The VeeCU

[WesRagle]

Hi Guys,

A quick update on the VeeCU development.

For those that don't know, here is how I installed the injectors for development work. They are cheap Chinese injectors, intended for motorcycles I think, but they are sufficient for their purpose. In any event, pretty much a direct shot at the intake port.





One of the primary reasons I wanted fuel injection on my VW is to protect my rear cylinder exhaust valves. My rear cylinders always lean out under high power causing me to have to reduce power, risk burning my rear cylinder exhaust valves, or run so rich on takeoff that power is again down. So, fuel balance is top of my priorities for fuel injection.

As for the VeeCU, I want it to be as simple and inexpensive as possible. So ..., no fancy/intelligent control head. That poses some challenges as far as the user interface is concerned. But, it forces me to think up simple solutions. And simple is good.

My first attempt at fuel trimming was to use a rotary switch to select a cylinder and then use the UP/DOWN switch to add or subtract fuel for that cylinder. That was a disaster. With no display it was way too easy to get lost, dazed, and confused.

Then it fully dawned on me that I wasn't trying to "Trim" the fuel, I was trying to "Balance" the fuel. That led to my second, and I think final, run at the user interface.

The rotary switch is a four position switch with two positions left of center and two positions right of center. What I have for engine instrumentation is a Grand Rapids EIS. It's a six cylinder unit that is too old to be updated to a four cylinder unit but it's what I have so I use it. I have wired the EGTs such that the readings match the cylinders if you stand up in the cockpit and look down on the engine.

Anyway, here is how the user interface for Balancing the fuel distribution ended up.

If you are trying to balance fuel on the left side of the engine, place the rotary switch one click left of center. If the EGT is high on the rear cylinder, click the UP/DOWN switch down, and here is the important point, that will add 1% fuel to the rear cylinder and subtract 1% fuel from the front cylinder. If you click the UP/DOWN switch up it will add to the front cylinder and subtract from the rear cylinder. Of course the right side works the same.

I presently have it such that if you turn the rotary switch all the way left and click up, it will add 1% fuel to the left pair and subtract 1% from the right pair. And similar on the right. I find this feature of limited utility and may remove it entirely. If you have bad balance left to right it's probably an intake leak or a clogged injector and you need to fix that.

Balancing the fuel this way never changes fuel flow, it just walks the EGTs into balance at whatever fuel flow you currently have. The sum of the trim percentages used to balance the EGTs is always zero.

I was fooling around today with the new Balance method and took a couple of pics.

Just because I can :) You will notice the trim numbers sum to zero. Also notice that it didn't take much trim to match up the EGTs.



Until yesterday, I had never played with the leaning functions of the EIS. This pic shows the peak EGT spread lean of peak.



With the balance method described above it's easy to get the "Gami Trim" to near zero. Just lean until the first cylinder peaks, click on that cylinder to add fuel, and carry on in a similar manner until all cylinders have peaked.

The reason that it didn't take much trim to balance the EGTs in the previous pic is because the EGTs were balanced during engine calibration, at each calibration point, and the trim numbers saved as part of the calibration. In normal operation those trims numbers are interpolated keeping the EGTs balanced over the full RPM range. You might also notice the Duty Cycle difference on the last line of the display to see how much imbalance there was that is compensated for by the calibration data.

BTW, at low RPMs it takes as much as 25% to keep the front to rear EGTs balanced.

I've also implemented the Cold Start fueling. I finally found a method I liked and copied it. Here is a link to the method I copied.

Ref: https://www.youtube.com/watch?v=jTL1OlmY3kU

One more test performed today was to run the engine to 2500 (the highest I'm comfortable with in the shop) and check the minimum secondary ignition voltage at which the engine ran smoothly on secondary alone. This test was performed at mixture points as lean as the engine would run smoothly and as rich as the engine would run smoothly. For both cases the engine began to miss just below 5 volts on the ignition coils.

Forever Forward,

Wes

[BRS]

Thanks Wes for the update. Out of curiosity, why not use the selector, one position for each injector, then the toggle UP/DN for +-1%. This seems simpler. But I admit to be conditioned to the way SDS does it. Old habbits... I don't see any need to keep the total fuel flow the same between adjustments.

[WesRagle]

Out of curiosity, why not use the selector, one position for each injector, then the toggle UP/DN for +-1%. This seems simpler.

