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

8565 Posts
Member #: 573
Formerly Axel

Podland

Now that Jean has developed the MS2/Extra code to handle fully sequential injection on the Megasquirt, I have decided to trial this option.

This involves some modifications to the MS board and the MS2 daughterboard, but these are all well detailed on Jean's website and the MS Forums. You also need one of Jean's Quad Injector Driver boards.

The problem that I have had with both the semi-sequential and the quasi siamese fully sequential modes is that I still cannot get enough fuel into the outer cylinders.

I have so far managed to get reasonably close AFRs up to 5000 rpm, but am still not really ready to trust it on a turbocharged engine.

The best results have been by using a single pulse that injects throughout the inlet phases of the both inner and outer cylinders.

This is a bit of a compromise as the code was not written to be used in this manner and although the car runs fine there are issues with acceleration enrichment. If you stomp on the throttle the AE will send all the extra fuel to the inner cylinders and there is a tendancy to bog.

So, I've got one of Jean's boards and also some LS2 type coils that mean I can run true fully sequential injection and coil on plug type ignition.

There is still the issue of how to time the injection to give the best results. I have come up with the following plan:


The red and blue lines represent the injection pulses with the dotted line being the mid-pulse timing.

Although the vertical scale reads RPM it could also read MAP, but hopefully you will get the jist of it.

At low revs or load the injection pulses are clearly separate, but as the revs or load rises, the pulses merge so that at full chat the pulses coincide.

From my trials with two pulses and the single pulse, I have sufficient data to develop the injection timing tables to give the desired results.

I will just use a single VE table to minimise the variables in tuning.

Any thoughts at this stage?

Edited by Paul S on 13th May, 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."


Jimster
Site Admin

User Avatar

9349 Posts
Member #: 58
455bhp per ton
12 sec 1/4 mile road legal mini

Sunny Bridgend, South Wales

I'm trying to keep up this time as I would still like to try this on a N/A 5 port A series

Team www.sheepspeed.com Racing

On 15th May, 2009 TurboDave said:

I think the welsh one has it right!


1st to provide running proof
of turbo twinkie in a car and first to
run a 1/4 in one!!

Is your data backed up?? directbackup.net one extra month free for all Turbo minis members, PM me for detials


Rod S

User Avatar

5847 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 13th May, 2009 Paul S said:
You also need one of Jean's Quad Injector Driver boards.


Or Jean's P&H board for those of us using low z injectors (sorry, couldn't resist that one....)

I'm still about two weeks away from running the engine on Jean's code (sorry Jean, I missed the date I said) but, the more I look at this as I piece it together, the more I think it's down to individual intake design.

Your's and mine are very different in that respect, especially location of injectors.

How much that difference makes, remains to be seen.

As soon as I finish the bathroom tiling (wifey has banned me from the garage until I've finished....) I'll catch up and wire mine up with the 4 injector P&H board. I've got so many different colour wires in my garage now, I can do any configuration and totally confuse myself.....

Anyway, back to your graph.... Is the bottom scale one engine rev (rather than on cycle) because it doesn't quite make sense at the moment - it looks like you are showing a 50% duty cycle at 6000 RPM

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


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland




On 13th May, 2009 Rod S said:

Anyway, back to your graph.... Is the bottom scale one engine rev (rather than on cycle) because it doesn't quite make sense at the moment - it looks like you are showing a 50% duty cycle at 6000 RPM


Yes it is just one rev and half a cycle. It would repeat for cylinders 3 and 4.

You are also correct in that I'm showing 50% duty cycle at 6000 rpm, but that may be at considerable boost. Also that may be 8000 rpm and 21 psi. The vertical scale is a little arbitary, but helps show the injection strategy.

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

On 13th May, 2009 Rod S said:
On 13th May, 2009 Paul S said:
You also need one of Jean's Quad Injector Driver boards.


Or Jean's P&H board for those of us using low z injectors (sorry, couldn't resist that one....)

