Thrasher Tech from 1996

[TexasT]

 TEP - Thrasher Engineered Performance
ECM Secrets & Tuning Hints

Through the years, we've consistently been asked the following questions in one form or another. We thought we'd address some common questions or misconceptions on this page. This information is specifically geared toward the SFI-M ECM used in the Buick Turbo Regals 1984-1987; other GM engine computers have similar characteristic s but may not adhere strictly to these guidelines.

 
1.  Does it take several minutes for the computer (ECM) to clear it's memory and "unlearn" everything?

No. Unplugging power to the ECM (The orange wire with connector behind the battery) immediately causes the ECM to lose it's memory - no need to wait any time at all before reconnecting it.

2.   Does it take several hundred miles of driving to "learn in" the computer?  Does the computer learn everything about the way I drive?

No. The ECM contains two things responsible for learning - a short term memory called the Integrator (INT) or Closed Loop Correction term (CLC), and a long term memory called the Block Learn Memory (BLM). The only thing that is stored and memorized is the BLM. There are 16 BLM cells numbered 0-15, separated into a grid by RPM and Mass Air Flow (MAF). As you rev the engine you "travel" through the BLM grid and hit various cells. 0 is the idle cell and 15 is the WOT cell. The BLM cell number is not a part of the serial data stream (in the SFI-M Turbo Buick ECMs) so scan tools are not able to indicate which BLM cell you’re in. They will however, give the value of the current BLM cell.

The O2 sensor will detect a rich/lean condition, and cause the INT to change. The BLM is simply a long term average of the INT. 128 is a normal number, signaling no change. A number greater than (>) 128 reflects a lean condition, and a number less than (<) 128 reflects a rich condition. Put another way, a number greater than 128 indicates the ECM is adding more fuel (going, or learning richer) to compensate for a lean condition, and a number less than 128 indicates the ECM is taking away fuel (going, or learning leaner) to compensate for a rich condition. The more we deviate from 128, the more correction the ECM is learning, and hence the worse the base cal is for that given condition.

Only the BLM is saved and memorized. The INT resets on start-up and is constantly changing. By clearing the ECM memory the BLMs are reset to 128. BLMs are the ONLY thing the ECM memorizes and are totally related to how the O2 sensor detects the fueling to be, thus the ECM does not learn or detect things such as driving habits. The O2 sensor considers a stochiometric air-fuel ratio (AF) of 14.7:1 to be optimum.

3. What are "good" BLM numbers to see? How far can the ECM compensate?

Generally, if the base open loop cal is correct for the size injectors used and compensated accordingly for the flow characteristic s, BLM should not deviate more than 10 points on either side of 128. This means you shouldn’t see numbers much beyond a range of 118-138, and the majority of cells should range more within 121-135. Remember that fuel pressure (FP) GREATLY affects the BLM values, and a deviation of 1 psi can throw BLMs off by a few points! If you see the majority of BLM numbers skewed in one direction, a slight tweak in FP may be warranted. Weather conditions affect BLMs drastically too, and this is precisely why they exist, and why closed loop operation with the ability to learn is preferable for part throttle driveability.

The factory calibration limited BLMs from 110 to 150, which is more than enough to compensate from any aging effects or weather conditions. In fact, it is due to this wide learning capability that some folks can passably run blue top injectors on a chip made for stock injectors, as the BLMs simply compensate for the incorrect base cal. The BLMs have limits because if a sensor malfunctioned, forcing BLMs to incorrectly skew one way or another, you’ll want to limit their authority somewhat so that the engine would still run somewhat decently. Most aftermarket manufacturers have widened this authority to 90-160. We can only guess why they’ve done this - it makes for a more forgiving calibration, as you can have the base calibration deviate from the optimum by a much larger amount and the ECM will still be able to compensate.

4. I’ve been told that I should drive fast and accelerate hard to learn in the computer.

No. By always accelerating hard you’ll never learn in the computer. This is because you’ll be in Power Enrichment (PE) mode. Once you’re in PE, Learn-mode is turned off. About 10 minutes of normal, easy driving in stop & go traffic is sufficient to learn in the majority of the BLM cells.

5. What is Power Enrichment (PE) mode? How do I know I’m in PE mode?

PE mode is where additional fueling is given to run a richer AF mixture than stochiometric. The system remains in a "Closed Loop" mode but ignores the O2 sensor, essentially running "open loop" under a "Closed Loop" mode for lack of a better term. On a scan tool, the "Closed Loop Mode" indicator will still indicate closed loop, however, the INT will be reset to 128 and Learn-mode will be off. The BLM value displayed will be the value used in the fueling calculations, and will usually, but not always, reflect BLM cell 15 (depending on RPM and MAF). It is the last value learned in before entering PE.

6. What does it take to enter PE mode? Do I need to set the Throttle Position Sensor (TPS) to read 4.8v at WOT?  I've been told from what I believe to be reliable sources that it took a MAF reading of 255 and TPS of 4.8 volts or so to enable PE mode.

That is an old wive's tale that's been propagated through the years and has unfortunately been taken as gospel.  We believe we know how that rumour got started, but we won't go into that now!  PE mode is actually entered quite easily. A certain throttle position is required based on RPM and a certain load must be reached, or given enough TPS, PE will enable regardless of RPM or load. Basically, about 20% TPS is needed to enter PE under most conditions.

The subject of max TPS reading is a sticky one, as there are folks that are convinced that they have seen an increase in performance by adjusting it higher. While we do not refute this, and can come up with some possible explanations as to why, these are the facts from the ECM software point of view:

For the 86-87 controller, there is no need to exceed 80% TPS in the stock chip. All PE fueling multipliers are identical after 75% and above, and for those of you racing with the AC on, the AC will shut off after 80% TPS. There is *nothing* in the cal that does anything beyond 80% TPS. For EFI Performance chips, there is no need to exceed 75% TPS.

For the 84-85 controller, there may be reason to go as high as 87.5%TPS as the PE fueling leans out slightly at that level in the stock factory chip. Again for our chips, 75% is fine.

