AFR Distribution with TBI setup
Turbo Mini Forums

Hi All,

I've been considering converting my Mini to fuel injection. Largely for convenience of tuning, easy starting, avoidance of vapor lock etc.

I was hoping to use a wet manifold / throttle body injection approach similar to the Specialist Components kit just to keep things as simple as possible for my first EFI conversion.

I was originally discouraged after reading a lot of concerns that AFR distribution could be significantly worse with TBI than with a carb.

I then saw Paul S's response in this thread:http://www.turbominis.co.uk/forums/index.php?p=vt&tid=611550

"Yes, ignore that earlier thread. We now know that the risks were over-stated."

Based on that comment, I was just wondering if any AFR distribution data exists for a TBI setup and what is actually considered acceptable AFR variation on a boosted application?

The car in question is running 15psi boost and about 140 "bananas" with a 5600rpm rev limit. If I do an EFI conversion, I'd like it to work with the above specs as I obviously don't want to turn down the boost

Any pointers you guys can provide would be greatly appreciated :)

Thanks

I’m not sure if anyone has fitted dual wideband sensors with a wet manifold setup, or a least not anybody that’s shared the results on here.
This is something that has been discussed many times but I don’t remember seeing any genuine data.
I’d be interested to know as it’s something I plan to do myself as I’ve all the parts sitting here.

Phil.

WWW.TURBO-MINI.COM

Cheers Phil. I figured there was a reason I couldn't find anything!

Has anyone even run TBI with >15psi boost before? I saw a post from wil_h saying that he ran 12psi with the SC TBI kit but I couldn't find anything else.

I could start by doing some trials with TBI on my brother's NA mini using twin widebands. Not sure how representative it will be for a turbo setup but I'm not keen to fit dual widebands to my turbo due to needing sample chambers.

I suppose data from an NA motor would be better than no data at all...

Grant

Edited by gcos93 on 23rd Sep, 2020.

As Phil says, no genuine data on this forum (there may be elsewhere but I doubt it).
Someone did try to use EGT data to do the comparison but it was proved worthless because the centre exhaust runs hotter anyway because it's siamesed.

Whilst it's easy enough to put dual widebands on a N/A engine, the main problem with dual widebands before a turbo is their pressure dependency - if you don't run them at consistent and equal pressures, the results will be meaningless. And they may also be killed by the temperature.
There has been some talk, both on here and other forums, about the latest generation from Bosch being less pressure dependent but, again, no real user data.

But, as far as I can see, using the sample chamber technique is the only safe way on a turbo.

Otherwise, on a throttle body/wet manifold, just put a single wideband after the turbo and use other signs - spark plug colours or holes in your pistons, to gauge the offset.....

Schrödinger's cat - so which one am I ???

Thanks Rod!

Yes, I've seen some pictures of Graham T's sample chamber setup. Makes sense but does add some complexity.

I'll have a think about ways of making it a temporary fit as I'd rather not have all that pipework hanging around permanently.

One thing you will need to do is to have 4 injection pulses per engine cycle. That's the only way you can have an AFR distribution that's close to correct.

And you will want to use more than one injector to cover the entire rpm/load range. How you do this will depend on your physical setup and what your ECU can do (same size injectors or one big one small, set in alternating mode or staged,...)

http://www.jbperf.com/

Thanks Jean :)

I was thinking along the lines of 2x 440cc injectors mounted side by side in alternating mode.
I'm planning to run 4 squirts/cycle as you suggest so it's good to hear that this is a sensible approach (for TBI).

Regarding ECUs, Speeduino looks promising but I still need to add MS3 to my spreadsheet.

The only potential issue with Speeduino is lack of siamese code in case I can't get any results with TBI.
To be honest it might be a fun challenge to add the siamese functionality to the Speeduino as I've got some embedded programming experience... although not on AVRs!

The MS3 also doesn't have a siamese code option. You need to use sequential injection and single pulse setup (not an MS3 option but the way to wire injectors and how you tune the injection timing); you also lose 2 injector channels. So if the Speeduino has sequential injection, you have the same option. The only ECU that has a complete siamese-port option is the MS2.

http://www.jbperf.com/

Ah ok, interesting. I'll do some more digging around here on the siamese code to better understand what I might be missing out on. Speeduino does seem to support sequential up to four cylinders.

