[bugeye59]

PSIG - Your input and experience is very much appreciated and has directed me to a closer analysis. I am grateful.

Let me reflect this at you and see what you think. Perhaps I may have missed something that you could point out.

At 6000 rpm I believe we have 25ms between ignition events on a 4 cylinder. By extension we have 25ms between intake valve openings.

Now, I understand that the firing order implies alternating two successive shots on each siamesed port. But these shots are still separated by 25ms.

Granted it is close but there seems to be enough time to fire a typical 2-15 ms injector pulse on each injector - if I follow jonbill's recommendation of OR'ing the drivers to double fire each of the two injectors.

It would seem I would not need 5x sized injector but perhaps a 2x size injector? Like a 24lb?

This is a street application, In fact unlikely to see above 5500rpm. Also do we really ever see a full 15ms injector pulse around max rpm? Even on acceleration load drops off as you max out revs.

Have I missed something here?

[bugeye59]

Putting the injector sizing issue aside for a moment...

The problem I see is phasing. As mentioned above we must fire alternating two shots in each siamesed port.

But there is the potential (even with a crank sensor) that this alternating pattern could end up opposite phase.
In other words 50% of the time when you start the car you could be out of phase.

Out of phase would be bad with the siamesed ports because I believe the double shots would happen against closed valves, and then, when one intake opens it would gulp most of the double charge.

I recon Speeduino does not have a cyl 1 TDC input.

Back in the 80's Fords EECIV would trigger off of the distributor, where the optical disk had a fatter pulse for cylinder 1.
The particular engine I am referring to is the bank fired 2.9L BTW, This engine would intentionally fire the injectors against closed valves, apparently it made more power that way. But it was separate ports with each having an injector.

It would seem that the best solution for the siamese port problem is to just fire the injectors as rapidly as possible.
Even two injectors per port does not entirely solve the problem but it makes it easier to fire more rapidly.

[PSIG]

Yes. 6000/60 seconds = 100 rev's per second. 1000ms / 100 = 10ms per rev, /2 firings per-rev = 5ms per-firing. So, at first glance that puts you at 5ms between the first cylinder and second cylinder valve closings for the second to intake all its charge. Let's look at one pair of Siamese cylinders, and we can call the first to fire #1, and the second to fire #2.

While that's the concept, reality is much different. For example, intake duration (valve open) on a stock cam may be 270°. That means #2 valve will be open and stealing #1's fuel for 90°. But there is no following fuel squirt for #1 to steal, and #2's valve is closed long before the next squirt for #1.

This is the primary compromise for Siamese ports. It works OK with a constant-flow carb or even a central down-throat top-injector TBI (1, 2 or 4 injectors) that 'spreads' the fuel flow, because fuel is relatively constant and there for either cylinder when its valve is open, overlap or not. Shooting individual short puffs of fuel becomes complex. Side-note, I would plan on per-cylinder EGTs to keep an eye on uneven fueling.

To qualify this, the first cylinder of the paired port gets all the fuel waiting patiently in the manifold, while the second cylinder steals some of the first's in the overlap, so it's not a super-rich vs super-lean result. But, it's far from even, and changes with conditions and flows. We do have an option to use sequential fuel trims (the best feature of sequential ) to alter the flow of each channel's injections, varying by load and rpm, just as something to stick in the back of your brain for later.

I won't go into it deeper, as I'm sure you're as exhausted as I am and still have the work to do. Hey, it is your project after all.

[PSIG]

The problem I see is phasing. ...

Absolutely, and I have assumed throughout that as timing is so important, that a cycle signal (cam) would be used in any case. Thanks for stating that simple thing I didn't.

It would seem that the best solution for the siamese port problem is to just fire the injectors as rapidly as possible.

While that would make more 'constant-flow', unless the intake runners were joined (allowing other cylinders to 'steal' the fuel), the first cylinder would get most of the fuel. The carbs work as the fuel flows only as long as the cylinders suck air through them. Injectors flow fuel whenever they squirt, whether a cylinder is pulling air or not.

So, if you had 12 injections from one injector on one Siamese port, most of the fuel would pile up in the runner before the first cylinder valve opened, but no equal excess for the second one.

