Using normal EFI on MPI A-series
Turbo Mini Forums

Hi,
I'm not sure this was discussed recently.
If I understand well most of us use special versions of FI systems programmed for siam ports.

I wonder if anybody here uses regular EFI systems made for 4 port engine with 4 injectors.

I spoke once with an AEM support engineer who told me he used their regular AEM ECU on an old US 4cyl siam port engine (20-30's Ford/GM, I don't remember). He'd put 2 injectors per port and just trimmed durations to balance cylinders.

The difference with A-series is that his engine had a very low HP/L ratio, which means he could use quite small injectors (for smooth idle) and still open them very short time (1/4 of rotation) in WOT.

We make 100hp/L which is a lot to be able to have good idle (and thus CO/NO level) as we need big injectors. On my 998 65lbs injectors are open more than 1/2 revolution in WOT and I'd need bigger injectors if I wanted more HP. It works great but having very special EFI is a problem to me. For example I can not sell my car as the buyer would be unable to tune it.

Edited by alpa on 13th Nov, 2016.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm

This isn't a direct answer (because I don't know what everyone uses, most use some form of wet manifold / throttle body / electronic carb rather than true port injection) but what may be of interest is if you look at the Rover MEMS (factory MPI) it is their standard four injector system but with the four outputs paralleled up to drive just two injectors in the specified sequence of four (two shots each).
But at higher loads when the two pulses got too close to each other, it switched to a single pulse covering both valve opennings.
Which is exactly what Jean included in the MS2-E siamese code as an option and what some people are doing with an MS3 anyway (because it doesn't have specific code for siamese).

So there are all sorts of possibilities out there....

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

A number of us use the MS3 with the MS3X expansion. The firmware does not have any specific siamese port features, so we can either use 2 out of the 4 injector outputs or use all 4 doubled up on to the 2 injectors.

To keep the injector size small for a good idle we use a second set of injectors staged in on boost.

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."

I see Rod. It must be tricky to adjust beginning and duration of this combined pulse right ? It must be changing depending on RPM and intake pressure.
I did not know exactly what Jean made in MS2.

Paul, so basically in the not staged mode it works like the AEM guy described, if 4 outputs drive 2 injectors.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm

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

Thanks
So (to make sure I understand well) basically in regular (not siam) MS mode you setup to open 4 injectors just before and during respective valve openning, right ? Which means you can inject during up to about 1/2 of revolution ?

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm

Well the whole idea is to inject during open valves - or actually enough before to allow for the transit time - but to make sure the fuel goes in through the right open valve.
You can't leave it sitting on the back of a closed valve like normal injection with high duty cycles.
So, yes, 1/2 a revolution (25% duty cycle but twice per injector) absolute maximum unless you merge the two pulses.
So it's like openning four injectors once every 1/2 revolution but actaully openning two injectors twice.
The staging means more injectors but still on the same 1/2 revolution.

Sorry if I'm confusing things but it's not easy to put into words.

The MS2 hybrid mode and the MS3 timed injection just complicates the explanation but basically you just want a system that puts fuel through open valves only and not leaves it sitting on the back of closed vales (like normal fuel injection) because if you put it on the back of closed valves, it will be the wrong one in a siamese port.

To do so means large injectors or staging to get duty cycles very low.

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

Thanks. Your explanations are clear.
I made my own ECU for MPI 998T in 2007-2008, so I did not know details about MS, I was just reading some discussions about Paul's trials. I'm actually injecting once on a closed valve and once (about 1 revolution later) into the outer cylinder. Works perfectly but it's not "standard". And I was pretty sure MS was trying to mimic Rover's solution which was a bit odd.

Edited by alpa on 14th Nov, 2016.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm

Co-incidentally I recently read the description on your website of how you approached the issue. I was a bit surprised at first about injecting on a closed valve but, as the inner opens first, so long as all that fuel has gone before the outer valve opens and you get the timing spot on for the outer valve, and the amount right, it obviously works.
In practice, my timing for the inner valve is slightly early (but not as early as you) but just to make sure the two pulses don't start to merge as I don't use (and don't want to use) the hybrid mode.
One other common thing between us is you say (on your website) you run an 85% difference between inner and outer - that is almost identical to me, I do it by offsetting the VE tables by 10% either way, in MS2 siamese mode you are allowed two VE tables, one for the inner and one for the outer. Not everyone uses it that way but I find it easiest.
We weren't actually trying to mimic Rover, at first we tried a semi-sequential setup (still an option in the code) to avoid the use of a cam sensor but when Paul tried it for real (my engine wasn't finished so I could only run it on a JimStim) it didn't really work so we decided on sorting out a cam sensor and fully sequential. The merged pulse hybrid mode came much later once we saw the issue and it was about a year after that I found a very old article on the Rover system that described how they also merged the pulses at higher RPMs. All we really knew about Rover initially was that it must be some form of fully sequential as they also needed a cam sensor.

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

I'm quite sure of my analysis about why in-outer are disbalanced. So injecting on the closed valve even helps.

I made my cam (phase) sensor with an old distributor, just cut its head and used the contactor. Still working after 15-20.000km. If it fails the ECU goes to semi-sequential mode in which there is no in-outer balancing.

Yes injecting with no in/outer balancing works quite well but it's not fun.

In the first (phase1) version I was not able to get power because I could not image the huge advance I needed for the outer cylinders. That required heavy firmware changes. Once that sorted-out the mapping was very easy as in/out timings are distinct. I believe having mapped everything in 4-5 hours of road rolling.

I think it's possible to use my approach (1 pulse per revolution) on a "normal" ECU if it can shift the start of injection depending on RPM and load. Which is obviously the case with MS. This is better for injectors that would run in spec. But the main problem of too big injectors remains.

PS: it's clear, we need direct injection .

Edited by alpa on 15th Nov, 2016.

std 998 A+, g295, MD266, RHF4, 109hp @0.8bar/5400rpm