Built completely new COP conversion wiring harness, and wired in two new 4700uF 30v caps (from RadioShack) in parallel across 12v and GND harness-side. Drove the car all day without issue and composite logger looks perfect. It appears that even though I was previously able to run without the 10,000uF cap, I don't have a choice now. Happy to be back on the road...and just in time for the good weather.
Here's the deal.
I started this related thread back in January: NO START - 95 Miata DIYPNP - RPM/Voltage drop?
In summary of that issue, it appeared that one of the two hall effect sensors inside the CAS decided to short to 12v causing a no-start condition. That was resolved by simply replacing the CAS.
A few months ago, I took the car for a spin and noticed that it was missing under boost after topping off at a gas station. Presumed to be a bad tank, the next move was to top off with very expensive 100 Octane and turn down my boost, which alleviated the issues. However, not before causing a leak in my head-gasket which subsequently let loose a few days later.
Fast forward to this month. Head is rebuilt, bolted up, and I'm starting the engine without the plugs in to build oil pressure. During this, geysers of gas spray all over my garage through the spark plug holes. During diagnostics, I discover that Injector bank 1 FET is shorted to ground pulling the related injectors to 100% duty cycle...good thing I didn't hydro-lock. Interesting that I've smoked two sensors that are related to the same WHT/RED 12V power source...although grounds are good.
FET fixed, and finally start engine. The thing is purring like a kitten, and then falls on it's ***. Start again, runs great...takes a couple of RPM hits and then stalls again. Diagnostics begin.
Here's what I found:
Some Trouble-shooting attempted:
- Datalogs show "Lost Sync Reason 39" which is related to the 4G63 CAS, and this causes timing to be completely lost AKA RPMs drop to 0. During timing loss event, MS thinks the engine was re-started and thus ASE ensues which causes engine to run unnecessarily rich - if during ASE another timing loss occurs, the engine will be too weak to recover and stalls.
- Acquire by oscilloscope, 12v signal feeding MegaSquirt appears to have noise which is causing disruption of the 12V (Opto -) signal feeding MegaSquirt. This is evident by probing the 12v power source (2B) and grounding pins inside the MS chassis, and comparing to 12v directly connecting to battery.
- Scope shows malformed square wave during engine run just prior to timing loss, which appears to be smoking gun. This is likely related to the aforementioned dirty 12v signal.
- Issue is intermittent, but is most prevalent during idle and low speeds.
- Re-did all grounds....three times. Under brake booster, behind intake manifold, at throttle body, at battery, and PPF.
- Replaced CAS with replacement unit and a known good OEM unit
- Checked all CAS and associated wiring
- Bypassed Main Relay and Fuel Injector related
- Unplugged alternator in case of bad regulator
- Charged car battery and kept on during run to prevent voltage drops
- Unplugged EVERYTHING not essential to running (all fuses, relays, etc.)
- Fed MS direct from JimStim to eliminate MS circuitry as culprit. No timing loss.
At this point I'm debating on re-running the CAS wiring direct to the ECU as a test. I'm at wits end and just want to drive this damn car again, because up until January it was daily-driver reliable.
Any ideas are greatly appreciated...so, thanks in advance.
Malformed square way before timing loss event
Overview of timing loss - note the waves spacing out starting from 25% in due to loss of timing and engine recovery
Example of 12v spikes looking at voltage supplied to ECU
More examples of voltage spikes
Composite logger showing tooth dropping out due to malformed CAS signal
Datalog showing timing loss event (sync loss reason 39)