After about a month of tuning I have a fueling map that seems close. Most of the map yields air fuel ratios of about 13.8. From what I have read this is about where the motor was designed to run and it runs very well.

However, I continue to have problems with overrun. Between 2600 and 3600 rpms under very light load the air fuel ratio is much richer. This is quite disappointing as I would hope to gain some economy here. The solution would seem to be closed loop with the oxygen sensor leaning out the air fuel mix. The problem is that the pulse widths are so short that O2 correction leads to large changes percentage wise in pulsewidth and fuel mix and the car bucks. I have one more idea to try. I am waiting on some new injectors since it feels as if one cylinder goes dry.

If that doesn't help, it appears to be run open loop and send fuel ou the exhaust or run closed loop and suffer some loss of smoothness.

I'd appreciate any advice,

Dave

I don’t have a clue if this will help you, it’s off the SDS site:

Closed Loop Feedback Control (CLOSE LP or CL)

CLOSED LOOP is in the OFF when shipped from the factory. CLOSED LOOP should be OFF if you are trying to program the system. When CLOSED LOOP is OFF then the SDS system does not need an oxygen sensor connected to the gray wire. When CLOSED LOOP is OFF the system runs from the values that are programmed.

CL1.JPG

To turn CLOSED LOOP ON push the +1 button. The gray wire must be connected to an oxygen sensor for proper closed loop operation. To turn CLOSED LOOP OFF push the +1 button.

When CLOSED LOOP is ON the EM-4 reads the oxygen sensor voltage and adjusts the injector pulses to attempt to hold the air/fuel ratio around stoichiometric (14.7 to 1 AFR for gasoline). Closed loop does not change any of the programmed values that you see in the programmer. Since there is a time delay time between sensing and correction, the air to fuel ratio will continuously fluctuate slightly to either side of stoichiometry. This condition can be seen when using a mixture meter for setup. Holding the mixture close to the stoichiometric range is essential for the lowest possible emissions when the engine is fitted with a catalytic converter. Closed loop operation is not possible with leaded fuels, since leaded fuel will damage the O2 sensor.


Many engines will not tolerate being idled at the relatively lean mixtures associated with closed loop operation nor will they safely tolerate these lean mixtures at full throttle, therefore the closed loop mode has rpm and manifold pressure limits, outside of which the EM-4 will switch out of closed loop into open loop (programmed values) mode. The EM-4 will also switch out of closed loop when the throttle is opened quickly. Closed loop will not be engaged by the EM-4 until the engine temperature exceeds approximately 35C (95F) and the sensor voltage first exceeds .625 volts. The rpm and MAP must also fall within the programmed limits.

CL2.JPG CL3.JPG

CL4.JPG CL5.JPG

The following limits are programmable: CL LO RPM LIMIT and CL HI RPM LIMIT, which determine where the EM-4 will discontinue closed loop control at. If you select 1500 and 4500 RPM respectively in these slots, closed loop operation will be discontinued whenever the rpm goes above or below these limits. The same thing applies for the manifold pressure limits, CL MAP LO and CL MAP HI. The engine must operate within these 4 limits or the EM-4 will revert back to open loop.


The O2 sensor will not supply reliable information to the EM-4 when cold (below 600 degrees F). A 3 wire heated sensor will give better results over an unheated one. Software limits prevent the EM-4 from adding or subtracting more than 25% to the primary pulse width in closed loop so it is important that the open loop fuel values are reasonably close to correct for proper functioning of the closed loop control.


Closed loop operation in most OE applications is generally limited to cruising conditions. Never program in limits corresponding to high power, high rpm conditions, since this will cause a loss in power and possible engine damage due to lean mixtures. A lean stumble is often apparent when the engine is unhappy about running in closed loop. Limits should be set to avoid these conditions. A mixture meter is highly recommended for system setup.


As stated before, many engines will not idle smoothly in closed loop mode. We recommend as a rough guideline, setting the CL LO RPM LIMIT no lower than 1500 rpm and the CL HI RPM LIMIT no higher than 70% of the redline rpm limit. Likewise, closed loop manifold pressure limits should preclude the low vacuum idle conditions as well as the high throttle ones. Set the CL MAP LO limit 3 to 5 inches above the normal idle MAP value and the HI limit around -5 inches for naturally aspirated engines and from -5 to 2 psi boost on turbo/super charged engines.


In CLOSED LOOP ON/OFF mode, ON/OFF is selected with the +1 and -1 buttons. In the other 4 CL modes, the +1,-1 buttons are used to select the HI/LO points which will be in the same graduations as your RPM and MAP ranges. Always leave the closed loop off if no O2 sensor is connected and always leave closed loop off until all normal programming is completed.

Dave,
I'm not familiar with the megasquirt, so what I say might not apply, but these are my thougths based on experiences tuning my MIC3.

1) your ve table is 3-d. So you can map different volumetric efficiency at different rpm/load points, right? It sounds like your ECU is programmed for to high of a VE in the 2600-3600rpm range at light load. Lowering the VE at that point in your map would lean out the mixture I would think, while maintaining the proper mixture at high load.