Note that we are quoting values from the stock chip or ours. We have yet to see any other manufacturers raise these TPS boundaries, but it’s possible some may have, although we have no idea why anyone would. Note also these are percentage based, and we assume a 5v reference. If by some chance this reference was higher on a particular car, then the TPS voltage would have to rise too. We suspect this to be the reason some folks have successfully raised TPS voltage to achieve gains, if these gains could definitely be attributed to the increase in TPS alone.

7. OK, what is Closed Loop, Open Loop, and Learn mode?

On a cold start, the engine starts out in Open Loop. This means that there is no feedback mechanism (from the O2 sensor) in place, and the ECM is not correcting or learning to try to achieve a stochiometric AF ratio. It is simply running off some preset tables in the chip. A carburated engine, or any engine without an O2 sensor, operates in open loop mode.

On a cold start, we have to run quite rich as a cold engine will not vaporize fuel very well so additional fuel is thrown in to offset the small percentage that vaporizes and does work - ultimately, this extra fuel is wasted. This open loop cold start fueling is responsible for how well the engine operates cold. A correct calibration for the injector size and characteristic s is important to achieve good seamless cold starts with a minimum of stumbles and hiccups. Factory engineers have an additional responsibility of meeting cold start emissions, which at times will hinder good driveability.

Closed loop operation occurs when the O2 sensor is used for feedback to adjust the AF ratio. Under most circumstances, when a preset coolant temp is reached and the O2 sensor heats up sufficiently to provide good signal, the ECM goes into closed loop mode, and the INT will begin to function. There is an additional coolant temp criteria that must be reached for Learn-mode to enable. This is when the BLMs begin to learn. Beta-tester Carl Ijames stated it particularly well - using a scan tool, "the Closed Loop flag is indicating that closed loop operation is now permitted but not necessarily in effect." Circumstances where closed loop is permitted but not in effect include PE mode and in the case of our chips, lean-cruise mode.

8. What good is the BLM during Open Loop, or PE mode? Are they being used or ignored? Learn mode is off under those conditions isn’t it?

Yes, Learn-mode is off during PE or Open Loop, but this only means that the ECM is not currently learning under existing conditions. However, whatever was previously stored in the BLM cell whenever Learn-mode was enabled is still being used! For example, if the BLM cell at 1500 RPM and 15 g/sec airflow learned to 130 under closed loop conditions, then 130 is still used to modify the fueling at that RPM/load point even under open loop or PE conditions. In other words, the stored values of the BLM cells are always in effect and will always affect fueling.

9.  But I’ve seen improvements / steady losses just by running down the track a few times with no other changes. Isn’t the chip "learning in"? How can it be learning if I’m at WOT in PE mode?

Maybe. Remember the ONLY thing learning in the chip is the BLM, and at WOT you’re only concerned with BLM 15. And that’s correct, the BLM is not changing during PE mode. However, with the stock BLM boundaries, BLM 15 is being learned at roughly 30-45 MPH with slightly heavy throttle, just before PE mode is entered. This is roughly the conditions seen while returning down the access road from the end of the track. Note we say stock BLM boundaries as it is possible to alter these, although we have not seen any other manufacturer do so.

What we feel is happening is that BLM 15 is being learned each time on the return road after the run! In fact we’ve seen this happen on our development cars. Remember also that throughout the entire run, whatever has been learned into BLM 15 is being used to affect fueling. This means that you can never count on having consistent fueling from run to run as BLM 15 has potential to learn on the return trip.

*** Warning - Blatant Advertising Ahead !!! ***

10. Is there a solution to this problem of inconsistent fueling from run to run? Also, why is it sometimes I tweak fuel pressure and get no noticeable changes?

Of course we have a solution! (Actually we have two solutions, but the second is much more expensive and has some additional drawbacks.) The problem of varying fuel pressure with no effect has been quite thoroughly investigated by Carl Ijames, one of our beta testers. Basically, by increasing fuel pressure to get more fuel at WOT, while cruising in normal closed loop learn mode, the BLM will adjust to this increased fuel pressure and lean out, thus effective negating your changes in fuel pressure (FP). As a result, you would need to keep increasing FP until the BLM hit it’s limit and stopped adjusting, and by then you’ve drastically affected part throttle driveability.

We have solved this problem by rewriting the code to reset the BLM to 128 in PE. Your BLM cell 15 used in WOT was actually learned under a non-WOT condition, not representative of WOT at all. Always keeping the WOT BLM neutral eliminates that variable, resulting in consistent fueling run after run. The only other variable affecting WOT fueling is MAF. Now you’re able to totally control WOT fueling with the fuel pressure without having the BLM potentially negate your changes.

Currently the only other available option is an aftermarket computer. They have a rather limited learning ability that has little affect on fueling, and will not cause the aforementioned problems. But they cost much more and are much more difficult to use. Actually there is a third solution, and that is to clear the ECM memory immediately before each run, thus resetting the BLMs to 128, and we know knowledgeable racers who have resorted to doing this with conventional chips.

11. Why did the stock factory calibration have this obvious flaw?

It’s unfair to the original engineers to consider this a flaw. No factory calibration is ever designed with racing in mind, and therefore it is not optimized for this sort of application. The factory cal was designed and optimized for the lower fuel pressure of the Bosch 233 regulator, lower levels of boost, and to still meet the stringent EPA emission standards for manufacturers. Under those conditions it performs extremely well.

On this subject, we should point out that EPA emission standards for vehicle manufacturers are much more stringent than even the California state emission inspections that we as owners must pass, which is why even some heavily modified vehicles do not have trouble passing as long as they have a cat - vehicle manufacturers could never get by so easily. Vehicle manufacturers must contend with federal drive-by noise level regulations too, which result in the more restrictive and quieter stock exhaust systems.

12. Is it necessary to clear the ECM memory every time I change a chip? Is there any special procedure for changing the chip?

