Flashing Check Engine Light, No drivability issues

[98legacy22stick]

As the subject says, I have been recently getting a flashing check engine light, usually at highway speeds of around 60-70, when decelerating to a lower speed, but with my foot still partially on the throttle. The flashing light is never accompanied by any change in engine behavior, vibration, rough running, or other typical signs of a misfire. It also never occurs when accelerating under partial or full throttle, or anytime when not at highway speeds.

Car is a 98 legacy 2.2 5 speed.

When I pull the codes, it is always cylinder 3 and 4 misfire. I would think that if I am losing half of my cylinders to a misfire, I would be able to feel it.

The only thing I could guess is causing this may be related to the fact that my EVAP system has been essentially inoperable for the past 4-5 months due to what I eventually found to be a leaking Vent Control Valve, causing the system, more often than not, to fail the diagnostic testing performed by the ECM and set a P0440 code.

Does the ECM attempt to trigger the purge solenoid, even when an EVAP code is set? If not, is it possible that I have a higher than expected concentration of vapors in my canister that is driving some kind of misfire every time the purge valve opens?

I took the canister off when replacing the Vent Control Valve and it didn't seem unusually heavy, around 5 lbs and definitely didn't have liquid sloshing around. Also, I confirmed that the EVAP problem was fixed based on my scanner reading an OK Status for the EVAP I/M which I was never able to achieve before without also setting the P0440 code.

The Car has 230k miles on it, but has been properly taken care of with up-to-date maintenance on all fluids, plugs, timing belt, valve lifter adjustment, etc.

Anyone come across this before?

[Legacy777]

At what RPM does the CEL occur and what percent throttle?

The ECU is supposed to open the canister purge valve under steady cruise. I'm not sure where that cutoff point is on throttle % open before the ECU closes the canister purge valve.

Have you looked at normal maintenance things like spark plugs & wires? When's the last time they've been replaced?

[98legacy22stick]

Thanks for the reply.

The codes didn't pop up today. But i'll write down the freeze data the next time it happens.

General conditions are around 3000 RPM, and maybe 10%-15% throttle.

Spark plugs were done last summer. Wires... it's been a while.

From some of the reading I've been doing on misfire detection, and how the data from the sensors is processed and analyzed against conditions that indicate a misfire, perhaps the sensor is sending a marginal signal or the signal is degrading between the sensor and ECM, or there is some variation in some aspect of the way the engine is running and at that operating condition, the threshold for indicating a misfire is exceeded, even though there is actually no misfire going on. If that is the case, it would take a lot more sophisticated equipment than a harbor freight scanner to isolate the problem. I don't know enough about how subarus specifically work with regard to this to do anything more than speculate.

[RoughDiamond]

Knock sensor issues were common on the OBD2 years, could be a "lying" sensor.

I would also check compression on those cylinders or even if you can get a "live data" scanner to see if the O2 sensor is freaking out as well. ECUs will monitor exhaust content to help detect misfires as well as specific pre-detonation frequencies. Sudden spike in either oxygen (causing voltage spike) or frequency will trigger a misfire detected signal.