From: Marc Thomas [mmthomas@ix.netcom.com] Sent: Monday, March 06, 2000 2:05 PM To: 928 Subject: [928] Bosch brains Folks, I hope that this help a little bit... ....We have installed, made harnesses and programmed sequential (motec) after market engine mgmt systems on 928 engines, made lots of power and have gotten them to pass calif smog! I have also bought, understood, installed, mapped, etc. other systems onto other engines. Are they better??...in general yes...but programming them to work for a daily driver takes about 200+ hours minimum. Programming them to race take about 20hours. And to race, you can get to the 90 percent level in about 8 hours. It is much harder to make a street car work correctly, due to all the factors that must be considered...cold start, hot start, part throttle decel, part throttle enrich, battery volt creep/duty cycle, canister dump, flappy, altitude start, warm up, smog...yes, we here in calif worry about smog...But a race car is easier. The stock bank to bank fire is a very good system....and it works well for almost all applications. Remember, The White Car generates between 425 + rear wheel hp and 470 + ft lbs (chassis dyno numbers) and uses a stock system with larger injectors and rrr. No chips, no fus..... Is this the limit, no, I expect the system to work into the 600 plus ground hp range due to the size of the MAF meter. There are other issues at that power level, and one is not the fuel/ign mgmt system. And the white car uses stock cams and passes smog. How does it work?...The 928 system works as follows.... When you start the engine, the maf takes readings to determine altitude, temp, etc. (mass air properties for the moment) and any changes from last stored data...it then provides a map offset factor to the fixed map. The fixed map is what is changed when a chip is burned. This is why the engine runs funky when you disconnect the battery..it must reset itself....take about 10 minutes. Every time you make a change to the engine, it takes about 10 minutes to reset..... The system runs on the maf meter up to some reading (approx. .xxx volts or so) or up to about 3000 rpm with full throttle, at which point, it switches over to map running and remains on the map till the rpms or maf reading drops below that above set point. O2 sensing/closed loop is done below some rpm...I have not found the point yet, but guess that is is around 3000 at full throttle or higher rpm at part throttle. The maf and O2 feed the ecu info to create the map offsets. The system uses all throttle positions (below the set point) to get the correct offsets for the map based upon many throttle positions/engine loads below the maf meter set point/rpm set point. The .xxx volts may be reached in many positions...full throttle at 2500 or 3000 rpm, or part throttle at 6000 rpm! With a bigger engine, the maf point is reached at a lower rpm (under full throttle), then with a stock engine. But that does not seem to affect the map, at least not on thw white car. For about the last year or so, our dyno guy has a pi a/f ratio meter that I just plug into the exhaust to get readings...and the curve looks very similar to a stock engine. The hardest part about working with a non tunable stock system is matching injector size with fuel pressure to get at a good a/f ratio through the range of rpms above the set point.... One question I am always asked is what really messes up the system?....cam timing....too much overlap raises hell...as a maf is unidirectional! I hope that this helps.... -- Marc M. Thomas DEVEK 650-592-5287 phone 650-610-0557 fax http://www.devek.net From: Rich [rich9928@earthlink.net] Sent: Saturday, November 04, 2000 8:14 AM To: 928 Subject: [928] LH fuel - EZK Spark Modules and Learning There has been quite a bit of Rennlist traffic about the Bosch LH-Jetronic fuel injection controller adaptive learning. There is no adaptive capability in the US '80 - '84's L-Jetronic injection systems. The US-spec 928s through '84 and Euros through '86 used the old vacuum/centripetal force-based spark advance. Some list members may not realize that for US models '85 and newer and Euros '87 and newer the 928 uses a completely separate spark controller, the EZK module, for spark control. The pair of controllers (ECUs) are what controls our engines. I don't have specific knowledge of our ECUs, I'll share what I know. The LH-Jetronic controller in our 928s is an early digital unit. I believe the first LH ECUs use an 8031 single chip microcontroller. The 8031 is the ROM-less variant of the 8051 that uses an external EPROM. The LH and EZK EPROMs are what you changed when "chiping" you car. [The original 8051/31 was designed in 1980. The 8031 has only 128 Bytes of Internal RAM]. I'm not aware of the CPU used in the EZK, it is likely also an 8031 or similar class of microcontroller. There may have been an ECU microcontroller upgrade when PSD was added and for the GTS when spark retardation was added for A/T shifting. The key issue back when the 928 ECUs were designed, no such thing as automotive grade EEPROM or Flash memory existed. There was no method to store data once the power was removed from the module when the battery was disconnected. Any "learned" data and diagnostic error codes are stored in the 128 Bytes of internal RAM. This RAM must also be used for all other "scratch pad" memory for calculations. The very limited amount of RAM means that there cannot be much learned data. Learning is limited to "finer tuning" of readings from sensors (non-linearity and trip points). Just to give a taste of what is in the current generation