If you’re changing chips from different manufacturers, it is advisable to clear the ECM memory. This is because manufacturers differ in their base fueling cal, resulting in possible incompatibilit ies in what is learned in the BLM cells. It is advisable to start "fresh" with neutral 128 BLM values and let the ECM learn in the direction the chip forces it. However, if you’re changing from a race to street chip (or vice versa) from the same manufacturer it MAY be OK to not clear the memory as hopefully the same base cal will be in place. Switching between various EFI Performance chips made for the same injectors is fine without clearing the memory. In fact, with a 16 position thumbwheel that is exactly what you are doing. Likewise, for those of you with a PromPaq switching between manufacturers could create problems.

There are no particular procedures to follow when changing the chip. It is advisable to do so with the engine not running; however, we have changed "on the fly" so to speak with the engine idling (no - not driving on the road or doing a WOT blast! We’re crazy, not stupid.) but we don’t really advise this. Basically the system goes into backup fuel mode without a chip, and when you insert one it usually continues on its merry way. Depending on circumstances, the BLMs may be reset, but occasionally something odd happens in the transition and the engine doesn’t run correctly or dies.

13. I’ve been told I need to raise the fuel pressure at the track and lower it while I’m out cruising, or my driveability/gas mileage will suffer.

This might have been true if you run into the scenario described above where the BLM was continually negating any change in FP until it was turned so high that the BLM were at the rails. This would definitely cause a rich condition during cruise and affect driveability and gas mileage. However, if the BLMs were not at the rails yet and could compensate, then driveability/gas mileage should not have been affected.

A common mistake is to run fuel pressure very high. Some folks seem to swear by this and love the way their car drives. While it’s true that running rich makes for very smooth engine operation, it also can result in very lethargic performance and slow turbo spool-up.

We have calibrated our chips to operate at 45 psi (vacuum off) so driveability/fuel economy is optimal with this pressure. Less pressure will simply cause the BLMs to learn richer to compensate, however, our open loop idle - optimized for 45 psi, cannot learn and will definitely suffer. This is the one disadvantage of our smoother open loop idle in that it cannot compensate for unstable fuel pressures! Increasing FP much beyond 45 psi is not recommended as that is about the limit at which most injectors can operate with a consistent spray pattern. Additionally, fuel pressure is inversely related to volume, so the higher the fuel pressure, the less volume the fuel pump is able to supply. Therefore, running less fuel pressure is often advantageous to running more. Our race chips run static (max) BPW. If you need more fuel, it’s probably time to move up in injector size!

[TexasT]

14. What about the Mass Airflow Sensor (MAF)? Why is it that I have a hard time peaking out the MAF? I have a highly modified / very quick car and I know I must be flowing at least 255 g/sec of air if not more? What are those screens for? Can I remove one or both?

Obviously, we’ve heard all sorts of question about the MAF. The MAF used on the Turbo Buicks is a hot foil bridge type sensor with a low frequency output of 30-150 Hz. The basic operating principle behind this system is King’s Law, which relates the mass of airflow over a surface to the power needed to keep the surface at a constant temperature. The sensing element, which is in the form of a bridge, is in the air stream with the hot foil on one side. To keep the bridge balanced, the temperature and resistance of the hot foil must be kept constant. As airflow changes, temperature changes and thus the foil resistance also changes, making the bridge unbalanced. An op-amp senses this unbalanced state and forces more current to the bridge from the driver, changing the foil resistance until the bridge is balanced once again. As current changes, so does voltage, and this voltage, referenced to battery voltage, is converted into a triangular-wave frequency by the Voltage Controlled Oscillator (VCO) and the charging/discharging of a capacitor. This triangle-wave is converted to a square wave and divided by 16 before being sent to the ECM.

This MAF is HIGHLY affected by changes in the ductwork leading to the MAF unit. The factory engineers calibrated the MAF tables using the factory metal can air cleaner setup, and any changes to that system by using an open element cone shaped K&N filter or a ram air system greatly affect the accuracy of this calibration. The honeycomb screens are present to create a laminar flow across the sensing element and create the "correct" turbulence for accurate readings. This system is so highly influenced by turbulence that even the orientation of the pleat where the ends of the paper element in the air filter joined together affects readings! In fact, the reason the stock canister used a hex nut to hold the canister together instead of a wing nut is that the factory engineers found that even the orientation of the wing nut had a significant effect!

So what does all this mean to you? Basically, the replacement of the stock air canister with an open element K&N or ram air disturbs the airflow enough to cause inaccurate readings. In our experience, with the stock air cleaner system in place we have had no problems achieving fairly steady max 255 MAF readings. By replacing the stock system with the K&N or ram air, MAF readings dropped and fluctuated more. However, our trap speeds and times were better, so we know we must be flowing MORE air, not less!

The solution - no, we don’t suggest you go back to the stock canister system, instead we have recalibrated the MAF parameters to accurately measure airflow with a K&N or ram air unit. As for the screens, we have seen no gains in removing one screen, and suggest you keep both in place. In fact, with only one screen the airflow "seen" by the hot film is no longer consistent, and we cannot guarantee that idle qualities will be up to spec. Removing both screens definitely affects the idle and driveability and will require a custom calibration. Unless you have an extremely quick car running the 1/4 almost in the single digits, there is no need to even consider removing the screens at all. For those of you in that league, we currently have in development a system to measure airflow indirectly, allowing the majority of the air to bypass the MAF, much like the Hitachi sensors used in the 3800 Series II engines. Look for an upcoming announcement, hopefully in the near future.

On the subject of air induction systems, we’ve seen many types, the most common one being the K&N underhood behind the headlamp. This is an excellent and simple system as it has a minimum number of bends and a short path, the only disadvantage being the hot underhood air it always has to contend with. A better alternative would be to have a ram air scoop funneling cool dense outside ambient air to the filter area, or to locate the filter below the headlight behind the bumper, making certain a minimum of bends are used. We feel that locating the filter in front of the radiator is not optimal, as the filter blocks a large portion of the radiator keeping it from doing its job, and more importantly, it necessitates a 90 to 180 degree turn in the ductwork, which drastically reduces flow.

Ram air systems are not really of much value either, their only advantage being the cold dense ambient air they draw. The consensus among various airflow experts is that ram air effects are negligible until speeds well over 100 mph are reached. At idle or low speeds, the canister is fairly cramped and restrictive, possibly impeding flow on the initial launch. A solution to this problem may be to retract the filter element so only a portion is inside the ram air canister. This frees up the filter element from the canister, while still reaping the benefits of having outside ambient air directed to it.

