Donations towards server fund so far this month.

 
£0.00 / £100.00 per month
Page:
Home > MS Code Discussions > Fully Sequential Injection

Paul S

User Avatar

8563 Posts
Member #: 573
Formerly Axel

Podland

I think that the problem is the injector itself.

My fuel rails are about 15mm bore and run at a very low velocity.

The narrow passages through the injector are where the high velocities ocurr and where the reverse pulses cause the problem.

This is interesting:

http://performancefuelsystems.com/Injector...-TechCorner.htm

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."


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

But if you have a reversing pulse, shouldn't you also get some reflection which means you have some sort of ram effect? And doesn't the shape of the injector pintle come into consideration (as well as the shape of the whole passage in the injector and fuel line)?

Jean

http://www.jbperf.com/


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

The article is interesting (no mention of low vs high z which I still think (hope?) makes a difference) but the lack of linearity is at low pulse widths.

At high load/boost we aren't talking about small pulse widths.

So back to my earlier argument, bigger injectors, accept poor idle (or use small/large staged) and just keep the pulses well apart.

And do everything you can to fuel rail sizing, prerssures etc. to minimise their effects.

My last 78 tiles are now looking very appealing for tomorrow, almost theraputic (twitter *happy*)

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


Paul S

User Avatar

8563 Posts
Member #: 573
Formerly Axel

Podland

My K1200 head also looks very appealing.

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."


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 15th May, 2009 Paul S said:
My K1200 head also looks very appealing.


Somehow I just know you won't give up on this :)

Otherwise next you'll be saying VEMS......

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


Paul S

User Avatar

8563 Posts
Member #: 573
Formerly Axel

Podland

No it's Megasquirt all the way for me. No point in having to learn something else.

I'm not giving up because I feel that the single pulse mode will make it work.

I'm just very frustrated that I can see that another summer is going to pass without having a turbo mini on the road.

All the effort recently has gone into the Miglia, but that will no be done in time for Avon and it's too late to get the 998 turbed either even if we had sorted the EFI.

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."


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

The frustration is obvious and I only came into this late so my input is minimal at the moment.

It's a shame others aren't trying it yet but I guess that's down to there being other "solutions" out there now and a normal human reaction of "let's see if it works first" and, of course, the cost of R&D.

I've got a couple more hours to do on my loom, a little bit more alloy welding on the runners and fuel rail and I have to build the high pressure fuel system.

The remaining 78 tiles may have to come first though.... but I could try Carl's
"you should do a bath/shower/wet room with 100mm hand made french tiles , with a colour mosaic not one straight edge or 90 corner ! "
as a getout though......

(EDIT for typos/grammer)

Edited by Rod S on 15th May, 2009.

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


Joe C

User Avatar

12151 Posts
Member #: 565
Carlos Fandango

Burnham-on-Crouch, Essex

I must say i commend you Paul, Rod & Jean for spending so much time, morey and effort on this.

right now i'm betting you'd bought an imp eh Rod, lol

as has been touched on before and you suggest Rod, there is an element of "lets see if it works" especially from us lesser mortals that are not "real" engineers :).

BTW Paul.... WTF is a Hydroynamacyst???


On 28th Aug, 2011 Kean said:
At the risk of being sigged...

Joe, do you have a photo of your tool?



http://www.turbominis.co.uk/forums/index.p...9064&lastpost=1

https://joe1977.imgbb.com/



Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

The thing is I'm not frustrated yet, but that's because I'm following and that is an envious position to be in because you learn from what the one(s) in front have done and their results.

If more people were "playing" with Jean's code, progress would be a lot quicker, but I fully understand why people are in the "let's see if it works" camp and I accept that and hopefully they will continue to still post their thoughts to help us along.....

Your 7 port project is an impressive achievement - I assume the fuelling setup is a lot easier - but how many have followed what you did ???

