Roman's 4GR V6 Toyota Carina

[Roman]

Because it's easier than doing any actual work on the car.
 thought I'd ask chatgpt what it thinks about exhaust design, with regards to preserving the high pitch of the engine that is currently getting nuked.
I was expecting it to trot out the usual stuff "make a 6-1 exhaust manifold" "fit an x pipe and an H pipe and a Z pipe" 

However it had some surprising suggestions. When I queried its sources it had some actual university studies etc to back up its findings.
I'm trying to have an exhaust that primarily makes zoomy noises in the 300-500hz sort of range.
The main things it stressed was that muffler length/shape/type/position are all critical for not destroying frequencies in this range.
Which is not something I've ever even thought about.

It says any oval muffler, (like I'm currently using)  is off the cards.
Mufflers must be round shaped to best preserve the pulses, and larger body diameter the better (better noise absorption with no downsides) 

It says aim for a straight through mufflers with no internal louvres.
A front resonator about 600mm past extractors around 180-200mm long.
Then a rear muffler, also straight through and circular with no louvres that is approx 300mm long. 
The first resonator near the extractors gets rid of all of the ear piecing high hz screeching noises.
Then the larger mufflers at the back do the bulk of the sound reduction without killing the pitch.

Surprisingly it also suggested that any addition of an X pipe or an H pipe will ruin how this works. 
So either it's scheme is fairly holistic and all needs to work together as one. Or it's fundamentally misunderstood what I'm aiming for. Haha.
It suggests keeping the banks totally separate which I found interesting as that's contrary to what seems to usually sound good.
Also a straight cut exhaust tip with no flare or rolled end. 

I asked it whether I'd be better off running both banks into an LFA style rear muffler box.
It said that this will achieve largely the same sound as that. But with less trial and error to get the box volume correct for an LFA style setup.
I'm guessing it will be a bunch louder though...

There's a huge amount of online real life evidence that points towards an x pipe making a motor like this sound way better. 
So ordinarily I'd just go straight for that.
However the reason I'll give this setup a chance is that keeping the banks separate for starters makes prototyping it exceptionally easy. 
There arent really any downsides to giving it a go, apart from a few extra hours. 
But it will scratch the itch to know whether or not chatgpts exhaust design suggestions are valid.
It's now made me curious about what sort of muffler lengths people have been using. 300mm max seems kinda short.

This is a before/after of what I'm going to do for starters
(before probably finding that chatgpt is full of crap and I end up fitting an X pipe)

[Roman]

Ethrottle fix 

I bought some different ethrottle motors that have a higher ratio gearbox on the front. So it can only move slower, but has better holding torque. 
So far so good, still really snappy on throttle. 
But now it only needs between 10-20% duty cycle to hold the throttle at any position. Previously at some points was near 50%. 
I changed the bracket design so rather than being single piece. 
They're two separate ones which slide onto a carbon fiber shaft. 
So there are no issues with the centers being off.



Works well, I have had it on a test mode all afternoon spamming throttle open and closed a whole bunch. 
I've also got the Toyota TPSs mounted on properly now, and its much much much better.
I've now got super aggressive PID settings and it's all working great.

Exhaust stuff

For the exhaust revisions, first thing was cutting off the extension part of my manifolds, and putting a V band on.
So they're easier to get in and out of the car, and also a lot easier to get test sections off and on.
A lot of people warned me about warping the flanges, so did it with the other half bolted on and slowly stitched around it while letting it cool a bit.
Seems good so far.



The next step  was to do some open header limmy bashing to make sure the neighbors horses are awake.

Will be interesting to do the same test once the 2-1 section is on there. For science, of course.

I am just going straight to making a 2-1 section instead of the chatgpt suggested dual 3-1 scheme. 
Like it would be interesting to see if it works but my patience is wearing thin for exhaust stuff at the moment.
Watching the video of the Stratos with the discrete 3-1 pipes vs the 6-1 and its absolutely no contest. The 6-1 sounds wild by comparison.
I will still be taking onboard its suggestions about muffler type/length/etc.

I started making a shallower angle 2-1 section. With about a 12 deg angle this time instead of about 45 deg. 
I've not had much luck trying to do cuts like this on the right angle and line, so printed a template that showed where to cut but also the angle to hold the grinder up against. 
Worked pretty good. 



It was a real bastard trying to weld up the gap in the middle though. It will probably still leak right at the top. haha. 




Then my next problem is trying to get the 2-1 section equal.
So one bank is further back than the other, adding to this is that my manifold design is different each side. So the collectors are at different distances back. 

The Vband on each of these bolts on to the collector, then I marked where the pipes line up with the gearbox crossmember.
So I can get an idea of the difference.