Or both if you want to use the 4-channel driver board for the 4 ignition drivers and the additional DB15 connector. *wink*

Paul, the accel enrichment you get with the single pulse with dual table mode is actually a bug in the main code and is easy to correct. I'll have a fix for that in the next release.

As for the graph, unless I'm reading this wrong, there might be a misunderstanding about using the 4 injector channels with the full sequential siamese-port mode. You indicate overlapping injector pulses which leads me to think that you expect to have each channel dedicated to a cylinder.

That's not the case. The 2 additional injector channels are simply there to allow staged injection in full sequential siamese mode. This means that the primary channels will still have an injection pulse for both inner and outer cylinders and the secondary injectors will come on at a certain load/RPM (as configured for staged injection) and will also have a pulse for both inner and outer cylinders. Also these pulses for the primary and secondary injectors will have the same timing and the same pulse width (except during the transition from staged off to on).

This means that injector channel 1 will inject fuel for cylinders 1 and 2 and channel 2 will inject for cylinders 3 and 4 with channels 3 and 4 inactive while staging is off. And when staging comes on, injector channels 1 and 3 will inject for both cylinders 1 and 2 and injector channels 2 and 4 will inject for both cylinders 3 and 4.

If that's what you understood and meant then you can just ignore the text above (and it will hopefully be useful to someone else). In any case, the way it's working will allow you to add more fuel to the outer cylinders because you will have 2 injectors providing fuel during the injection window.

If you meant to use the non-siamese full sequential then the issue I see from my understanding of your graph is that all injectors have the same timing with respect to their own cylinder. This means that you will not be able to get the overlapping you show. But as I said, the staged injection with the siamese full sequential should cover your needs (in a different way).

Jean

http://www.jbperf.com/


Turbo Phil

User Avatar

4378 Posts
Member #: 20
My sister is so fit I won't show anyone her picture

Lake District

You guys pushing this forward deserve a huge amount of credit. Well done. :)

WWW.TURBO-MINI.COM


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland

Bugger, that has stuffed that idea then.

Back to the effing drawing board.

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

It's always possible to implement another mode but I'd like to know the what is currently available doesn't work.

Is there any reason why you think staged injection wouldn't solve your issue? Staging can be combined with dual table mode so you have the same flexibility in terms of timing and setting the VE separately for the inner and outer cylinders.

Jean

http://www.jbperf.com/


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland

The major problem that I see is that using two pulses close together on the same injector does not work as the theory would suggest.

By the time the MS has shut off the injector, waited for the next opportunity to fire again, then fired it takes more than just the millisecond to open the injector to get the fuel moving due to reverse pulses in the fuel tract of the injector.

Reality is that you need 20% of the duty cycle between pulses.

So, at high revs you end up with much the same injection timing as the semi-sequential mode. Once you start injecting against a closed valve, it's a whole different ball game. I know because I spent 6 months or more trying to get it to work.

If we can get the single pulse mode to work correctly, without firing the second pulse under any circumstances, then there is still a solution with the staged injectors.

Ultimately we can still get four injectors firing at 50% duty cycle at full chat with a passing chance of getting the same fuel in each cylinder.

Rant over......... for now.

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

To be pedantic about it, if you really need 20% duty cycle between pulses (which is probably a good estimate) and if you limit the duty cycle to 50% or less, you can have the pulses closer than with semi-sequential. But that's not significant because you'll still be injecting mostly on a closed valve.

When you say you've been trying to make the semi-sequential to work without success, I think you're ignoring the fact that you can now have a different pulse width for the inner and outer cylinders when using the sequential mode (instead of semi-sequential). This should go a long way towards solving the AFR mismatch.

Anyway, the unwanted extra pulse will be removed in the next code release. And if you need it, I can send you a pre-release version.

As I said, if it's really needed I can add a new mode that will do what you wanted. It shouldn't be a huge deal because all the pieces are there in the code and is mostly a matter of putting them together. It certainly wouldn't be even close to what it took to create the original sequential siamese option.

Jean

http://www.jbperf.com/


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland

I'm happy to wait for the next release.

