The 5 port way forward with phase (cam) sensing.
Turbo Mini Forums

No content yet - waiting for two (or more) other interested parties to agree moving this information here.....

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

Well, I'll start by bringing this up to date with my results without phase sensing.

There are a couple of threads titled "Siamese Code Trials". Ive used the code on a 1275 and a 998 so far. Both have given similar results in terms of AFR distribution. I'm getting, typically 2-3 AFR points richer on the inner cylinders whilst injecting the same amount of fuel.

basically all injectors fire at the same point so that on one port fuel goes straight into the outer cylinder inlet valve. At the same time, on the other port, fuel is injected 180 degrees before the inner cylinder inlet valve opens.

Sensor error has been discounted by swapping sensors and controllers around, even trying a new sensor. The readings are consistently much richer on the inner cylinders.

Three (or more?) possible reasons:
1. Difference in VE of inner and outer cylinders.
2. Greater atomisation/vapourisation of the fuel destined for the inner cylinders.
3. Wall wetting.

I think that the inner cylinders run at a lower VE because they have to accelerate flow in the port, whereas the outer cylinders benefit from the inertia. Also, as the outer cylinder inlet valve opens, it stops flow into the inner cylinder whereas the outer cylinder benefits from a bit of inertia filling an the compression stroke.

It is a well documented fact that the A series does not like well atomised fuel. It produces more power with fuel "droplets" rather than a fine spray. Read Vizards latest.

Injecting before the inlet valve opens on the inner cylinder could cause the fuel to start to vapourise. Not sure about this as the AFRs are no closer on a stone cold engine.

Part of the problem may be that I am using MPi injectors that have a 30 degree cone spray pattern that may cause higher atomistaion than required.

I'm thinking of trying some smaller pencil stream injectors to see if that improves matters.

I was also thinking of modifying a head, with just bigger valves on the inner cylinders to verify the VE theory, but that may take too long. It is also a fudge as ultimately it is just papering over the cracks.

Im hoping that bringing in a phase sensor will allow us to use alternating injection points and time the injecting pulse at the inlet valve overlap so that we can time and trim to get the best out of the engine.

Edited by Paul S on 18th Jan, 2009.

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."

At the moment the phase sensor will allow the injection point to be moved (with a timing for inner and another timing for outer cylinders) but it will still have an injection pulse per cylinder so you could time it such that you always inject on an open valve (for inner and outer cylinders). It will also have a fuel table for inner and another for outer cylinders which will allow to correct for the different fueling needs.

It is not planned to have a "merged" pulse that will fuel both cylinders timed around the valve overlap. The reason I'm not fond of this is that the intake valve overlap also happens during the intake/exhaust overlap of the inner cylinders which means some of the fuel injected during that time will go straight through to the exhaust. That may not be a problem with cams optimized for turbo engines where the overlap is not big. The more annoying part in my view is the fact that you can't start the engine in that mode since timing at start up is much less precise and pulse width is very small compared to the cycle duration which will result in a grossly uneven fuel distribution. And I don't know at which point in rpm and load, the pulse width will be large enough to give a reliable fueling in that mode. Also, I don't know if timing is reliable enough to take into account all the different running conditions.

So the first solution will be to have one distinct pulse per cylinder timed to be during open valve for the outer cylinders and at any other time for inner cylinders with fueling adjusted with 2 different tables. This also has the advantage of having the possibility of adjusting the fueling of the inner and outer cylinders differently during transients (acceleration enrichment) which would be very difficult to do with a "merged" pulse because you'd need to adjust both pulse width and timing.

Jean

Edited by jbelanger on 18th Jan, 2009.

http://www.jbperf.com/

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."

Yes you can. Some people run 4 squirts per cycle so injections 180 degrees apart are ok. Of course, it depends on the injectors (opening time), duty cycle, and RPM. A sluggish injector ran at close to 50% duty cycle (for the siamese code) at high RPM will basically not do 2 squirts but a single long one even if commanded to do 2.

But it's also possible to set the timing tables such that the timing for the outer cylinders is slightly retarded and the one for the inner cylinders is slightly advanced (or even heavily advanced) such that the injection events are separated by 200, 220 or even more degrees as needed.

