Lucas' 1991 Mazda Eunos Roadster

[Themi]

I don't quite know if the head makes much of a difference, but it isn't well tuned right now.

It idles at about 12:1, is closer to 10:1 under WOT and cruises at about 14:1. I've done about 220kms like this, and really need to spend some time sitting down with logs and changing some cells. I also believe the VVT is still not working, and sits at full "retard" at all times. I've been having fun with things like launch control and some mild burbles during decel, rather than the important stuff 

Tuning with Speeduino has been great though, I believe it has most of the features of MS3, and more than MS2. Between when I started this project and finished, there were many many new features added to the speeduino, such as dual fuel/ignition tables for hybrid alpha-N/SD. This is great. There is also a huge community of people creating plug-and-play boards using this technology as it is all open source. Also ones with more fuel/ignition channels, or high current inputs/outputs etc. Seriously cool stuff, and still very cheap compared to the proprietary ECU's such as from MS.

The WOF guy thought it was cool  I don't think ITB's go against VIRM's, the head is questionable but I guess it is mostly the same stuff.

I've just purchased some trumpets from MAX Fabrication in Tauranga. Super friendly, good communications and ultra fast turnaround time. I would seriously recommend them to anyone wanting some fab done.

I also got a new steering wheel (Momo mod.78 320mm), as a treat to the car for getting a warrant. Off to Beach Hop tomorrow morning, hopefully the car makes it there and back in 1 piece! I think I will be the least appropriate car in Whangamata, almost the polar opposite of a tidy american muscle

[Themi]

Also, thanks for the support! Glad someone else is getting value from my work It's only fair given the inspiration/knowledge I get from your builds too!

 

Sneaky pic of the trumpets just sent from the fabricator...

[~Slideways~]

28 minutes ago, Themi said:

I don't quite know if the head makes much of a difference, but it isn't well tuned right now.

It idles at about 12:1, is closer to 10:1 under WOT and cruises at about 14:1. I've done about 220kms like this, and really need to spend some time sitting down with logs and changing some cells. I also believe the VVT is still not working, and sits at full "retard" at all times. I've been having fun with things like launch control and some mild burbles during decel, rather than the important stuff 

Tuning with Speeduino has been great though, I believe it has most of the features of MS3, and more than MS2. Between when I started this project and finished, there were many many new features added to the speeduino, such as dual fuel/ignition tables for hybrid alpha-N/SD. This is great. There is also a huge community of people creating plug-and-play boards using this technology as it is all open source. Also ones with more fuel/ignition channels, or high current inputs/outputs etc. Seriously cool stuff, and still very cheap compared to the proprietary ECU's such as from MS.

The WOF guy thought it was cool  I don't think ITB's go against VIRM's, the head is questionable but I guess it is mostly the same stuff.

I've just purchased some trumpets from MAX Fabrication in Tauranga. Super friendly, good communications and ultra fast turnaround time. I would seriously recommend them to anyone wanting some fab done.

I also got a new steering wheel (Momo mod.78 320mm), as a treat to the car for getting a warrant. Off to Beach Hop tomorrow morning, hopefully the car makes it there and back in 1 piece! I think I will be the least appropriate car in Whangamata, almost the polar opposite of a tidy american muscle

 

Awesome stuff, have you played around with the Autotune option at all? I was planning on trying that out on my mx5/speeduino/itb thing but its been about 14months since I started a 'quick respray' for it.

[Themi]

3 minutes ago, ~Slideways~ said:

... Autotune ...

TL:DR is that it is great, and friends of mine swear by it (even for ITB's).

My issue is that I am chasing a verrry intermittent misfire. 1 cylinder burbles a bit during idle and sometimes under acceleration. This I believe is either a vacuum leak on a single cylinder, or they aren't properly synchronised (could also be injector imbalances). Misses read lean so autotune pumps more fuel in. Additionally, a feature I would like added to the autotune is for it to not correct cells if acceleration enrichment is running. Currently under TPS smashing the car is a bit rich and leans out the cells, but if you are static in that cell (say going up a hill with a steady throttle) it will then be quite lean.

Whereas if you tune it yourself you can ensure you are only seeing data from the logs where acceleration enrichment isn't happening. Finding the actual AFR during static loads and do the "autotuning" yourself.

Admittedly if you are a sensible driver, you can be very gentle on the throttle and autotune will work great

[Roman]

Yeah autotune can work great if it's got good filters for steady state. 

But usually doesnt exclude enough stuff. 
Or sometimes makes massive waves in the fuel table as it fluctuates half way between two cells, and keeps making one richer and one leaner to try get target.

