The motor cooked and I rebuilt it and replaced all sensors that were melted except the O2, TPS and AFM. Since then, which was 9 months ago, we have done 4000km, and the engine has been perfect. Idles correct, quiet, fuel economy fine, power and torque great.

Then a couple weekends ago, after 80km of driving on the freeway at 3300rpm and the engine just died. In 15 seconds we were in the emergency lane.

It will idle smoothly, however at 1000rpm, and if you touch the throttle even 1mm the engine will cough and die. If you take your foot off the throttle straight away it goes back to a smooth 1000rpm idle.

We have checked
1) Cam belt is still correctly aligned
2) Fuel pressure and flow and return, correct
3) Crank angle sensor replaced
4) TPS checked by Auto elec to show correct resistance values from rheostat
5) new spark plugs fitted
6) s/h AFM and air filter fitted

Could it be the ECU? Or what else?

My money would be on the afm,what air filter r u running?
The oiled filters tend to cover the element and stuff up the readings,
Clean it first or replace with known working one.
Most of the afm of early 90's will work,wrx ones too,as long as the plug fits.
Some afms have an extra tine in the plug but will work
(green sticker from memory)
Good luck

The AFM fitted was S/H but working, and a new non-oil filter is fitted. The old filter was thick with dust.
Just picked up an ECU out of a running car, so will try that on Mon. Our old ECU was not bolted down, so there is talk of diode failure due to shock. Will keep you posted of results.

probably MAF. MAP does similar thing on dumb ECUs which done have a default lookup table based on TPS values in case of MAP or MAF failure.

Is there any software and interface cable you can hook up and look at the data stream on these things?

Maybe time to move to a different system or if possible ( doubt it given lack of custom code for these cars) incorporate MAP or more reliable delphi MAF?

I spoke to mechanic today. He fitted the new s/h ECU that is guaranteed from a running vehicle. Still no go. He now wants to remove the injectors and test them, and check the timing belt and the compression. I can see he is trying to cover all bases and making sure there is fuel and compression. I still have a gut feeling it is related to the signal from the TPS or AFM, not necessarily the senders, but maybe a pin in the wiring plugs. The only other thing I can think of is the relays. Is there specific relay for the idle air circuit solenoid? if that is staying open then the Throttle is pressed, then maybe that is why it dies as soon as the throttle is pressed, as it is holding both circuits open at the same time, making it lean.
What about the O2 sensor?

Some of the jdm subarus have the cam and crank sensors wires at the ecu swapped around,happened to me years ago on a legacy gt
It would idle perfectly,rev up to 2000rpm and cough and run rough,swapped them over and fixed it
Couldnt believe the car could run with wires swapped over

Could it be a faulty spark igniter or one of the loom relay's?

It seems everything mechanical is correct.

JDM - Japanese domestic market

I learned that about 2 months ago

I bought a jdm sti forester half cut and fitted it in my audm gt turbo forester,
Kept the same harness,replaced ecu with the sti,swapped drivetrain over,
One thing i had to do is swap the crank and cam sensor wires over at the ecu
Go figure

We are going to try another TPS

Here we go. Put on the new s/h TPS (Throttle position sensor). It is all good and running. The PCV (Positive crankcase ventilation) valve was blocked and stuck shut. It is running smooth as silk. All seems to be good.

Except when you rev it up to about 2000rpm and want to go more, it all of a sudden goes asthmatic and struggles, until you back off the throttle. We checked the tail pipe, and more gas comes out at idle than at this asthmatic point. There is a clunk in the muffler, so it seems a baffle has come loose.

This is quite feasible, as I bought a new muffler only 8 weeks ago, as there was a chronic rattle comming from themuffler that is 12 years old. So now that it actually will accelerate up to 2000 rpm is a bonus, and I will put the new muffler on, and that may be it. I will definitley cut open the old muffler to inspec the inside.

Cut the old muffler off, and it was blocked solid with catalytic converter insides. It was full to from the in side of the muffler flush to the top. Cut the cat off, and it was emty except for a golf ball size chunk of ceramic comb. So with a new cat and muffler, it is running all good again. The mechanic said he had never seen it before, and was stumped.