The throttle switch on CIS cars doesn't activate until WOT, so the car should remain in closed loop until it is activated. Part of the beta testing will be to determine if the full load throttle switch needs to be disconnected.

I agree that timing has a good bit to do with power production but at this point I'm not too concerned with it. Open hemispherical combustion chambers (like open chambered 103s) tend to not tolerate much spark advance so I don't think there is a lot of room for improvement. For now I'm just concerned with control over fuel, given that the ignition trim resistor give a small bit of control over the spark map.

Take it from me, this car will love advance, enriching wont give u much pwr especially if u consider how restricted the air flow is in those cars at low engine speeds. what the M103 lacks is pull low end, and advance will improve that. IMHO. I admire your efforts and i understand the drive behind tuning this system, good luck

Modifying the temp sensor signal works to some extent since unlike the lambda, the ECU doesn't ignore the input for the temp sensors at WOT. I'd guess what happened in your case is that you changed the resistance to report colder temp which in turn caused the ECU to advance the ignition and add a bit of fuel :)

On the topic of ignition advance, I found that my m103 engine is happiest with about 28-29 degs at WOT (road tune, haven't verified this on a dyno yet), which is not that much of an advance.

P.S. Jay, do you know exactly how much total ignition advance you run on the dyno when you got 203hp ?

I see one problem here and that is that if the WOT switch is disconnected than the ECU would think that it's at part throttle and will not enrich the mixture, you could go around this problem by enriching the mixture with the controller but that would mean that you'll be also running very rich a part throttle i.e. when cruising ... what you could do is maybe disconnect the WOT switch from the ECU and then have it activate the controller, it might work but it's not the best solution. Best way in my book is to find a way to modify the signal to the EHA, say monitor rpms, engine load (map sensor) and modify accordingly .

unfortunately no :S

but every car is some what specific, and mine was swis specs, slowest as they come (so i've heard) any way any 103 should be close to 200 hp with a free flow and a bit of advance.

but i have a few vids in my utube chan to show how sweet the engine was in N/A form (they are the early vids, new ones are with TT)

be careful dux not to do any damage running w.o.t while too lean.
rules of thumb=
Air/Fuel Ratio Limits.------------------- lambda
6.0:1 Rich run limit
9.0:1 Low power, black smoke------- .61
11.5:1 Rich best torque at WOT ------ .78
12.5:1 Safe best power at WOT ----- .85
13.2:1 Lean best torque at WOT --- - .89
14.7:1 Chemically ideal ----------------- 1
15.5:1 Lean light load, part throttle ---- 1.05
16.2:1 Best economy, part throttle ---- - 1.1
18-22:1 Lean run limit

i dont think u will be able to get out side the 14.2-15.2 range because this is the range in which a standard narrow band o2 sensor reads, which wont make bugger all difference,
and i think if you try and trick the computer it will just vary massively while "searching" for its middle point of 14.7
i have a piggyback hooked to my ecu and a wideband o2, and i know that when the ecu cant equalise easily enough into the narrow range it just starts varying massivly and goes real rich then real lean looking for the medium,
so if u set the new medium at say 20% richer it at say .4v it isnt going to read down to -.1v so you still only have the same scale,
in closed loop my car will bounce continuously between 14.2-15.2 the whole time when parked at idle, and accelerating at low throttle positions,
it doesnt settle at say 14.7 like youd think, if it did then yes i think u might be able to adjust it, but its a constant yoyo.
but using ur technique in my mind i think u will be only able to adjust within the range that the standard o2 can read,
are u proposing that ul get it to bounce within a smaller range instead, say 14.2-14.7? for example?

^that was all in my head and i wouldnt of been happy without saying it because i wasnt sure if u got what i ment last time,

feuw what a story.

I'm not surprised that you made best power below 30 degs timing. The open chamber just doesn't support much timing. Some closed chamber engines are happy at 40 degs.

I'm not worried about disconnecting the throttle switch and cruising rich b/c half of the function of the controller is to be able to dial fuel out as necessary. Given that my car has a catalyst and has to pass annual emissions testing, running it really rich will only happen in short bursts. Running 20% for very long will likely melt down the catalyst.

I've been an M/B tech for 12 years and worked on a lot of CIS cars. I've had 100k engines come in that have never had a lambda adjustment and were running at 80% with no C/E light. I've also seem em come in running less than 30%. Trust me, from behind the wheel you can tell the difference.

I don't doubt that my controller will leave some power on the table. I'm not shooting for the be all, end all of CIS tuning. Maybe some day, but for now I'm just getting proficient at programming PIC microchips in Assembler. For now, a simple interface that gives a fair adjustment over the normal range will be what I consider a success

Delayed again
 

:mad: Well after waiting a week there is still delays. For some reason that I havn't figured out the new PICs I'm using are write protecting after only one write. So basicly I waited a week to change the code 3 times before I ran out of PICs. I don't have the PICs configured to write protect or code protect so I'm pretty stumped. I programmed the old PIC dozens of times and didn't change anything except a few instructions in the code and the actual chips themselves. The lesson... don't get careless and short out your working PIC. :o Back to the datasheet I guess.

