gemellocattivo.com

Mark's setting up a Pseudo-DI port injection system for his V12.
The idea interests me for my Northstar.

A normal port injection system puddles fuel on the back of the closed intake valve. When the valve opens, the cylinder slurps that fuel in... hardly optimal mixture quality, and costs a port EFI system 3-5% power relative to a Pseudo-DI system. In the port system, a small and somewhat unpredictable amount of the fuel even goes out the exhaust pipe during the overlap period.

However for the Pseudo-DI system,

If the injection doesn't start until the exhaust valve closes, then some of the charge air has escaped out the exhaust valve... this is the difference between Volumetric Efficiency and Trapping Efficiency. An engine with 120% VE pulls 120% of its displaced volume through the throttles, but if it has a 83% trapping efficiency, then 0.17 of the 120% goes out the exhaust pipe and the cylinders are actually being filled with 100% of their volume. (1.20 * 0.833 = 1).

Won't the extra charge air going out the exhaust confuse the wide band O2? If the 100% of the air in the chamber is burning stoich, then adding the untrapped charge air will shift Lambda by 20%, right?

How should that effect be compensated without knowing what the trapping efficiency is?

Plug reading? Knock sensing via plug gap impedance? Educated guesswork? Establish a tune that reads stoich on the WBO2, then lean it out for max power?

Interesting points.

On the tuning I think all engines that lose air out the exhaust will have the "looking lean" problem because fuel available or not the charge shouldn't combust so the O2 sensor sees O2. So that should depend on the charge loss not the injection type.

Which leans to the second point.....contrary to popular belief you really can't tune with an O2 sensor. You just can't know what mixture makes the most power without running the engine on a dyno and measuring the power. For sure on a naturally aspirated engine the answer will be in the 11:1 - 14.5:1 range and probably close to 13:1 so 13:1ish is a great starting point, but it's just that, a starting point. Once you get on the dyno you add or remove fuel and timing based on the hp/torque and don't really pay any attention to the O2 reading or at least I don't

Here's a video of a V8 with 8 O2 sensors doing closed loop separately on all 8 from a thread on speedtalk
https://www.youtube.com/watch?v=6kQGp-f ... tube_gdata

This is a feature that the MS3pro has appearently and the guy in the video, who I've always found to be pretty knowledgeable loves it and I guess seeing they are all the same should kind of indicate they are all the same, but most people use EGT or plug reads for that because they both tell you about the combustion process directly not the cycle byproducts so they tend to be more helpful for extracting hp. I guess either way having data is better than not having data.

Then there is that 3-5% gain I've heard of and am chasing...that still confuses me. In theory I think it can be explained by sending the fuel in as liquid....but also in theory vaporizing the fuel should cool the intake charge causing it to become denser by a similar %. I don't know. High injection is on all the sport bikes now and helps hp.....but my motec dealer insists he's removed it and gone short pulse low injection and made more hp so I'm going to try

One thing I stumbled on the DI injectors that the guys I know claiming the short pulse injection all use have a VERY narrow spray pattern, like 5-10 degrees. that might be part of the answer, I'd think a low mounted injector with that little spray angle has a much better chance of delivering liquid to the cylinder.

They also say the only way to time the injection is on the dyno and getting it right is critical to seeing the hp bump. That is odd and makes me wonder a bit about what's going on exactly.

I guess it is WBO2 101 that the free O2, and not the raw fuel or lack thereof, that has the reducing effect on the Nernst cell

So in terms of WBO2 readings, a Pseudo DI engine won't be any different than a port injected engine in a the same state of tune.

That brings up another question in my mind... Engines with different trapping efficiencies will show different lambda readings for the same actual "combustion" lambda... so the sensor never reads "combustion" lambda, which is always richer than the sensor reading due to lost mixture or charge air. I understand that AFR by itself is not a tuning tool. It's like EGT, in that it's a guideline for the safe operating limits of the engine.
The way to deal with the variability of mixture from "actual" is just to adjust mixture to peak torque, and go through the entire RPM range, adjusting mixture for peak torque at each RPM point to account for changes in the engine's trapping efficiency with RPM. Obviously this isn't as big a job as it sounds like... tune for a flat AFR curve, then adjust the entire curve up or down in 0.1 or 0.2 increments to see where in the curve the torque goes up or down, then adjust the fuel (and ignition) curves accordingly.

Regarding the high mount injectors...
Typical fuel injectors at 40-50 psi are lousy atomizers. They really just squirt streams of fuel. Mounting conventional injectors high gives the mixture motion in the runner more time to reduce the droplet size and homogenize the charge, resulting in better mixture quality and more consistent cycle to cycle operation.

While EFI meters more precisely than carbs, a well set up carb can delivery a higher quality mixture than a production based port EFI system, in terms of droplet size and mixture homogeneity.

I've read some topics and posts on Speedtalk discussing high pressure port injection--running 200-300 psi fuel pressures with specialized injectors in the intake runners. These achieved better mixture quality due to better atomization and showed a power increase relative to typical 3.5 bar fuel pressure.

Fuel vapor fraction going through the valve is another tradeoff. Fuel vapor displaces MUCH more air than the same mass of liquid fuel, so you want as little fuel vapor as possible going through the valve, BUT you need some amount of fuel vapor in the chamber to ignite the mixture... so the fuel needs to start vaporizing before it goes through the valve.
OTOH, vaporizing fuel drops the temperature of the entire mixture, increasing air density (and reducing velocity), so there's probably a very specific vapor fraction that results in the greatest mass of air in the chamber... BUT that vapor fraction will depend on some other factors and may be different with high mount vs. low mount injectors, due to the differences in interaction time that the fuel and air have before going through the valve.