[Themi]

2 turbo (1.6 and built 1.8) and 2 ITB (both 1.6) mx5's got out to Hahei beach from Whangamata, great stretch of road to find redline on up and down the windy sections. First time the four of us have been in the same place since a meet at Kelly Tarlton's carpark in what must have been 2018 or 2019.

My car did great, it really roars with the new trumpets. Downside is that it never got hot, and we are now assuming there is no thermostat in the back of the head oops! Another issue is that my throttles got very stuck and wouldn't return once opened. It turned out one of the adjustment grub screws in the linkage was loose, and tightened down on to one of the linkage tabs locking the throttles together. Due to my shit welding this misalignment was enough to hold the back two throttles open independently of the front two which was not good. We managed to fix it at the beach and made it back in one piece.

Next thing to do is re-wrap all of the new bits of loom, replace thermostat, work out some sort of filter which fits the trumpets, and possibly get an undertray sorted. As it stands, there is a lot of area under the car which is open direct up to the throttles.

All in all I am glad the car made it to beach hop, although it does feel like a fish out of water  Looking forward to the drive home on Monday.

[Themi]

Hiccups!

Here we see VVT target Angle (green), influencing the intake cam advance (red). As green increases, VVT duty increases (not shown) and the cam advances (red). White is RPM.

 

In this image, we see cam angle (red) displaying as essentially a mirror image of RPM (white). This is not correct.

 

I cannot even begin to hypothesise why sometimes each of these two scenarios occur, but it is a pain. As the closed loop VVT control does not work when the cam angle is out of reasonable range, I must operate as an on/off "vtec" style vvt. This definitely gives a bit of kick up the rear end at high load and medium RPM, but is not perfect.

 

If anyone has a MS3 or speeduino running a late model NB with VVT, I'd love to hear from you about your wiring and VVT settings.

 

[Roman]

I know someone that got Toyota VVTI working with a Speeduino.

It could be that the PID settings are a bit whack?

Not sure if you've seen this video before, but it's my go-to when I forget (often) how PID works! 

Brilliant vid.
 

 

[Themi]

7 hours ago, Roman said:

PID settings are a bit whack?

They most definitely are, but the cam angle reading is derived from the difference between cam trigger and crank trigger angle. I don't know if PID would (could) mess that up. Odd regardless but I'll give the video a go to help me fix the pid

[Roman]

On the toyota running speeduino.

With the factory loom the earth for the cam and crank triggers tied together and ran to the ecu as a single wire. The only way to get vvti / cam angle sensor to read correctly was running a new wire all the way from crank trigger to the ecu.

As I think the filter for interference or whatever on the speeduino wasnt great.

[Themi]

I'm running 3 core shielded wire from both cam and crank all the way to ecu, where the ground wires go to the ecu and shields tie together to ground (on ecu side only ).

I do think interference could be an issue, no scope though. Maybe it should be an investment. Thanks for the advice Roman I'll keep digging.

Pic with trumpets for interest. 970kms done on this head now.

Also purchased a thermostat so temps go above 50 while cruising whoops. And some new hoses for the radiator setup, it is made of used and chopped rubber hoses right now.

[Themi]

Here is my thought process, looking at the speeduino firmware and doing some brain thinking (usually a terrible idea).

This is an annotated version of lines 2232-2236 from github.com/noisymime/speeduino/blob/master/speeduino/decoders.ino

//Record the VVT tooth time
//toothCurrentCount = what crank tooth we have last seen. Values 1-8 for 4 tooth crank wheel over 720* cycle.
//curTime2 = what time the cam tooth was last seen
//toothLastToothTime = what time the crank tooth was last seen.

// if (we have just seen crank tooth 1) AND ("current time" is greater than "the time that the last crank tooth was seen")
if( (toothCurrentCount == 1) && (curTime2 > toothLastToothTime) )
{
	lastVVTtime = curTime2 - toothLastToothTime; // recalculate VVT angle
}

What I am hypothesising is that the VVT angle code for the miata99-05 trigger was written/tested only for the stock setup (i.e. crank tooth 1 passes the sensor at exactly 0*-ATDC). This portion of code is on an interrupt when a cam tooth passes the cam sensor, however the VVT angle is only re-calculated if the last crank tooth to pass the trigger was tooth 1.

In my setup, crank tooth 1 passes the sensor 14* BTDC.  Crank tooth 2 is 70* after that, so 56* ATDC.

I believe at high RPM, oil pressure is high enough to push the cam as far into retard as possible. Crank tooth 2 passes the crank sensor before the proper cam tooth, causing the VVT angle (intake cam angle) to not be re-calculated.

This means that the old calculated angle remains the same in the ECU. However at high RPM, the time between crank tooth 1 and cam tooth represents a larger degree of rotation (rpm increases = time between cycles decreases = increase of angle).