Socket one of your PIC's in upside down and power up your board for a second. That'll generally knock out all the values. Or fry it. Trust me, done it hundreds of times with success.

You can also try a different programmer. I eventually broke down and bought a $600 BK Precision programmer. I never have any trouble with that sucker.
Good Luck!
Adam

Well it was a programmer problem, but it wasn't a big deal. I had a marginal ground which would pass enough current to read and to write but not enough to erase, which of course has to happen before before a write, but not an overwrite of the same data. A little touch up with the soldering iron got it straightened out. As luck would have it, I had bought this programmer just for this project. It turns out that my old programmer didn't support the 18LF2525 device. Needless to say, this isn't a 600$ programmer.:rolleyes:

With my programmer issues sorted out, I got the code in the chip and functional on the board this evening. I'll start the beta testing over the next couple of days and let you guys know of my successes and failures.

Keep your fingers crossed for me:D

I did my first testing today:D I logged about 20 miles without a C/E light. Most of that driving was with the controller set for no adjustment, outputting the same O2 voltage as was input. I did most of the driving with no adjustment making sure that my circuitry would be able to mimic the O2 signal sufficiently to fool the CIS. So far, the CIS module is interpreting my output fine. I put the Lambda tester on it at the shop and it had good oscillation. After work I pulled it in and put it on the gas analyzer. Turns out it worked better than I though it would. I recorded my CO and HC values at idle and at 2000 RPM. Unadjusted my idle CO was ~.40 - .70% and at 2000 my CO was ~.55 - .85%. The Lambda controller would lean the idle to .12% and lean 2000 to .16% CO. Not bad, I must say, but it turns out it will richen much better than it leans. At full rich adjustment idle would go over 7.0% CO and 2000 would go to ~6.6% CO. I really didn't expect it to richen that well but my ASSumption is that the CIS is willing to work harder to prevent a severe lean condition than a rich one. Remember that I get an adjustment by fooling the CIS lean to get it richer and vice versa.

I did get one good stoplight to try an enriched acceleration. I'm not going to say that it was night and day but I did notice a bit better pep. :) I am not confident saying that there is more peak power but it felt like the power band had gotten wider. Typically my TE doesn't get any steam behind it until the revs are up quite a bit. Enriched, it felt like it was willing to pull with gusto sooner and longer.

Having the O2 sensor voltage display is neat to watch but also distracting in traffic. I was surprised how much time is spent in a decel fuel cut. If there is one thing CIS does well, it is cut and apply fuel without hiccups. Maybe that smoothness is why Mercedes stuck with it well after most manufacturers switched to EFI.

I did all of my driving today with various amounts of negative enrichment. At the higher settings the power loss is evident. I didn't notice any pinging but it is pretty cold here and the air is thin. Pinging isn't normally a problem for my TE, even in the summer with 85 octane, but I do see 103s that will ping on cheaper fuel once in a while.

With another 20 miles under it's belt, still no C/E light:)

Today was a mix of driving with fuel being added and subtracted. No real problems to report. There is a bit of a delay from setting the controller to add fuel until the CIS actually adjusts it. Since the controller affects the CIS's feedback system, it has to see the adjusted values for a bit before the mixture actually starts to change. I noticed this on the gas analyzer and now during actual driving. The dealy seems to be around 10 to 20 secs. I think that when the O2 sensor is really good and hot the delay is less. At stoplights I adjust when the crosswalk signal changes and that seems to work well. I have gotten a couple of good enriched accelaration runs today. It definately likes the extra fuel. I have a fair hill near the house that has a 30mph limit. At 30 mph, I normaly am just on the border line of being able to climb it in high gear without a down shift. When leaned out a downshift is certain, when enriched it pulls the hill in high gear with ease. Not emperical data, I know, but proof of concept none the less.:P

I've been talking with my co-worker about trying it in his stroked 500SL. That puppy really likes to run rich but he keeps it tuned pretty lean most of the time.:rolleyes:

I tanked up today so that has been one tank full through the car with the controller providing the O2 signal. That's a bit over 200 miles and so far so good, no problems or c/e light. It cruises quite well when leaned out. Most of the driving so far has been spent with it leaned out. Why burn more gas than I have to right. A couple of times I've dialed in more fuel but that has only been in short bursts. When leaned out I just back off on the fuel until I feel a slight power loss and leave it that way. I havn't pulled the full load contact and as of yet don't think I need to. Since most of my driving is with it leaned out I like having the full load contact there to bring it out of closed loop when I need the power back for merging and such. I may still experiment with removing the full load contact but don't see it being critical to the function of the controller.