Fyrestrike

We've been struggling with this problem for 2 months now and at this point I'm open to any and all theories and suggestions. Alright, some background information:

Car

-Race Miata with F20C (33k miles), only mods are AEM 1012, Hondata IMG, custom intake header and exhaust, supertech retainers, spring locators, double valve springs.

Problem

-High RPM misfire after a few corners on the track. Rev limit set to 9300RPM, engine will rev to 9300 3 or 4 times before it starts missing over 9000. If revving to redline is continued, the misfire gets worse and the engine then struggles to rev over 8800. Then, it will almost flat line at 8500, and so on. If car is cooled down, it will start that cycle over.

-We were able to reproduce the problem on the dyno (love Dyno Dynamics). After 7-10 loaded revs from 6000-9300, the problem begins and will cycle the same as above. If revving is continued, the A/F ratio drops to 10:1 and the car dumps fuel out of the exhaust.



What we've tried and logged

-Different AEM - no change
-Different fuel injectors - no change
-Different ignition coils and spark plugs - no change
-Different battery - no change

On the dyno, we logged:
-Battery voltage - values within range, no change when trying conventional wet cell
-Coolant temp - values within range, no pattern as to when the motor starts misfiring. Happened once at 194F, next time at 205F, etc.
-Ignition Timing - Remains consitent with tune even during the misfire, no obvious problem here
-Intake air temp - values within range, no pattern
-Air fuel ratio - values within range until misfire is provoked several times, at that point it goes pig rich.
-All 8 fuel injectors we tried read within honda's acceptable range of resistance.
-All 8 coils tried were from working cars without this problem.

Some notes

-Compression is 225 +/- 5psi across all cylinders
-The motor has run for 3000 track miles without this issue and with the same tune.
-We have not swapped the two cam angle sensors or the crank position sensor yet, any logical reason these would be the culprit if the timing is spot on?
-We've checked over the engine wiring harness and there is no obvious signs of damage.

Am I missing something here?!

luisS2K

the map sensor readings are what they should be while WOT at the RPM range??

INTJ

Dang, great project.

High speed miss was always an electrical problem for me. Put a simple RF shield around the ECU. Recheck the wire harness, and particularly the grounds. Vibration shorts should be minimized by the dyno run, so that eliminates a huge chase.

The symptoms lead me to believe that the ecu is cutting back. Doesn't that one have a fault code system?

The othe track is that heat is causing something to short. Sometimes the coil on plug can have a crack that is hard to see cold.

Fyrestrike

Sorry, forgot to mention that. We logged MAP voltage and pressure and everything looks fine. The pressure numbers are not exactly what the dyno dynamics sensor reads but they're +/-5%.

Fyrestrike

Haven't tried shielding the ECU or checking the grounds, will look at that. I agree, the ECU seems to be cutting back but the AEM only does what it's told, it's not as "smart" as the stock ECU. We've dug around the map to see if it's pulling fuel/timing for a specific reason and we've come up empty so far.

Heat causing something to fail is where I'm thinking the problem is but we've swapped coils and injectors already, is it possible for the coil wiring to fail from the inside out?

Fyrestrike

Bump for more thoughts.

Crusin_x1

You mentioned that you changed the coil and plugs, what about the wires?


.ah nuts... THEN I realize... never mind....

CKit

The AEM replaces the knock sensor, right?

INTJ

Well both of those are easy to check.

The coil is unlikely, but look for fine carbon traces/hairline cracks and do an ohmmeter test. You can bypass the injectors into clear poly pipe or buckets if you want. Presumably this would just give you a hot cyl. which the plug would show right away.

I like the knock sensor idea, which I discounted with the dyno and assumed you could get the timing code. Just take it of the head and safely secure it. Easy enough to test.

Good luck.

Fyrestrike

No, it can log and use the OEM knock sensor but we have it disabled because it's hard to tune with it (picks up too much noise in VTEC) and our timing is conservative.