Hi Brock,

It would seem that way, but believe me, it's not. I've calibrated the engine several times. Consider my previous post where I said:

BTW, at low RPMs it takes as much as 25% to keep the front to rear EGTs balanced.

With direct port injection, using standard intake manifold logs (which have no independent intake runners), there is a lot of time and opportunity for the front cylinder to scavenge fuel from the rear cylinder. Using the balancing method I outlined it takes 25 clicks in calibration mode to make the EGTs meet in the middle. With the "one at a time method" it would take 50 clicks to make them match ... somewhere.

Calibration of the engine has two distinct and separate purposes.

1) Build the volumetric efficiency (VE) table. The VE table will be initially populated such that the engine will run, but not well.

2) Balance the cylinder to cylinder fuel distribution.

The calibration points are at a series of increasing RPMs. Each point is calibrated by first walking the EGTs into line and then leaning to "Lean Best Torque" followed by a press of the Store button.

Remember that the user will not have access to a VeeCU display. He only has EGTs and RPM as references.

Now pretend that you are calibrating the engine. You see that the EGTs are out of whack. What will be your first instinct? My first instinct was to add fuel to the lean cylinder until it matched the rich cylinder. Well, that throws the fuel flow out of whack. So the next move is to adjust the mixture with the mixture knob to lean the engine back down to Lean Best Torque. With that done you press the Store button. What you have just done is mix apples and oranges. EDIT: I should have noted that the new VE is calculated as the old VE * MIX Knob position. There is now a distinct discontinuity in the VE table. Sure, the calibration trim numbers will lie for the bad VE number, but now the VE for that RPM is no longer usable independent of the trim numbers. Also, if the system achieves it's ultimate potential there will be ground support software available to download the "tune". I don't want to have to explain over and over why there are "bumps" in the VE table.

Every number in the VeeCU has units (except VE which is has the unit of unitless :-) If the system ever makes it to production it will ship with a list of fully characterized fuel injectors embedded in the software. That way I don't have to use "magic numbers" where bigger is more and smaller is less. I even modified the Startup Fueling example I copied to change the primer pulse from mSec to micro-liters to keep the tune/calibration independent of the selected fuel injector. In short it makes development easier when there is a place for everything and everything is in its place.

From a user perspective consider this from the GAMI web site.



Notice the unit of "Spread" is GPH. In normal operating mode, what better way to achieve the perfect "Gami trim" than holding fuel flow constant (kinda) and walking the EGT to peak at that fuel flow?

Wes

[WesRagle]

Hey Brock,

I just re-read my previous post and realized it probably sounded a little weird. You raised a question and I went off on a tangent arguing with myself as to whether or not I had made the right decision. Sorry.

So, let me give the answer another shot.

Out of curiosity, why not use the selector, one position for each injector, then the toggle UP/DN for +-1%. This seems simpler.

After a short time working with the user interface it becomes quite natural. Just point to the hot cylinder and give it a click. The hot cylinder gets a little richer and the cold cylinder gets a little leaner. Balancing this way always pulls the mixture toward center.

Commentary: Keeping the mixture centered keeps the Volumetric Efficiency more relevant when it is used to calculate the fuel volume needed to produce the fuel mixture pulled from the fueling map. The VeeCU will have a conventional fueling grid with MAP on one axis and RPM on the other. Another decision I've argued with myself about :-)

Wes

[WesRagle]

Hi Guys,

I was cruising the internet this evening and came across this paper. It's funny how the more things change the more they stay the same. The price of the paper was 10 cents a copy when it was published :-)

Ref: https://ntrs.nasa.gov/api/citations/19930091255/downloads/19930091255.pdf

The paper does a good job of graphically depicting "Lean Best Torque". If you look at the graph at the lower left corner of page 10 (Fig. 14) you can see the abrupt drop in power at Lean Best Torque. That makes it easy to locate with the mixture knob using only RPM as a reference (if using a fixed pitch prop). That's why I use it for calibration. For our VWs LBT is achieved at an Equivalence Ratio very close to 1.1 (10% rich of peak). The graph also shows the large amount of fuel flow held in reserve at ER 1.1 for fuel cooling. Lots of extra cooling available for take off and climb.

Wes

[WesRagle]

Hi Guys,

I made a short video to demonstrate the VeeCU's tolerance to sensor and wiring faults. This is only the second video I've ever made, so be kind.

Ref: https://youtu.be/0HNxLckm0nQ

Thanks,

Wes