As for the Imp engine as an alternative project..... my engine has to use very expensive Wills Rings (instead of a head gasket) just to hold it together for 80HP normally aspirated - the thought of rebuilding it as turbo is seriously pushing the boundaries - although Fuel Injection would be stupidly easy with 8 ports - but it's only 998cc.

Mind you, really oversquare and revs to 8K plus easily.......

If only the remaining bodyshells wern't so rotten or the fibreglass Clan/Davrian/Ginetta so expensive....

EDIT - because it read crap first time...

Edited by Rod S on 15th May, 2009.

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


Gerald O

8 Posts
Member #: 5706
Junior Member

Raleigh, North Carolina, US




On 15th May, 2009 jbelanger said:
...In any case, I agree that the way forward is the hybrid mode with all the flexibility that will bring. And it will simplify tuning by removing some variables, some fiddling, and some unintended side effects.

I'll see where I'll fit the additional tables and how the transition will be handled. Hopefully I'll be able to do that relatively quickly and have a new release soon.

Jean
What about making it so the two pulses merge in a graceful and predictable manner? A third table just seems like too much complication. I can think of better uses for the memory.

I'd be interested in trying a configuration where the start/middle/end pulse modes could be set independently for the inners and outers. One could set the inners to run in end-pulse mode, and the outers in beginning-pulse mode, thus being able to guarantee that they remain separate by a known amount. I'm still focused on exhausting the possibilities of fully exploiting the timing method that I proposed on the msextra site for my B.


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

I would like to do it without having to use a third table but how? I tried to think of a way to actually do it but I don't see it. You could add the 2 tables contribution but then from the merge point on, you'd need to tune 2 tables to get a single pulse which doesn't make sense. And you'd still need a third timing table.

As for the memory, there is already a third VE table used for another purpose that could be used for this instead. And there is some memory available for the third timing table.

The idea of doing the end and beginning of pulse for the inners and outers is an interesting one. I'll have a look at it.

Jean

http://www.jbperf.com/


Rob H

4308 Posts
Member #: 700
Formerly British Open Classic

The West Country

On 15th May, 2009 Rod S said:
It's a shame others aren't trying it yet


I think this morning is the first time I've read one of the EFI posts and actually understood it, but I'm still not at a level of understanding where I can really add much.

On 15th May, 2009 Paul S said:
The problem is that the injection window for each cylinder is no bigger than the intake stroke, so maximum of 25% for each pulse.

Then you have to take away the minimum rest period between pulses which leaves you with 15-20% depending on how much your injector needs.


I understand where you're coming from and how you go 25% but wouldn't the actual value be considerably less and driven by the cam choice, assuming you want to avoid having fuel loitering in the ports?

Isambard Kingdom Brunel said:
Nothing is impossible if you are an Engineer


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk


On 16th May, 2009 Gerald O said:
What about making it so the two pulses merge in a graceful and predictable manner? ....etc.


Having had a good night's sleep I am now thinking the same.

Assuming what I saw on the scope was representative of all possible setups, although the point of merge was a bit messy and random, once they had fully merged, they seemed to fall into a perfectly symettrical patteren - it just wasn't what we were expecting with alternating short and long lengths.

The end points all seemed right (as expected at least) but they alternately started at the wrong point.

Is there a fundamental reason for this that could be "fixed" easily.

Although my intention remains to have large enough injectors (or staged) so the merge is avoided, I think I could live with the messy merge so long as the resultant pulses were wht I expected (mathematically).

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


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

On 16th May, 2009 Rod S said:
Is there a fundamental reason for this that could be "fixed" easily.

At the moment, each pulse is scheduled individually and the start of a pulse is scheduled at the wheel tooth before the desired angle. When the code wants to schedule the start of a pulse and sees that it's already injecting (collision between pulses) it waits for the following wheel tooth and forces the start at that tooth no matter if the previous pulse has ended or not.