Probably a silly question but I've been playing around in tuner studio again and notice there's a mappable fuel trim for each cylinder if using sequential injection: Is there a reason why this can't be used to help reduce charge robbing?
Obviously it would be necessary to use a cam sensor and run twin widebands to tune the fuel trim so I'm not really intending to use it this way if I go TBI... just curious.

Nothing preventing you from using the fuel trim tables but you might also need to tune injection timing to get a perfect AFR distribution. But if you start tuning the injection timing, you might be able to simply use that with the main VE table to get the same result which would be much less work than playing with timing, VE and trim.

There are potentially many different ways to get to good fueling and the best will depend on your actual setup and which one makes the most sense to you (which way to tackle the problem is more comfortable to you and how that relates to analysis of the datalogs). But there are also other correction and calibration factors that need to be correctly set which will affect fueling differently depending on which tuning method you use (injector parameters, voltage corrections, baro corrections, ...). So it gets a bit complicated to predict all the potential interactions.

http://www.jbperf.com/

If you go for a rear mounted (ie, more conventional) turbo setup the sample chambers can be quite unobtrusive.
All my original build photos went from this forum when photobucket committed commercial suicide but I still have them all on my server.

Mine's a rather crude Metro T3 turbo and cast manifold.
It's not ideal as the small bore sample tubes both have a dip in them which means they can collect condensation so I have to delay the power-up of the LSU heaters until the engine is running but the physical setup is very un-obtrusive because I run port injection so there is a large plenum chamber above that location.
Re. the MS2 siamese code it was really only intended for port injection with timed pulses to co-incide with the individual inlet valve openings so the injectors have to be as close to the valves as possible.
I might be wrong but I don't think it would work very well with a TBI setup because (a) the injectors would be a lot further away and (b) they wouldn't be physically associated with only one runner each.

Re, Speeduino, I had a play with it a couple of years ago. It appealed to me because of the TunerStudio interface. It is fully sequential (or batch if you want). At the time I would have said it was half way between an MS2 and an MS3 and although I never pursued it (just built one on their own board and played with it on a JimStim) but I still occasionally look in on their forum and it seems to have advanced quite a bit since i had a play.

Schrödinger's cat - so which one am I ???

The problem with TBI with sequential, timed injection as I see it is that the injectors are sitting at the end of a manifold and port that is greater volume than the cylinder. So, theoretically you need to start injecting before the cylinder starts filling. Even then some or all of the fuel will be left sitting in the port.

But, at the point you want to inject, the TB is feeding the other port so is going to go the wrong way.

It's a can of worms that I dont want to start to get my head around.

I did port injection using an Mpi manifold and an MS3 kit for less than the cost of an off-the-shelf kit, so that would always be my first choice.

Edited by Paul S on 24th Sep, 2020.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Thanks everyone for the helpful responses!

Jean: Interesting. Fair point that it may not be the best way to achieve even AFRs if other parameters are available (e.g. if using the siamese code).

Rod: Your sample chambers look very tidy! What sort of fittings have you used to connect the pipes to the sample chambers & manifold? Some sort of compression fittings? Flared pipes?

Just to clarify: If I go TBI, I'm not intending to use the siamese code. I was just thinking that if I start off with an MS2 then I have the possibility to upgrade to port injection at a later date depending on my success with TBI.

Paul:

Thanks for confirming that you were talking about port injection when you mentioned sequential. That's what I read in your previous message and what I considered when I replied. I also don't see the point in using timed injection when using a TBI setup and that's not what I was advocating, if that was not obvious.

http://www.jbperf.com/

Ah sorry Jean. Were you responding to this comment from my previous post?

No I was responding to Rod and Paul who both mentioned that TBI and sequential don't mix well which neither you nor I were proposing. I just wanted to make that clear to both and that I agree with them.

Edited by jbelanger on 24th Sep, 2020.

http://www.jbperf.com/

Schrödinger's cat - so which one am I ???