2) Most ECU's have a "minimum injection pulsewidth" paramater. Assuming the megasquirt does too, If your minimum pulsewidth is set too high the ECU won't be able to follow portions of the map because the "minimum pulsewidth" setting will prevent the pulsewidth from being shortened to the proper value. Maybe on overrun or a extremely light load your map is "bottoming out" on the minimum pulsewidth parameter.

Just brainstorming. I don't know much, but I'll help if I can.

-Ben M.

Mark and Ben

Thanks for the help

I do have the closed loop set to turn off at 1200 rpms and it is hard coded to turn off above 5000 rpms There is no adjustment governed by MAP I am aware of in Megasquirt.

The pulsewidths in the area I am having trouble with are 2.3 to 2.5 ms (1.0 ms of this is a constant applied to the whole map they call openning time.) 2.5 ms is too rich ( about 12 afr) and 2.3 ms is too lean (about 17 afr). I realize that as pulsewidths get shorter each ms chage in pilsewidth will cause a larger charge in afr, but this amount seems too large to me. That has led me to suspect a bad injector ( maybe it is openning very slowly) causing a cylinder to drop out. It seems it is worse when the car is warm.

I was having this same problem with the D-jet prior to converting so I doubt it is the ECU or software. I already replaced one injector and this helped some.

At any rate I'm getting a little frustrated. I want to do this right, I really believe in the megasquirt system, but I'm not sure I have the expertise to get it done.

Dave

Of course you do! Don't let this get you down! Believe me I know how you feel. I had one little issue with my ECU install that almost made me throw it all in the trash more than once. Figuring out the problem took 3 months. Fixing it took 10 minutes. I've been all smiles and miles ever since. Keep at it. You will get this all worked out and it will be worth it in the end.

Ok, back to the problem. Are you saying that you can get it either too lean or too rich and that right now it just happens to be set too rich? Or is it always rich in that range? 2.3-2.5 ms seems like a really small window for a large change in mixture. Have you considered using different injectors? A lower flow injector should give you more tuning resolution. The d-jet injectors seem to be way overkill IMHO.
I am using injectors from a Subaru 1.8 turbo. $5 each at the wrecking yard. They are working very well for me. The longer pulsewidths make tuning easier, and they are nowhere near max duty cycle.

-Ben M.

Ben

That silly little 0.2 ms pulsewidth change results in a change in an afr change of 4 or 5 parts air to fuel. In other words, tuning resolution just craps out. Thats why I am betting one injector may be quitting. The new injectors just arrived and I intend to change them for the ride home from work. Hopefully it will run better. I know my fueling map is close except for the area where it cruises on the highway.

I really want to avoid changing to smaller injectors as I would like to work out a program that everyone with a 2.0 L could use to get a running start in converting to Megasquirt. I doubt if I am alone in believing the D-jets days are numbred. For me, Megasquirt was the only choice because of cost.

I also recognize that the Bosch 280 150 019 is not a very good choice for this motor. I wonder what the people who designed this engine were thinking of. Was the belief at the time use the biggest injectors available, were their choices so limited that they had to settle for these injectors, didn't they know better. It sure is a pain in the butt. The engine designers stuggled with it too, why else would the ECU have a circuit to cut off fuel and why put in a decel valve except to lean out the mixture. I' m hoping that more modern brains can offer better contol and I can work around this problem. If not, I may have to choose between a compromise in the way the car runs or changing to smaller injectors. At any rate I committed to getting them as close as I can.

Dave

I'm not following along (MS denial)

What range injector did the Type3 VW's use in 1969/70 ?? It was a 1600CC engine with almost identical injection.

I fully understand your issue. I hope it is an injector gone bad. It sounds like it (or one that needs to be cleaned)


B

Turned out it was an injector. # 2 hole had a bent pintel. I replaced it and the car was much better coming home. I'll max the VE table and make some corrections to the air density table (too lean when warming up ). Then add in O2 correction and we'll see what happens. I think it will still be a little rich at very high vacuum (20 Kpa and below) but not too bad.

Sorry I ragged on the engine designers. They were smarter than I gave them credit for. Maybe I should have given them credit for injectors that worked well for 28 years. Guess I'll take my crow with a little humility.

Dave

I have absollutely no idea what you are talking about. open loop, closed loop, blah blah blah. But it really sounds cooool. I'm going to have to re-read it a couple of times so it sink in. I like that word... stoichiometric

Open loop means the fuel injection ECU does not use an oxygen sensor to correct the fuel mixture. Closed loop uses this correction. Most modern fuel injection systems use closed loop to correct to stoich ~ 14.7 parts air to fuel which burns all the fuel and yields few emissions. The catalytic converters require this to be very close.

I would like to run most of the time at ~ 13.8 afr since our air cooled engines need to run rich. Under acceleration I would like to see about 12.5 or so for max power. This replicates at least to some extent what the stock D-Jet does. The challenge is that this requires a system that allows the target air fuel mixture to vary. The firmware I am trying to use allows the mixture to vary with load (manifold vacuum) so I hope to have a 914 running closed loop with variable afr (more or less a modern D-jet) that can accept almost any engine change I would throw at it with little tuning.

Dave

Exact same injector as the 914- 1.7 d-jet