Laminar's RX-8 EcoBoost swap thread

[Laminar]

I'm not having Link output OBDII as I don't need everything it sends and it doesn't send everything I need, so I'll just set up custom CAN frames from the Link with the stuff I'm interested in.

[digitalsolo]

Oh, that's way easier. 

[Laminar]

Big step forward in CAN decoding - I was seeing the CAN message in the ESP32 but needed to parse out each byte and then somehow turn those hex values into numbers I care about. I was getting stuck on how to separate the byte out and was tired of Googling. My son's birthday was this weekend and I got him the HackPack from Mark Rober. Part of the fun is that the coding interface includes an AI chatbot that helps you learn what the code does and offers suggestions on how to modify it and make it work. So I fired up ChatGPT and asked about how to parse my data. After a couple of questions, ChatGPT let me know that the CAN message is already an addressable array and I can pull the equivalent integer values straight out without any extra work, then provided sample code that I could copy and paste straight into my program. Genius! Combined with a helpful post on the Link forums, I found I can take the first byte, multiply it by 256 and then add the second byte and I get my integer value! I'm successfully reading battery voltage, one of the few parameters that means anything on the bench.

So now I need to set up the Link to send all of the goodies I'm looking for, the ASL board to read those values and split them between the three screens, and each screen to direct the right value to the right variable.

The last thing I want to do is add display switching - maybe add some secondary displays that I can check when needed. Maybe put VVT position/target on a screen so I can verify that's working, or a screen for fuel pressure target/actual/duty cycle to monitor that system, etc.

[Laminar]

And then I discovered the downside of cheating on your homework. If it doesn't work, I don't know how to troubleshoot it because I didn't write it.


I tried using ChatGPT's more complicated suggestions for lumping all 8 frames of CAN data together in an array or a dictionary or something weird, but it didn't work well. I ended up just storing it all in a single bytearray object that is only the size of the data being sent. ESP-now can do up to 250 bytes per message, and my 8 frames of 8 bytes of data, after I pulled out the frame identifier in byte 0, is only 56 bytes long. So if I wanted to, I could send a lot more data from the Link than I currently am. Here are my frames. Each byte can be a value of up to 255, so for data that's less than 255 or doesn't need any resolution beyond the decimal point, I only gave one byte. If I needed more than that, they got two bytes.





I worked through the scaling and rounding for the values I could simulate, and then I went ahead and combined all of my previous screen files into a single screen program and set up screen switching based on a byte from the Link. This way I can use buttons to actively switch each display through any screen.


https://www.youtube.com/shorts/9D_BodTXOZ8


I'll still need to make up screens for VVT and VSS and anything else I added along the way like idle state and boost state and fault code count. The ASL can ESP does all of its receiving and calcs in less than 50ms, all of the data easily fits in a single ESP-now packet, so the only limit to the system is screen update rate. Here I have a timer running in tenths of a second, showing how it's not perfectly keeping up with every 10th, but if it needs to skip ahead it does. This might be enough to provide decent reaction time to things like overboost or knock retard.


https://www.youtube.com/shorts/k3ic-LwP0xY

[Laminar]

Spent some time playing with it over Christmas break, (semi-) finalized the screen configurations and got screen switching set up.


https://www.youtube.com/shorts/9ueBwVCoZVs


The screens now show and hide things at will, like launch control setup and traction control setup will appear when those features are turned on, when you turn cruise on, it moves the speed indicator over and brings up a cruise set speed and status.


I still want to make some diagnosis screens, like one showing fuel pump pressure vs. target vs. pump output, the same for oil pressure, another for VVT target vs. actual, etc.


I also want to set up a "race mode" with only the bare essentials displayed and obvious warnings for just the engine-damaging vitals - fuel pressure, oil pressure, lean lambda, and knock. I haven't fully fleshed out what I'm going to do for drive modes with my multi-position dials, but it would be easy to make one of those modes switch each screen to its race mode screen.

[digitalsolo]

The "pop up" for traction control/etc. is a great idea/feature.

[Laminar]

With the gauge programming starting to fall in place, I pivoted toward the physical installation. I decided to go with an a-pillar gauge pod, and got to measuring and designing. I whipped up an initial revision with a single gauge pod on it for test fitting. I want to make sure the screen sits in there nicely, nothing vibrates, and I have room for my cables. This test fit will let me figure out exactly how much room I need for everything, if my profile, fitment, and angles are good, so the final version can be sleeker.





I went to print this out, only to have my printer jam up and create a giant ball of ABS that borked the extruder.








