Jamie's 914-H6 Subaru Swap, Palm Springs California Options

[jamlip]

Its official - BMW 320i tie rod ends are an interchange for turbo tie rods. Good riddance, Uro.

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1756610194.1.jpeg)

[jamlip]

Current status - the engine cranks and I'm getting feeds from the sensors. I can't get a nice enough signal from the crank position sensor yet. Working on that.

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1756610311.1.jpeg)

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1756610312.2.jpeg)

[jamlip]

Have crank signal and, I think, now a decent-enough cam signal that I can do sequential ignition. For the sake of simplicity, I plan to get it running on wasted spark first.

However, the injectors and coils refuse to fire. They're all together on one relay and I suspect I'm going to be digging back into the harness to correct a mistake I've made somewhere. Annoying. Early start at work tomorrow so project for another night.

[jamlip]

Injectors were wired wrong. Coil packs were plugged in in the wrong firing order. Probably some decision I made in the small hours of the night when I’d have been better off asleep.

Anyway, fixed it all and managed to get it started and holding an idle. Was getting a decent cam angle signal so went straight to sequential ignition.

It sounds really serious. I like it.

[gereed75]

congrats Way to persevere!!

[East coaster]

Congrats on getting it running!

[jamlip]

Thanks guys. I swapped over two of the coil packs today (back to front!) and got a smooth idle. I must have wired this thing drunk.

Next step is to get it to rev. Right now it just bogs down if you give it any revs. I'm getting missed VR (crank) sensor triggers, so there might be more to it than fuel / ignition timing.

The thing about tuning an engine from scratch is that you start with four dozen variables, of which perhaps half a infinitely variable, and you just have to smash away at them with some half-baked idea of what is affecting what until you see an improvement in the way the engine is supposed to run. Hair falls out and bad words are shouted till the long after you've stopped really caring. This is what we call a hobby.

[jamlip]

Well, I wasted a lot of time chasing the rev problem and it was entirely my fault - the coils were not grounded to the engine. Instead, Capt Dickhead here had decided to ground them to a wire that ran back to the ECU. Insane. Anyway, fixed now.

It now starts, idles and screams like an absolute banshee. It sounds bonkers and I love it.

This is my third Megasquirt install. First was ignition control only on a 1972 911 (back when they were still cheap, the owners were cool, and the cars fun to play with), second was spark and fuel on a twin-plugged 912 (that was after they had become valuable, but I stopped caring), and now this beauty. The other two installs were straightforward, but Individual Throttle Bodies (ITBs), which I have installed on this engine, are difficult to tune since they don't give a very good MAP (vacuum) reference once the throttle butterflies open up.

When tuning a car, MAP signal is easiest to work from because for a certain amount of vacuum (ie. engine load) at a certain RPM, the engine requires a certain amount of fuel. That's really all there is to it - you put those numbers in a table with MAP on the Y axis and RPM on the X, and the ECU uses that data to modulate the injectors. Most engines create a nice strong vacuum, which the ECU can reference all the way through the rev range. Engines with ITBs do not. When the throttles are open past 20 or 30%, measurable vacuum ceases to exist and the ECU no longer knows what the heck is going on.

So to counteract this issue, you have to blend the signals from MAP, at lower throttle settings whilst the throttle butterflies are still creating a vacuum, and Throttle Position when the throttles are opened up.

That's the theory. I've never done this before. I suspect it will require a lot of work to get the thing drivable.

[76-914]

I suspect that you will figure that out too, James.

[jamlip]

Finished the exhaust. I wanted the pipes to exit together, in the center of the bumper, so I made this. It's some extra bends for the exhaust gases to navigate, but it looks cool and takes quite a lot of the sound out. Used mild steel (for acoustic reasons) and painted it black. The tips are 2" to 2-1/4" adapters in stainless steel.

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1758730394.1.jpeg)

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1758730395.2.jpeg)

These are the mufflers I used - quite large. Mounted them up above the axle, at an angle to allow for articulation.

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1758730395.3.jpeg)

I currently have the engine running on Alpha-N (throttle-position only) mapping. I'm finding ITB mode (blended Alpha-N and MAP) too difficult to digest at the moment, but I have tee'd all the intakes into a metered plenum for when I'm ready to work out how to blend it all together. At the moment the MAP signal does nothing except tell the ECU when the engine is at idle.

Idle with TBIs is difficult. Most engines use an Idle Air Control Valve to bypass a small amount of air for idle control, but they don't work great on Alpha-N because the throttle position sensor has no way of knowing what's going on and the ECU cannot measure this airflow.

Luckily, Megasquirt has a system for controlling ignition advance to target a desired idle speed, and I have managed to get it to hold a nice steady 800rpm. The engine sounds very smooth at idle and is about as quiet as I could hope for.

Lots of other little projects since the last update. I added a coolant reservoir, wired one radiator fan to the ECU (the other will have a Davies Craig fan controller, set to activate at a temperature slightly higher than the ECU). Added an AC evaporator under the dash.

I also had to reroute the accessory belt as I have eliminated the power steering pump. My solution so far is to also remove the spring-loaded tensioner pulley and use this Amazon Small Block Chevy alternator adjuster to tension the belt. Being SBC, the adjuster is in Freedom Units and would not take the 10mm bolt, so knocked out the bearing and found an old socket that fitted snugly in the heim joint. I cut down the length a bit, and though-bolted it to the engine case. Cannot work out if this was resourceful or hillbilly, but it seems to do the job.

I think the belt needs to be about 46 inches long, but I'll find out later. I'm on my fifth attempt at it now and I'm pretty sure my local Autozone think I'm running some kind of return scam (IMG:style_emoticons/default/wacko.gif)

(IMG:http://www.914world.com/bbs2/uploads_offsite/i.imgur.com-26060-1758730395.4.jpeg)