Lambda Controller - PeachParts Mercedes-Benz Forum

duxthe1 · #1 12-01-2008, 02:12 AM

Over the past few months I've been working on designing, building, and programming a lambda adjustment module. It reads the O2 sensor voltage and gives a lambda percentage readout on 10 LEDs. It also outputs the O2 sensor voltage which is controlled by a potentiometer. In the center range of the pot it outputs an unmodified O2 signal. From center you can adjust the mixture either leaner or richer. The adjustment occurs in six steps as you turn further away from center. It has colored LED readouts for the adjustments....green for no adjustment, red for a rich adjustment, and blue for a lean adjustment.

The project is based of of a Microchip PIC18lf2525 microcontroller for which I've written a little less than 1800 lines of code for. Awaiting delivery of a few last minute parts, I'm planning to start the beta testing in my 90 300TE by next weekend.

300EVIL · #2 12-01-2008, 10:55 AM

Sounds interesting.... You'll obviously be using a wide-band O2 sensor, correct? A narrow-band won't do what you are proposing. Will your output be controlling the EHA? I'd also consider using a digital readout for better accuracy.

duxthe1 · #3 12-01-2008, 01:40 PM

No wide band, it's using the factory sensor. The range of adjustment is only 30% lambda either way. It doesn't affect the EHA directly. It only changes the O2 signal enough that the CIS-E will make the fuel adjustments automaticly. The management does have to be in closed loop before the controller can adjust it.

I didn't do a digital readout for 2 reasons. 1. The lambda signal is constantly moving so tracking it down to the milivolt would just keep a fluctuating readout. The 10 LED readout acts as a buffering interface. If the lambda is in the 50% range the 50% LED will be lit. I don't see the need read out the lambda within a closer tolerance. A 10% discrepancy in actual Lambda would just barely be within the range of being able to notice a difference from behind the wheel. 2. Keeping it simple. A digital readout would add to the complexity and after 1793 lines of code in Assembler, I'm not looking for more complicated.

Thanks for the input, though. All ideas are welcome. I hope to be able to develop it to the point of having a good product that will perform well in the field.

c280nz · #4 12-01-2008, 02:33 PM

im not sure how well this will work if at all,
the standard o2 sensors only read from about 14.2-15.2 and the whole time the standard computer bounces between these two figures in closed loop for fuel economy, i cant see any benifit from making it richer while at light load?
and wont it still just adjust within this range?

what im saying is can you explain a bit better what it will actually do? and how it will adjust to make the car richer or leaner, as the computer automatically adjusts to small differences, or is it to be basically just a guage.
possibly your car might have a different ecu with different fuel adjustments to myne and hence i shouldnt have posted anything?

duxthe1 · #5 12-01-2008, 08:26 PM

The idea behind the project is to be able to make adjsutments to the fuel mixture on the fly while in closed loop. With the lambda set to ~50% for normal unadjusted driving, the controller should be able to move the lambda up or down ~30% in 5% increments. So with a good 50% basic adjustment you could adjust the lambda to either a maximum of 80% or a minimum of 20%. I didn't see any situations where a normal engine (non forced induction)would want or be able to operate outside of those tolerances.

The controller adjusts the mixture richer by outputting an O2 sensor voltage that is leaner than it actually is, thus allowing the CIS-E to adjust it richer. The inverse is also true, if you want to adjust the mixture leaner then the controller outputs an O2 signal that is interpreted as rich allowing the CIS-E to adjust itself leaner. The CIS-E module tries to oscilate the mixture around 50%. The controller is designed to basicly allow you to select different values (other than 50%) around which it's adjustments are centered. Such that if I want to center the CIS-E around 40% then I decrease the O2 sensor's voltage by 10%. This lower O2 sensor voltage is interpreted as 10% lean, and the CIS-E will adjust it 10% richer. All the while the CIS-E thinks it's still centering around 50%

The concept behind the controller is to be able to dial in a bit more fuel at the stoplight when there is a BMW next to you, or to be able to dial the mixture leaner when you are on the highway at a nice cruise, and then be able to set it so there is no adjustment for general putzing around. All with just a turn of a knob.

300EVIL · #6 12-01-2008, 09:16 PM

Hmmm... I'm a bit skeptical about the whole idea but "maybe" it will work as long as the altered signal looks exactly like the NBO2 output and shows a slightly lean condition. It may throw a CEL though....