This works well with a normal engine where the collisions occur only when approaching 100% duty cycle. It does affect timing (delayed until the following tooth) but at that point it's not important. And it does cut short the previous pulse but since you're starting a new one, you're at 100% duty cycle or more. And once the pulse width goes down to a "normal" duty cycle, the timing is re-established together with the pulse width.

The problem with the siamese code is that the pulses are not evenly distributed and they are not for the same cylinder. So when they merge, the inner cylinder pulse can be cut short by the forced start for the outer cylinders. Also, the timing for the outer cylinders is delayed by up to a wheel tooth (or 2 in some cases).

The problem about fixing it is that it would be necessary to compute both the beginning and end of the inner cylinder and check if there is an overlap with the outers start. That's not trivial and it's computation intensive. Also, what do you do with the collision? Merge the pulse width and advance the outers pulse and delay the inners pulse? Advance the whole thing to preserve the overall pulse width? Then there is the issue of the 2 opening time durations that need to be re-adjusted to one but then there may not be an overlap after that. And so forth.

As you can see, handling a graceful correct merging of the pulses is a very complex computation intensive process. And there would still be a need to have some user inputs to resolve certain parameters that would be different for different setups. So the 3 tables for going from 2-pulse to 1-pulse mode and having beginning-of-pulse timing for outers and end-of-pulse for inners would be the cleanest solution I can think of.

It's still not an easy setup for tuning but leaves all the flexibility to handle any setup from the almost standard engine setup which could get by with only the 2-pulse mode and 2 injectors all the way to a very highly boosted engine which will need 4 injectors with the transition to single pulse mode and staged injection.

Jean

http://www.jbperf.com/


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

Jean I (think) I understand your explanation but it's this part that doesn't make sense

On 16th May, 2009 jbelanger said:
The problem with the siamese code is that the pulses are not evenly distributed and they are not for the same cylinder. So when they merge, the inner cylinder pulse can be cut short by the forced start for the outer cylinders. Also, the timing for the outer cylinders is delayed by up to a wheel tooth (or 2 in some cases).


If the inner gets cut short why is it that what I'm actually seeing is the inner delayed but the outer apparently OK even after a ragged merge



The top trace is the cam pulse (from JimStim) so at a fixed point in the cycle. So I only show one injector.

Unless I'm totally confused, the purple circle shows me the outer pulse always finishing at the intended time but the green lines show the inner starting alternately on time, then late.

I've picked the trace I got just as the pulses were merging (the one in the middle isn't quite there yet) but the symettry is too much to be random.

Also, even if "cut short" could become "started late" depending on the timing setting choice of "start of pulse/mid pulse/end of pulse", it doesn't equate to a tooth length - there must be 72 teeth between the cam pulses, the difference in green lines is much more than 1/72nd.

Rod.

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


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

Rod,

I agree that there is something wrong with the picture above and that it doesn't match my description. I'd like to have a look at it myself if you could send me the msq. What were the conditions (MAP, RPM) when this was taken?

As for the timing, the pulses would all end at the same place if they all got pushed back and at that scale, it would be very difficult to see if they are actually exactly the same or vary by some degrees. Don't forget that it could be pushed back by much less than one tooth if the planned starting angle was already very close to the next tooth for the forced starting.

Jean

http://www.jbperf.com/


Rod S

User Avatar

5844 Posts
Member #: 2024
Formally Retired

Rural Suffolk

Jean,

It's from part way down page 2 of the thread so Paul's msq (although from early in his testing) and at 6K RPM with some boost added to get the fuelling up and pulses longer.

Later I tried my own (yet to be run) msq and got the same.

I agree it would be hard to see 1/72nd on the end of pulse on that plot - but I had pulled the timebase in to show how symettric the start error was.

PM me your actual email address (I cant see it on your web site, probably me not looking hard enough) and I'll email it over.

Rod.

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

Home > MS Code Discussions > Fully Sequential Injection
Users viewing this thread: none. (+ 1 Guests) <- Prev  
To post messages you must be logged in!
Username: Password:
Page: