MPi injection characteristic data
Turbo Mini Forums

Well, this superceeds everything we have thought / believed we knew in the past about the MEMS MPi.

This is a 3.5Mb Powerpoint that shows the traces.
http://www.davecoxon.co.uk/FIXED_DATA.ppt


Hopefully you guys appreciate it. I'm realieved we finally have the MEMS answer at least!!!

Edited by TurboDave16V on 11th Oct, 2007.

Thank you for doing that work. This is very interesting and intriguing data.

However, it would have been very interesting to have both the MPi crank trigger information and the cam trigger. The cam information would be needed to know where in the cycle the engine actually is because it's not obivous at all to me where it is. Also, the crank information would have been nice to finally know what the pattern is with respect to TDC.

So,while this is very valuable information, it still doesn't answer everything.

I don't want to be negative. It's just that it's so close that it's frustrating to be missing some crucial bits.

In any case, thank you for the information. It will make me think about a few things. I still think the current approach is valid but it may change future development priorities.

Jean

http://www.jbperf.com/

This did come across as negative
:(
I'll see what I can do...

Sorry about that, that was not my intent. I'll elaborate a bit on the why I say I would need more information.

First, on the positive side, the very interesting and unexpected result of the presented data is that the pattern goes from a single pulse per cycle per injector at low revs to 2 pulses per cycle at higher revs. This is exactly the opposite of what I would have expected to see.

Another interesting aspect is that is does do a simultaneous pulse of both injectors once per cycle when it starts (from 6 to 6.15) which is the same pattern as what Marcel proposed and what is in my MS2 code. Of course the difference is that the MEMS just uses this mode until the sync with the cam signal is done (I assume).

Where I start having a problem, is starting at 6.20 where it starts to have a single pulse per cycle per injector. You say it is just after TDC which is what is seen on the graph but then that would feed only the outside cylinder (1-4). Is it really TDC? Or is it TDC of the other pair and it takes a long time for the fuel to reach the cylinders?

The only way to have a single pulse per cycle is to inject when the inside cylinder is finishing its intake cycle and the outside cylinder is starting its own. What the traces show seems to go against that.

In any case, this single pulse per cycle indicates clearly why any cam with some significant overlap will work miserably with the MPi as you said previously. I assume that this method, as well as the fact that at higher rpms the single pulse becomes individual pulses for each cylinders, has been done to minimize the emission. It certainly is not for ease of managing the injection timing. Two very distinctive and well spaced pulses would be much easier to manage to make sure that the injectors have the time to close when going to 2 pulses. Otherwise it becomes very difficult to predict the actual fuel injected if the injector is in between closing and opening.

Another interesting find is the additional pulses that show up occasionally. Without having more information, I would assume that this might be due to some enrichment of some kind (acceleration?) which requires the addition of some fuel before the next pulse is due. It's probably done this way for driveability reasons.

So my comment in my previous post was not meant to criticize what you've done but meant to say that _I_ have still a lot of unanswered questions and that some more information could help.

cheers,
Jean

http://www.jbperf.com/

fascinating,thank you dave .thats very usefull.
the single yellow pulse reminded me of something i have on the tvr dfi injection called the asynchronous pulse ,that oyu increase to cater for flow anomalies,where it would be lean ,and resides outside the normal pattern .
i notice that the priority on the timing seems to be the closing point ,and they are extending the openning point up the scale so the closing point is staying roughly the same .
regards robert

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM

Dave, thanks for the data ..... certainly has me thinking!!! I am with Jean however that TDC / cam triggers would have been useful (reading the other post I see you tried to get a TDC signal but it didn't work out for you).

As you have several charts with multiple events on it should be possible (if we know the revs) to calculate the timing to a rough level based on the pulse spread ..... still would be good to get a TDC / cam if we could.

I know have an MPi engine spare, an MS2 and all I need to do is rig up a reasonable level of mixture measurement (dual wideband, tripple EGT etc.) and we can have a play (Rob H .... I know your reading this .... I need your help ..... well at least your wideband ..lol)

Jean .... hoping to have a play soon with a real engine (already scrapped off in my mind so if we lean det it who cares) on real rollers.

Every day is a school day ...........

How fast and how expensive ...... the same question...

Very interesting.

Is the pulse break that starts at 6.52 due to the ECU applying PWM?

The idea of PWM injectors is that the voltage is pulse after a certain time.

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."

Some more thoughts on the data:

Basically I'm assuming that the apparent dual pulse is actually PWM. Anyway, the pulses are so close together they will apear as a single pulse from a fueling point of view.

I'm also ignoring the odd extra pulses at various points. These may be just a bit of Warm Up Enrichment or due to Lambda feedback.

There is only one injection event per port per cycle, except at idle when there are two (MOT test point).

The start time varies with rpm.

The end time appears to be static at WOT, but varies with MAP. I'm assuming the data is all at WOT until the upshift and then the end time advances.