Here's your biggest problem.... The optimum AFR for making power is around 12.5:1 the stock NB02 sensor won't even come close to reading that.

In my opinion, the best way to do this would be to bypass the factory ECU and use a WBO2 to control the EHA directly. Optimally you'd also need load and RPM inputs but by then your basically looking at EFI. Theirs not many hidden patches to engine management other than reconfiguring everything. With what you are proposing, it sounds like a minimal power gain with the possibility of a CEL.

Take my post with a grain of salt however. I'd hate to get your hopes down and would like to see your results. I'd personally love to see you prove me wrong. Gonna need to see some dyno charts though!

Good Luck!
Adam

duxthe1 · #7 12-01-2008, 11:27 PM

Man, no sweat, I appreciate your input. The controller should very closely mimic an unaltered O2 sensor. The controller reads the O2 sensor digitally, decides to modify or not, and then outputs the new voltage through a MAX5354 DAC, once every repetition of the program. I havn't decided on maximum clock frequency yet, and have only simulated at ~125khz, but 8Mhz is theoreticly possible with the hardware I'm using. I'm going to determine clock frequncey once I can test it in operation.

I'm not too convinced that it'll need a wideband O2. Remember that I am getting a richer mixture by fooling the ECU leaner. Although I'll admit to not doing the math to figure out how much my 30% lambda ajustment will alter the mixture. FIgured it easier to measure it once I have it functional in the car. I'm lucky enough to have a gas analyzer at work.

I know as an experienced mb tech that a car running at 20% lambda is pig rich. I'd expect peak power to have fallen off somewhere before 20%, though maybe not.

Don't forget that the controller is designed to lean out the mixture too. This idea was hatched back when gas was still near 4$ a gal. After a 14hr@4000RPM road trip last summer when gas won't cheap, I was really into the idea of being able to run lean on the long hauls.

Granted there are a lot of shortcomings compared to a stand alone EFI. My design goals were to keep it a simple interface that requires no other mods than to plug it in to the cig lighter and O2 sensor. Technicly, you won't even have to open the hood to use it. There's all sorts of tweaker gizmos for the EFI guys but almost nothing for us CIS-E tweakers. What I'm shooting for is a simple, easy to use interface that gives the CIS user control of the fuel mixture over most of the usable range of a/f mixture during closed loop.

Whether I pull it off or not remains to be seen but so far, so good.

300EVIL · #8 12-02-2008, 12:19 AM

Quote:

Originally Posted by duxthe1

Man, no sweat, I appreciate your input. The controller should very closely mimic an unaltered O2 sensor. The controller reads the O2 sensor digitally, decides to modify or not, and then outputs the new voltage through a MAX5354 DAC, once every repetition of the program. I havn't decided on maximum clock frequency yet, and have only simulated at ~125khz, but 8Mhz is theoreticly possible with the hardware I'm using. I'm going to determine clock frequncey once I can test it in operation.

I'm not too convinced that it'll need a wideband O2. Remember that I am getting a richer mixture by fooling the ECU leaner. Although I'll admit to not doing the math to figure out how much my 30% lambda ajustment will alter the mixture. FIgured it easier to measure it once I have it functional in the car. I'm lucky enough to have a gas analyzer at work.

I know as an experienced mb tech that a car running at 20% lambda is pig rich. I'd expect peak power to have fallen off somewhere before 20%, though maybe not.

Don't forget that the controller is designed to lean out the mixture too. This idea was hatched back when gas was still near 4$ a gal. After a 14hr@4000RPM road trip last summer when gas won't cheap, I was really into the idea of being able to run lean on the long hauls.

Granted there are a lot of shortcomings compared to a stand alone EFI. My design goals were to keep it a simple interface that requires no other mods than to plug it in to the cig lighter and O2 sensor. Technicly, you won't even have to open the hood to use it. There's all sorts of tweaker gizmos for the EFI guys but almost nothing for us CIS-E tweakers. What I'm shooting for is a simple, easy to use interface that gives the CIS user control of the fuel mixture over most of the usable range of a/f mixture during closed loop.

Whether I pull it off or not remains to be seen but so far, so good.

I completely agree with your train of thought. It sounds like you did your homework and plugging in a piggyback controller between the O2 sensor and ECU would be a way simple mod to "trick" the ECU into performance mode.

However.... look at this. The factory ECU was designed to see a signal from the stock NBO2 sensor as is. If you have ever wired an LED meter to your stock O2 sensor (which I have....) you will notice the reading goes off the chart on a regular basis.

