alternate idea for accel enrichment

[BMW528e]

After I started using warmup enrichment I noticed a loss of some of the MPG gain I had made when I first converted to MS. Apparently, even while the car was running like crap and could barely get out of the driveway without stalling, it used less fuel overall. This led me to believe that the main issue with cold running is actually the whole fuel-condensation-in-the-port issue which is explained elsewhere in the forum. In the interest of efficiency and throttle response, I came up with some new code to address this. I just got it running in my car today and it seems to be doing the trick.

The idea is this: When MAP rises, and stays above a baseline value for x number of consecutive ignition cycles, accel activates and extra fuel is injected with an amount proportional to the increase in MAP. The additional pulsewidth is based on microseconds per 1 kPa rise, plus another value for cold temperatures (decreases linearly from -40c to 70c just like the cold pulsewidth from the normal TPS accel code)
When MAP falls, the same thing happens in reverse, with fuel being subtracted.
Here is a link to the modified source if anyone is interested... (based on HR10g!) http://www.hyakushiki.net/motronic/tjsmod.7z
NOTE: I did not go through and change all the staged injection code to accomodate the new pulsewidth modifier variable so that isn`t going to work quite right.
Also, I changed the spark advance table to use 50RPM resolution instead of 100RPM.

[ihateforums]

I've made further modifications to this code for my use on my 633CSi. Updated version here
A group of settings for "spark angle fudge factors" has been added. The purpose being to correct the timing inaccuracy that results from using the flywheel VR sensors (116 teeth) to run a 6 cylinder engine.

I've increased the magnitude of the MAP-based enrichment feature described in the above post, as this engine seems to be very sensitive to changes in MAP/throttle. Someday I would like to try wiring my injectors in batches of 1,3,5 and 2,4,6 to run semi-sequential. In the meantime, I tried increasing the sample rate for the MAP ADC, and moving the KPA code to later in the program to minimize any lag between reading the MAP sensor and recalculating the pulsewidth. Honestly, it didn't seem to have much effect though.

I made two other changes. First, the ADC routine had 5 or 6 variables that were consecutive in RAM, but the remaining ones were elsewhere and had to be handled as a special case in the interrupt code. I changed it so all 8 are consecutive and shortened the interrupt routine. Second and more importantly, it was adding the previous value to the new value and dividing by two. So if the reading changed from 40 to 41, the result of this "averaging" would keep it at 40, basically throwing away 1 bit of precision. I changed it so when the value is rounded, it will be in the direction of the NEW value.

[PSIG]

It's nice to see more tweaking of MS1, as there are still a lot of users (old and new) out there. Has the staged portion of the code also been updated so the package is as functional as HR10g originally was? I'm thinking this code mod would be useful for a staged 8-injector 4V TBI project I'm considering. No reason to retire an MS1 that's been running perfectly for over 10 years, right?

David

[jsmcortina]

No reason to retire an MS1 that's been running perfectly for over 10 years, right?
David

I retired my MS1 a number of years ago. Why use old technology when we've spent hundreds (thousands) of hours making MS2 and MS3 better. The processor for MS1 is an obsolete part now too.

James

[PSIG]

No reason to retire an MS1 that's been running perfectly for over 10 years, right?
David

I retired my MS1 a number of years ago. Why use old technology when we've spent hundreds (thousands) of hours making MS2 and MS3 better. The processor for MS1 is an obsolete part now too.

James

I understand your point completely, James. Without straying from the thread subject; what I am saying is that I have a perfectly good ECM, already paid for, and fully capable of what I am wanting it to do. Why trash-can it and spend more to get results that are not necessary for this project? Additionally, the MS2 and 3 are an unnecessary increase in expense for many projects, such as the 32' inboard boat carb-conversion I am considering. While the conversion is currently under $500 for many of these conversions, I can't keep it under that magic number with the more expensive brains that provide no equivalent value of increase in performances. I hope the availability of the less expensive but still useful MS1 does not end anytime soon, at least without an equivalent low-cost replacement, or MS will lose much of it's fan base in-favor of the least-expensive replacement that will emerge to fill the vacuum.