In any case, I'd start with timing set such that it is similar to the semi-sequential timing, i.e., inner cylinders 360 degrees before the outer and then I'd bring the timing in and see how much improvement I could get and where the best timing is.

Jean

http://www.jbperf.com/

Great stuff Jean.

I look forward to testing the new code.

Presumably I need to move the cam sensor input to JS10?

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."

Right,

Apologies for starting this topic then dis-appearing - I needed to go away and understand where my original assumptions had gone wrong.

I also had a play with my Blue Peter carbdoard mockup to make sure I can actually squeeze 4 injectors in - possible but the angles will be a bit funny and I'll have to make a heat shield up to protect them from the standard Metro exhaust manifold.

So back to the physics, "As for the mods needed for the four injector drivers, you don't lose the IAC stepper motor. What you lose is the ability to use PWM to control the low-impedance injector's peak and hold phases. So you need an external driver that doesn't use PWM which is what my board does. The standard MS injector drivers already use 4 CPU pins. I just change how these are used. And there is a need to bypass a chip on the MS2 card which combines signal from the 4 pins (NAND gate). I'll have more details on the mods in the coming days and I'll take pictures. ..."

First, my mistake over the IAC stepper motor - I was getting confused with a lot of other hardware mods in the MS"-Extra manual where the stepper drive is sacrificed.

I had looked at your board (Jean) a long time ago and refreshed my memory today, but in its original implementation (according to your site) it simply takes the two outputs from U4 and pairs them to the new board so although there are 4 physical drivers, there are still only two inputs.

I assume what you refer to above is a new implementation allowing 4 CPU outputs to contol each of the new drivers direct... I can't find a schematic of the MS2 daughterboard but with my strongest reading glasses and a magnifying glass, I can see a nice little 74HC00 SMC chip on the board - is that where the proposed mod is to be ??? If so, let me just say I HATE soldering (or un-soldering) SMCs...

Presumably what you have in mind is a pretty fundamantal change to the code which, along with the hardware changes on the daughterboard, make it an "all or nothing" decision - the MS becomes specific to, and only operable with, that particular code ???

So, before going down that route, what are the other options......

The fundamental principle now is to inject only through an open valve. So no more squirting at an open outer and closed inner simultaneously. So the cycle (1,3,4,2...) becomes 3,4.. 2,1.. 3,4.. 2,1.. etc ie, each injector (assuming only one per port initially) has to give 2 squirts (each of less than 25% duty cycle) in quick sucession, then the other one does the same. The phase sensor is needed so 3,4 can be differentiated from 2,1. Then, as I understand it, your new code, Jean, will allow each of the two quick squirts to be different in start point and duration.

It is therefore also pretty fundamantal that with the existing hardware, one driver will be dedicated to the port for 1,2 and the other dedicated to the port for 3,4.

So no staged injection without Jean's new board, and the two squirts, Jean from an earlier post "However, the available duty cycle is about 40-50%. Don't forget you need to inject on an open valve which lasts for about 25% of the cycle but you feed 2 cylinders hence the 40-50%...." so let's be pessimistic and say 40%.

So two very large injectors required, twice the size I had originally thought....

So the first thought is would two medium sized ones wired together and operating together, give any better results at idle than a single large one ???

Paul, you said "You will only need staged if your injectors cannot give a reasonable idle with all four in use. I have had all four of my 480cc/min injectors running at idle without a problem, certainly no discernable difference with just two..." What I'm not clear on though is how you had all 4 running at idle with the existing siamese code ??? Did you just have the two injector channels fire together throughout the range (but at half the fueling).

Enough for the moment, my head is hurting.

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

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."

Excellent....

With Jean's new code (if I've understood it right) I'm assuming you will then need one driver for port 1,2 and the other for port 3,4.

If I'm right, you will have to rewire either with just two injectors, or them paired together on the same port rather than paired as uppers and lowers.

It looks like I shall now start with 4 but wired together as a pair on each port which will hopefully counteract the weird angles they will be facing.

At least I only have the 3.5% CO limit to deal with.

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

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."

Ah yes, I remember you saying before about how you had tried it "phased" but with no phase sensor and a 50/50 chance of it being right !!!

