DIY Fuel injection thread.

[yoeddynz]

Hell I dont know . Ive never investigated it. In fact I have learned more from the last few days on this thread about it than ever!

 

Keep the experiments coming Dave and every time you take out an old biddy at a crossing from some voltage surge add a sticker WW1 style to your rear window.

[Lith]

I haven't had the opportunity to play with E-throttle directly myself, either - though I was involved in a project using an E-throttle and it was pretty exciting/alarming seeing some of the behaviour that happened before the PID tuning was dialled in right.  Scary.  Keep us up to date on progress, and good luck!

[Roman]

I can see why it's a bit of a rock and hard place for e-throttle support from ECU manufacturers. On one hand more and more cars have e-throttle from factory so the vendors need to cross that bridge at some stage. But on the other hand, incorrectly setup and it can be very dangerous and it's left to the end user to tinker with... I guess they dont want to provide people with suggested settings either, because if the settings are bad the consequences can be greater than just blowing a motor up. Which is generally the worst case scenario of most other variables you can tinker with.

I think the instructions from Link about what proportional gain / derivative gain / integral gain do to throttle plate movement are intentionally obtuse. In order to put people off the idea of "Having a go" when they dont know what they are doing... Then stoving into a wall haha.

Adjusting the settings one by one, with the TB in front of me in clear view it was pretty easy to see the effects of each though and get it setup nicely. But I'd be concerned if someone was trying to set this up "blind" or just using default settings which were way off the mark.

It seems a lot of people have trouble at the calibration stage, it's very fussy. The first TPS that I tried gives a perfect signal on the primary TPS output but a very shaky looking secondary output. Second TPS I tried was perfect on both. On a non E-throttle car this TPS still would have been 100% fine when using primary output only.
 

[mjrstar]

E throttle is all good, my coon is now tuned to drop throttle % to prevent impending transmission doom when the trans is cold or on high torque shifts.

Its not quite active enough to drop wheelspin like the traction control probably should but at least its not all soft like the OEM ignition cut for tcl which makes it feel like someone turned the key off.

Also look up sprint booster for options around external control for OEM e throttle setups to make more or less aggressive.

[Roman]

Some Virtual Dyno shenanigans:

 

https://innovativetuning.wordpress.com/our-experiments-with-virtual-dyno/

[cute wee gem]

I could never get Virtual Dyno to work properly with either MS or Emanage. Apparently my Mark II makes like 39hp and my old Gemini makes like 500hp lol - Tried all sorts of settings to fix it but no dice.

 

I think I updated it and it changed the HP readings of both dramatically but still not correct =(

 

Only thing was to bodge the final drive to something ridiculous to get any where near normal readings, but then that defies the point really.

[downtrail]

Trick is when you export the log as xyz.csv open with exael and find your starting point of your run when tps hits 100% then delete all rows before that tgen find the end of the run ie when engine rpm hits 7000rpm or what ever your testing to then delete all after that.

Tps needs to be calibrated correctly or itll just make stuff up.

And on the subject of e-manage..... Ewww lol

I could never get Virtual Dyno to work properly with either MS or Emanage. Apparently my Mark II makes like 39hp and my old Gemini makes like 500hp lol - Tried all sorts of settings to fix it but no dice.

I think I updated it and it changed the HP readings of both dramatically but still not correct =(

Only thing was to bodge the final drive to something ridiculous to get any where near normal readings, but then that defies the point really.

[downtrail]

Trick is when you export the log as xyz.csv open with exael and find your starting point of your run when tps hits 100% then delete all rows before that tgen find the end of the run ie when engine rpm hits 7000rpm or what ever your testing to then delete all after that.

Tps needs to be calibrated correctly or itll just make stuff up.

And on the subject of e-manage..... Ewww lol

I could never get Virtual Dyno to work properly with either MS or Emanage. Apparently my Mark II makes like 39hp and my old Gemini makes like 500hp lol - Tried all sorts of settings to fix it but no dice.

I think I updated it and it changed the HP readings of both dramatically but still not correct =(

Only thing was to bodge the final drive to something ridiculous to get any where near normal readings, but then that defies the point really.