15. If I’m actually flowing more air than the MAF can read, how am I getting enough fuel? I’ve heard that once the MAF maxes out at 255 the ECM is no longer able to control fuel? Should I shoot for 253 or 254 instead?

You are definitely able to get enough fuel provided your fuel pressure is sufficient and the injectors are capable of flowing enough fuel. It is NOT true that the ECM is no longer able to control fuel once the MAF maxes at 255. The correct statement would be that the ECM can no longer accurately meter fuel after max MAF. At 255 g/sec and less airflow, the ECM will deliver exactly the amount of fuel needed to achieve the commanded airfuel (AF) ratio. For example, if you command 12:1 AF ratio, the ECM will look at the MAF, RPM, and injector flow rate (other factors such as BLMs and INT are ignored for the sake of simplicity) and calculate the correct Base Pulse Width (BPW) of the injectors (how long the injectors stay on) to give just enough fuel to achieve this AF ratio.

Once your MAF is peaked at 255, and you’re really flowing more air, the trick is to command an even richer AF ratio to make up the difference. For example, say you’re actually flowing 300 g/sec of air. In this case, we may need to command a ridiculously rich 10:1 AF ratio to get the amount of fuel we need. What happens is once again, the ECM looks at RPM, inj flow rate, and the maxed out MAF reading 255, and calculates a BPW capable of supplying enough fuel to achieve 10:1 AF. However, since we’re actually flowing much more than 255 g/sec, with the amount of fuel delivered we’re actually only achieving a true 12:1 AF, and everyone is happy. Note that if we actually could achieve 10:1 AF this is way too rich, and the car would lay down and probably blow some black smoke.

There is nothing "magic" about the 255 g/sec MAF number. The ECM doesn’t treat 255 any differently from any other number, it’s simply the max value the MAF variable can read, and it’s used in the formulas and equations just like any other value. There’s no reason to aim for anything less than 255.

*** Warning - More Blatant Advertising Ahead !!! ***

16. I’ve noticed on shifts my MAF drops quite a bit, and I’ve a mega HP setup shifting at 6500 rpm etc. - I know I’m not flowing that little air even on shifts!

We can help of course! (Or we wouldn’t have posed this question would we?) This is not only a problem for the mega HP race machines, as it occurs on mild stock-type street setups too with the same consequences. On shifts, the sudden change in airflow can cause the MAF to momentarily "bounce" regardless of actual airflow, the result being as sudden lean spike in the fueling as the calculated injector BPW follows the dip in the MAF. This contributes to the detonation that often occurs on the shifts.

Our recalibrated MAF parameters help this problem somewhat, but does not completely eliminate it if the "dip" is severe enough. Therefore, to eliminate this phenomenon completely, we have developed a "Max MAF Enforcer" algorithm that disregards the actual MAF reading and forces MAF to equal 255 after a predetermined RPM under WOT conditions. By doing so, once the RPM and WOT conditions are met the MAF can bounce or even be completely disconnected without affecting fueling at all, as the MAF variable is forced to 255 for all fueling calculations.

What are the advantages of this "Max MAF Enforcer" algorithm? More consistent WOT fueling at high RPM, with less possibility of detonation (and resultant spark retard) on shifts. Disadvantages? Well, if you have a really slow car that for some reason doesn’t actually flow 255 g/sec or more, you’ll run too rich and slow down even more. However, we consider that your problem and not ours! Really, any car running in the 13’s flows at least 255 g/sec from our experience.

Important: The "Max MAF Enforcer" will not show up on a scan tool consistently! Do not expect the MAF reading on a scantool to stick at 255 when the "Max MAF Enforcer" is enabled! We know this may sound like a cop-out, but there is a legitimate reason for this. The MAF is read on every 6.25 msec loop in the ECM, however, the fueling output is calculated every other 6.25 msec loop, on a 12.5 msec loop. Basically the MAF is read twice but only used once every other time. The software is setup this way to ensure the most recent MAF reading available is used. We alter the MAF after it is read and before it is used to compute fueling, therefore, the 255 we input into MAF is only resident there a short period of time before it is wiped away by the "real" actual MAF reading. Don’t worry, we stick 255 in again every time on the 12.5 msec loop before fueling is calculated, so the fueling calculations only see our 255 and never the actual MAF (when our WOT & high RPM conditions are met). Therefore, especially with the relatively slow update rate of the serial data, the scan tool will more often than not miss our forced 255 value and capture the actual MAF reading instead, forcing us to take great pains via other means to ensure that our software patch does operate as designed.

17. I’ve a Knock Detector and it has often detected knock yet my scan tool shows little or no retard! What’s going on? How does this ESC system work?

Ahhh. That mysterious ESC knock module. In a nutshell, the system works like this. Every engine/trans combo in a particular platform will have a certain characteristic sonic signature. The general engine noise of the pistons and valvetrain have their own characteristic s, and the sonic signature of knock is very distinctive as well. The engine is mapped out via accelerometers and spectrum analyzers to determine the best practical location to locate a knock sensor, where the signal to noise (SN) ratio of knock to base engine noise is best. A particular knock sensor (Piezoelectric resonant in the case of the turbo Regals) of the right characteristic s is chosen, generally one whose resonant frequency is compatible with the center frequency of that engine’s knock. In the case of the turbo Regals, a 6kHz sensor is used. This knock sensor hears knock and engine vibrations, and provides a voltage input to the ESC module.