All this means that from now on, it appears there can be no staging without Jean's additional driver board and daughterboard mods.

I'm interested to hear more about it although I shall start with just the two drivers.

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

Paul,

I'm really surprised that you can run with the 4 injectors active. How small is the pulse width at idle?

Rod,

You got it. As for using 2 medium injectors instead of 1 big, I don't think it would be better. It would actually be worse since you have 2 slightly different reaction times and and you still need very short pulses at idle so you have less control.

The modifications to the MS2 card are intended to be as non-intrusive as possible. The schematic for the MS2 card can be found here. The intent is to run in parallel to the current driver to minimize the needed changes. So it would require soldering small wires to connect pins 1 and 4 of the 74HC00 to pins 4 and 5 of the card. These last 2 pins are connected to the crystal which is only required for MS1 so they can be used for another purpose (as long as the crystal and the associated components are left out).

Then 2 of the 4 drivers of the p&h board would be connected to these pins. The other 2 drivers would be connected to pins 39 and 40 of the MS2 card or to R32 and R36 on the MS V3 board. These resistors could be removed or left in and the MS injector drivers could also be left as is but they would not be functional while using the siamese code in the 4-driver mode. However, they would still be functional (if the resistors are left in) with the normal code and the siamese code in 2-driver modes (no staging or semi-sequential mode). In that case, the p&h board drivers would not be functional (actually 2 will and 2 won't be).

I don't know if that all makes sense but hopefully it does. In any case, I will be doing a web page with the schematic and pictures to show exactly what I mean.

Jean

http://www.jbperf.com/

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."

Jean,

You're a star....

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

I've just bought some high impedance 385cc/min Siemens Deka injectors with a pencil stream.

I'll try and find some time to fit them to see if that has an impact on the current Siamesed Code, just to add another piece to the jigsaw.

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 found this link on one of the "other" Mini sites

http://www.miniestate.de/MPI_technology.pdf

I'm sure most of you have seen it before (I had't but I'm relatively new to this!) and it confirms - as has already been established on this forum long before I became interested - that they used different injection timing points. It also says they stretch the pulse forward and backward, which I may have also read on here (can't remember).

The most amusing bit though (this one for Paul) is halfway down the penultimate page... "On the prototype we had no camshaft sensor so the car would only start 50% of the time."

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

Taking this forward on the basis that ultimately 4 individual injector drivers will be required (each port being driven seperately, and the additional staged ones being driven seperately) I've looked at two possible physical installations (with the standard Metro exhaust manifold).



Option 1 puts the injectors at wierd angles to the runner (circumfrential-wise) so could add to all the other variables as either injector could favour one valve. Also, making up a fuel rail will be a nightmare as the inlet connections are no longer parallel.


Option 2 allows the first injector to be as close as possible to the valves to overcome wall wetting issues and the extra distance of the second injector probably won't matter too much if it only comes in at very high flow rates and boost. But it puts the plenum 35mm further back ( and slightly higher) which is a nightmare to fit in....

Thoughts ???

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

As long as it will fit in the engine bay then it should be OK. Once it is in and fitted you do not need access to the rear of the plenum.

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."

heres one way to fix the rail problem rod on the tvr v8.

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

Robert, how have you got the injectors "clamped" in there ???

I thought about flexibles to allow different angles on each injector but most designs I've looked at use the "rail" to physically hold the injectors in place against boost pressure.

The only OEM design I've seen with flexibles on the inlet side has clamps on the base of the injectors (and that was an N/A car...)

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

It looks like they're held in place with bits of welding rod / wire...

For the exact reasons you mention, I would favor option 2 also.

As for flexible injector hoses, you have some options for holding the injectors in place such as what is available here.

Jean

http://www.jbperf.com/

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

My bad - you're right!

I know the older 8v 1.8 nissans had injector holders around the body of the injector...

Webcon also used to sell some nice 'top feed hats' for injectors...

With option 2 you could make up a couple of rails from a bit of Aeromotive billet fuel rail available from DemonThieves.

That's what I going to do this time, rather than machine up parts and braze them all together.

Edited by Paul S on 26th Jan, 2009.

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."