[Roman]

Yeah when my car's running again I'm going to set the datalogging to switch on at WOT, and then end when off the throttle. So it just records little snippets that are a good size for it.

 

My existing (full length trackday) datalogs work awesomely with it, but I have to end up snipping out the WOT runs and exporting first. It goes completely bonkers if you try put in a full length run.

 

Your gearing info, vehicle weight, aero drag, rolling radius  etc etc has to be spot on or it'll give you whack results.

 

If you show your car is in 3rd gear but it was actually in 2nd, it will say that you made zillions more horsepower. Or if you were driving up a hill or something. The drags are a perfect place to test.

[cute wee gem]

Yep all info was correct including gearing. I tried multiple pulls and cut them as per downtrails methods. Still no workie.

 

And yes, Emanage is fucking horrible but it's only temporary. MS3X is on the cards, along with a bigger turbo and COP etc

[cute wee gem]

Only difference being .txt not excel

[downtrail]

Ohhh yeah export it to .cvs and try

[downtrail]

Sorry .csv

[Roman]

E-throttle dramas, sadface.

It seems that my TPS, and my spare TPS, are both mildly faulty to the point where they give error codes when operating.
I'd suspect the wiring or something else instead, but one goes through calibration fine and the other doesnt.
Doing a bit of a search on the net and it seems to be a relatively common thing for Altezza TPSs to wear out, but the FPS is usually fine.
Thinking about this it make sense.
As the Foot Position Sensor only 'wears' when you press your foot on the gas.
But the ECU moves the throttle plate around way more than what your foot moves, because it uses it to control idle speed and a few other things. So the wear rate is probably 5 times higher, and perhaps only at certain points where it 'hovers'.

Since the TPS is now so critical (rather than just fuel compensation previously) I bought a $$$brand new one$$$ today for peace of mind.
Hopefully this fixes it!

[Roman]

In other news, I disabled the E-throttle so I get the car idling and carry on with some other start up tune bits.
Getting my head around the knock sensing setup, it's a lot more primitive than I would have thought.
When using a wideband knock sensor, it only gives two options to listen for frequencies over either 1-10kz or 10-20kz (or something like that)
My engine's theoretical knock frequency is about 6.7khz. But if other engine noise (valves, injectors etc) is within that 1-10khz it gets picked up all the same.

All you can do is log the strength of the signal from the knock sensor while it's not knocking into a table (as it varies with rpm) and then if the strength increases a whole lot above that threshold it knows it must be knock and pulls timing.

In it's defence though it does monitor for knock only during a time window around each cylinders ignition period. And it's apparently pretty good at detecting knock even if it's method seems crude. ( I would have thought I would be able to filter to a narrower frequency)

Need to do some more reading to make sure I understand it correctly and get it setup right though.

[Lith]

Bummer, that's interesting that it's pretty common.  What happens on Altezzas when it happens, do they throw a check light and/or go into some kind of limp mode?

 

I have been having a complete nightmare of a time with tuning lately, so far over half of the cars I've tuned in recent months have had fuel delivery issues and I am getting over it.  

 

Two Skylines, one with Trademe "brand new flowtested high flow injectors" and one with an engine from another car - both with lean misses at idle with reasonable rich measured AFRs, check plugs and things seem off.  I bailed on both tunes, both guys understandably unhappy and it's a hard call being a non-professional/mate doing it for the love of it and saying "I'm pretty sure this thing which shouldn't happen is happening but honestly I know what I'm doing k?" - one of them sent his injectors to get flow tested and cleaned and semi-frustratingly the company testing them stopped before doing the full set because apparently they were all over the show but we didn't get a full set of data because they just decided to clean them and say they were all good.  Oh yeah, one of these Skylines it was my second attempt on the dyno at as the previous attempt was aborted due to the old fuel pump not keeping up.

 

A WRX that someone played around with a split fuel rail setup did basically the same symptoms, both running stock FPRs but one had a kink in the return line causing a creeping fuel pressure on one bank.  All sounds like it wasn't stressful when I announce what the issue is, it's not fun when you have f-all time to get the car tuned and ready for an event and at the time you are trying to tune it you don't know that's what is going on, but you need to know and obvious end up working it out.  

