Laminar's RX-8 EcoBoost swap thread

[digitalsolo]

Oil lines and ECU work looks super fiddly.   Nice work pushing through!

[Laminar]

More progress. Finished up the ECU connector and heat-shrinked all of the pins I used. I need to figure out what to do with the remaining pins - some may have voltage on them and some may be grounded and I wouldn't want those touching each other, so I'll have to think about that.



Pins I'm not using:
Fuel pump +            86   Ye/Vt
Fuel pump -            107   Gn/Vt
Evap purge            46   Wh/Bn
Turbo bypass            80   Bn/Wh
HP fuel pres            71   Bn/Bu
LP fuel temp            31   Gn
Turbo boost p            94   Ye/Bu
CAC temp            10   Bn/Wh
EGR pres            49   Wh/Vt
EGR v -            19   Bn/Ye
EGR v +            40   Bu/Gn
DPF            32   Gn/Og
Exhaust Pres            97   Gy/Bu
CHT2            8   Bu/Gy
EGR temp            96   Gy/Vt
Sigret            99   Gn/Bu
Inj1            2   Gn/Bu
            23   Ye/Bu
Inj2            66   Gy/Ye
            45   Bu/Og
Inj3            44   Vt/Gy
            65   Gn/Vt
Inj4            1   Ye/Og
            22   Bu
Inj5            43   Bn
            64   Wh/Gn
Inj6            41   Gn/Wh
            20   Bn/Vt

From there I got the harness divided between stuff going into the engine bay and stuff staying inside the car, then separated into the engine bay harness into the ECU plug and other engine bay stuff - flex fuel sensor, cooling fan control, A/C control, alternator control, etc.



Time to test my plug.



I confirmed spark on all 6 coils, but couldn't tell if the injectors were firing. I couldn't hear anything clicking but the injectors are buried under the intake manifold so I couldn't get to them. I also saw readings coming from the sensors for fuel pressure, oil pressure, crankcase pressure, wastegate control pressure, MAP, IAT, and coolant temp.

Then I hooked up the external DBW box to the RX-8 pedal and the Megasquirt CANbus and confirmed that all worked like it's supposed to.

Since my engine originally came with a new version of DBW throttle that combines a single analog feedback with a digital/serial feedback, I had to swap to the earlier model TB with dual analog feedbacks in place. I took out the sixth wire from the throttle and routed that back through one of the now-unused pins from the external coolant pump pigtail to get my signal back to the DBW box.



After all of THAT I hooked up the two CANbus wideband O2 controllers (14point7 Spartan3) and made sure they were configured and talking. Upgrading their firmware required running a shady app in the command line and my MacBook Air with virtualized Windows couldn't do that - it kept complaining about missing DLLs. So I grabbed my son's cheap Win11 laptop and got the firmware updated, then used the MacBook to send the required serial commands to change the CAN ID on one of the boxes so I can run them both simultaneously.



Weirdly, all of that was fun. I needed to confirm injector operation, so I went ahead and pulled the upper intake.



Port injection rails up top, direct below. At this point I realized I wired up to the wrong fuel pressure sensor. The drawings showed one combination pressure/temperature sensor, and one standalone pressure sensor. I had guessed that the combo sensor would be in the direct rail but I guessed wrong, so I'm going to have to swap my pressure reading over a pin. I'm already getting fuel temperature from the flex sensor so I won't waste another ECU input on redundant info. Fuel temperature at the rail is most crucial for starting a direct injected engine, so that doesn't really do anything for me.



Upper injectors are 29lb/hr I believe. Good enough to supplement the DI system, but not enough to support good horsepower on their own. I'll confirm they work for startup, then swap them out for some bigger boys.



Looking down the port you can see the port injection doing its job of keeping things clean:



With the injectors accessible I could confirm all six were clicking when told to, so that was great news.

