Donations towards server fund so far this month.

 
£0.00 / £100.00 per month
Page:
Home > Paul S trials and testing > Siamese Code Trial - Take Three

Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 14th Mar, 2009 jbelanger said:

Can't you suck on the MAP nipple? :) :)


No the tube isn't long enough :):):)

I just wish I had a real one at the monent intead of an MS/PC/scope *happy*

Here is Paul's .msq (without MAP correction but still WOT as far as I can simulate it at present) at 3k RPM



Injector pulses similar BUT, of course the RPM is lower so..



The gap between pulses is so much bigger.

EDIT - I should point out the difference in photo display is beacuse I've lost daylight and my camera can't cope with that so you will just have to guess where the timebase markers on the display are.....

The sooner I get the fancy USB scope, the better,,,,

EDIT 2 - also I obviously halved the timebase to keep the display showing one cycle (2 revs) to try to make sense of this so a 5 mSec pulse is now half the length on the display compared to the previous photos....

Edited by Rod S on 14th Mar, 2009.

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


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 14th Mar, 2009 robert said:
sounds like the higher fuel pressure may pay off then to shorten the pulse .. how high can it go ,,,70 psi ?what is the inj flow and pressure on the mpi mini?


Robert, I think it's the pulse that needs to be shortened but I think higher flow injectors is better than higher fuel pressure - earlier I simulated the higher flows in MT with good results but the flow from an injector is not proportional to pressure - a large pressure increase shows non-proportional flow gains - so I would like to keep pressure for "fine tuning" if that makes sense.... (it almost does to me:) )

Edited by Rod S on 14th Mar, 2009.

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


James_H

User Avatar

3692 Posts
Member #: 1833
Formally mini_majic

Auckland, New Zealand

so higher pressure and shorter opening times should prove interesting....


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

One further thought....

Paul, do you have an .msq of your best "pre cam sensor" siamese code so I could run that and display it.

Just a thought....

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


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 14th Mar, 2009 jbelanger said:

However, you're mistaken somewhat in your comment to Rod. If the pulses are too close to one another the 1ms of dead time is no longer valid since the injector hasn't had time to fully close before voltage is applied to it again. Then all bets are off as to what the actual dead time is. I would be very wary of anything less than 1ms between pulses and would actually personally prefer to see more unless tests are done to establish what's the safe interval.

Also if the overlap is quite short, you can actually have more fuel than what you'd expect because you get a single 1ms of dead time instead of 2. So overlap, or anything close to it, means unexpected behaviour. Hence my repeated (and unwanted :) ) comments about spreading the pulses.

Jean


Right, I think you may have hit on something here.

From an electrical point of view a 1mS opening time is possible and it may even be possible to have a gap in the injector open period of less than that.

However, hydraulically, the injector has to give the fuel time to accelerate though the nozzle. This probably takes place during the injector opening time and carries on for a short time thereafter.

Now when the injector closes it will create a reverse pulse. If you try to re-open the injector within a very short time after it has closed, it is very likely that it will take even longer to accelerate the fuel because of the reverse pulse.

This may explain why I cannot get any more fuel into the outer cylinders. It is probably due to a dead point in the fuel delivery due to the closing then opening a short time after.

So, we either shorten the pulse widths with higher fuel pressure or more injectors, or we get rid of the gap between pulses altogether.

Thanks for the help guys. It's been getting me down and my IBS is howling at me tonight, so apologies for sounding terse.

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

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

On 14th Mar, 2009 Rod S said:
One further thought....

Paul, do you have an .msq of your best "pre cam sensor" siamese code so I could run that and display it.

Just a thought....

You won't be able to use it with the code you have because this was the old siamese code and there has been loads of changes in the msq format and content. Unless Paul did port his old data to the new code.

Jean

http://www.jbperf.com/


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 14th Mar, 2009 jbelanger said:
and there has been loads of changes in the msq format and content. Unless Paul did port his old data to the new code.


I don't think so as there was an earlier comment about not being able to import the VE tables and having to enter them manually...

A shame (every little bit extra helps) but I still think the progress is forwards....

It is just going to take some time but I'm sure the end result will be worth it...

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


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

yes rod i agree ,....but as a quick and dirty 'see what happens' without any more expense its worth a try .. i seem to remember a calculator on ms forum for finding flow increae with pressure change.. ill see if i can find it .