The only thing the ECU is looking for is what you can see on the LED display.

What you are proposing.... (for a legitimate performance gain) Will create a condition for the stock NBO2 to go "off the charts". If you leave it at this condition (what the ECU sees) for long enough the ECU will just look at the reading as being false and send the management into LHM and give you a CEL.

In my opinion, the factory ECU will see that the O2 sensor is "telling a lie" and ignore it. Never tried it myself but that's the "ECU logic" norm for what you are proposing.

300EVIL · #9 12-02-2008, 12:28 AM

If you'd like, I can "loan" you a WBO2 sensor with homemade controller (you have to put down a deposit though) to properly configure, test and tweak your system. It's one accurate mofo!

duxthe1 · #10 12-02-2008, 01:20 AM

I think the big thing ECUs validate is whether the O2 signal oscillates and the time from peak to peak. The controller will track the oscillations and mimic them. Part of the "beta testing" will be determining how far from 50% I can deviate the signal before the ECU calls shennanigans. The CIS cars can get away with a pretty poor tune and don't set C/E lights easily. Hopefully I can run it right to the edge of the chart and not "off the charts" so to speak. The neat thing about having the LEDs display the Lambda is that it will let me know when the O2 signal is called out for a fault with a fixed 50%.

I'll keep you guys posted with progress. Right now I'm waiting 5-10 days on a delivery of components, which is torture since I've spent the last several weeks getting the code written and debugged.

300EVIL · #11 12-02-2008, 01:38 AM

Quote:

Originally Posted by duxthe1

I think the big thing ECUs validate is whether the O2 signal oscillates and the time from peak to peak. The controller will track the oscillations and mimic them. Part of the "beta testing" will be determining how far from 50% I can deviate the signal before the ECU calls shennanigans. The CIS cars can get away with a pretty poor tune and don't set C/E lights easily. Hopefully I can run it right to the edge of the chart and not "off the charts" so to speak. The neat thing about having the LEDs display the Lambda is that it will let me know when the O2 signal is called out for a fault with a fixed 50%.

I'll keep you guys posted with progress. Right now I'm waiting 5-10 days on a delivery of components, which is torture since I've spent the last several weeks getting the code written and debugged.

Sounds pretty good to me! You've obviously done your research. I can't wait to see your results. If everything turns out okay; I am an expert in board development!

My prototype...

http://www.putfile.com/pic/8453430#

My final product....

Let me know if you need any help in this area....

Adam

duxthe1 · #12 12-02-2008, 01:36 PM

That's a nice clean organized looking board. If everything works as planned I am going to be looking into making PCBs. If that time comes, you're my man. Etching a board is something I don't have experience with. All of my projects to date have been on prototype boards. I've tried downloading a few free schematic editors that can design the PCB layout but havn't had much luck with any of them. There always seems to be a catch, like needing to define component footprints and such.

RBYCC · #13 12-02-2008, 03:14 PM

Quote:

Originally Posted by duxthe1

The concept behind the controller is to be able to dial in a bit more fuel at the stoplight when there is a BMW next to you, or to be able to dial the mixture leaner when you are on the highway at a nice cruise, and then be able to set it so there is no adjustment for general putzing around. All with just a turn of a knob.

Enrichment alone at idle and initial acceleration is not going to make your car any faster.

If you want to work on something for performance then why not look at a timing adjustment.
Fuel enrichment and timing on a NA engine go hand in hand...

And then there is the ECU.

Joreto · #14 12-02-2008, 03:40 PM

I don't think it'll work even for initial acceleration since opening the throttle passed a certain % the car goes into open loop and doesn't listen to the signal from the O2 sensor. The greatest benefit from this may be for fuel economy when cruising (leaning out the mixture) and maybe a bit more performance at part throttle (when enriched) but then you'll lose fuel economy.

JayRash · #15 12-02-2008, 04:00 PM

on my old NA setup i had 2 variable resistors which if i recall did alter the resistance of the temp sensor input to both ecus injection and ignition. Any way did dyno the car and ended with a 203 hp M103. Later on with a more flowing exhaust i got it up to about 210 but then i had to move the R16 euro ECE SWITCH to inside the car as that also changes timing advance / retard. Did that cause i had become so used to knowing to which position the R16 needs to be at depending on ambient temps. Winter / summer and sorts.