I would like some feedback from other testers to see if they can get the sequential siamese mode to work. It may be just my particular inlet design or cam selection that is causing the problems. I don't know.

However, I am currently of the opinion that the sequential siamese mode is unworkable in it's current guise using dual VE tables.

Hence, instead of a new mode, I would like to see this mode based on the single pulse method, so that you do not have to fudge around with settings to get it to work.

Finally Jean, I would just like to say that the work you have done is great and without your efforts this would not be possible. It is very much appreciated.

But...... we have a bit more fine tuning of the code to do before we can make it work properly.

Edited by Paul S on 13th May, 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."


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

I agree that there's more work to be done. And I would also like to see results from others just to see how much difference there is between different setups.

I can see the new mode being able to do both of the configurations you want: with 2 drivers, it would be a single pulse without having to use dual table mode and with 4 drivers, it would be the per-cylinder injector channels with different timing for the inner and outer cylinders. The single pulse is really the same thing where 2 drivers are ignored (unless you still need staged injection).

Jean

http://www.jbperf.com/


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

I think that ultimately what will be the best solution is something close to what your graph shows but in a slightly different way (and something I had in mind when I first started thinking about this code) where you have 2 pulses at low load/RPM and a single pulse at high load/RPM. All that with a single injector driver per intake port. And then over that, the possibility to add staged injection for even more fuel if/when needed. With this you could be injecting on an open valve only and maximize the duty cycle.

What is needed with this is a way to handle the transition between the 2 modes. There's the issue of when to change mode (with some hysteresis) and how to handle the timing and pulse widths without having to fiddle with too many tables.

One advantage of a mode like this is that you could have 3 mode in one: one pulse per cylinder, one pulse per intake port, and a mix of both. All that would be required is the setting of the transition point to have one of the 3 sub-modes.

Jean

http://www.jbperf.com/


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland

I think that the transition from dual to single would be a nightmare to set up.

You would have to ensure that one of the VE tables doubled up exactly at the same time as the other zeroed out. The timing table would also have to be set up so that the timing changed precisely at the same time.

Synchronising all this would be very difficult. In my opinion.

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."


Andy500

User Avatar

812 Posts
Member #: 2443
Post Whore

Cheshire

wow, my heads gone now!


Rod S

User Avatar

5847 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 13th May, 2009 Paul S said:
The major problem that I see is that using two pulses close together on the same injector does not work as the theory would suggest.

By the time the MS has shut off the injector, waited for the next opportunity to fire again, then fired it takes more than just the millisecond to open the injector to get the fuel moving due to reverse pulses in the fuel tract of the injector.

Reality is that you need 20% of the duty cycle between pulses.


Sorry, been doing something else the last couple of hours.....

But, surely this simply calls for larger injectors or "basic" staging.

Whilst mine isn't quite finished yet, from earlier posts/discussions when I was being educated about injector duty cycle, I have bought some very high capacity injectors to do my testing with.

So I hope to do my testing with the two pulses per injector, per cycle...

I am also hoping (refering to my earlier comment) that having the injector boss virtually touching the head, will help in terms of wall wetting and transit time.

Also, one of the reasons I have stuck with low z is because they are alleged to open/close faster (remains to be proven....)

I just need a big kick to get it finished....

Edited by Rod S on 13th May, 2009.

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


pinkyandnobrain

User Avatar

720 Posts
Member #: 2588
Post Whore

Pretoria South Africa

(RUNS..............And drop-kicks Rod S in the ass !!!!)

LOL

Guys awesome stuff ! The 5 port EFI stuff is the main reason I joined the forum and I must say im still getting to grips with in all but what you guys have achieved is really amazing !!
*Clapping*

"So wat we gonna do tonight Brain?"
"Same thing we do evernight Pinky!"
"Try to take over the world!"


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland

The problem with big injectors is that they are supposed to be problematic at idle. Hence, I'm reluctant to take too great a step in that direction.

However, I've not had an idle problem with four large (480cc) high z injectors, although pulse widths were very low, less than 2mS.