Next up was testing crank and cam signals. I bought the Megasquirt used, so I had to check and see what kind of sensors it was set for. I got it configured for a Hall sensor on the crank and cam, which took some fiddling and/or soldering.





The 2.7 has no timing marks on the crank pulley, so I needed to add some. I made a little timing pointer and then pulled out my piston stop.



I rotated the engine in both directions and marked where the piston stopped in each direction, then set the center of those marks as TDC on the compression stroke.



I printed off a timing wheel to get my 10/20/30 degree measurements and used a Dremel to etch those marks in place, I sure hope they're right.





From there I was able to turn the engine over with a ratchet and see corresponding tooth inputs from the crank. 

Next step will be figuring out how to see the cam inputs, then pulling the engine out, putting the flywheel in place, priming the oil system, and confirming full crank/cam sync when turning it over with the starter motor.

[Laminar]

Last night I managed to sneak out and buzz all of the control solenoids in test mode - VVT 1-4, wastegate vacuum pressure control, and oil pressure control. I still need to figure out exactly what frequency control they each want, but for now I know the wiring is good. All that's left to test on my plug are the knock sensors, but there's not really a good way to test those - they're looking for a certain frequency during a certain cylinder's ignition window, so that's hard to simulate. Oh, and the cam sensor inputs - I need to build a couple of circuits to bring those signals in, then I can figure out a way to test all four of those.

[Cobranut]

Last night I managed to sneak out and buzz all of the control solenoids in test mode - VVT 1-4, wastegate vacuum pressure control, and oil pressure control. I still need to figure out exactly what frequency control they each want, but for now I know the wiring is good. All that's left to test on my plug are the knock sensors, but there's not really a good way to test those - they're looking for a certain frequency during a certain cylinder's ignition window, so that's hard to simulate. Oh, and the cam sensor inputs - I need to build a couple of circuits to bring those signals in, then I can figure out a way to test all four of those.

If your knock sensors are wideband, you should be able to tap on the block nearby and see some signal at the ECU.
That would at least test the hardware and wiring.

[Laminar]

The ECU doesn't display the raw output of the knock sensors, it only increments a knock count if it sees a knock during a specific timing window.

https://www.msextra.com/forums/viewtopic.php?p=574498#p574498

[Exidous]

You can actively see knock in the later MS stuff like the Pro within the logger. Didn't know it was limited on the MS3X.

[Laminar]

The boost regulation system acts on the same principles as a dome control setup, except it uses vacuum instead of a bottle of CO2. The driver's side intake cam drives a vacuum pump, and that vacuum is regulated by a boost solenoid to the wastegate actuators. More duty cycle is more vacuum, more vacuum opens up the wastegates. So more duty = less boost.

Megasquirt can do dome pressure, though it expects positive pressure, maybe not vacuum. I still need to confirm if it will actually work or maybe I'll need to do some creative scaling of the pressure sensor, but I was able to do a little testing last night. The shop vac can pull about 3" Hg.



The tubing after the solenoid has a built-in pressure sensor so I can monitor actual pressure going to the wastegates. With the vacuum cleaner on and drawing vacuum, I can pulse the boost control solenoid and see it regulate pressure to the wastegates. In an attempt to get a little more head, I pulled out the real vacuum cleaner and it pulled about 5" Hg.



From there I was able to land on a pwm frequency that felt good (26Hz), and determine the control range of the solenoid (4% to 85%).

[digitalsolo]

Can you basically "invert" the ADC graph on the sensor and lie to the MS about boost vs. vacuum?

[shainiac]

That'd be pretty cool if you could get vacuum dome control working. I used on-board air dome control with my MS3Pro and it worked great. After you get a realistic reference table using open loop, the closed loop worked great.

Does the factory vacuum pump also help pull vac on the crankcase?

[Laminar]

The way it is now, if I have 100kPa in the wastegate system, that means I have no vacuum, the wastegates are closed, and I have max boost. If I have 77kPa in the wastegate system, my wastegates are starting to open, and at 33kPa in the system my wastegates are fully open and I have minimum boost. So in that case, the higher the wastegate pressure signal, the more boost I have - that is "correct." The part that I think is backwards is the solenoid control - it defaults to full boost, and only if I apply some duty to the solenoid do I reduce the boost level. I could possibly replumb it to where the vacuum pump goes straight to the wastegates and the solenoid is used to bleed pressure off of the system, in that case increasing duty bleeds off more vacuum and closes the wastegates, increasing boost.

[Laminar]

That'd be pretty cool if you could get vacuum dome control working. I used on-board air dome control with my MS3Pro and it worked great. After you get a realistic reference table using open loop, the closed loop worked great.

Does the factory vacuum pump also help pull vac on the crankcase?

I'm not 100% sure. I know it supplies the brake booster, but I still have to finalize that and the PCV system.

[shainiac]

It's been 3 years since I've messed with an MS3, but do they have a generic PID output option? On my MaxxECU, I use a generic PID output that references dome pressure to maintain a target with the drain and fill solenoids. It works great, but I had to build a dummy table that references a "boost target" for every dome target. It's technically an open loop boost control strategy, but is within a PSI of target pretty much always. That might be another option for you if MS3 can handle it.

IIRC, you also want the source for your dome control to be 2-3X the min/max of your control range. For me, I never needed more than 15psi of dome, so I ran the system onboard air at 45psi. This gives you faster reaction with the fill/dump. If you need 30kpa for your target boost and your pump only draws 20kpa of vacuum, that's not a ton of "extra" to move the air in/out. Not saying it wont work, but it may be less responsive. Also, adding volume to the dome system helps avoid over/under shooting the target. I drill and tapped an AC dryer cylinder and tee'd that into the dome feed lines. The solenoids have to fill that volume as well and it makes the whole system more stable. I think it has more to do with the sensor not seeing spikes/dips whenever the solenoids open/close.

Another side note - if you can draw a vacuum on your crankcase with the pump, it'll help the turbos drain and reduce the possibility of smoking. The vacuum force over the oil level in the pan helps draw oil out of the CHRAs. On my low-mount turbos, I had one that would smoke a little. Adding scavengers to the downpipe and pulling ~1psi on the crankcase fixed the issue. 1psi of draw on the CHRA drains is equivalent to over 2' of water column. Hope that makes sense.

[Laminar]

Totally makes sense - thank you for all of the info.

The MS can do a generic PID loop, and actually includes options for more solenoid frequencies and even inverting the output to the solenoid. I may have to play with that if the stock stuff is going to give me trouble.

I have a buddy with a Fusion Sport that put his oscilloscope on his boost control solenoid last night and said the factory ECU was controlling at 250Hz. MS assumes a normal boost control solenoid and only offers low frequency PWM for that loop. I definitely have to play with that now.

And since I don't have full factory PCV/purge valve control I'll probably be running a catch can. I think it would be trivial to pipe that into the vacuum pump to keep some draw on the crankcase. The low mount turbo drains are nagging at the back of my mind, at least a little. 

[Laminar]

Welp, the Megasquirt developer that offered to make up a decoder for the Ford V6 engine isn't responding, so I went ahead and ordered a Link G4X XtremeX. Upsides are built-in DBW control, cruise control, charcoal purge canister control, better CAN support for the RX-8's stock systems. Downside is cost, but hopefully I can resell the MS3 and the DBW box and recoup a little bit.

Of all of the things I would have expected to prevent the MS3 from working in this scenario, it sure wasn't basic crank/cam detecting.

[shainiac]

My friend has an RX8 with a Link G4+ and the CAN communication with the stock body module is pretty convenient. It should work well for you.
I used to be a big Megasquirt fanboy, but they seemed to have basically stopped developing the product. It used to be that there was multiple firmware releases a year, but I think there have only been two or three since 2017! I switched to MaxxECU and love it. Having robust CAN coms and DBW are awesome.