High dwell on 911 SC ‘81 lambda (US)

[mysocal911]

Originally Posted by AndrewCologne

The classic Retrofit is a superb solution and Im thinking of purchasing one as well.

Surely your Nano processor has enough memory left that you could include a digital ignition system in your present
Lambda controller, right? It's not that difficult, actually less complex than controlling the mixture.

[Sub8]

Originally Posted by AndrewCologne

Hi Sub8,

well, till now no production in series is planned. This is a private handmade prototype and it's tested since about 18 month. It's developed concept and coding routines are done by myself with assistance of an engineer of electrical engineering regarding the electrical assembly.

Further questions regarding costs or purchasing in this state via PM.

The classic Retrofit is a superb solution and Im thinking of purchasing one as well.
But regarding this ECU concept the thought of a "pseudo mapped" injection with still a K-Jet is not possible.

The solution above till now is totally lambda based, means till now no further hall sensors, no further mapping sensors or whatever need to be installed.
For optimal plug'n play compatibility without any further electronical modifications I kept the original wiring and sensor signals. Means the existing throttle switches i.E.
The only further swap which is needed is the little 15c switch on the right chain housing which will be changed to a 120°c sensor, enclosed with the ECU unit.

So the old ECU unit will simply changed with this one where the LCD-display – if connected – can be put into the ash tray for realtime reports while driving. Adjustments via software will be done via USB.



The optimal mixture will still be achieved by the frequency valve which is driven by this ECU as before. But as here a CPU is handling the whole job, its possible to optimize many engine running states like a smooth idle, direct and optimal lambda 0.85 enrichment at acceleration or WOT and ... the support of Bosch LSU based wideband sensor controllers on the market like Innovate, PLX, AEM is given. With that it could also be possible to achieve specific mixtures for idling, cruising and accelerating/WOT. I think thats your point.

It will work with the original unheated o2 sensor, the heated o2 sensor from the 3.2 Carrara and as mentioned above with wideband sensors/controllers.


While developing and testing I and two other testing people with the same 930/16 engine figured out many interesting behaviors of the K-Jet of these US SCs.

Yes that exactly what I meant!

Let the CIS give you lambda 1 and your box allows everything (well almost) that you would want a 3D mappable solution to do without any of the hassle.....

Sorry for thread hijack!

[AndrewCologne]

Originally Posted by mysocal911

Surely your Nano processor has enough memory left that you could include a digital ignition system in your present
Lambda controller, right? It's not that difficult, actually less complex than controlling the mixture.

Hi Dave,

thats what I've been thinking about also.
Reading/passing the (TD) trigger signal between the dizzy and the CDI.
But as till now I was totally in that process of coding/testing/coding etc. of the lambda core, this could be a forthcoming option.

[mysocal911]

Originally Posted by AndrewCologne

Hi Dave,

thats what I've been thinking about also.
Reading/passing the (TD) trigger signal between the dizzy and the CDI.
But as till now I was totally in that process of coding/testing/coding etc. of the lambda core, this could be a forthcoming option.

Using the Nano, my ignition design for a six cylinder BMW took about 60 lines of C code with Nano's interrupt input sensing
the distributor signal.

[AndrewCologne]

Originally Posted by Sub8

Yes that exactly what I meant!
Let the CIS give you lambda 1 and your box allows everything (well almost) that you would want a 3D mappable solution to do without any of the hassle.....

With a narrowband sensor you can only hit the lambda 1 ballpark as it only provides you the "spot on Lambda 1 window" signal within 0.1 and 0.9 Volt.
So with this sensor in regular the lambda 0,85, needed for Acceleration, can only be guessed but I found out how much more duty cycle is needed for a proper hit. The "static 65% duty cycle way" often results far off and for shure with a factory initial CO setting of 0.4 - 0.8% Vol. at 50% duty cycle.

Your wanted approach can be achieved with the wideband sensor, but also here for a functional "pseudo mapping" many more signals must be read like the rpm one from the speedometer/CDI, a manifold vacuum sensor signal etc etc. This is possible but makes the unit very user specific addressed, means a typical 911 DIY "make the best of it" user who is very in mechanics electrics etc. ... lets see :-)

I already temporarely used the rpm signal in the ECU for playing a bit with it – i.E. switching the FV off when deaccelerating – the whole SSIs changed into a nice popping beat box animal which sounded like hell – well at least for 911 sound freaks heroes

Sorry for thread hijack!

I think Funracer will forgive us :-)
As its still dealing with Dwell, Duty Cycle, ECU and more

[AndrewCologne]

Originally Posted by mysocal911

Using the Nano, my ignition design for a six cylinder BMW took about 60 lines of C code with Nano's interrupt input sensing
the distributor signal.

Yes, I follow you .... and could be the next approach :-)

[AndrewCologne]

@ Sub8
Here a video where the needed lambda 0.85 spot for accelerating is perfectly hit and kept when using a wideband sensor with this ECU.

[mysocal911]

Originally Posted by AndrewCologne

@ Sub8
Here a video where the needed lambda 0.85 spot for accelerating is perfectly hit and kept when using a wideband sensor with this ECU.

Great effort on your part. You should have no problem designing an ignition system. A simple design without ignition
maps is possible by using an equation to replicate a basic distributor advance curve.

[Funracer]

Originally Posted by mysocal911

Was this done? With that done and with the O2 sensor disconnected, the duty cycle should be close to 50%, if not the Lambda ECU is bad.
Again, the Lambda system is extremely simple to diagnose!

Car running, O2 disconnected, AEC unplugged DC did not go to 50% and did not change at all.

Checked connector outs 12 and 14 got OL on both. No ground.