In light of that, I *think* Psuedo-DI with conventional pressure relies on the high air velocity through the throat/valve, which is of course the highest velocity in the system at any point on the load map, to help atomize the fuel right as it goes into the chamber, avoiding both the big SLURP of conventionally sized injectors, and the mixture quality degradation from high mount injectors that spray longer than the intake valve is open and experience static air in the runner.

Sooo.... tuning fuel delivery is an iffy, finnicky operation.

However, a DEFINITE advantage of Pseudo DI (PDI? PsDI?) is that it doesn't send raw fuel out the exhaust, so it should show improved fuel economy, particularly at low RPM & light loads for engines in high states of tune, relative to conventional port injection. It should also be easier on the catalysts during extended operation at high engine loads because they don't have to deal with as much raw fuel in the exhaust stream.

That all sounds right....it's frikin complicated.

I'm operating on the belief that BIG carbs on drag engines generally make as much if not more HP than EFI because the fuel is basically pouring into the engine. I read someone on speed talk on about "super cycles" and I know he was working with carb engines so I'm thinking he's seeing the mixture variations that you get from fuel just running in, kind of repeatable but not really.

The important point I think though is the fuel goes where the air is going so you get the pseudo DI effect....but you get it with mixture variations cycle to cycle. With huge injectors spraying at the valve you should get the gain but with much smaller cycle to cycle variation.

Maybe?...that's what I'm thinking anyway.

On the fuel economy point that is what I had in mind too. But my motec guy though is saying he only does valve open injection at high power and at low power sprays on the closed valve because it runs smoother...which is the very worst thing for economy and emissions with high over-lap engines. WTF????

Looking at my dynomation sims the highest charge loss to exhaust I see is under 1% but the loss to reversion (into the intake) hits 20%! So on my engine, according to the simulation there is not a lot of fuel loss to the exhaust no matter what I do with the injection so maybe the motec guy isn't nuts....unless that 1% the very beginning of the intake cycle and is almost entirely fuel because all I'm spraying on the closed valve so the fuel is sitting on the valve ready to go out the exhaust

My plan is to fire the thing up with injection end time at 400 degrees BTDC.....so on the valve. Using MAP load sense. All know, all easy and just get it running decent. Once it's running I'll map the MAP vs TPS values and create a new TPS load based tune to get my ECU to stable injection time vs load...then start playing with the pseudo-DI and I have.

It is threads like this that make my head swell and feel like it is going to explode.
Having no engineering education it is fascinating to read and actually follow the logic of most of it.

_________________
Wade Williams
Master Ferrari Tech
Restoration Tech
Engine Machinist
Fabricator
Racecar Driver

LOL

Engineering is mostly just common sense and 9th grade algebra...but don't tell anyone I let secret out

Heard time and again in the Navy: "Be brilliant at the basics"

K
I
S
S

Keep
It
Stupid
Simple

Yup.
I never had Algebra, so it is a little foreign for me.

_________________
Wade Williams
Master Ferrari Tech
Restoration Tech
Engine Machinist
Fabricator
Racecar Driver

I know it's an old thread, but it's cool so I thought I'd post to it.

When fuel or fuel/air 'short circuit' through the exhaust valve, you can no longer use an O2 based lambda sensor to measure lambda. You can use a gas analyzer which measures CO/CO2 to measure burned lambda (in cylinder lambda if nothing burns in the exhaust), or CO/CO2/THC/O2 to measure exhaust lambda (which includes burned gases and unburnt O2/THC for final air/fuel ratio). This is generally regarded as the 'best way' to measure lambda since it isn't cross-sensitive to anything (wideband O2 sensors are cross-sensitive to high pressure among other things). However normal gas analyzers are really, really, really slow (mine outputs data at 2hz, with a several second sampling delay through a 20ft heated sample line).

2-stroke engines usually suffer from this problem. You can't use a wideband to calibrate a 2-stroke since quite a bit of the air goes right through. 2-strokes rely on exhaust resonance to keep the streams separate, this isn't very effective at most speeds. So looking at 2-stroke tutorials might be useful. Virtually all are carb'd though since they are used in powersports and boats. One company makes a 2-stroke DI system but it doesn't use a conventional (common rail) design, so pressure is built with a huge electric solenoid in each injector.


As to DI,
The loss of power due to PFI vs GDI is partly due to in-cylinder evaporation of fuel increasing the knock limit. Fuel requires energy to evaporate, so when it evaporates it will cool its surroundings. For PFI this is the valve and intake port, along with some of the air. For DI, this is all air or combustion chamber, all of which helps improve knock limiting. However, GDI systems have poorer atomization and much poorer mixing of fuel, so modern GDI systems are starting to fail the ever-lower soot emissions limits (Designed for Diesels), since soot is formed at locally rich pockets in the cylinder and poor mixing causes rich and lean pockets. There's also a little bit of extra air which can go into the cylinder, because it is not displaced by evaporated fuel.

Usually the proper method to calibrate is to enrich/enlean and look for either best torque or best BSFC. It's possible that at O2 reading of stoich you will actually be close to LBT in the cylinder, which is the point of best torque,

_________________
"Sometimes, the elegant implementation is a function. Not a method. Not a class. Not a framework. Just a function." ~ John Carmack

"Any sufficiently advanced technology is indistinguishable from magic" ~Arthur C. Clarke

The important piece of this thread is optimizing the air cooling prior to the intake valve closing.....to stuff more air into the cylinder. With DI you know you are cooling the air but the intake is closed so while you may be able to run slightly higher CR you don't make a lot more power. With standard port injection more of the cooling is done on the valve and port walls which also doesn't help make HP. But with a short, well timed fuel pulse most of the latent heat to vaporize the fuel comes from the air and you gain .....5%ish on air density and therfore hp & torque.