So I need to squeeze about 140mm into one side prior to the 2-1 merge. 
Will take a bit of stuffing around. But hopefully worth it.

Injector issues

Something else that is annoying after reviewing things from the drags. 
My injectors are at 96% duty at 9750rpm. 
So absolutely completely out of injector. 
"They say" you should aim to not exceed 85% duty cycle. 
I'm now getting to 85% at only 7500rpm.
So definitely need new injectors now.



With my old intake, I was only getting to 80% duty right at the limiter. Only 68% duty cycle at 7500. So a big difference.



These are around 330cc at the fuel pressure I'm running, and so its chugging down about 1.7 litres per minute at full steam.
Annoyingly the BMW V10 uses a non standard injector length, and has a longer than usual pintle on the front. 
So it might be a bit of a pain finding some bigger injectors.
But need to figure it out.
Not the worst problem in the world to have? As it means the engine should actually be making some alright power now.


Gearbox issues 

I pulled my other gearbox out as it wasnt shifting properly, and the problem would happen even when the motor was off. 
When putting the factory 3S bellhousing back on, the old gearbox now works fine. 
So I'm thinking there are a few possible issues... 

1. When using a rubberband to hold the reverse fork thing on, this could be causing the issue. Maybe. 

2. The weight of motor and gearbox is flexing the bellhousing because I've got no bolts around the bottom. Will test this by jacking it up by the center and see if anything changes.

3. The bellhousing welding warped the front casing slightly. 
apparently it takes hardly anything, to pull the shift forks or front bearings out of alignment and cause issues. 

4. Something else. 

Will pull the motor and gearbox back out once I've finished with exhaust stuff, and do some more testing. Hopefully there's an easy-ish fix, but it was interesting to see that the factory bellhousing back on, fixed the problem.

[Roman]

Slow progress on the exhaust, but I've manage to cobble together a 2-1 section that is nearly equal (Still needs about 75mm more, on the wiggly side) 



After the 2-1 join it'll revert back to a single 3" pipe and a single barrel muffler for now.
Here's a neighbor's horses awakeness test with the 2-1 mostly finished. 
 

I think I've cured the 3 cyl sound at least. 
So its a good starting point, can hopefully fine tune it from there without having to stuff around with that part anymore.
Will just have a single 3" pipe from there back, and a single barrel shaped muffler.

Theennnnnnnnn
I need to get some air filters on at some point. 
Now that I know what the runner length is going to be, more or less. can plan an airbox.
After an extensive amount of fucking around with test prints and so on, have got an airbox shape sorted.

I'm thinking the red backing plate bits will be alloy plate (less the radius shown) then can hopefully get a nice hidden detail with cap screws coming in from the back (rather than studs)



It took 30 something hours to print the two parts, not including the 2x failed attempts due to powercuts.



I was initially planning to just print PLA and then wrap in carbon.
But, I think I'll try make some proper molds this time and try get a nice surface finish. 
Then maybe just wet layup on the inside. not sure yet.
Just get this bit finished first.
Sanding sucks.



Please ignore the fact that I say this is about fitting air filters, yet have no obvious way to fit an air filter to this so far. 
 

[Roman]

Alright, so had to pause on the airbox stuff because my ethrottle motor shit itself again. 
As I'm not sure how new system would work they might clash with airboxes.
However, I found some similar motors to my existing ones. But from RS components instead of aliexpress.

They have motors of similar design to the ones I'm using, but they better list the specs including max current.
My existing motors are rated to 1.2 amps, where as ethrottle will push out around 2 amps at max. 
So I found some replacements that are rated to 2.8 amps. So fingers crossed thats going to fix the problem.

Next  thing, is that my motor still sounds quite rough and not very "v6 smooth" kinda thing. 
I've been wondering if the small amount of imbalance in 2-1 section is the cause, and if so, how I could test that. 
Since my engine has VVT on the exhaust side, I can adjust the timing of when the exhaust valves open. Usually all at once, but if I unplug one side. 
then I can move the opening/closing timing and see if it eliminates the imbalance.

I am guessing there are only a few degrees of imbalance, but I can swing it by around 40 crank degrees which is huge.
So I did this, and the motor seemed really insensitive to it.
Apart from starting to glug when you retarded the cam too far. 
So is something else wrong?
I unplughed all of the coilpacks 1 by 1 while the motor idled. It turns out that cylinder 1 and cylinder 6 both have bad coilpacks.
So I replace these, and now the motor is back to firing up near first crank instead of taking ages. Then also sounds a bit different.

So here is were we are at with the 2-1 section finished, and a 300mm long straight through muffler stuffed on the back for testing. But all coilpacks working ha.
There's no pipe length after the muffler itself at this point.
 