I'll try recompile the firmware to temporarily accept toothCurrentCount as 1 or 2 to recalculate VVT time. I don't see any ill effects of this but it might solve my problem.

 

Phew?

Update:

I've left the text there for future reference, but I solved this issue by changing cam/crank triggers and changing a few VVT settings. I never got VVT working on the miata trigger so it might still be a bug introduced when the cam/crank trigger offsets are modified like in my example.

 

[Themi]

Not a lot has changed I replaced the radiator hoses as previously mentioned (in early december?). I didn't touch it over the Christmas/NY break, although I have downloaded some photogrammetry software (recap) and will give that a go, trying to model a new intake manifold.

I will throw some measurements up here, in case anyone in the future wants to know. Note: only the dimensions in mm have been measured by me (with micrometers). Any dimensions in degrees are third party, and will need to be verified in future.

Intake cam duration	230° @ 0.003"
Intake cam opening	-3° - 32° BTDC
Intake cam closing	53° - 18° ABDC
Intake cam lobe height	41.50mm
Intake cam base circle	33.00mm
Intake valve lift	8.50mm
Exhaust cam duration	240° @ 0.003"
Exhaust cam opening	53° BBDC
Exhaust cam closing	7° ATDC
Exhaust cam lobe height	42.00mm
Exhaust cam base circle	33.10mm
Exhaust valve lift	8.90mm
Valve overlap (Separation angle)	10°-39° (115°-98°)

Note: Exhaust cam does have the tabs for a CAS at back of cam

The stock Mazda B6 engine has only 7.8mm lift! This confirms my suspicion that the ZL-VE has bigger cams than the B6 did. Awesome.

Pic of new radiator hoses (the aluminium tube slides all the way up to the 90* bends, so I can trim them back for a cleaner look later)

[Themi]

Sooo I made a bracket for the IAT sensor, so it actually reads intake air temps, not fuel rail temperature(?). Can you tell I'm not a fabricator?

The one downside is now there is for some reason a short to ground somewhere up under the dash blowing an engine bay fuse... I verified what wire was shorted, disconnected everything and still shorted so it must be contacting ground somewhere along the wire. After fiddling, it is no longer shorted but I'm not sure why, which makes it harder to fix.

I did want to try take this to the os meet tomorrow but that's looking like a more distant goal each minute. Seat and cluster had to come out for easy access, still awkward up under the dash.

[Themi]

Found the short. A poorly modified door buzzer (I promise it wasn't me) had frayed wires, and was right behind the drivers side internal fuse panel. The ground was touching the wire which had a short to ground. All fixed!

I did make the meet although stupidly did not take a picture of my car there. Tidied up the wiring over the coils/injectors also (finally).

I think my to-do list now has dash gauges as the highest priority. I have speedo, fuel, handbrake, high beam and indicator lamps. Every other dial/warning lamp is inactive but should be used.

Tacho won't be too hard, ecu has an output for that. Oil and coolant gauges will be tricky, I can either use the OEM sensor and run the wire to the dash gauge, or I could find a way of driving them with an external Arduino or similar. Fun future project? For now I plan to use a programmable output to control the "o2 temp warning" bulb if the coolant temp climbs above 92° (or below 0 indicating coolant temp error = no fans).

Although I do have an oil pressure sensor, it doesn't seem to read properly and I think there is a wiring issue to be dealt to.

[Themi]

So I gave photogrammetry a try to start the CFD process, using two different tools. My camera settings were incorrect (ISO too high, and I corrected for lens distortion and chromatic abberation which the software will do itself) plus the head itself was wet. I had just given it a rinse off with some degreaser. A tip is that I did this in bright sunlight but under shade, so there was a lot of diffuse lighting in the area. You want the object evenly lit with no shadows for best results.

Autodesk Recap Photo:

Paid-only tool, which requires cloud credits (paid) to use. Students get unlimited cloud credits but suffer an extended queue. It took about 2 hours to generate this 3d model using the student queue. Honestly, it is far better than I expected.

Agisoft Metashape:

Also a paid tool (30 day free trial) but can process locally (Arrrh me mateys!). This took about 30 minutes to process, with some babying required. I believe Metashape is slightly less accurate than Recap especially on sharp edges, such as the transition into the port from the intake manifold (these get rounded slightly in Metashape). I chose Metashape over something like Meshroom as the support for openCL (amd GPU's) is better. This model I had already "cropped" away the environment and rest of the head to reduce edges as this was imported into fusion (worst tool for meshes...).