of Porsches, here we go: - Controller Area Network (CAN), modules and sensors "share" information and one module can issue commands to another [control and instruments]. - Fuel and Spark control integrated into one module, Tiptronic electronic Transmission control. All using CPUs with 1000s of times more processing power than our 928s - Variable boost, valve lift and timing is included with the Turbo's control - Traction control integrated with ABS, with electronic throttle actuator [the computer controls how wide the throttle is opened]. - Sophisticated stability control to monitor skid and add braking - limit power to limit spins The new BMW 7er series will have over 100 CAN connected electronic modules. Without CAN, over 2.5 km of cable would have to be used to make all the connections. Over 25% of the value of the car is now electronics. Rich '93 GTS '83S -----Original Message----- From: Richard Carter [mailto:drcarter@columbus.rr.com] Sent: Tuesday, March 12, 2002 4:21 PM To: 928 Subject: [928] Bosch secret code cracked (long) Car and Driver May 1985 has an article on Mike Valentine (radar detector guy) cracking the Bosch brain code. He did to get a Gray Market BMW to pass Federal Emissions testing. "From being interested in Porsches, I kinda knew how Bosch fuel injection works. Basically the fuel is supplied at constant pressure. For each cylinder there is a solenoid injector. How much fuel goes in is determined by how long an electrical pulse the injector gets. And that's the brain's job. But where does the brain get its information? You're not supposed to pry the top off the airflow meter. But I did it anyway. Inside there's a flap in the airstream pivoting on a shaft. It's held shut by a cuckoo-clock spring. When air comes in, it blows open the flap, kinda like a screen door with no latch. And the harder it blows, the harder the door goes out against the closing spring. On the shaft is a potentiometer. Looks like a volume control on a radio. As soon as any electronics buff sees that, it's 'Aha! Now I see where the signal comes from.' It's like an air-volume knob. The more air coming through, the louder it turns up the knob. That's easy. So I was poking around the engine compartment with my oscilloscope. Hook up an injector wire, then play with the throttle and see what happens to pulse width. The brain is responsible for holding idle speed steady. When I pushed the flap to lean it out, the rpm would drop and the brain would say, 'Hey, it's slowing down, we gotta richen this beauty back up,' and the pulse width would go up. You could push the flap around, and the idle speed wouldn't change at all. The brain would fight until you got completely outside its tolerance limits." Valentine concluded that the only way to deal with a Bosch brain was to think like Bosch brain. "Since the brain determines the mixture, I had to figure some fake-out, some way to send it bogus information so it would make EPA mixture instead of German-market mixture." (Or in our case max hp mixture.) "I found that the brain sends a constant reference signal to the potentiometer on the airflow meter. the meter processes the signal according to how much air is going into the engine and sends the manipulated signal back to the brain. Out the other side of the brain goes the signal to the injectors telling them how long to stay open. If you think about the problem, you realize that you want to add fuel in percents rather than increments. One increment at full throttle may not be very much, but the same increment at idle may drown it. So the input side is the right opening. If you fiddle with the reference going into the airflow meter, then everything coming out is proportional to the airflow.You want to find the leanest part of the map; that's probably going downhill with the throttle just cracked open. And the richest part; that should be cold starting. Then if you can add or subtract from the reference to get the mixture right in those two extremes, that's all the voltage you'll ever need. And this establishes the limits for the control circuit you're going to build. Then you work out the in-between points. If it's lean at part-throttle cruise, just how lean is it? I figured this out on the road, running on cruise control. I had a stack of batteries and a bunch of resistors on a board. And some wires into the passengers compartment from the fuel injection. Then with a meter hooked up to he oxygen sensor, I could tap around on the battery stack until I found a control voltage that would make it go rich.The only anomaly I ever found was 25% rich then stomping on full throttle. The signal back to the brain was higher than the reference, and it hadn't been taught to think about that. So it crashed. There's really not too much to the circuitry. There are some operational-amplifier chips and some little decision-making circuits called CMOS analog switches. You need a loop amplifier for stability. And, of course, a box and connectors and stuff. You get out your semiconductor catalogs and try to find things that don't cost $30 apiece. you can't quite buy all of this from Radio Shack, but it's close, just the barest level above. For the prototype, in fact, I got about 75% of what I needed from Radio Shack." Valentine has adapted his box to work on other BMW models, including the M635 and the Alpina B6, as well as the Porsche 911 with Motronic II, and even the Porsche 928 with the hot-wire airflow meter. ----- Original Message ----- From: Mike Valentine Sorry Rick, All that stuff is long gone. It's been 15 years since I've seen it. Cheers, Mike Valentine