[bugeye59]

I was considering modifying the manifold by welding in a crossover tube. SU manifolds have such a crossover. My thought was that this would provide a more stable manifold pressure reading. But, as you indicated, it may help the mixture stealing issue at lower rpms. However my gut feeling is that it will have little effect at the mid/higher rpms at light loading where injector pulses are short.

The Weber manifold under consideration is by no means a straight shot. It may be possible to strategically place the crossover pipe such that it lies partially in the target zone of the injector. Thereby directing some of the spray pattern into the crossover pipe. But I suspect air flow is ultimately going to determine where everything goes.

Last night I did some basic calculations on air velocity through the manifold in an effort to get a rough idea of the lengths of the fuel/air pulses that would result at different RPMs and different injector durations. I need to double check those numbers.

On last question PSIG. and then I will leave you while I go back to contemplating my project.

Considering the following:
A four cylinder with a separate port and injector for each cylinder and a using a crank sensor.
Is it the case that Speeduino does not know firing order (because it does not know TDC on cylinder 1) and it may be that the sometimes injector fires against a closed valve and sometimes not?

[bugeye59]

Seems Marcel has done an interesting study on the siamese port problem...

https://www.starchak.ca/efi/siamese.htm

But he does not consider the balance tube.

PSIG, just post that link if anyone else inquires about siamesed ports

[PSIG]

Thanks for the link, and it is very similar to another that I could not find to post. I guess it's gone. Although it also covers my specific points to describe specific concerns earlier, it goes deeper into the overall dynamics. Yes, it will save me typing in the future, so thanks again.

Considering the following:
A four cylinder with a separate port and injector for each cylinder and a using a crank sensor.
Is it the case that Speeduino does not know firing order (because it does not know TDC on cylinder 1) and it may be that the sometimes injector fires against a closed valve and sometimes not?

First, yes, if limiting to only the crank sensor, as it does not know which of the two crank rotations of the cycle it's on. Speeduino can know where cylinder 1 is, as a basic option in setup with some form of cycle signal. This is not difficult or complex, and would be a requirement with most forms of timed injections on Siamese ports.

Second, the timing of each injection channel is fully controllable for where they end (individually or all-same in older code) - whether it knows Cyl1 position or not. This may force older code to use this valuable function. Most engines benefit from same angles on all cylinders, but Siamese would benefit from individual-cylinder control.

Note to @noisymime - this important function needs to be maintained, in-general for best operation, and specifically required for others. For examples, I can tune engines better with it (always a good thing), but without it I cannot (or very difficult) to tune most engines to pass certain emissions requirements. This is a big deal.

Injector close angle.png (54.83 KiB) Viewed 241 times

Also if using fuel-only, injections can be offset using odd-fire code to stagger injections within the cycle; an option to look at for Siamese. With sequential or multi-squirt options, many systems purposely inject on closed valves, or twice-per-cycle (open and closed), while others at some point of open. The timing of injections affects fuel distribution and vaporization, affecting efficiencies.

The factories invested in sequential mostly for efficient and effective emissions control. In racing and competition, sequential is used for entirely different purposes related to fuel timings and trims. We can use them for power, fuel economy, emissions or other specific efficiencies, each different from the others. These are all options in Speeduino.

[bugeye59]

Thanks for the tips.

It is definitely doable with two injectors...
https://www.youtube.com/watch?v=rL4seHvYS_c

Scroll to 3:56 in the above video. Granted it is a single entry split manifold but notice the large balance passage casted close into to the head side of the manifold.

In addition, the distance of the injectors to the head on this setup are very similar to the throttle body and weber manifold that I am suggesting.

I think adding a large diameter balance pipe to the Weber manifold will be very similar to above setup. Not only does it smooth out MAP, but at lower rpms It acts as a plenum that allows each siamesed port to pull additional mixture from the opposite side.

He also mentions - "Two injectors fired twice per cycle" (without defining what a cycle is)

[PeterJones]

BMW K100 head would solve some of these issues (and add a few, but...)

[bugeye59]

If I hit a wall setting up Speeduino there is a $25 option of last resort...
https://store.classicminidiy.com/collec ... i-base-map

But I think I can manage.

This guy is also selling a dual injector throttle body for $600.

You can see now, why the Sherryberg $95 manifold and $118 throttle body are so attractive.

I've started soldering up my Speeduino 0.4.3d pcb.