Honestly, both of these are good enough for what I want. You can see on the closeups in the above image that even the water droplets are shown in good detail on the surface of the intake manifold mating surface. The stud threads leave something to be desired, but I think using this to model threads might be asking too much

Anyway, in the next episode of this I will be modelling a port as a single body (not mesh) so fusion doesn't throw a wobbly. I will also make sure the camera settings are adequate and I will try to set up some white objects in the sun around this to increase the diffuse illumination especially at the bottom of the ports. I may also try the baby powder trick, dusting the inside of the ports with baby powder to make them brighter.

I do also want to use photogrammetry on the car as a whole, run CFD on the mx5 and compare its drag coefficient to a lobster

[Themi]

Sort of big news, sort of mundane technicalities. I installed the 36-2 trigger wheel I have had lying around for ages. In doing so I needed only a single tooth cam trigger wheel. I removed the rocker cover, made a cardboard cover and used an angle grinder to cut the two teeth off (pic taken half way through the process). I was very careful to have no metal possibly enter the head (there is blu-tack around the base of those teeth).

Settings:
Missing tooth 36 with 2 missing
Single tooth cam trigger
190* ATDC primary trigger angle
Rising edge (leading) cam/crank
Weak filter

Pic of setup (36-2 wheel shown for posterity).

And pic of the 36-2 trigger beside my crank position sensor:

I also got VVT working, properly and consistently! It was brought to my attention that solenoids such as the VVT actuator solenoid require flyback diodes. I added a 1n4004 right at the solenoid, and once that was working, VVT was almost perfect. A few tweaks of the PID values and it was mint! I need to re-pot the diode and connections into the back of the VVT solenoid but it runs very well, and the tune feels more responsive with VVT not just on/off like it was previously.

VVT settings (only the important ones):
Closed loop
300Hz
Increased duty = retarded cam timing (backwards of the Miata 99-05 trigger)
-159* cam angle
PID: 0.81, 0.47, 0.023
Min/max duty: 40/60% (for now)

I removed the engine strut brace as that needs tidying up, plus swapped VVT valve body(?) to the non-painted one, so that I can re-paint the other one.

[Themi]

Someone asked, so I thought I'd just put it here for anyone in the future looking for this information. Units to the thousandth are micrometer measured, anything in the 1's is ruler.

ZL-VE valve and spring information:

Valves:
Intake:
29.25mm head diameter
91mm overall length
6.00mm stem diameter
3mm tip length

Exhaust:
26.00mm head diameter
91mm length overall length
6.00mm stem diameter
3mm tip length

Springs:
Intake:
42.50mm length
23.25mm outer diameter

Exhaust:
42.50mm length
23.25mm diameter

So yeah, smaller intake valve than the B6 by 1mm. Darn, not what the trustworthy internet had initially lead me to believe! I should have checked before installing the head but what can you do? FS-DE valves are 31.5mm, which look like they would fit... but it would be darn close! I might have to contact some engine builders and ask how feasible it is to put seat inserts in so close to the edge of the head? 32.5mm (FS-DE +1mm) would be a definite NO as the valve would actually start outside of the bore of the engine.

Looking at some catalogues, toyota 3E motor in the tercel seems to have 31.15mm OD exhaust valves, 6mm stem and about 92mm total length. I need to get my hands on one! Why on earth did a toyota 1.5l have 31.1mm exhaust valves?

[Roman]

Yeah but if the port shape is better then it's not going to matter if the valve is smaller. 
Think about it, why would a bigger valve flow more, it's just a bigger thing in the way of flow? 
Put a big valve in there and you get problems like the valve being more shrouded by the bore wall and edge of combustion chamber.
It needs a bigger seat but then unless the size or the port close to the valve is bigger too then you've not achieved much. 

Does this head have a different valve angle, or port shape or whatever? 

You still might be better off with a better shaped port with smaller valves TBH.

With VVT you could easily punch ahead of a fixed timing motor with fixed valves, taking whole powerband into account. 
 

[Themi]

8 minutes ago, Roman said:

...it's just a bigger thing in the way of flow...

Good point, I guess I was just angry at the b6 having a larger intake valve!

2 minutes ago, Roman said:

... problems like the valve being more shrouded by the bore wall and edge of combustion chamber...

Yep after looking at this, that would definitely be an issue.

Shit photo but the clearance is only about 1.5-2mm and any larger you would need to remove a significant portion of the head material around the valve. Even from factory they have made clearance in the head for the valve, you can see that it is slightly larger than the "round bore" of the head.

 

5 minutes ago, Roman said:

You still might be better off with a better shaped port with smaller valves TBH.

Probably yeah. It is hard to know what exactly flows better, but this head does theoretically have a better port shape. I think just by looking at the clearance to the cylinder wall, larger valves might be off the table.

Thanks for the input, I think it has been made clear that larger valves would be a losing game. Those 2zz valve springs and working on tuning and intake manifold would probably be better uses of my (lack of) money.