Then it all gets a bit confusing. The end timing at WOT appears to correspond with 75 degrees after TDC on the outer cylinders. At low revs, the start timing is at approx. 30 degrees after TCD. This is more or less when the inlet valve has closed on the inner cylinders!

Given that the inlet valve on the outer cylinders stays open until approx 220 deg after TDC, how does the inner cylinder get any fuel?

At higher revs it appears to provide a single pulse covering the inlet opening of both the inner and outer cylinders as expected.

Are you sure that the TDC point is correct? (with respect)

I think that MEMS was a fudge to get around legislation and was specifically aim at passing MOTs rather than providing optimum AFR throughout the load/rev range.

I'm sure we (Jean) can do better.

Edited by Paul S on 13th Aug, 2007.

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."

Awight Guys!

As i mentioned - I wanted to get the TPS onto the signal aswell.
I will do some searching, but I think that the acquisition unit i have is limited in 'overall' capability - ie it'd do 10μs overall, but once you start adding channels, the resolution drops...
I also think the pulse channel capability is 50μs max... This should probably still be ok though...

I'll see if i can get the unit again next weekend, and also tap into the TPS aswell as crank and cam phase units.

Re TDC - Just to confirm, there was a single 'peg' bonded onto the pulley. I made this purposely 'thin' so the pulse spike would be clearer. This was as close to TDC as i could 'eyeball it' with a long-rod DTI poking through a spark plug hole, and rotating the engine via a long bar on a wheel nut with one wheel off the ground (in 4th). I'd hazard a guess that this is a max of 5, maybe 6-degrees out, as i spent a long time trying to get this right - trying to carefully position the crank when what you are turning is applying 2.75 increase in ratio is a PITA let me assure you!!!

I had earlier assumed that the TDC at or before the injection point was the TDC when the intake valves were open.

However, is it possible that due to the inlet tract volume, the fuel would not get to the valves until just before the next TDC?

It would then make more sense.

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 showed this to my mate Graeme earlier - and a lengthy discussion started, where we suspect we migth have an answer.

Firstly - remember that the MEMS in the Mini MPi was (IIRC from the miniworld article where Mike Theaker was interviewed) was generated in a weekend by himself.
Next - He didn't create MEMS - he merely took the (then) latest MEMS and did something with it to make it work in a mini.

Hence - this leads me to a conclusion:
There are four injector drivers in the MEMS K-series ECU; and this is probably the same base ECU one used for the MPi mini aswell.
In seeing the injector pulses that appear to be one, which later converge - what we are actually seeing is two seperate injector drivers operating the same injector. The first half of the pulse is (say) driver A, the second half of the pulse is (say) driver B.

There are two ways this could be done - either it's done through Code in the ECU, or they're wired together inside the ECU - OR - they're wired together inside the MPi loom.

As regards the advancing of the injection event - why is it possible this couldn't actually be inside the (k-series) MEMS anyway? It kind of makes sense as if you have any engine you want to inject truly sequentially (fuel only injected into a moving, not static airstream) and want to be able to make full use of your available injection window, then you have to have an advance table for the start of injection (assuming your injection pulse is the standard - not symetrical as discussed on here) pulse type.

Another thing that makes me think this - is the occasional 'spurious' injection pulses in the traces. Consider a setup as i described above, with the 8-port K-series - these spikes are basically throwing extra fuel into each inlet tract - firing each injector at a time where the fuel would go into the open valve on an 8-port. Wether these are an enrichment from MAP, TPS, or Lambda is not clear - but equally - I suspect it is not important...


Hmmm. I think I need to do some more data acquisition - including the crank and cam phase sensor. LOL

Does anyone have a wiring diagram for the K-series and A-series including pin-outs on the ECU for 1997 model year MEMS?

Hi,
Yes. Let me sort it out....
Download from my sever here: http://www.teamtorque.co.uk/pics/rave.zip
Sorry it's only slow!
Cheers,
Gavin :)

Edited by Bat on 13th Aug, 2007.

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

Dave, they are wired in the loom

Incidently, the ECU is the MEMSJ2 ECU which was first used on the Rover KV6, this was the only engine that used the cam sync at that time in the rover range, it also used and oil temp input, that which, I think, when Rover re designed the A series block incorporating it into the design, were intent on using it. Some where along the line they decided against it.

On the KV6 the ECU may be capeble of duel table, and dual O2 sensing, thus making it a little easier to use in the five port aplication and true sequential injection

This MEMSJ2 ECU was then re developed for the rest of the range, with no internal MAP sensor and still included the cam sync.

You and your colegue may well be right as if there are two fuel tables, and two of the injector drivers linked to each table, then crossed over by doubling up the injector outputs, giving you the different timed injection periods?? am i babbling yet, oopsie

Thats my mind on that.

Edited by Sprocket on 13th Aug, 2007.

if you're looking for:

Firstly, good work TurboDave!