I took that as a sign to go ahead and upgrade to a better direct drive extruder, and why not go ahead and redo the entire control system. With that running and like, 80% good, I went for another print. This thing is difficult to orient because no matter what there's a ton of overhang. On this run, I went ahead and printed upright, but forgot to add supports to the overhang. Unbelievably, the thing actually almost worked  . The gauge pod cantilevered out into space and it printed a bunch of it. Unfortunately, with no support, the ABS actually curved upward as it cooled, eventually going high enough that the extruder caught it on the way past and pulled the part off the bed. Fortunately, it was enough that I could verify the profile around the a-pillar and kind of gauge if the angle looked good.








I'm happy with the fitment around the a-pillar trim.








The angle toward the driver looks good. I'll do another print (with support this time) so I can confirm size and fitment and then finalize the design.


As work winds down this week, hopefully I can get back out to the garage and keep pushing forward with finishing the engine bay, electrical, and fuel system. I want to drive this thing in 2025!

[Exidous]

Find someone with a Bambu printer. They aren't perfect but they are as good as it gets so far.

It'll spaghetti rarely and areas near support won't look great but otherwise usually perfect.

Shaniac and I both have the the X1C at our works. They're fantastic.

[Laminar]

We have a Markforged at work if I need something serious, but this was just an issues of remembering to add supports. I added supports on Saturday and got it printed no issues. Switched to some blue PLA because the roll of ABS in there was about to run out.




Screen-switching button fits nicely.





Popped a screen in there to see what it looks like.





Fitment of the board and cable is good, now I can optimize that routing and maybe get rid of the step for the cable tunnel. I can reduce the overall length and make sure I can stack three of them.

[Laminar]

Got back out to the garage and tried to get some movement on finalizing test fits before pulling the engine again.


The AC compressor was pushing into the frame rail. I had already trimmed back the pinch weld, but it was still interfering with the rail. I took the whole assembly out and cut down the lower tabs by 1/4". I mounted it back on the engine and confirmed clearance to the frame rail. "Clearance" meaning it doesn't actively touch, but there's not enough room to even slip the belt in there with the compressor installed.








I'll need to drill out the spot welds and pull off the remainder of that old cross brace, that will get me a little bit. I may end up notching the rail just a tad so that I can get the belt on and off without pulling the compressor and also so nothing rubs as the engine rocks.


I did get a belt on and confirm I still have good tension after moving the compressor a bit closer. I want to put the laser and angle gauge on there once more before final welding the mounting bracket, I don't think I messed up the angle but I want to be sure before I shred a belt.

[Laminar]

I put the level on the compressor and confirmed my angles are still good. Then I pulled the compressor and took a bit more out of the pinch weld and drilled out the remainder of the crossmember.








I ended up with about 1/4" clearance. That might be enough...






Last week I went in and modified the new fan shroud to fit the radiator, so I went ahead and reinstalled that assembly.



I have an F-150 coolant expansion tank that I've been using to run the engine, but it's pretty wide. I picked up a Mustang tank that's more compact and I'm not 100% where I want to put it.



Another option would be on the passenger side right by the firewall, but that's where the stock washer bottle goes. I deleted the washer bottle on my M3, which means that before every race I have to go to the gas station and wash the windshield, so I'm trying to keep it on this car. I suppose a washer bottle can go almost anywhere, but the coolant tank does a lot better if it's up high.


In the Mustang, it's above the level of the radiator but not higher than the heater hoses, so I suppose it's not the highest point in the engine bay:



I still have to figure out some place for the air intake to go. The weather looks nice tomorrow, maybe I'll make a junkyard run to grab an airbox and see if any of the RX-8s have a steering angle sensor I can steal.

[cholmes]

I ended up with about 1/4" clearance. That might be enough... 

Just thinking like the redneck I truly am: with the engine out, how about heating that area of the frame with a torch to a dull red and beating on it a bit with a sledge hammer?

[Laminar]

Just thinking like the redneck I truly am: with the engine out, how about heating that area of the frame with a torch to a dull red and beating on it a bit with a sledge hammer?

That's probably all I'll need. I'm hoping the engine doesn't move too much with the poly mounts.


Picked up a couple bits. First is an electric cutout. I want a nice quiet daily but I don't want to kill a bunch of power when I need it. Still haven't completely settled on exhaust configuration but I have a couple ideas.



Did a junkyard trip and couldn't find the RX-8 they said they had, but I grabbed a nice compact coolant tank from a Mazda6 and a combined air filter/coolant tank from an F-150.





I won't use the coolant side of the F-150 unit, but I at least want the filter box for an example of how mine will work. I think I can pull in air through the front from above the passenger side of the radiator, up through a filter, then split out the top two ways, one to each turbo. This at least gives me ideas for fitment and mounting, and that box supports an easy 500hp on the Raptor so I know it won't limit me.