The ESC module’s function is to discriminate between base engine noise and knock. It accomplishes this by first running the signal through a band pass filter to remove unwanted high & low frequencies. Then a custom IC known as a Signal to Noise Enhancement Filter is used to condition the signal. Knock is characterized in a knock filter so that only waveforms typical of knock are passed to a knock comparator, where a noise channel produces a dc reference proportional to the rms value of the knock filter. When knock occurs, the knock channel exceeds the noise channel, and a comparator generates an output pulse whose duration is proportional to knock intensity. Shorter pulses are generally due to extraneous non-knock signals and an ignore circuit is used to eliminate all pulses shorter than a calibrated time duration. An add-on timer then re-establishes the necessary pulse duration to provide sufficient spark retard after extraneous noise is eliminated by the ignore circuit. The resultant output is a pulse whose duration is proportional to knock intensity - the amount and rate of retard and restoration of normal spark advance is accomplished by the ECM software, where ESC spark retard is applied to the Electronic Spark Timing (EST) signal proportional to the number of delta counts accrued by the detonation counter. The detonation counter is examined every 12.5 msec, a delta count is calculated, and the amount of retard determined by taking the retard from the previous 12.5 msec loop and adding the difference in time between the detonation count of the current 12.5 msec period and the detonation count of the previous 12.5 msec period, multiplied by the application rate.

In general, with maybe a few exceptions, the majority of knock detectors are too sensitive, and it is doubtful whether those manufacturers have a clear understanding of how the ESC system really works. Some have failed to understand that it is the duration of the signal, not the intensity or magnitude, that is proportional to knock intensity. Those knock detectors will tend to detect a large number of short duration bursts, as the sheer number of high-low-high transitions tends to give a false rms type value which they detect as knock. Likewise, they miss the long duration low pulses that are truly indicative of real knock.

Most knock detectors that operate on the correct principle are set to detect signals too low in duration. These low duration signals are usually nothing more than typical engine noise, and the ignore circuitry in the ESC module is calibrated to disregard signals of this duration, therefore no spark retard is actually activated while these detectors are lighting up like Christmas trees. We hesitate to publicly recommend a specific vendor’s knock detector, but we have seen one that is reasonably accurate. Note we are NOT referring to knock detectors that operate via the ALDL diagnostic link that show exact amounts of knock retard as scan tools do - those are all as accurate as the serial data.

[TexasT]

18. I’ve often seen my scan tool indicate knock counts, yet show no resulting spark retard. Sometimes on a scan tool I don’t see a parameter update for a few frames, I see things get out of sync, or I see obviously erroneous data.

First, we must make clear we are unable to give a comprehensive answer to this question as we have not studied the serial data code in the software sufficiently to know how the knock count field is being updated. However, based on what we do know there are at least two explanations for this phenomenon. The first and most likely answer is that the knock counts seen were of insufficient duration to cause any spark retard. This means that the noise causing the knock sensor to trigger wasn’t very intense, and probably not really detonation. The second reason would be the detonation was very light and the knock signal duration very small, resulting in very little spark retard which quickly restored to zero. With the slow update rate of the serial data, this small amount of spark retard could easily be missed. Additional factors we cannot completely explain involve this slow update rate causing the scan tool to "miss" some data. Sorry, but we definitely do not have an answer for everything!

 19. Why do 84-86 cars have "BLO" on the ESC module and 87 and 89 TTAs have HKP? Are they interchangeabl e?

Yes, they are interchangeabl e. The 84-86 BLO modules had a 4 pole filter, which was determined to be overkill and changed to a 2 pole filter for 87 and 89. This change necessitated a new broadcast code (HKP) to be used. Functionality and calibration are identical between the two. It has often been asked whether broadcast code BLO was purposely chosen for this BLOwn engine, and the answer is no. It just happened to be the next 3 letter combination in the system - the Lord has a sense of humor too.

*** Warning - Yet More Blatant Advertising Ahead !!! ***

20. Whenever I launch, my ESC knock (spark) retard goes to max and pulls out 10, 15, 30 degrees of spark! What can I do? What’s going on?

Do you hear any audible detonation? Are you sure you aren’t really detonating? If you’re sure it’s false retard, and you’ve checked out the ESC system and find nothing wrong (no bad sensor, bad module, bad ECM or wiring) you probably have a really quick car with a big 3" downpipe and you’re launching with sub 1.6 60 fts, probably twisting the frame as well? A typical problem is the large downpipe banging against the chassis on the launch. If a tie-strap doesn’t help, and you can’t figure out a way to keep something from physically banging, then of course, we have a solution for you!

Other manufacturers have an electronic device that ties to the TPS as a trigger and into the ESC output line or ESC sensor that effectively shuts off the ESC system for a second or two after launch. We see nothing wrong with that solution except for the cost, but we decided it would be much more elegant to offer a less expensive software solution. We offer the same exact function all in the chip with our "Launch False Knock Retard Eliminator" algorithm. No bulky external box, no wires to splice, nothing to hook up - it’s all in the chip! On launch, ESC knock retard is disabled for as many seconds as you request, and re-enables as soon as that time ends. Obviously, we can only program in one time period per chip - again, a 16 position thumbwheel chip would be ideal for multiple choices.

We only offer this feature as an additional option and do not include it in our normal line of chips, for the simple reason that you should only use this alternative as a last resort, and only if you are currently experiencing this problem! We strongly urge you to try your best to eliminate the cause of the false retard first, but if you continually have problems we offer the best solution! This feature can be added to any of our Turbo Buick chips for the 86-87 1227148 ECM.

21. What about 7th injector systems? Are they useful? Do I need it?

We have no direct experience with 7th injector systems as we’ve never felt the need for one. In our opinion, a properly sized injector for the application with the correct calibration to take advantage of all the injector offers is always the best option. This is a key point, as we’ve seen too many chips for higher flow injectors that do not come even close to maxing out the BPW of the injector! We perceive the root cause of this problem is that most other manufacturers are not able to determine the commanded injector BPW that their fueling curves deliver. We have seen many chips for higher flow injectors where the injector flow rate was raised to compensate for the higher flow injectors, yet the fueling curves remained identical to a stock injector curve! This accomplishes nothing except allow the part throttle driveability to be in the ballpark. (Changing the injector constant is not sufficient by itself to compensate for larger injectors, but it will usually get you close.) What the programmer fails to understand is that the injector flow constant is merely a multiplier, and by raising the flow rate by say, 10% and not adding that 10% back into the fueling tables, he is effectively lowering the fuel curve across the board by 10% - and thus effectively making the high flow injector output exactly as much fuel as the stock injector did!