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


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

newRate = oldRate * sqrt(newPressure/oldPressure)


even better heres a calculator

http://www.csgnetwork.com/fiflowcalc.html

oops heres a better one

http://www.injectorrx.com/sizing.html

Edited by robert on 14th Mar, 2009.

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


jbelanger

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

The flow increase is equal to the square root of the pressure ratio. So, for example, if you go from 45psi to 60psi fuel pressure you increase flow by 15.47% (square root of 60/45 = 1.1547).

Jean

http://www.jbperf.com/


robert

User Avatar

6743 Posts
Member #: 828
Post Whore

uranus

so even 70psi is only going to give 468 cc odd,25 % more ...
if the pump can do that ..worth a try to see if the logs change though .

bigger injectors deff worth trying .

trouble is it all costs !

Edited by robert on 14th Mar, 2009.

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


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

OK,

Paul, if you are only testing on two injectors, what about trying putting the other two wired in parallel and halving the VE tables ???

That should put you back with a big gap between (like my 3000 RPM photo).

Although I'm probably missing something obvious as the alcohol kicks in.....

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


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

Increasing the fuel pressure is not going to give much of an advantage. Plus I want to leave enough for the pump to cope with up to 30psi boost!

I still have the other two injectors connected to the fuel rail so I can easiy bring them back in.

Ultimately, the only solution may be to go full sequential with Jean's new board.

Rod, can you please try and merge the two pulses? Just need to be sure that the ECU fires the injector when the first pulse starts and stops when the second pulse stops and not before.

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."


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland


On 14th Mar, 2009 Rod S said:
OK,

Paul, if you are only testing on two injectors, what about trying putting the other two wired in parallel and halving the VE tables ???

That should put you back with a big gap between (like my 3000 RPM photo).

Although I'm probably missing something obvious as the alcohol kicks in.....


Ha Ha thoughts crossed in space.

But you just need to halve the Req_Fuel, not the whole 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."


jbelanger

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

Paul,

If you want a single pulse don't do it with 2 pulses merged but with one table set to 0. Merging 2 pulses has side effects in the code because it does delay the second pulse to try to cope with the overlap.

With a single pulse from a zeroed table you keep full control over timing and pulse width. Of course you have to either up the VE table or the req fuel.

Jean

http://www.jbperf.com/


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 14th Mar, 2009 Paul S said:

Rod, can you please try and merge the two pulses? Just need to be sure that the ECU fires the injector when the first pulse starts and stops when the second pulse stops and not before.


Not sure what you actually mean there...

I'll try whatever you want but I can't alter the VE tables easily for paired injectors because the injector "size" isn't entered directly into MT, it's used to "calculate" the fuel tables but you can't recover the information after the calculation - ie, I can't see the injector sizes you used in your .msq, only the end result as a "ReqdFuel" figure and associated VE tables....

However, that may not be what you are asking... (by "merge the two pulses"...)

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


jbelanger

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

After writing the last post it just reminded me of something. Another reason not to have pulses too close together is due to how the injection timing is done. It starts a timer at the wheel tooth just before the injection angle and then fires the injector for the pulse duration using the same timer. However, if the injection from the previous pulse is still going, it can't use the timer so it sets a flag which says to fire immediately at the next tooth.

So this has a side effect of changing the timing of the second pulse and potentially reducing the first pulse if the overlap is too big (more than the equivalent of 1 tooth). This method was chosen because it handles 100% duty cycle very well in a normal engine sequential injection and it should be ok with siamese engines if people are careful about it (since you don't more than 60% duty cycle if that).

I'm sorry not to have mentioned that before but there are so many things in the code that some aspect of it still need to be written down.

Jean

http://www.jbperf.com/


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland




On 14th Mar, 2009 Rod S said:
On 14th Mar, 2009 Paul S said:

Rod, can you please try and merge the two pulses? Just need to be sure that the ECU fires the injector when the first pulse starts and stops when the second pulse stops and not before.


Not sure what you actually mean there...

I'll try whatever you want but I can't alter the VE tables easily for paired injectors because the injector "size" isn't entered directly into MT, it's used to "calculate" the fuel tables but you can't recover the information after the calculation - ie, I can't see the injector sizes you used in your .msq, only the end result as a "ReqdFuel" figure and associated VE tables....