It's not easy trying to take these kind of measurments on an engine in the real world, it's always simpler in theory than in practice. Whilst I agree that traces of crank and cam sensors would be great I think there is an awful lot of useful information in the traces you have.

I believe that Jean is correct in stating that the simaltaneous injector operation during the startup phase occurs until cam syncronisation is achieved. This provides the engine with a faster startup time since otherwise potentially it could have to wait two full crank revolutions before even trying to inject. We know the MEMS ecu and twin point injection system was fitted to reduce engine emissions which explains the switch to syncronised injection as soon as possible.

The fact that each injector only operates once every other engine revolution means that the pulse must be split between the inner and outer cylinders otherwise as stated the inner cylinders would recieve no fuel and we'd have a two cylinder engine! I agree that the injection timing looks a little strange, perhaps later than one would expect if we wanted it split across the inlet/inlet overlap period but maybe there's an explanation for this. I've found that my own engine at idle (1200rpm) has an injection advance upper limit of 70deg BTDC1/4 and a lower limit of around 40deg ATDC1/4 for the outer cylinders to recieve the full fuel pulse, outside of this window and the outer cylinder starts to run lean, I think this is similar to what Marcel discovered. Looking at the timing on TDs traces I would suggest that the late start of injection (approx 30deg ATDC1/4) would result in some fuel (lets assume half) entering the outer cylinder towards the end of the intake stroke and the remainder sitting in the inlet port, waiting for the inner cylinder to open its inlet valve. This seems at odds with what I would have expected for good emissions since you are delibrately wetting the inlet port for the inner cylinder but maybe this gives better mixing characteristics thus improving idle quality? By injecting later on the outer cylinders you are at least ensuring that the small amount of fuel injected at low speed/load resides near the spark plug, if you inject early it may be that the mixture within the cylinder becomes slightly stratified rather than homogenous and hence less ignitable. Just a thought!

Regarding what appear to be back to back injection pulses I would question what this achieves. Looking at the traces on page 34 the injector voltage only drops from 13V to -2V and then rises back to 13V in a few hundred microseconds. The injector reaction time is around 1.2ms at 14V (I measured this on my own injectors) so a drop in supply voltage for this short time will not allow the injector current to fall far enough to let the injector shut-off mechanically, result is that your two pulses actually appear as one continous injection event without interuption. I think as TD says this is likely to be an anomally of the MEMS ecu being adapted (poorly) to suit the A series engine requirements.

with regards the adaption of the ecu, i jsut had a quick look over the diagrams, the MPi mini does indeed have 4 injector outputs, paired together in the wiring loom and one pair drivign each injector. Wiring diagram for the VVC mems unit show the same injector pin numbers but one for each injecor. jsut to confirm what you already thought!

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

this is the difference though, the MEMSJ2 ECU and KV6 engine was designed to run two seperate cylinder banks and as such two seperate fuel tables. The wiring for this ECU also shows two lambda sensors to suport this. In escence its two seperate 3 cylinder engines.

The A series is in theory two seperate 2 cylinder engines

I think it is very feasable that the two fuel maps are combined by crossing over the injector drivers simply by pairing them up in the loom. I can see how the timing shift might work in a fairly simple maner, though probibly not what it is. If you set one map to give say 1 third the fuel required at low rpm and the other the remaining 2 thirds, then say at mid rpm its half and half and at high rpm its 2 thirds and 1 third, with the extra events on the crank position comined with the cam sync, you could move the injection event or at least the biggest chunk of the required fuel into a better time period.

Does that make sense, lol i lost my self there. I shal leav it for posterity and for people to laugh

Edited by Sprocket on 13th Aug, 2007.

MPi flywheel data

Missing teeth at 30, 60, 210 and 250.

There is 2 teeth between 30 and 60, 14 teeth between 60 and 210, 3 teeth between 210 and 250, 13 teeth between 250 and 30. 0 being TDC on #1/4 cylinder (TDC1) 180 being TDC on #2/3 cylinder (TDC2)

First event at 30 degrees after TDC1 and TDC2 with the second events at 60 degrees after TDC1 and 70 degrees after TDC2

First event most likely ignition and secon events injection?

Dont forget that this is purely determining theTDC of each pair of cylinders, the actual engine events could be configured within the software to fire at any point for instance with VEMS, it doesnt matter which tooth, as long as you tell it which tooth to trigger on, not necisaraly the missing one.

Edited by Sprocket on 13th Aug, 2007.

Edited by Rob H on 14th Aug, 2007.

Rob .... thanks for that, it's a great start.

Will also need an extra pair of hands to make sure I make progress on this ..... as soon as the kitchen is finished .. lol

Every day is a school day ...........

How fast and how expensive ...... the same question...

Mine is available for testing if required.

why has everyone on this topic started talking fooking Russian,
i cannot understand a bloody word,


carl

WWW.FORCE-RACING.CO.UK PLEASE CLICK HERE

Which is why I wired the ECU in The DON and you didn't........LOL!!!!!