Doesnt sound particularly pleasing to the ear and its also still god awful loud. 
Howeverrrrrrrr  its making the right sort of noises to start with, that can be beaten into shape so that's good. 

So I think the 2-1 section is good now. So I can work on beating the rest of it into shape from here. 

This is the plan roughly: 




The two main ways I can adjust it a bit are with muffler dimensions and adding a helmholtz/branch resonator. 
Adjusting the muffler controls the high pitch stuff, and overall volume of the system. 




Adjusting a branch resonator can eliminate an annoying drone or idle noise.




There's a program called Audacity that is a free download, than can show you a frequency plot but also do a bunch of other stuff. 
So looking at a plot from this can give you an idea of which frequency you need to eliminate if wanting to do a branch resonator.






But what is also handy is that you can use the graphic equalizer in the program to eliminate that frequency first and then play the noise back.
So you can see what sort of change it would make before committing to it. 




 

[Roman]

Since it was just laying there, I tried bolting my old 2-1 section on backwards, to form an X pipe of sorts instead of a 2-1. 

Instantly better, definitely needs this and then 2x pipes to the back.
 

So since that was looking good, I committed to the idea of 2x 2.5" pipes to the back of the car and bought mufflers to suit. And a prefabricated 2-1 section as it was cheap enough.

So something like this (But welded up)




annnnddd then it went back to sounding like pooz again? 
 

Some further investigation revealed that the reason for this is that the motor had spat a few rockers out. Woops.

So it was only running on 3 1/2 cylinders. 

Why did it do this? 
Something stupid that I brought apon myself. 

I was testing my new ethrottle motor setup, which is a few iterations better looking than this:




The problem is that the ethrottle motors now interfered with the VVT solenoid plugs. So I undid the bolt and rotated the solenoids out of the way. 
I forgot to factor in that this means there's no oil going into the pulley anymore. 
So the cam would have been pushed to home position by valve spring pressure, then whipped forward really quickly once past the cam centerline, faster than the springs could react so the rocker flicked out. 

Thankfully there's no major damage, it's just an absolute ball ache to pull the motor back apart to fix it. 

So, uttering the most tragic and soul/budget/project destroying words ever uttered:

"Since I'm doing that, I may as well do this"

Since it's about 4 hours worth of effort either side of removing the cams. I may as well get a few other things done at the same time. 

So I'll pull the heads back off and port them. I didnt want to go through that effort again, if this setup was just going to break stuff the same as last time. 
I'm also fairly sick of dealing with some of the annoyances of having the cam timing being off by 1 tooth. 

So I'll leave the exhaust cam as it is, but retime the intake so I can get more advance. 

I'm currently capped at +20 degrees advance right to 6500rpm, it probably wants 10-15 degrees more advance than that for best power through midrange. 
Then it will idle a bit nicer too, as I'm not having to run the solenoids at idle condition.

So yeah! Fun times. 

Rough order of operations:

-Pull heads off and port gross intake part as per last time
-Make solid lifter / hydraulic bits so I can find out how big a stopper needs to be in the pulley
-make stopper 
-redo cam timing etc
-Reassemble and refit everything 
-buy and fit new injectors
-Then hopefully dyno time again. 

The one piece of good news is that it looks like my gearbox woes have been fixed with a new plastic bush in the shifter. Phew! That was the job I was least looking forward to figuring out.

[Roman]

I stripped the motor back down and ported the heads. I was just going to fix up the mainly gross intake part, but, I found my favourite dremel bit.
It didnt take long to do everything the same as what I did on the first engine. So I opened up the exhaust ports a little too at the outlet, and blended the bowl cuts on everything.

Two of the valves were bent, from where a rocker had dislodged and then just kinda pushed it over slightly. Only very slightly, the valves still came out without too much issue. 
However it's a good thing that I decided to port the heads (which necessitated removing the valves) as otherwise I'd probably not have noticed. 
So its made me realize that even if it was possible to just pop the rockers back in, its not a good idea. 

Anyway, I got all that finished and the motor all back together, and back in. 
With new 670cc injectors, thanks to @chrisR for helping me get some.
However it wouldnt rev over cranking rpm. I tried a bunch of different things but concluded it must be a physical problem. 
So motor fully back out, motor fully apart again. Nothing appeared to be wrong, apart from the cam timing must have been out. 
As there are 3 chains which never line back up once you first rotate the motor, and each pulley needs to go on non standard positions. 
It's near impossible to check the timing once it's been rotated. So just pulled it all out. 

As per usual this involves about 300 litres of silicone as the sump and front cover need to come off for anything timing chain related. Ugh. 
It's nearly good to go back in, weather and WW3 permitting. 
Fingers crossed its good this time. 

 

[Roman]

Alright, so, back up on blocks for some driveway tuning haha. 
I will front load this post with loud noises and then regress into nerdy details.
 

So this setup is as per before, with the X pipe setup and 2x adrenalin R mufflers. But now running on all cyls. 
Sounds a lot better but still a bit of tinny noise that could probably be reduced with some helmholtz resonators. 
I've not retuned any of the part throttle map either, so it sounds a bit grot as it passes through some of the areas where it's widly rich or lean.

Anyway, the total sum of the following changes since dyno:
-new intake manifold 
-runner length corrected
-ported head
-x pipe
-660cc injectors 

There is a pretty hectic increase in the fuel map values. (Not just because of injector change, it was like this before)
The ECU did a surprisingly good job of adjusting to the 660cc injectors, at full throttle on the first run it was fairly close to target.
But basically, there are fairly meaty improvements everywhere.



I did a VVTI sweep pushing the numbers on the table forward and backwards by 5 degrees. 
But interestingly it seemed to like it mostly exactly how it was already. 
there was only 1 spot at 7500 where it liked having somre more cam advance. Everything else the same. 

I've got a fairly decent prototype for the ethrottle motor holders now, but will just have to get them printed from metal before I can drive it. 

Then my current x pipe setup doesnt fit the car at all, it's just to test if the sound was actually going to be any better. Which I think it is. 

So next month's car budget will be getting ethrottle bits printed, and some more exhaust work. Somehow fit 2x pipes over the diff! 

 

[Roman]

E Throttle mounts

It's been tricky trying to figure out the ethrottle motor mounting before committing to a metal print. 
My main concern is that I've only got 2 pickup points on the side of the throttle, and it's normally just meant for holding a TPS sensor rather supporting a cantilevered weight. 
It would be considerably unfun to strip the threads or break the side off one of my throttles.
So I've given it some extra support/triangulation with some arms that link up to the main throttle attachment points.
It looks kinda weird having these big tubes just sticking out, but I'm not really sure what else I can do here.
Anyway, I'm hopeful that this concludes the prolonged fuckaround relating to the intake and ethrottle on this motor.



I've sent the parts off to be made so it'll be 2-3 weeks till they're here.
I've got a tig setup here again so I'll be doing some more work on the exhaust while I wait.
So hopefully it'll be drivable again in a few weeks time once intake and exhaust are sorted.

I've had some downtime where progress on the critical path has been delayed, so have been keeping busy working on some side quests.

Side Quest Project restart #1: GPS data into ECU

I've learned a fair bit about writing code over the last few years.
However usually I come up with a concept thats interesting to me.
But I get bogged down in the maths and other uninteresting drudgery of spending 100 hours to finish something to a polished standard.
Now however. Chatgpt can do all of the heavy lifting of maths and writing boring code, and I can concentrate on generating concepts and testing them which is much more fun. 

So the first of my unfinished sidequest projects that I'd like to have another go at.
Is getting some GPS data integrated into my ECU.
Historically I've had ECU data showing engine stuff, and then a completely separate system for showing lap times, cornering speed, GPS data etc. 
Which is fine, but it would be nice to be able to see information like throttle % and rpm overlaid onto the GPS data. 
My ECU doesnt specifically support having GPS data come in.
But I've got 8x generic analog inputs to use for things.
So I can just say that one is X axis, one is Y axis, one is altitude, one is vehicle speed, one is vehicle direction (or whatever) 
The limitation is that each of these can be a 16 bit value. Meaning the number has a range from zero to 65,536.

16 bit value is cool but it's not big enough to fit a full latitude and longitude in there.

Example: 
Latitude: -3721.24130
Longitude: 17504.60600

If I were to remove the decimal points the number is way too big.
In the past I just tried truncating the GPS values to show the last digits.
However if one of the digits changes that you arent including, your map ends up wrapping around.
You can see where I ended up going off one side of the map and coming back on the other on the bottom left.

So I think the best solution to this, is to pick a latitude and longitude value that is near where you are, and just set that as a fixed zero point. 
Then subtract your current position from that, and just send the difference to my ECU.
So for example, the values might be that you are 50,126mm away from zero in the X axis, and 20,024mm away in the Y axis.

Last time I got as far as realizing my truncated GPS data was going to wrap around, and then my brain hurt from the maths involved in trying to fix it. So I shelved it.
But this time around I just need to come up with the general structure of how it works, and the copy/paste machine does all of the stuff that hurts my head.
Then I'm just validating it and setting up the canbus values. 
I've spent a few afternoons on it, and this is working nicely.

I've got some plans to add extra bits once my car is closer to being able to attend track days though.
I want to add stuff like:
-Geofencing to automatically set a fixed zero point each time you're at a particular track (So the zero point is consistent, every time you're at Hampton for example)
-Be able to setup splits when going to a new track, in the car. 
-Real time view of laptimes, splits, predicitive lap timing. 
-Have the ECU's onboard accellerometer data help take care of low speed stuff (like exact time you launch at the drags) to mitigate the low speed problems of GPS.
-Save full GPS data stream and track configs onto the Teensy's SD card.

Most of these features above are already working, but just need some validation once my car's mobile again.
This would be months and months and months of work writing all of the code manually.
It's still a lot of work, but mostly around testing and validation which is a lot more fun.


Side Quest #2: Make Widebands trim nicely

My ECU does not have a particularly great way of dealing with dual wideband / trimming when you have 2x banks.
Which is a very important feature when realistically, for a variety of reasons. A 6 throttle setup with 2 banks/motors is never going to read identically on each side. 
My ECU basically just says "Is bank 1 lean? yes. Add some more fuel" over and over.
And then there is a global gain value that you can adjust to make it more sensitive, or less sensitive.
The problem here is that this does not come close to dealing with the wide range of engine conditions effectively. 
At full throttle there is a large amount of gas coming past the sensor, and very little delay in getting an accurate reading.
So you can have very aggressive and fast acting settings, and it'll trim on target nicely.
However at idle or low rpm/load. There is such little gas volume flowing past the sensor, that there is a significant delay. 
If you have a steady idle, and then start/stop a fuel trim that dumps in 20% more fuel (or whatever) at very low rpm and load it might be over 1 second second until the wideband sees the spike in results. 
If you are trying to trim huge amounts of fuel, and adjusting as though there is only a 20ms delay. It's going to oscillate like crazy.

So rather than using the ECU's onboard logic which is a bit lacking. 
I'm outsourcing it to canbus and can come up with my own methods.

The new method is based around adjusting the delay and magnitude of the trim, based on the amount of gas flowing through the exhaust.
It's calculating a grams/second value in real time then working from that.



I've got this working, and trimming the banks.
I had it record all of the results to serial data, then recorded results under different load and rpm combinations.
Sometimes it was not reaching target at steady state, somtimes oscillating, sometimes pretty good. 
So I just kept feeding the logs back through the chatgpt copy/paste machine and it would make iterative improvements to its PID values and delay time tables.
It's got some failsafes and checks built in, so it doesnt go completely bonkers if a sensor blows up or goes offline or whatever.
It throws an error code if it hasnt seen a canbus value from the sensor within a certain timeframe.

So yeah! I came up with the general concept and how it can trim the tables. Then the general logic of how it works. I did all of the validation testing and logging data.
Chatgpt did the coding part in the middle.
Saved me a whole bunch of hours trudging through boring code, and has made this type of project actually be quite fun. 
This is currently working great and I'd consider it finished.
I just need to figure out the best way to turn the whole system off or on, when I'm wanting trims or not.

Side Quest #3: Digidash in the bin
I've had various iterations of digidash ideas that I've been working on, probably for the last 10 or 15 years. 
I have a burst of motivation to do something, or write code for a feature. Then get bored or distracted again. 
Then change the hardware and have to rewrite everything. etc etc. 
A lot has changed in the last 10-15 years though, regarding how absolutely saturated out lives are with digital screens. 
Initially I thought it would be super cool to have a digital dash, and try keep it to an 80s aesthetic. 
But, now, honestly. I dont want to look at yet another bloody screen while I'm out driving my old car! 

I think one of the best looking analog gauge clusters in a factory car, is the Type R DC2. 
Has a nice clean layout, nice font, good rpm scale etc. 
So I'm going to try and copy the general idea of this, mostly. 



I think analog gauges have a lot of positives compared to a flat screen.

Firstly, looking at the gauge cluster is the main point of interaction between the driver and the car, and one of the first things that you see before you get in and turn the key. 
Having a digidash where it's just a satin black sea of nothing. Is a disconnection from that first excitement of getting in your car and looking at the cool dash.
It makes me sad to see that (for example) an RX3, 240Z, NSX, R35 GTR, and anything in between. Have all had their unique character homogenized with the same aftermarket dash cluster. 
When a car is at a show or something, people want to peer in and see what the dash cluster looks like. Its lame when its all turned off and it just a black blob.

Analog gauges are a lot easier to see in varying light conditions as well, you dont get absolutely obliterated by low angle sun over your shoulder.
Another benefit of analog gauges is that they are a lot easier to glance at quickly. I'm not sure if this is because there is a physical depth to them, but I find it way harder glance down at a digital display and read a value.
I like the idea of having a big center mount tacho, and then the max rpm occurs when the needle is pointing straight up. So it's easy to glance at or see with peripheral vision.

Something else I realized while developing a digidash, is that if you make a context senstive display. There arent ever really many things you even want to see at once. 
If you only display something when there is a problem problem (low oil pressure) there's not much you need to look at. 
If I make my own analog dash. I can still have a small digital display as a portion, without it dominating the aesthetic. There will still be cool stuff to look at.
Also a lot of the features that I wanted to display digitally I can build into an analog setup. 

Example, on a digital dash it's very easy to rescale the tacho to suit a reduced redline if the engine is too hot/cold/whatever. 
On an analog gauge, it's fixed as it's printed on.
Then also, with a digital dash it's easy to show something like the current rpm at which launch control is set. 
On an analog gauge, a bit harder. 
But when I'm controlling all of it, I can build these features into the background. 

I've ordered some addressable LED ring lights, so you can make each segment a different colour or brightness. 
So I'm going to backlight each segment of the tacho. So it can show normal backlight, redline area, or launch target, or whatever with different lighting.
Then my launch rpm can be on a dial on the dash, and it'll light up the particular segment that shows launch rpm. 
Or, I guess when adjusting launch I could just have the needle move to the current target instead of showing current rpm.
Probably nicer that way.

To get an OEM sort of look I'll get a front face lasercut and then use some resin printing to get the transparent areas for back lighting everything. 
Then a combination of straight and ring light LEDs to light it all up. 

To get a super fast and stable response on the needles I'm gonna use some small 270 degree servo motors for the tacho and speedo. 

I'll do some more testing once the parts turn up, but I think I've got the general concept locked in. 






 

[Roman]

I've got a weird old car and now a weird engine in it.
This is fine, but sometimes it would be nice if there were a whole bunch of other people with the same thing so we could compare notes.

My existing process has been a lot of trial and error at trackdays to get the car handling nice.
The problem being that over the course of my ownership, at best I could probably get to a track event 2 or 3 times a year.
So it's hard to get a feel for the effect of subtle changes.

My new engine (not track tested yet apart from drag strip) revs a lot higher than the old one.
Which is problematic for my existing gearing. I've got a 6 speed box but now 2nd gear will run right to 130kph which is fairly stupid. So I've been considering my options for changing final drive ratios.
It's all well and good looking at the graphs and numbers, but I dont think this gives a particularly good idea of how the car might feel.

So I've setup my car in a simulator to try and get a general feel for how various changes might impact it.

The PC game called Assetto Corsa is very highly moddable which is great. There's no Toyota Carina model in the game.
But each car essentially has all of its important stats set via a series of text files that you can edit.

I've taken the base model of an AE86 coupe and then revised it to the known details of my car which are slightly different for wheelbase, weight etc.

What's cool though, is that using Chatgpt/codex I can have it sift through a whole bunch of real life data, to input into the game to refine the model.
I figure that if I can tip as much real life information into the simulator as possible, I'll get a baseline that is as close as possible to reality. So any changes that I make from here, will have a reasonable chance of being realistic.

I've got four pretty good sources of historical information about the car. Corner weights, Dyno results, ECU logs, and GPS data.

I've fed the dyno results image into chatgpt to build the power curve of the motor in the game.
Then it's used the corner weight info to set the weight distribution and overall weight.
I've fed my ECU logs into it, as these have a very good record of gear shift times, engine free revving rate, coast down rate etc.
Then I've fed the GPS data in. From which it analyzes average corner speed, low corner speed, high corner speed, braking rate etc.
To develop a tyre model for the car based on real life data.
Then also total braking force.

I had a really good laugh when the best lap time I've managed after about 30 laps in the game so far. Matches my PB in real life.



How this has been useful:
I was initially thinking that I would need around a 6:1 final drive ratio to make all of the gears usable.
But with some testing it was actually incredibly annoying, as it needed a lot of mid corner gear shifts.
I've found that my diff size (Toyota F series) has an off the shelf 5.3 ratio that is commonly used for off roading.
This is looking like the best overall compromise and gives a top speed of around 220kph which is plenty enough for an Toyota made out of bean cans.

My car has always been very softly sprung compared to what other people run. I believe this is because I've had an excessively stiff front swaybar, and I've just made the rest of the car work around that.
So I've been using the sim to get a feel for what sort of front spring rate might pair well with a softer bar.
It looks like going to around a 200lb front spring will work really nicely. Or at least be a very good starting point compared to a stab in the dark.
I cant wait to finally get my car back out there again, and see how the new setup works out.