However, that may not be what you are asking... (by "merge the two pulses"...)


What I meant was to advance the second pulse so that it starts before the first pulse has finished.

Just to see how the code handles it. It may still close the injector at the end of the first pulse.

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."


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

On 14th Mar, 2009 jbelanger said:
Paul,

If you want a single pulse don't do it with 2 pulses merged but with one table set to 0. Merging 2 pulses has side effects in the code because it does delay the second pulse to try to cope with the overlap.

With a single pulse from a zeroed table you keep full control over timing and pulse width. Of course you have to either up the VE table or the req fuel.

Jean


With setting the table to zero, would the code not still add the injector opening time and part fire the injector?

My preferred method would be to time the second pulse inside the first by adding 180 degrees to Injection Timing Table 1.

EDIT: After reading your other post Jean, I now realise that my preferred method will not work because of the timer issues.

If Rod can still see what happens on the scope when the pulses get too close, that would confirm.

Edited by Paul S on 15th Mar, 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."


Rod S

User Avatar

5988 Posts
Member #: 2024
Formally Retired

Rural Suffolk

On 15th Mar, 2009 Paul S said:

If Rod can still see what happens on the scope when the pulses get too close, that would confirm.


I'll give it a try later - just got back from the Gardening Centre (wifey dragged me as it's apparently Spring now) and as a result I now have "things to do" in the garden....

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


jbelanger

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

On 15th Mar, 2009 Paul S said:
With setting the table to zero, would the code not still add the injector opening time and part fire the injector?

No. The code simply doesn't fire the injector if the VE table is 0. You can see that in MT as the reported pulse width will be 0 and not 1ms.

Jean

http://www.jbperf.com/


jbelanger

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

One thing to add. If you do the single pulse with 0 VE in the second table and you want to have 2 pulses at low RPM, you have to be careful how you do the transition.

Because the code interpolates between the VE table cells, you can't simply keep the table you have now and fill it with zeros starting at one column. That would result in a diminishing pulse width between the RPM before the table transition and the RPM with 0. Since you will need to increase VE (essentially doubling it) in the first table to compensate and if you have the same RPM points then it should compensate to give the correct overall fueling. But you will have to adjust the timing table such that you get the fuel where it is needed which might be tough.

So what you could do is to have the transition as small as possible with the first column with 0 being 1RPM more than the last column with non-zero values. Then the transition would pretty much be like a switch from one mode to the other.

Jean

http://www.jbperf.com/


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

We have had some success today.

We are still working with single injectors per port and 3 Bar fuel pressure.

We have spent considerable time adjusting both timing tables in each direction to find the optimum.

We have advanced the first pulse so that the mid-pulse is at 145 degrees and retarded the second pulse a bit so that is at 35 degrees, both at 6000 and 100% MAP. Hence the pulses are 290 degrees apart at peak.

Still with 20% less fuel in the first pulse.

There is an obvious benefit from separating the pulses at high rpm, but not as much as I had hoped.

The AFRs from 3800 to 5500rpm are now at 11:1, inners and 12.5:1, outers.

At long last we have managed to get some more fuel into the outer cylinders, but still cannot lean out the inners.

We might try playing a bit more with the VE tables now that we seem to have optimised the timing.

Edited by Paul S on 15th Mar, 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."


Paul S

User Avatar

8604 Posts
Member #: 573
Formerly Axel

Podland

A little bit of log for your enjoyment *happy*


Attachments:

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

1267 Posts
Member #: 831
Post Whore

Montreal, Canada

That's interesting. First, congratulation on the progress.

Second, it's interesting to see that during fast transitions (around 223s and, less visibly, 292s), the AFRs get reversed in that the inner cylinders get richer than the outer cylinders as opposed to what we see at high RPM.

I'd really like to see a Megatune log to see what is happening within the code because, while this really shows what is happening in the engine, it gives absolutely no reason why the code is doing what it's doing. It could provide great clues as to why you can't lean out the inner cylinders because they obviously can be leaned out as shown by the transitions.

Jean

http://www.jbperf.com/

Home > Paul S trials and testing > Siamese Code Trial - Take Three
Users viewing this thread: none. (+ 1 Guests) <- Prev   Next ->
To post messages you must be logged in!
Username: Password:
Page: