Description of the Rover MEMS (MPI) injection....
Turbo Mini Forums

This was termed 'bi-directional pulsewidth stretching' by the Engineer Mike Theaker who created the MPi system for the Mini.

The Rover patent doesn't actually cover any of this - it doesn't need to as it's protected the basic principle of injecting a siamesed inlet engine...

I understand that basically, a set point (lets say 140 degrees ATDC) was set as the 'point of injection'. The injection event grew both forwards and backwards unequally about this 'fixed' point.

Hence - a typical injection figure could be starting at 130 degrees ATDC and ending at 160 degrees ATDC (basically stretching asymetrically about a fixed point with a 1:2 ratio about the 140 degree point)...

That is as much as I understood from an article, and a few subseqent emails with Mike Theaker.

Sadly - I don't know if the fixed point was nearer TDC, in the middle of the 'window' or nearer BDC (that might have explained something)!



You know what - I'm going to try and track him down. I know where he used to work, and our company is just about to have some deallings with this organisation, so i'll have to see what i can find out!
Now that'd be smart if he was willing to give us some pointers! I can't see that he's bound by any Rover agreements anymore - and we wouldn't be asking him for everything, just a few pointers...

Edited by TurboDave16V on 4th Jan, 2006.

TurboDave,

This is something that should not be too difficult to do in the code. It would however require to know quite precisely the opening time of the injector to have the precise effective pulsewidth. Once you have that, you just divide the effective pulse width in 2, translate that in degrees at the current rpm, and deduct that from the set point.

Do you think that the set point is fixed for the whole rpm range?

http://www.jbperf.com/

Another point: with staged injection the set point would have to change unless the injectors are set such that both injectors are the same distance from the intake valves. That is if I understand how this works.

http://www.jbperf.com/

I think that sounds exactly like what I was describing as a poor solution before. The fixed point is somewhere in the overlap, I would imagine. The asymmetry makes sense because of ramcharging, or scavenging making the beginning of the cycle be more efficient than the end. I just don't know how to deal w/ that, really.
Apparently it's a better idea than I thought, but I guess they have more resources to address the concerns I had. If we could find the values they used, that'd be a good starting point.

-Tyler
DO NOT TOUCH MY HORNS OF DOOM!

I think that the fixed point is actually for the other solution you were describing, i.e., one set point per cylinder. Although at higher rpm it could also be used for the solution you describe. And it could also be one set point per 360 degrees but set for the outside cylinders since the inside cylinders don't really need their injection timed for an open valve.

Another point is that I think I saw somewhere that the MPi cam has very little if any overlap between the intake of the inside and outside cylinders which would simplify what concerns you.

http://www.jbperf.com/

Well I will have on my desk tomorrow, an MPI camshaft.... Which i will be sending to Marcel so he can do a measuring job on it - so we'll know the truth for sure!

The injection 'point' definately happened at a point somewhere in the injection window. Basically - the ECU was sending out a signal EVERY 180 crank degrees to one of the injectors - the cam and crank sensor dictated which one...

I note from the Rover Patent it states that the 'inner' valve injector switches off before the outer valve opens - basically when both valves are open - there is no fuel leaving the injector...





I'm really liking this section.

Edited by TurboDave16V on 4th Jan, 2006.

Hi,
The way I understood the Rover MPI to work was that the pulses were stretched either side of the fixed point. So the first pulse starts earlier with more fuel demand, but finishes in the same place. The same for the second pulse but starts in the same place and finishes later. Here I'm talking about one port only. So I would imagine the pulses are stretching out either side of the overlap point with both valves open? So it appears were all singing from the same sheet!
@ Dave... If what Jbelanger is saying about the MPI valve timing is correct, that would tie in with what I said to you previously about the MPI valve timings being altered to allow time for the injector to switch off and on again?
Cheers,
Gavin. :)

Edited by Bat on 4th Jan, 2006.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

I've added another feature to the code. This is the 'bi-directional pulsewidth stretching' as TurboDave call it and that I've called mid-pulse timing in the code. So, instead of using the table, the injection is timed with respect to a single advance angle but it's the middle of the injection pulse that is timed to that angle so as the pulse grows, it increases as much before and after this reference timing point.

Again, I've put the files in the Meagsquirt Forum:
http://www.msefi.com/viewtopic.php?t=14617

This is as far as I think I can go with the code without access to hardware or to a new official release. I hope that someone will try this and let me know how it goes. I'm sure there are bugs in the code so if you find any I'll do my best to help.

Regards,
Jean

http://www.jbperf.com/

Good work Jean...

Edited by TurboDave16V on 13th Aug, 2007.