switch inside). You don't have to take sensor apart to check it. Just put a multimeter across the switch terminals and check that it switches just as the throttle gets to idle.

[silverbullet]

Latest update, idle getting progressively worse. Anything from 800-2000!
Blipping the throttle helps to drop it a bit, but not enough to stop it being embarassing then it will suddenly drop back to 900 for a while.
An interesting tell-tale is that it's far worse when hot, like when caught in traffic and the fan cuts in and out (at least that works on one speed) BUT the fan pulls down the rpms a good bit so...
IACV or it's wiring must be at fault, otherwise the idle would be stable, regardless of load.
Could just be an earth fault on that circuit (I hope) otherwise it's start hunting for another valve.

[RJES]

So the throttle pot switch is definitely switching OK to turn the idle speed control circuit on? If so, the ISCV could be sticking. There are some very tight toleranced parts in them, and the mini throttle inside is only moved by electromagnetics (a bit like a moving coil meter mechanism operating a barrel throttle), so the slightest bit of dirt in there and it'll stick. Also, if the dust is ferrous (such as angle grinder debris), the Subaru technique of flushing the valve out won't help much as the dirt will be stuck to the magnetic components. Take the valve to pieces and check for dirt and that everything moved freely, but be sure to mark the electromagnet actuator before you take it to bits. There is no procedure for resetting these valves if you lose the position. They were supplied preset.

[silverbullet]

Hi Richard
Yes, I got desperate and opened up the throttle switch to check that there was still a carbon track left at the idle position. 3mm drill bit and steady hands to remove the two brass rivets. Quite a bit of carbon dust/muck on the contact finger/brush (and some wear) so I cleaned that off with great care and put back on the throttle body. Ran the engine with the throttle pot cover off to check idle switch contacting ok then set and put cover back on.
No different at all.
When I took the ISCV motor off previously to check for seizure, all was well but there was a bit of magnetized muck on the valve spindle. I didn't check the valve internally for ferrous contamination as I didn't appreciate that the whole valve barrel is magnetized, so will do this next.
Have to drain coolant system anyway to change radiator - found small split coming from bleed screw tapping...
PS Starting ok now that I have jury-rigged the cold start wire as advised!

[RJES]

It could be more dirt in the valve then. Hopefully that's all it is. They are more complex than they look. I initially thought that the coolant pipes at the bottom were just to preheat the barrel throttle inside so it can't ice up in freezing damp weather. There is actually another valve in the bottom operated by a bi-metalic strip, which is heated by the coolant flow according to some Subaru service info I found. I've never taken the bottom half of one to pieces to have a look though.

The electromagnetic part of the valve is operated by two coils. One pulls the valve open, and the other closed. They are both fed with PWM signals (preasumably with opposite duty cycle) to alter how open or closed they are.

The wear on the throttle pot tracks sounds rather like the old Bosch 'moving flap' airflow meters, which were renowned for the tracks wearing out.