Out of a bit of frustration I then unplugged the lambda ECU, car almost died, richened the mixture and set mixture by the “lift the air plate 1mm” method. Took it for its first drive in a month and it ran great.

So is my lambda ECU bad? If so what next?

Thanks

[mysocal911]

Originally Posted by Funracer

Car running, O2 disconnected, AEC unplugged DC did not go to 50% and did not change at all.

1. What is the duty cycle?
2. When running with the ECU connected and the back of the connector cover removed, measure the voltage
at pins 2, 7, 12, 14.

[Funracer]

Originally Posted by Bob Kontak

Another simple test is to check the unplugged O2 sensor side of that connection while the car is running. Let it warm up and it should read between 0.1V to 0.9V. 0.1v indicating lean exhaust. 0.9V indicating rich. You literally can get your idle AF mixture roughly close to 14.7:1 by adjusting the 3mm adjusting screw until it reads 0.5V.

Did this today. It showed .8V, too rich. As I turned the mixture CCW towards lean it went lower .7V and then dropped instantly to .3 MILLIVOLTS! I tried to reduce it to .5V many times but it did this every time. There was no reading between .7V and .3 millivolts. This is a brand new sensor I installed back in February.

I have 2 other O2 sensors that tested good today. I will put one in tomorrow and see if it works better.

[AndrewCologne]

Originally Posted by Funracer

Did this today. It showed .8V, too rich. As I turned the mixture CCW towards lean it went lower .7V and then dropped instantly to .3 MILLIVOLTS! I tried to reduce it to .5V many times but it did this every time. There was no reading between .7V and .3 millivolts. This is a brand new sensor I installed back in February.

I have 2 other O2 sensors that tested good today. I will put one in tomorrow and see if it works better.

Yes, its very difficult as you must hit the CO mixture that precise so it exactly results in the ballpark for a proper Lambda 1 reading, means 0.1-0.9v. Just a 1/16 turn already affects the mixture significantly seen with the eyes of an O2 sensor.

But I guess here your focus should be set on the ECU – just follow what mysocial911 suggested 2 posts before and let us know your readings.

[mysocal911]

Originally Posted by Funracer

Did this today. It showed .8V, too rich. As I turned the mixture CCW towards lean it went lower .7V and then dropped instantly to .3 MILLIVOLTS! I tried to reduce it to .5V many times but it did this every time. There was no reading between .7V and .3 millivolts. This is a brand new sensor I installed back in February.

I have 2 other O2 sensors that tested good today. I will put one in tomorrow and see if it works better.

Why are you now worried about the O2 sensor voltage? Focus on getting the DC to 50% without any inputs on pins 2,7,12,14
(pins 7,12,14 must be above 5 volts, pin 2 at about .50V).

[AndrewCologne]

Are you sure with 12 and 14, ... as these lead to the Acc. enrichment unit and AFAIK this includes Schmitt Trigger components, so for me this sounds like affecting the PWM signal where no constant voltage is present. Or am I on the wrong way?

Also 7 leads to the "15°C temp"/"35°Throttle" switches which simply lead to ground when cold or accelerating, which would cause a short circuit if voltage would be present on this wire on the side of the ECU.

[mysocal911]

Actual voltage measurements from key inputs pins of various Porsche 911 Lambda ECUs:

0 280 800 055
2 - .53, O2 input
6 - 12, no Lambda effect (NLE)
7 - 12, enriches at ground (EAG)
9 - 5, NLE
11 - 12, EAG
12 - 12, EAG

0 280 800 006
2 - .53, O2 input
6 - 12, NLE
7 - 12, EAG
9 - 5, NLE
12 - 12, EAG

0 280 800 037
2 - .53 O2 input
6 - 12, NLE
7 - 12, EAG
9 - 5, NLE
11 - 12, EAG
12 - 12, EAG

So for the Lambda ECU (055) in this thread with no O2 sensor and pins 7/11/12 open, a 50% DC should result. It's that simple!

[AndrewCologne]

... wrong thought, .... deleted.

[Funracer]

Originally Posted by mysocal911

Why are you now worried about the O2 sensor voltage? Focus on getting the DC to 50% without any inputs on pins 2,7,12,14
(pins 7,12,14 must be above 5 volts, pin 2 at about .50V).

Because a previous poster had suggested it and it was easy and quick to test as I had limited time yesterday. Given the results I got I am not sure the O2 is outputting correctly. I will change it to a known good one today and then do the pin voltage tests.

[Funracer]

Originally Posted by AndrewCologne

Yes, its very difficult as you must hit the CO mixture that precise so it exactly results in the ballpark for a proper Lambda 1 reading, means 0.1-0.9v. Just a 1/16 turn already affects the mixture significantly seen with the eyes of an O2 sensor.
.

I was turning it far less than 1/16. It went from .7 to less than .1 after barely touching it and did so every time I tried to lean it out.

[Funracer]

Originally Posted by mysocal911

1. What is the duty cycle?
2. When running with the ECU connected and the back of the connector cover removed, measure the voltage
at pins 2, 7, 12, 14.

Car running with Lambda ECU plugged in, AEC unplugged, O2 unplugged shows 80% plus minus a 5%


Car running, ECU connected, AEC unplugged:
Pin 2 is .5V
Pin 7 is 13V
Pin 12 is 13V
Pin 14 is less than 1V

With AEC plugged in:
Pin 14 less than 1V

[mysocal911]

Ringing from the frequency valve pulses can distort DC measurements with some meters, here DC is actually 50%,
but some meters could indicate 83%;



Closeup of ringing;



Proper DC (50%) using a test light;



Therefore, disconnect the FV and use a test light as the FV, alligator clip to +12V and other end to FV connection at pin 15,
and measure the DC.