We realize that some chip manufacturers will take offense at the above comments. Unfortunately, this is a fact that we have observed for quite some time. However, we do not claim to have studied every available chip on the market, and some programmers could very well have a clear grasp of this concept. We are simply pointing out this potential problem for the benefit of racers so that unnecessary spending is avoided. Only parties guilty of this error need feel slighted.

Assuming this is not your problem and you do have a chip that is opening the injector for a sufficient amount of time, you may need to step up to a larger injector. The only time we can see using a 7th injector is if you decide not to increase the injector size for some reason (or for some of you running the 1/4 in the single digits, if you can’t find larger injectors!) - then the 7th injector may be the crutch necessary to Band-Aid the problem.

22. What about alcohol or water injection? Is it helpful? Do I need it?

Again, we have no direct experience with alcohol or water injection. However, we can definitely see the benefits if you’re trying to run mega amounts of boost, especially on pump octane levels with relatively high spark advance. We personally have not felt the need as we generally run boost in the high teens (low 20s in 1st) with pump gas, but we would definitely consider it if we wanted to regularly run with higher boost, or run a more aggressive spark advance curve on pump gas.

23.  I’ve been told that by tweaking the cam sensor I can change my timing and/or fueling.

Sorry, this isn’t possible. Well, actually it’s possible, but you don’t want to try since the only change you can make is to make it fire & fuel the wrong cylinder. You cannot alter WHEN it fires/fuels the correct cylinder.

The ECM controls all timing and fueling. Its trigger is the Crank Sensor, which is non-adjustable. The cam sensor’s job is to sequence the ignition and injector firing so that each "pulse" of the crank sensor is synched to the correct cylinder. This means that as long as you’re in the "window" of correctness where everything is in sync, you’ll have exactly the spark and fueling as programmed in the chip. Turn the cam sensor far enough to get outside that "window" of correctness, and you’ll sync the crank pulses to the wrong cylinders, and the engine will backfire or worse. If you wish to alter the timing, it must be done in the chip. Our timing tables are optimized for a given octane, and we offer 16 position thumbwheel chips for those who wish to switch on the fly.

24. So what exactly is the relationship between the cam and crank sensor? And how is the spark and fuel delivery tied into them? How exactly does this Distributorles s Ignition System (DIS) work?

The DIS system used in the Turbo Buicks is a 3x system, meaning that 3 reference pulses occur from the crank sensor for every crankshaft revolution. These reference pulses occur 120 crank degrees apart, with 60 degrees between the rising and falling edge. The falling edge of the crank signal occurs exactly 70 degrees BTDC, meaning the rising edge will occur exactly 10 degrees BTDC.

There are two modes of operation, bypass and Electronic Spark Timing (EST). Bypass mode is used during cranking and in limp-home operation. In bypass mode the coil dwell time is defaulted to 60 degrees, as the coil begins charging on the falling edge of the 3x signal and fires on the rising edge. Therefore spark advance in bypass mode is always defaulted to 10 degrees BTDC, a reasonable number for cranking and limp-home operation. Fuel delivery during crank is in batch mode instead of sequential. Once RPM exceeds 400, the system switches to sequential fuel delivery and EST mode, where spark advance and dwell are controlled by the ECM. In EST mode, spark advance is triggered on the falling edge of the crank sensor. Basically, the EST software algorithm looks at desired spark advance and RPM and computes the length of time from the crank sensor falling edge (knowing that this falling edge is at 70 degrees BTDC) that would be required to "wait" in order that spark delivery occurs at that desired spark advance. Additionally, the start of injection occurs on the falling edge of the crank signal as well. Therefore, it is obvious that the crank sensor alone is responsible for ignition and injector timing. However, it is also evident that the 3x trigger wheel mounted on the harmonic balancer must be positioned correctly; likewise, the balancer must be keyed correctly - if either was off, the crank sensor rising and falling edges would no longer occur when they should, thereby causing ignition and fuel delivery timing to be inaccurate. Although we have not seen it for ourselves, we have heard reports of some balancers having these problems.

So, what use is the cam sensor then? Well, if you only had this uniformly spaced crank signal, you would never be able to tell exactly which cylinder each crank pulse corresponded to. This is where the cam signal is needed, as it occurs once every 2 crankshaft revs, that is, once every 6 crank signals. The cam sensor should be adjusted so that the falling edge occurs at 25 degrees ATDC of cylinder #1. Now we are able to sync each crank pulse to its respective cylinder, because after detecting a high to low transition of the cam signal, the ECM knows that the next hi-lo transition of the crank signal will be cylinder #6, which will be the first cylinder to fire. Thus, it’s evident that adjusting the cam sensor so that the falling edge no longer arrives at 25 degrees ATDC does absolutely nothing, until it is adjusted so far off spec that the next low-going crank signal is no longer cylinder #6, but #1 or #5. The problem is that the ECM still thinks that pulse should be #6, and it will fire the 3-6 coil, right in the middle of the intake stroke!

The cam sensor resyncs every 2 crank revolutions. If you happen to lose the cam signal while the engine is running, a malf code will set, and the system operation will default to simultaneous double fire operation instead, but the engine will remain running until you shut it off. Once you kill the engine, you will not be able to start it again.

Ever wonder why the crank time varies so much from start to start? It all depends on how the cam sensor is positioned when the engine finally stops after a keydown. If it stops immediately before the cam sensor transition, then the next start-up will be very quick as everything can sync up in less than 1/2 a crankshaft rev. However, if the engine stops after the cam sensor transition, a full two engine revs may be required before everything is able to sync.

Side note: If you are consistently plagued by slow starts, power up the ignition only (turn key to run, but not start) and listen for the fuel pump as it should be powered thru the fuel pump relay for approx. 5 secs to prime the fuel rails. If you never hear the fuel pump prime, suspect a bad fuel pump relay - the cause of your slow starts is this bad relay, as you are relying on the oil pressure switch to turn on the fuel pump, which it will only do after sufficient oil pressure is achieved.