On a more modern tyre and with the new engine, should be a few seconds quicker than a 1.21 hopefully.
It's felt well worth the effort of a few rainy days inside to get the car setup this way to test some iterative changes and see what's worthwhile or not.
Unsurprisingly, putting a grippier tyre on seems to be exponentially more beneficial than any other slight fiddling of suspension settings.

 

[Roman]

As per usual with this cursed machine, some good and some bad. Haha.

My ethrottle bits turned up, and they had accidentally printed them out of stainless. 
So they're way heavier than expected. 
Not only that but they were absolutely impossible to try and drill or tap holes into. 
I broke a few taps, then just tried drilling. Ended up melting some drill bits. 
Anyway, they're going to reprint them for me from aluminium. 
But in the meantime I fitted them up so I'm not stuck for another 6 weeks waiting.
Looks kinda weird but they are working great. 



Next thing, I finished doing a 1-2 pipe over the diff. Was a bit fiddly making space for 2 pipes but everything's good.



Then checked for leaks and what not and back to the usual jack stand hoons. 
Doesnt sound quite how I want it yet, but I am sick of doing exhaust work and just wanna go drive it. 
So thats the next plan! 
While its up on the stands I just need to add a few heat shields and move some lines to keep fuel stuff a bit further away from the hot pipes. 
But hopefully not too long until it's good to go.

 

[Roman]

Alright so this is the final in my punishing Car is up on stands series of posts. (maybe)

Video first then nerd stuff second 
 

After all of the headaches and frustrations I've had from persisting with ethrottle.
I'm finally starting to see some of the rewards.
My ethrottle tracking is stable and reliable enough that I can start doing some part throttle tuning decently.
Ethrottle makes it super easy because you can make it so any pedal travel over 20% limits the motor to 20% throttle, if you want to tune the 20% row. 
This engine has been real fiddly in a few areas at low rpm and low load where it goes atkinson cycle and/or transitions out of it. 
I've also got some bank to bank imbalance. Although that's annoying, it's at least consistent now. So I'll just build a compensation table for it.
It might seem silly having the car up on stands compared to driving it or going to the dyno. 
But using the wheels and gearing as a flywheel allows you to reach some low load and low rpm spots that are very hard to get to otherwise.

I spent some time going through the 3%, 5%, 10%, 20%, 50%, and 100% rows and the interpolated between.
So it's good enough to take it for a drive without any surprises. 
It hasnt had any issues with vibration etc causing problems anymore which is great.

Some of my previous VE improvements were a little misleading, because it had jacked up the injector spray time to pretty much fully open. Which doesnt flow proportionally as much as 85% or whatever. 
So with the new intake, diddling the ports a little bit, bigger injectors and the new exhaust setup. 
This is the full throttle VE table differences now that the motor is tuned back to target.

Pretty hectic at 6500rpm for reasons unknown! But it does the same thing every time. Maybe adjusting injector timing or VVT angles will drop the number there to something more realistic.


Supposedly you can estimate power increase by taking between 50% and 90% of the VE increase as % of power gained.

So splitting the difference and on the basis of 75% of VE increase = power increase, it's probably maybe looking something like below.

I'm guessing the 6500rpm thing is some sort of weird VE oddity and it wont actually make power peak there.
I will be annoyed if it's peaking at 6500 on the dyno next time haha. Will be weird to shift to the next gear and end up with better power...
If the rest of it is somewhat accurate, will be happy.



I'm circling the drain towards having smaller and smaller problems to resolve which is great. 
I'm currently mucking around with the VVT PID settings a bit more, as one side is still oscillating slightly. Then both exhaust pulleys are a bit sluggish to reach target.
This is what the intake is doing while cruising:



PID stuff is annoying to figure out, so I've just taken a log of the results and dumped it into chatgpt along with my existing settings. 
It's given me some new numbers to work with, hopefully it'll hone in on some good settings in fewer iterations than using my potato brain to guess at it.

I took the car for a drive up the road and nothing fell off, and all of my rockers sound like they are still in the right place. 
So, in the context of this absolutely snakebit project that is a huge win. Haha.

[Roman]

 

 

[Roman]

Cool stuff first the before nerdy infodump:
 

 

[Roman]

Definitely feels like I'm circling the drain towards smaller and more satisfying problems to solve. 
One "problem" is that its annoying to have a bunch of different programs I regularly use, some of which need a CSV export from my ECU file.

The regular stuff: 
PClink for viewing ECU files and adjusting tune 
MegalogviewerHD for figuring out complex stuff with scatter plots or whatever (.CSV format)
Virtual dyno for power estimations (.CSV format) 
Racechrono for GPS / trackday stuff (It's own file format) 

I dont mind so much exporting things to CSV from the main log file, but its a pain having the GPS data completely separated. 