Much of the success of MS over the years has been due to the trade of study and labor against system cost, allowing many to try EFI that was otherwise unfordable, or enticing them to participate for learning because it was affordable. With the public interest in EFI and overall rising fuel prices - that's a huge segment! How do you see the future of EFI systems that are not for emissions-controlled vehicles, and also not to replace existing ECMs?

David

[racingmini_mtl]

Ignoring all the features and code aspects, the main problem with the MS1 chip is that it is discontinued by the manufacturer. Once the stocks available at different distributors run out, that's the end. It would be possible to make the equivalent of an MS2 card with the SMD version of the MS1 chip or port the code to a new chip but that costs money and you would likely end up in MS2 price territory without the advantages.

Jean

[seishuku]

Didn't James already make a post saying the code will not be licensed for any other 908 chips other than the one currently in use?

[racingmini_mtl]

Didn't James already make a post saying the code will not be licensed for any other 908 chips other than the one currently in use?

Yes. I was just talking about a hypothetical possibility that doesn't make sense anyway due to cost.

Jean

[ihateforums]

It's nice to see more tweaking of MS1, as there are still a lot of users (old and new) out there. Has the staged portion of the code also been updated so the package is as functional as HR10g originally was?

I haven't taken the time understand how staged injection is supposed to work so I haven't done anything with it. But, the problem with that was related to the new accel scheme. Other changes like the ADC routine could be put into the original code while leaving all features intact.

From what I've seen in the MS2 docs it would not be possible to run my 6 cylinder with a 116-tooth wheel. And while I have some familiarity with 8-bit assembly language from doing NEC TurboGrafx development, making modifications to MS2 code would be beyond my ability.

It's too bad about the 908 chips running out. If a port of the MS1 code to the CPU12 architecture was allowed that would be interesting.

[dontz125]

From what I've seen in the MS2 docs it would not be possible to run my 6 cylinder with a 116-tooth wheel.

I've just taken a moment to refresh my memory on how the MS1 handles a toothed wheel, and I don't see how it handles your 6-cyl with a 116-tooth wheel. Is it 116 teeth, with none missing? 120-4? 117-1?

[racingmini_mtl]

For a 6-cylinder engine, you still need a wheel with a number of teeth that is a multiple of 3 even on MS1 (and MS2 and MS3). And if you have 116+1 as I think, that won't work. Moreover, I seriously doubt that the MS1 processor would be able to handle the number of interrupts that a 116-tooth wheel would generate (but I may be wrong because I don't know how the code is written; I also don't think it supports anything over 60 teeth).

Jean

[jsmcortina]

If a port of the MS1 code to the CPU12 architecture was allowed that would be interesting.

I'm sorry, but why on earth would we want to port 10 year old code to the current architecture when we've worked had to create a far superioir product on S12???

MS1 has served its time and is ready for retirement.

James

[ihateforums]

For a 6-cylinder engine, you still need a wheel with a number of teeth that is a multiple of 3 even on MS1 (and MS2 and MS3). And if you have 116+1 as I think, that won't work. Moreover, I seriously doubt that the MS1 processor would be able to handle the number of interrupts that a 116-tooth wheel would generate (but I may be wrong because I don't know how the code is written; I also don't think it supports anything over 60 teeth).

It does work, I've dynoed the engine up to 6000RPM with no problems. It's 116 teeth plus a separate reference pin (dual VR sensors).

I'm sorry, but why on earth would we want to port 10 year old code to the current architecture when we've worked had to create a far superioir product on S12???

Well, I didn't say that you would want to port it. And I'm not knocking the newer products (I have no experience with them). Personally, I'm still interested in MS1 is all. 10 years old or not. My car is over 30 years old and I still find it interesting as well.