[TexasT]

25. But I know cam sensor can be installed 180 degrees out of phase and the car will still run - it just won’t run right. Why is this?

Now, it is very possible to install the cam sensor 180 degrees out of phase and still have a running car! It makes no difference to the ignition system that the cam sensor is 180 out, since it's a waste spark system. You're still firing the same coil pack - from an ignition standpoint if the cam sensor is 180 out there's absolutely no difference as the current still runs out the same terminal, thru one plug from center to ground, thru the block and back from the ground electrode to the center electrode on the other plug and returns to the other coil terminal. This pattern occurs the same way regardless of whether the cylinder is under compression or exhaust. The plug (cylinder) that fires the "usual" way does this every compression and exhaust stroke, and the plug (cylinder) firing "backwards" always fires backwards on its compression and exhaust stroke.

What changes is the fueling. With the cam sensor 180 out you'll fuel the cylinder on the wrong stroke, and the fuel will sit around puddling on a hot intake valve for 1 engine rev until the valve opens and it's forced in.

26. I’ve heard that I can short out a terminal on the ALDL connector (diagnostic connector where a scan tool plugs into) and make the car run in batch mode to help performance.

Yes, it is possible to force the system to operate in Simultaneous Double Fire (Batch) mode rather than in Sequential mode as it normally does, however, this hurts performance rather than helps it. Basically, rather than giving a complete squirt of fuel as the intake valve opens, all the injectors simultaneously deliver half their pulse width once per engine revolution, allowing it to sit and puddle on a hot intake valve as in the above scenario. Having the injectors fire twice per combustion event giving half their usual amount of fuel each time requires turning on/off the injectors an additional time, and since fuel flow is minimal during the ramp up time this cannot provide more fuel than having an injector open just once.

27.  I’ve seen some racers use a different coil pack with the 3 individual oval coils instead of the stock rectangular one with the 3 coils connected. What’s the deal?

In 86-87 there were two suppliers for the ignition system - Magnavox and Delco. The 3.8 Buicks used the Magnavox unit, which is the stock rectangular one with the 3 connected coils. In the same model years, the 3.3l engines and the 3.8 in the Olds apparently used the Delco unit - the one with the 3 individual oval coils. The 2 are interchangeabl e, the wiring harness is the same and plugs into either one, but an adapter plate must be used to attach a Delco unit onto brackets made for the Magnavox. Additionally, the ignition modules (the base that the coils attach to) only mate with their respective coils packs (i.e. the Magnavox ignition module only fits the Magnavox coil pak and the Delco ignition module only fits the Delco coils.) Note also that all Delco individual coils are interchangeabl e regardless of which make car or model year they originated from (i.e. coils from a Cadillac Northstar engine can be used on a Buick) but the ignition module (the base) MUST be from a 86-87 system and NOT from a 88 or newer 3800 system. This is because the 88 and newer 3800 engines have a 18x ignition system, while the 86-87 3.8L engines had only the 3x system. A module designed for a 18x system requires a 18x input signal which the 86-87 engines do not have.

So, why do some racers bother to switch ignition systems? It’s a little known secret, but we’re in a generous mood - the Delco coils have lower inductance (approx. 3.5 mH) than the Magnavox (approx. 8mH), and provide slightly more energy. Primary winding resistance of the Magnavox unit is 0.7 ohm vs. 0.34 for Delco. Magnavox current limit is 6.7 amps with a 3.5 msec risetime vs. 9 amps @ 3 msec for Delco. This translates to a risetime energy of 8.58 mJ for Magnavox and 9.87 mJ for Delco. The Magnavox current limit power is 65 watts times current limit duration, while the Delco current limit power is 102 watts times current limit duration. Is this slight energy increase significant? Probably not, but every little bit helps.

28.  What about scan tools?

Are you serious about having an optimally running car? Then you need a scan tool. We highly recommend Ken Mosher’s TurboLink from TDS Technologies if you’ve access to a laptop computer. He can be reached at 402-292-6707, email him at Kenmosher@nfinity.com, or visit his web page at http://nfinity.com/~kenmosher/tlink. This is by far the most advanced scan tool available for the Turbo Buicks, as it displays everything at once on one screen in either numerical or graphical format, has extensive storage capability for storing a run or for diagnosing a problem, has a multitude of triggering options to begin recording, and the ability to save data into files for later review. The latest version has the ability to record boost as well, something no other scan tool can do. It is simply an excellent product at a very reasonable cost. The only disadvantages are the fact that you must have a laptop computer to use it (Ken often has used laptops for sale as well cheap!), the fact that it currently does not have the capability to read data from cars other than those that use the Turbo Buick ECM, and it’s next to impossible to mount the laptop on your dash to read as you go down the track!

Any scan tool would be an excellent investment for the serious Turbo Buick racer and we strongly urge you to become proficient with a scan tool if you are not already. As you can see already if you have read this far, it is vital that you have a good grasp of scan tool data to achieve optimum performance from your car, and to know if your tuner is being straight with you or feeding you a line!

29.  Is there any way to increase the update rate of my scan tool?

Unfortunately, the serial data baud rate of the Turbo Buick ECMs is a disappointing 160 baud. This translates to an update approximately once every 1.44 seconds. There is no way to increase this, and it is not the fault of the scan tool manufacturers that the data rate is so slow. Remember this is a very early controller, designed in the early 80s, but was cutting edge technology in its day when the fastest computer modem operated at only 300 baud!

However, there is one solution available an innovative scheme of removing some of the more obscure and seldom used data from the data stream and replacing those with the more important parameters has been developed. This effectively makes the data update at a much quicker rate. Disadvantages are that some parameters are lost (usually not a big sacrifice), the code changes must be incorporated into the EPROM, and only a few scan tools can currently read this altered data stream - all other conventional scan tools will not be able to decode it.

30.  Will running "lean cruise" mode negatively impact valvetrain durability due to increased Exhaust Gas Temperatures (EGT).