 

Another Skyline which I have been helping for ages now and have systematically found ALL the problems you could think of with it, none engine related (yet? :/) due to shit parts and dodgy workmanship.  Most recently it seemed like everything was finally sorted so did a check over, set base pressure to 45psi and did a road tune on it to make sure it was driveable and all systems seemed go - was happy with that so set aside another session to get it dialled in with a soft road tune at full boost to make sure everything looked good to book a dyno and get the damn thing out of my hair.   Yesterday was the follow up session, was excited that it was going to be it - hooked up the laptop and didn't really do a full check over because the car supposedly hadn't been touched since last time I was there.... car hard started then idled lean and the more I looked into it the clearer it was that the whole map was out of whack in a lean sense.  WTF?   Checked the fuel pressure, it was ~6psi lower than the last time I was there - apparently nothing had been touched, everything seemed tight.  WTF.  Trademe-spec FPR, not sure what the cause is but threw my toys politely and have left him to work out why it's changed or replace everything or maybe gtfo and find a professional who gets paid good money for this kind of thing.

 

THEN a Silvia (flex-fuel, yay!) which I have done tuning duties for from the point it was put together in 2011 and I had recently been happily celebrating the fact it has been running reliably for several years on a stock motor and probably one of the first (if not the first?) aftermarket flexfuel cars in NZ and it's been so long since I've touched the fuel tables etc that I hardly remember how they all look - decided to set up closed loop boost control to experiment with as I've never bothered with it before and we thought there could be some benefit, he took it out for a thrash with my changes and the boost control experiment seemed to work but the thing was running ~13.4:1 at WOT.  Wtf????  

 

Checked the logs, injector duty cycle was where it should be for the boost, rpm and ethanol content - something mechanical has changed.  Great.  Checked base pressure, that was also where it should be.  Greeeeeat.   He pulled a bunch of stuff apart and checked over things and everything seemed fine, replaced fuel filter etc then took it for a test run and still lean.  Grrr.

 

Decided to replace the vacuum hose to the FPR just because with all other things eliminated the symptoms pointed to fuel pressure but we had no means of checking it on boost at this stage, took it for another run and BOOM! all is mint again, though there was no obvious signs on the piece of hose that there should be an issue.

 

Apologies for the rant, but bloody hell...wtf have I done to ask for this?   

 

People, don't skimp on fuel system parts.  Buy "genuine" injectors and ideally make sure they all flow evenly, same goes for good fuel pressure regs, use suitable hoses and whatever else you have involved in fuel transportation, if you can get a fuel pressure sensor and wire it into your ECU for logging/warning lights, and make sure the fuel system you have chosen is suited to the application.   If there is ANYTHING you spend the money on with a build you want to not have problems with, get the fuel system sorted.   

 

<deep breath>

 

Hope everything is going a bit better with everyone else!

[CamB]

What a tale of woe... Hang in there.

On one hand its scaring the crap outa me on my car and where I am heading. On the other hand, great positive reinforcement on me buying a fuel pressure sensor a couple of days ago. It was $180 plus some fabrication to come that I didn't want to spend but kept having a nagging feeling I better do it (because race car). It's still not ideal as I have a Link G4 Storm so am limited on inputs, and the fuel pressure has ended up as an auxiliary in to a VEMS Round wideband/EGT and will switch a warning light if pressure deviates drops below a set level (I would prefer the ECU cut boost/timing). Cost of sensor reflects a need to buy a differential one (ie should always show FPR spec as its differential between fuel rail pressure and intake) as in my opinion absolute fuel pressure is meaningless in an emergency, so hopefully will be easy to see any deviation.

