Cold start : cuts out after ~7 seconds. Did this once every few months, but now totally dead.

One factor landspeed has mentioned gets my attention: the vehicle is operating (barely) without a catalytic converter. There are sensors fore and aft of the cat, so I would suspect something is going on with the exhaust system. I think that would account for many of the symptoms described.

As for the MAF sensor, just because it looks clean doesn’t mean that it is!

 

Yep; the forward sensor is a requirement, the aft is the feedback loop to manage electronically controlled fuel mixture. The ECU will run the engine for a bit on base factory settings - then it should throw codes and possibly throw it into limp mode. Don't know how they have it wired up to spoof those sensors.

 

I bailed from this after giving my advice, which was essentially the same advice you've received multiple times now. I would add that there are gas tanks available here in the US from dismantlers, most of which are in California. Check car-part.com to find all of those. I've replaced the tank in one of mine with a used tank and haven't had any new problems.

Just because you get some gas from the pump doesn't mean you get enough fuel pressure to not cause problems. And, did you check the fuel filler neck like I told you. Maybe you don't have snow, but these are old cars and it could still be rusted through. Literally takes 30 seconds and no tools to eliminate that as an issue.

I also am leery of advising on the early Japanese models. Not sure how different they may be, especially the ECU software.

Otherwise, you seem to simply have typical ICE issues. Do you have a service manual so that you can follow standard diagnostic procedures? If not, go to http://www.charm.li and download it. They only have the 2001 and up manuals, so again, not sure any differences with your Japanese model which may have come out before that.

 

For the spark plugs, I strongly recommend that you use ONLY the OEM Denso iridium-tip SK16R11, as listed in the owner's manual. The manual also lists NGK iFR5A11. I have used none but the Denso, and am now on my third set @176K miles. The manual calls for changing plugs at 90k miles.

 

I wonder if this description might be considered off by one. There is a sensor that's forward of the whole engine: that'd be the MAF sensor, in the air intake. Then there's a sensor forward and a sensor aft of the first catalytic converter.

The MAF sensor, forward of the whole engine, is a definite requirement; the engine can't even do good open-loop injection control without that.

The next sensor, aft of the engine but forward of the cat, is used for feedback to achieve closed-loop mixture control.

The final sensor, aft of the cat, is used for monitoring the functioning of the cat.

 

OT, but wasn't the designation NHWxx, not NWHxx?

 

NHWxx is correct.

 

Congratulations on finding the solution to this puzzling problem!

I wonder whether there were any sub-codes along with the "engine system failure" code in the Hybrid Control ECU?

 

These hybrids have to have their engine rpm perfectly match the motor generator's speed control of the planetary gears or else you have severe damper slipping. A total cylinder outage like yours can cause a no start or stall because the engine rpm becomes erratic. It is not good to drive this way since the damper can self destruct which is a major repair generally requiring subframe and drivetrain removal if you go by the book.

 

Like the discs in real clutches, that one has friction surfaces and also a shaft connection surrounded by coil springs that can bounce to take up small irregularities in the rotation being transmitted. Those springs always have some job to do, because the engine output is inherently a bit bouncy (two power strokes per revolution) while the rotating masses in the transmission preferably rotate more smoothly.

The friction faces in a car with a clutch get used regularly, as you use the pedal to engage and disengage the drivetrain. The friction faces in a Prius see very little action ever; they are clamped together under a constant pressure that would only let them slip at all to prevent a very high torque from doing damage, say if one of the other of transmission or engine were to fail and seize while the other was turning, it you were spinning wheels and suddenly got traction, that kind of thing.

When the engine misfires, its output is even more uneven than usual. You hear noise because the gears in the transmission have a slight amount of lash, and the MG rotors are fairly massive, and all that stuff is getting whacked back and forth as it rotates, because of the uneven crankshaft rotation. It is the springs in the damper that really get overworked in that situation, and if it goes on long enough, they can get pounded to smithereens, and you can reach down in the bellhousing through the inspection hole and feel broken pieces of spring.

Of course, once the springs are smashed, the metal-on-metal clatter gets even worse, because the damper is striking the ends of its travel with no spring in the way. The friction surfaces might then begin to slide a tiny bit at each whack, because those springless metal-to-metal whacks produce a high-enough instant of torque.

Nothing you do at the driver controls is going to make the damper disc slip. It's just there to slip at high torques that exceed what the car is normally capable of, like the cases described above.

Shifting between N and D is a purely electromagnetic affair. The power management control ECU is always in control of how much electromagnetic force it applies at the motor/generators, and it always synchronizes the force it applies to what the drive train is currently doing, and isn't going to make any abrupt change that would make the damper slip.

The main worry about driving too long with misfires is beating up the coil springs in the damper.