My ECU is ~10 years old and doesnt support GPS data coming in at all. 
It can have some custom data come in over canbus, but it's limited to 16 bit values.
A 16 bit value cant fit a full latitude and longitude string into it. 
Previously I was just trying to shorten these down, but decided it makes life easier later if I've got the full data.
So If I want to stuff these into my ECU I need to break them in half and send it in bits. 

The next problem with this idea is that sending just latitude and longitude would consume 4 out of my 8 inputs that I can send this way, and I need more on top of that. 

So I had a look at what other variables I can send over canbus. 
There are 8x thermocouple inputs over canbus which are 16 bit values. 
So basically I am sending a bunch of shit into my ECU that my ECU believes are temperatures. But once the data is in there, I can unpack it back to something meaningful later. So long as its there.

So it comes into the ECU like a chaotic garbage can full of junk temperatures, but its actually GPS data.




The next part of the problem is I now need a program that can read LInk ECU's .llg log file format so I can turn it back into useful info.
These are somehow compressed or whatever, so need to figure that out. 
If working from just that file alone it would be hard to decode. 

But I can export a full log file as a .CSV format, acting as a rosetta stone for Codex to decode the original file while I just sit back drinking some coffee.
Living the dream. 

That worked, so then I had it make a program where it views the data and can run through it. As well as start/stop line, splits, etc.
This is a visualization of the path and then the ribbon below the car is throttle position. 

The reason this is real handy, is that if you had a problem like left hand corners causing oil pressure drop. 
It's very easy to see it happening this way, rather than trying to marry up an ECU log file to where you were on the track. 
 

So since that worked and I was feeling drunk with scope creep possibilities. 
I added a second page which is a 2D plot that can have things either shown separately or overlaid. 



Then another page which is a scatter plot, which is the main thing that I ever use Megalogviewer for. 



I cannot understate how excellent it is, to be able to do all of this from a standard log file rather than exporting it to CSV every time.
The fact that this all runs straight from my ECU logs now is bloody excellent. 

Later on I might try add in some functions that replicate the Virtual Dyno program. 
But this is great for now. The major win was getting GPS data in there. 

I can add a start/finish line and splits, and I can have these setup in my dash. 
It can overlay a comparison of a second lap, show min/max times and whatever else. 
My ECU has a 3 axis accellerometer so I can integrate that somehow if I want.
My dash screen isnt grunty enough to show a real time map of GPS data but that's not really useful while you're driving anyway. 
I just want a green or red light that shows me if I was faster or slower on a particular split and/or lap.
So that's easy.
The bulk of it will just be ticking away in the background recording stuff.

I've already got some GPS data from old trackdays which show the coordinates of the start/finish lines and splits on all of the tracks. 
So I can select a track from the list on my dash and it should just work. Maybe. Haha.

Gonna be cool to test it out. 

 

 

 

[Roman]

I Have been trundling up and down the road a bit to see what falls off / what sucks / etc. 
What falls off: Exhaust off the one hanger at the moment. So will add a few more. 
What sucks: Getting a cramp in my leg because the ethrottle pedal is on a stupid angle. 
So I shortened that, and now its mint. 

I'm getting a general idea of fuel economy now that it's driving nicely.
Including cold start, I'm getting around 10.9 liters per 100km when leaving my driveway and doing a few laps up and down the road driving gently. 
Not great, not hugely awful I guess.
Will probably see some improvements diddling the VVT, afr target and ignition timing a bit.
I could probably get it to mid or low 9s with some effort... But then I'm gonna jam in a 5.3 final drive ratio so it's all going to turn to crap anyway haha.

I'm a bit better prepared for the dyno this time, I've got a full set of each of these to try out:



doing some testing where I'm just looking at air fuel ratios though, it looks like the absolute shortest ones might be the winner. 

I keep getting sidetracked from finishing my airboxes, but a big milestone is having proper baseplates finished, and figuring out how it can go together and still be easy to put together. 

I have finalized something where there will be two alloy base plates that sandwich together. One half will be fixed onto the throttles, which will help make the whole assembly rigid which is good. 
Then the other half will be bonded to the airbox. 

Both made out of 6mm plate but milled down so that the perimieter profile locates the box correctly, then it only needs a few studs and cap screws to hold the halves together.
Big thanks to Stu who is cutting these for me, much appreciated. 

Fusion 360 is looking preeeeetty fancy with its rendering these days! 
Unfortunately the real life versions are unlikely to look quite as pretty. But I'll do my best. 








Also I've been working on some branding for my products

[Roman]