Help diagnose my Megasquirt - OR - What the heck am I missing?

Last weekend my RX-7 ran progressively worse and worse until it was barely running by the time I got home. I found that at least one of the ignition coils (FC leading coil, one FC trailing coil through distributor, wired in parallel to an MSD 6AL) was arcing to the fenderwell. That would explain... well, all of it. At first it ran great at WOT but barely at part throttle, which usually means the trailing is dead on a bridge port. That coil was definitely bad. The engine was running on one rotor at the end of the trip so the leading is probably shot too. So I ordered some different coils that should work better with the MSD and also allow mounting in a better spot.

A couple days ago, the new coils weren't arrived yet but I wanted the car mobile, so I went to connect a pair of MSD Blaster coils I had lying around and run both coils through the distributor. Problem was, the engine was STILL running on only one rotor when I did that!

Compression is really good by the "pull the leading plugs and crank it" test. Spraying windshield de-icer down the intake (nothing else flammable handy) made the engine run on both rotors and sound pretty happy. So we (probably) have good spark, we (most likely) have good compression, but adding fuel makes it run happier so we must have a bad injector or driver, right?

Tonight I got down to bidness and pulled the carb body off so I could access the injectors. My recollection was that the two injectors were on separate drivers. My recollection was wrong. Both injectors (low impedance) are on Inj1m and share the same power circuit, which goes through a 6-7 ohm resistor block ganked from the FC that donated its coils. So total resistance on that circuit is (2.5 in parallel = 1.25, plus resistor) about 7.5-8ish ohms. Easily dealt with by a single driver and has worked fine since I installed Megasquirt something like 100-120k miles ago. Most of that on a 1.01 board, but now I have a 3.57 that I think has been in there for 10-20k or so. Either way, if one is getting power then they both are, and I tested wire integrity with my Power Probe to be sure.

Then, I pulled the fuel rail, secured the injectors to the rail, connected my laptop and went into test most. My deadtime is .825, arrived at by complete BS methods but it's worked fine for a long time. 2ms pulses gave me tiny little dribbles. 10mm pulses gave me beautiful three-element spray patterns from both injectors. (RC engineering 1000cc/min) And they looked about even. So we are getting equal amounts of fuel from the injectors.

Here is the part where my brain hurts. I started to think, maybe when it was running poorly one of the presilencers collapsed internally and now it has a severe exhaust restriction on one rotor. (Nevermind that the presilencers don't look burnt, and it runs better with additional fuel) So I disconnected the plug wires from the front rotor, gave them suitable ground, and cranked it. No run and barely any kick. A-ha! So I reconnected those and disconnected the wires from the rear rotor. No run and barely any kick. Uh? I reconnected the plug wires. Start and run (barely) on one rotor.

Grasping at anything, I thought, maybe the Inj1 driver is HURT but still sort of functioning. I happen to have a pair of injector connectors powered up and wired to Inj2. After verifying that the computer is running simultaneous, I swapped connectors around so Inj2 is connected to the injectors. No change at all!

It's about this point that it was getting to be 9pm and I really needed to stop annoying the neighbors, so I cleaned everything up and went home.

What am I missing??? I don't have a fuel pressure gauge, but I don't see how a failing coil could kill a pump. (It sounds the same as it ever has, really). Fuel composition is probably not an issue. I have a friend with the Mazda factory compression tester, which I'll throw on there for giggles, but it runs better with more fuel. I reloaded my last known-good tune in case somehow all the stray voltage corrupted it, but that made no difference.

I dunno. I dunno at all.

Forgot, one other thing I tried. The MSD gets its trigger from one of the pickup coils in the distributor. The Megasquirt gets its crank signal from the MSD's tach output. The other pickup coil in the distributor has an ignitor that goes to a little junction block with a 1000 ohm (I think) resistor between the two old trailing coil wires. Its only purpose in life is to drive the tach. I removed the junction block, installed an ignition coil in its place, and ran that through the distributor. Still only running on one rotor! So it is (probably) not an MSD box that has stuffed itself.

Do you have a leak or restriction in your MAP line? I got a little chunk of tube stuck in my line from when I cut it. It eventually managed to completely block the line.

Plug in tunerstudio and double check that all of your sensors are doing the right thing. Is RPM correct and steady? IAT and CLT are reasonable? MAP does the right thing when you are stopped/crank/run/throttle?