No. Our "lean cruise" mode enables only at cruising speeds at light throttle, and varies the AF ratio from a normal stochiometric 14.7:1 up to 16.1:1 depending on load. One of the reasons this is not done on production vehicles here in the US is due to the higher NOx of lean combustion. However, many vehicle manufacturers outside the US have this feature in production, due to different emission regulations. Typical EGTs at light cruise with stochiometric AF ratios of 14.7:1 are around 1000 degrees F, while a leaner 16.1:1 AF ratio results in approximately a 75 degree increase based on our observations. This is the real reason we run lean cruise - to raise EGTs slightly to aid turbo spoolup; the improvement in fuel economy is a secondary benefit.

31.  Regarding chips, I’ve been told since I have a big turbo or cam, I need a special "big turbo" and/or "big cam" chip. How does one of these differ from a "normal" chip?

A big cam/turbo chip is a misnomer. There may be some minor tweaks such as raising the idle to cover up the lopey idle of a big cam, and some additional fueling thrown in under PE to account for the increase in air that is compressed into the cylinders by a larger turbo, but that’s about it. If you can call such changes a "big turbo/cam" chip then we suppose it’s justifiable. We’re sure we will have disagreement on this point, but these are the facts:

The Turbo Buick is not a speed density calibration (MAP based), it is Mass Airflow (MAF based). Yes, we agree with a MAP based cal a special "big cam" chip would be necessary to compensate for the different vacuum conditions on the MAP that a big cam imposes. Fueling is directly related to MAP readings, so a big cam’s effect on the MAP definitely has to be accounted for. However, this is not the case for a MAF based system as it is measuring the exact mass of air that the engine ingests. No matter how big or small the cam is, it still measures the exact amount of air going in and fuels accordingly. Timing may need to be altered slightly, and big turbos may require some additional help to spool quicker, but not enough that one could consider these changes a "big turbo/cam" chip. Besides, the factory experimented with some rather large cam and turbo combinations, and did not significantly alter the cal.

Unfortunately, most racers still consider the chip to be "black magic", and they give the chip (or blame the chip as the case may be) much more credit than it’s due. There is nothing mysterious or magical about the chip, and we hope that by reading this (you did actually read this in its entirety and not just skip to the end?) you have a better understanding of what the chip is and is not capable of. This particular engine management system was designed by a group of engineers (OK - a rather talented team of engineers) in the early 80’s, nearly 15 years ago. You can expect engineers are going to do things the easiest way that works, despite what most people think! Then we make it look hard to justify our jobs! For real magic, consider what God did with living creatures - that’s what’s incredible.

32.  Can I disconnect and remove my O2 sensor?  I run leaded race gas and don't want to poison an expensive O2 sensor.

Yes, but only if you have a chip that's programmed to allow you to do this, and to the best of our knowledge, ours are the only ones that do!  Conventional chips will set a code and go into a default fueling mode if you disconnect the O2 sensor.  We detect that the O2 sensor is not present, and simply continue on in open loop mode and do not set any malf codes.   Being the cheap racers we are, we thought it was foolish to keep poisoning expensive O2 sensors when we raced with leaded fuel, so we decided to solve that problem right then and there! 

Now, some folks will say they rely on the O2 sensor for tuning purposes.  That's good to an extent, but remember the leaded fuel is poisoning the O2 sensor as you're running, causing your readings to become more and more inaccurate and suspect.  Heated O2 sensors seem to resist poisoning and last a bit longer, but they'll eventually become inaccurate as well.  We simply plug the O2 sensor hole with a plug that has a thermocouple installed, so we can monitor EGTs (exhaust gas temps) and tune with that instead!

Some folks will claim that the O2 sensor is required for proper operation, and their car will run very strangely when it's disconnected.  Well, this is what's going on.  First, if they're saying their cars run very odd under cruising conditions with either a bad O2 sensor or a disconnected one, they're correct - but only with a conventional chip.  A conventional chip will detect a disconnected O2 sensor, set a malf code, and run on some default  limp-home fueling which is far from optimal.  The Thrasher chip will detect a disconnected O2 sensor, but it will not set any malf code, and will not revert to a default limp-home fueling mode, it'll simply continue on in open-loop mode.

Do not disconnect power to the ECM and clear it's memory before running a Thrasher chip with a disconnected O2 sensor!  You can do this, but your engine won't run optimally.  Remember the block learn memory (BLM) discussed in question #2?  Remember that after driving around a bit, that BLM table has been learned in.  Thus, if you don't clear the ECM memory, those BLM values stay where they learned to on your last cruise, and are still in use after you disconnect the O2 sensor.  Yes, because you've disconnected the O2 sensor, the ECM will not get out of open loop mode and will not continue to "learn" anymore as you drive; however, the "old" BLM values are still there are being used, and are probably very close to being where they need to be for optimal operation anyway.  Thus you at least retain the last learned information from the last time you drove with the O2 sensor connected.

If you DO clear the ECM memory before disconnecting the O2 sensor, then all BLM cells are reset to their normal, nominal value of 128, and no corrections to the fueling will be made.  Thus, if there were conditions where the ECM had learned in a significant amount of fueling compensations, this information would be lost, and your fueling will no longer be close to optimal anymore.

Remember too that Thrasher chips reset the BLM at WOT to 128?  (See question #10 in this guide for a detailed explantion of why this is a good thing!)  Thus at WOT, the Thrasher chip is operating under a true open loop condition anyway, so you'll see no WOT differences in running with or without an O2 sensor connected!  Remember, as discussed in question #10, a conventional chip's WOT operation may still be affected by a poisoned O2 sensor because the WOT BLM cell 15 will continue to learn, and whatever random value it's learned to (random because the O2 has been poisoned) is used to alter WOT fueling.  A disconnected O2 sensor will affect a conventional chip's WOT operation as well simply due to the fact a malf code will set and a default fueling model will be substituted for the proper chip calibration.

[TexasT]

Apparently there is a 20000 character limit on each post so I broke it up.

[daveismissing]

Good stuff
Carl really gets around

[Be4u]

This is great!

[good2win22]

Nice find Rich!  Thanks for sharing!

[nocooler]

I loved my thrasher chips. The race chip was bad ass

[motorhead]

I stopped reading at #14 because I got all misty eyed for simpler/not simpler times.