I've just bought a set of four 630cc high z injectors for the 1293 turbo. So I could try those on the 1030 engine.

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

On 13th May, 2009 Paul S said:
I think that the transition from dual to single would be a nightmare to set up.

You would have to ensure that one of the VE tables doubled up exactly at the same time as the other zeroed out. The timing table would also have to be set up so that the timing changed precisely at the same time.

Synchronising all this would be very difficult. In my opinion.

I wouldn't do the transition that way. It would probably be best to have 3 tables: 2 for the 2-pulse mode and 1 for the single-pulse mode. You could then tune the 2-pulse mode to some 100s of RPMs above your intended transition point and the single-pulse table below it. Then the code could make the transition with some hysteresis so as not to jump from one mode to the other and back too quickly.

So you would end up with, say, 2 tables for 0 to 3000RPM and one table from 2000 to 8000RPM (for VE and timing) with a transition at 2500RPM (which could actually be at 2700 going up and 2300 going down).

This dual mode would mean a bit more tuning than a single mode but might be worth it and would be an "easy" way of getting all the options in a single sequential mode instead of having siamese a, siamese b, siamese c,...

Jean

http://www.jbperf.com/


Rod S

User Avatar

5847 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 13th May, 2009 Paul S said:
The problem with big injectors is that they are supposed to be problematic at idle.


Agreed - but lumpy idle compared to a melted piston..... I know which I would take :)

Sorry, very flippant...... I think that is why the staging is the answer, one of us has to actually build the four driver injector system (low z or high z) and see what that does with small/large injectors.

I only have 78 (very large) tiles to go.....

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


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

On 13th May, 2009 Rod S said:
I only have 78 (very large) tiles to go.....

Do they really have to all be even? Randomness has it's charm... *wink*

http://www.jbperf.com/


Rod S

User Avatar

5847 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 13th May, 2009 jbelanger said:
On 13th May, 2009 Rod S said:
I only have 78 (very large) tiles to go.....

Do they really have to all be even? Randomness has it's charm... *wink*


Unfortunately, wifey borrows my Rubert's Gauge just to check the quality of my grouting.... so random tiles would not work *frown*

I've actually been finishing of the new wiring loom over the last couple of days and the thing I now REALLY hate about MS is the use of a Dsub37 connector - that connector is for small PC type wires, NOT automotive sized wires......

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


jbelanger

1244 Posts
Member #: 831
Post Whore

Montreal, Canada

On 13th May, 2009 Rod S said:
I've actually been finishing of the new wiring loom over the last couple of days and the thing I now REALLY hate about MS is the use of a Dsub37 connector - that connector is for small PC type wires, NOT automotive sized wires......

That's why it's a very good idea to buy a ready made wiring loom. Unfortunately, none of them are complete for an MS2/Extra setup.

http://www.jbperf.com/


Paul S

User Avatar

8565 Posts
Member #: 573
Formerly Axel

Podland


On 13th May, 2009 jbelanger said:
I wouldn't do the transition that way. It would probably be best to have 3 tables: 2 for the 2-pulse mode and 1 for the single-pulse mode. You could then tune the 2-pulse mode to some 100s of RPMs above your intended transition point and the single-pulse table below it. Then the code could make the transition with some hysteresis so as not to jump from one mode to the other and back too quickly.

So you would end up with, say, 2 tables for 0 to 3000RPM and one table from 2000 to 8000RPM (for VE and timing) with a transition at 2500RPM (which could actually be at 2700 going up and 2300 going down).

This dual mode would mean a bit more tuning than a single mode but might be worth it and would be an "easy" way of getting all the options in a single sequential mode instead of having siamese a, siamese b, siamese c,...

Jean


I can see that would work as long as you have enough memory left for the extra table.

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."


Gerald O

8 Posts
Member #: 5706
Junior Member

Raleigh, North Carolina, US

What happens with the existing code when the timing for the two pulses overlaps? Does it effectively result in a single longer pulse?

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