To clarify your last post, for that setup the ignition control was wired to take the MS out of the equation, correct?

MS doesn't do ignition. Fuel only. I guess I'm old, when I started playing with it, fuel only was the norm and ignition control was exotic and something only hardcore experimenters did, and I still take that mindset

When I stopped at patgizz's house during the ordeal drive home, I checked the hose going to the MAP and fuel pressure regulator with a Mityvac and it held vacuum. With the engine even in perfect running order it only runs at 70-80kpa at lower RPM and it's very ragged, so observing anything other then "yep, it moved" is all you can do.

I have had the MAP hose blow off before. That makes it idle with authority and rev like nobody's business when you touch the throttle. It's a bit wasteful and the exhaust gets a bit visible and eye watering when you do that though.

Have you tried swapping around spark plugs/wires/coils? My Volvo needed about 15% less fuel at idle after replacing spark plug wires and coil to run smoothly, so this could still be ignition and not fuel related.

Like mck1117 said, sanity check the live data on all sensors, or just start unplugging stuff until it runs better.

Also check the fuel pressure, failures of multiple parts do sometimes line up just to berkeley with you. My favorite was a friends car that blew a fuel pump and had the upstream O2 go crazy at the same time, took a while to figure that one out.

Fuel pressure is one of the few things I haven't checked. Makes no sense to me why it would cause the engine to only run on one rotor.

Weird thing, at the rallycross the engine didn't seem to want to pull past 8000, although I chalked that up to having an open diff and getting a lot of single tire spin and the car was just catching up to the engine. Maybe. On the way home, close to the end, I had stopped at a rest area to diagnose and some Puerto Rican guys were also there and we ended up talking rotary for 10-15 minutes. When I left I ran the first three gears out to 8000 just to show off... sounded and pulled pretty good actually. But NFG at part throttle, which made me think even more that the trailing ignition had quit.

But it wouldn't be the first time I had a fuel pump take a dump on me. Most of the reason I came up with my somewhat bizarre fuel surge tank setup was to keep aerated fuel from getting to the fuel pump. But usually the pump would not allow WOT and would come back to goodness after cooling off.

Could be one side had a little stronger spark or something and is more tolerant of a lean mix.

In reply to rslifkin:

That's what I was thinking with the plug/wire swap.

Cap, wires, and plugs were replaced when I removed the dead ignition coils because they would not be usable with the new setup. Cap looked new, wires looked new (were borrowed from my '81, the PO did a lot of parts-hanging when the engine died), and the plugs were brand new NGKs that I had sitting in my spares box. And I verified that spark was actually getting down to the plugs. And at one point I reinstalled the electronic Kettering style ignition on the leading, leaving the MSD still controlling the trailing (MSD can't be removed entirely)

The bizarre part is that pulling the plug wires from EITHER rotor makes it not run at all, but when it is running it sounds like it is running on one rotor.

When I get back to the garage I am going to see if I can rig up something to test fuel pressure. (How that would keep it from running on two rotors I dunno, but it's still an Unknown) And then see about reloading the computer's firmware, which honestly has me a little jittery because I've never messed with firmware on a Megasquirt before and I don't want to brick the computer. In the interest of not making potential new problems, I'll load the same MS2/Extra 3.3.3 that is in there now.

How's the compression on the non-running rotor?

That's the thing. Both rotors have equal sounding compression, and I can't tell which one isn't running, because disabling EITHER rotor makes the engine not run at all.

This is why my hairline receded three sizes this weekend. And I'll blame my white chin stubble on this, too

I am wondering at this point if the firmware got corrupted somehow (there was a lot of lightning flying around under the hood for 3+ hours) and 2 squirts/cycle is now 1 squirt/cycle, which would starve every other revolution of fuel. Which, given that I am running batch fire, would mean that the engine is firing the rotors alternately and it would sound like running on one rotor.

I think.

The last time I tested compression was quite a few years ago and it was something like 80/80/80 and 60/60/50. But this is with way the wrong tool and I've had engines start and run just fine at 25/25/25. I am curious to try my friend's Mazda factory tester to see how it compares to whatever I get. But by the "Polish compression test" it sounds pretty good.

Try a temperature gun on the headers. My guess is that it's running badly on two rotors rather than only running on one rotor.

Yeah I agree that it's probably running poorly on 2.

You have a wideband O2 hooked up?

mck1117 wrote: You have a wideband O2 hooked up?

Won't help. It'll read stupidly lean with a bad misfire regardless of how rich / lean the thing is actually running.

Knurled wrote: I am wondering at this point if the firmware got corrupted somehow (there was a lot of lightning flying around under the hood for 3+ hours) and 2 squirts/cycle is now 1 squirt/cycle, which would starve every other revolution of fuel. Which, given that I am running batch fire, would mean that the engine is firing the rotors alternately and it would sound like running on one rotor.

Is the crank position sensor for the fuel separate from the whatever triggers the spark?

BrokenYugo wrote:

Knurled wrote: I am wondering at this point if the firmware got corrupted somehow (there was a lot of lightning flying around under the hood for 3+ hours) and 2 squirts/cycle is now 1 squirt/cycle, which would starve every other revolution of fuel. Which, given that I am running batch fire, would mean that the engine is firing the rotors alternately and it would sound like running on one rotor.

Is the crank position sensor for the fuel separate from the whatever triggers the spark?

MSD does spark. Fuel is triggered by MSD tach out. MS's RPM is same as the tach (which is driven by a separate ignition system) so I don't think there's a half signal happening.

MadScientistMatt wrote: Try a temperature gun on the headers. My guess is that it's running badly on two rotors rather than only running on one rotor.

I'd love to but it doesn't run long enough for the headers to even get hot to the touch. Frustrating.

The sound of running is very characteristic "rotary on one rotor" sound. I've heard it enough times.

I'd still wonder about fuel pressure then, makes sense in my head, something like the engine is so starved for fuel it has to load up a revolution or two before it hits.

Can you send a datalog of it trying to start to see if the MS is doing anything funky?

I could. All the data looks fine to me, or at least looks nominal.

I remember the event where I burned the PWM transistor out. (I was running close to 100% duty cycle for extended periods, the 1.01 did not like that...) I got it running by disabling PWM entirely, and a friend (a calibration engineer) saw the MAP bouncing between 70 and 80 and said something's wrong. Nah, that's where it idles, and you can ignore the O2 readings at idle too, lots of free oxygen in the exhaust...

I will preface this by noting that I have a bad head cold and don't know anything about Wankel ignition, but... Is there wasted spark anywhere in this sytem? What happens when you pull the plug on half a wasted spark system? Something I just read said they were in series, so it seems like pulling either plug could cause both not to fire. So one side isn't working regardless of spark, and if you pull a plug, then neither side works, regardless of which side you pull it from (because it stops the working side running).

This may be at right angles to reality. Apologies.

The leading coils can run wasted spark, and this was the way I had the car up until last weekend. For "get the car moving" purposes I reverted the leading ignition to going through the distributor. My ultimate plan involves the MSD driving three Accel coils, one for each leading plug and one for the trailing.

Regardless, when testing, I always give any free plug wires a spark plug in contact with a good ground.

Update: Front rotor is 70/70/75, rear is 60/60/50. At least I know the engine is aging gracefully This may sound horrifyingly low but it's pretty good for my tester, which seems to read 10% lower than anyone else's anyway. Must be the 4' long hose. I'll see if I can bum the Mazda tester sometime this week to compare numbers.

One thing weird though is after I connected everything back up and I tried to start it on external fuel, it would sometimes pop back through the intake. Curious.

So, I removed the header. It's a long primary exhaust where the collector is up over the rear axle. My first thought was that I collapsed an exhaust sleeve. Nope, the sleeves are intact. (WOW I forgot how large I made the exhaust ports) Next I grabbed my vacuum cleaner, switched it from Suck to Blow (Spaceballs is a national treasure) and blew through the exhaust system. On both sides I had an equally strong wind out the tailpipe, so I will reasonably assume that it didn't collapse a presilencer or something.

At least yanking the header will allow me to belt-sand the exit flange flat and put a new gasket on it. It's actually a different brand of header than Racing Beat altogether, and it was originally a 12A header anyway, so I had to do a LOT of bending/warping/shaping/welding and that flange never was very flat.

Do a "real" comoression test. Eliminate some of the guess work