While I am at it, what's OS opinion on my last digital input. It's a BMW M10 with a turbo. Link will have usual wideband / TPS / wheel speed inputs, plus I am adding oil pressure, a rotary switch to go from not many boosts to lots of boosts (as its a stepped switch I may actually configure it for progressive boost on some clicks - sort of a rain mode), and a toggle switch to turn logging on. That leaves one digital input - options are:

- start position switch (basically only useful for pre crank prime in cold start - not exactly a priority)
- speed limit (useful if I ever do Targa again)
- clutch switch (launch control) - dubious benefit on a circuit car. Might also allow flat shift but I don't overly value this as a feature
- anti lag on (not gonna do it - bad for turbo) (edit - and don't have idle valve)
- switch to E85 map (I can also use half the rotary switch for this with a virtual output, so not my preference, plus see below)
- non-driven speed sensor (theoretically allows traction control - good theory, but I have a G4 not a G4+ so traction control not properly implemented)
- Ethanol sensor (flex fuel). This is the obvious choice but very expensive (sensor $200+ is a down payment. Then, go through and work out if fuel system ok with E85 $unknown, probably need stronger fuel pump and bigger injectors $lots). It won't get driven much so I see this as a factor on whether E85 is a good idea.
- some other idea, as long as it doesn't require any other ECU inputs or outputs as I am out of both...

- (edit) like I could have the output from the VEMS gauge that triggers a warning light as an input into the Link. Basically a "something's wrong" input, although I'd want to be pretty sure it was reliable. I don't understand PWM setup well enough (and know if the Link is flexible enough) to make a PWM output that reflects fuel pressure, but I guess that would be a potential way to display actual differential fuel pressure in the Link rather than an alarm.

Current preference is either start position switch (so boring) or run the wires for an ethanol sensor and worry about it later. (edit - or that fuel pressure alarm idea). Thoughts?

Edited September 29, 2015 by CamB

[Lith]

Making flex fuel work nicely with a G4 Link is non-trivial, I'd only do it if you were fully prepared to go with all the bells and whistles (big pump, injectors, PTFE lines etc) and ready to do a whole pile of tuning work.... rewarding if you are down for that, but.

 

I'd be tempted to go warning for fuel pressure if the thing is primarily a track car, pretty boring and maybe I am just in a paranoid kind of mood after fuel pressure night mares but in terms of something you are going to be caning hard and doing lots of g-force "boring" and "predictable" is exactly how you want your engine function to be.  

[CamB]

Thanks. Good feedback on the e85 route. Not a cheap exercise.

I am definitely keen on the Fuel Pressure alarm idea. I just hadn't thought of it when I first wrote the post..:

... I also googled and I think it can only be an alarm:

- using the boost control output the VEMS Round could probably be convinced to have duty cycle that related to deviation (above or below) in fuel pressure. Maybe not precise but at least enough to be able to distinguish between a little and a lot; but

- link DIs are frequency only (fair enough)

I will hook it up and log how often the alarm is triggered before I think about putting a safety in there. Or maybe on a timer (2 seconds of warning - invoke boost limit).

[Roman]

Praise ye Gods, I got the E-throttle working properly!
Turns out I had two of the TPS wires around the wrong way, problem solved.
I have now been able to increase the PID values DRAMATICALLY and so the response time is virtually instantaneous. It's awesome.

The next thing I'm going to work on is idle speed settings.

Keeping in mind what Lith said about X-Tau and the article he posted.
When the motor is very cold in my (Factory ECU'd) Toyota Echo, it has a crazy high idle. Maybe 2000+rpm. This is only for that first tiny bit where the motor is very cold at 6am, and it slowly drops down to say 1200-1300 for a while then when fully warm is at about the 800 mark.

My thoughts are that since the fuel condenses so readily when the motor is very cold it wants to keep the airspeed higher than you'd expect.

This (maybe?) has a knock on effect of minimising wall wetting which causes you trouble a few seconds down the track with fluctuating AFRs and then the accumulative problems of adding too much/not enough fuel to seesaw around the changing rate of evaporation of the pooled fuel as it warms up.

My plan is to wait for a reaaaalllllyyyy cold morning and let my Echo  idle all the way up to full temperature while datalogging it with OBD. (And draw a graph, of course)

Then have these rpm speeds / temps as my goal for the cold idle speeds on the Carina. Then from there, find the leanest setting that lets it run nicely at each point to minimise wall wetting as best possible.

It's a lot of work for not much improvement, but one of my annoyances is having a few things that worked better on the factory ECU and are now a bit so-so.