www.cp-e.com
Member
MODS: Sponsorhip fee has been paid via paypal 1/24/07 @ 11:56am
Custom Performance Engineering has been anxiously awaiting sponsorship status on Mazdas247 for some time now. We've had a lot of great successes in the past few months and we wanted to share some of our latest developments with the community.
We are currently completing the development of our full tuning solution for the SPEED6. Our unit is a piggyback air/fuel, timing and boost controller with datalogging capabilities. The MSRP for a non-PnP unit is $600, or $635 with two pre-loaded base maps. It should be available for pre-order in the next couple weeks.
We recently wanted to demonstrate the capabilities of our piggyback, so we installed it onto a SPEED6 with a turbo-back exhaust and cold air inatake. We also wanted to see what we could learn about the factory ECU. Here is what we found...
Throttle Position and Boost Pressure
We decided to put the piggyback on Seoulfuls (Brians) SPEED6 so we could get a better idea of what the car is doing with timing, the throttle plate, the wastegate, and so on. Brians SPEED6 had a turbo-back and an intake installed at the time of testing.
Whats interesting is that the SPEED6s throttle plate does a similar dance as the SPEED3s at wide open throttle. The difference is that the SPEED6s ECU doesnt create any appreciable performance limitations like the SPEED3s ECU does. The SPEED3 uses the throttle plate as a power output governor, which was likely implemented to reduce the amount of torque steer and wheel spin in the lower gears. Since the SPEED6 doesnt have the same traction or torque steer issues, it isnt clear what the purpose of the throttle plate actuation is during wide-open throttle. Theory dictates that in order get the most out of an engine, the volumetric efficiency should be as high as possible to reduce pumping losses. The fact that the SPEED6 tapers its throttle opening towards redline and reduces boost to some degree goes against convention, and likely has another purpose. The good news however, is that the throttle plate in the SPEED6 doesnt seem to limit peak boost like it does in the SPEED3. Here is a chart of the boost pressure versus the throttle position at WOT. Take note of the peak voltage of the throttle sensor, and the boost pressure (we dialed it in for 15psi in this case)
Notice how the throttle plate is NEVER wide open (wide open would indicate about 4.7-volts). Not even close. Also, as the rpms begin to climb, so does the throttle position. One may argue that this slow increase in the throttle position is to increase intake air velocity as rpms climb, a la the manifold runner control, but notice that the turbo is already at full bore pushing 15psi. At about 5500rpm the fun begins to end and the throttle starts shutting things down.
Whats interesting in this next chart is that we dialed in 19psi of boost to see how the car reacted. Is the turbo capable of holding 19psi to redline? How is the throttle plate going to affect this?
Again, take notice of the throttle position. You can see how it ramps up quickly between 2300-2700rpm, and then the slope is decreased as soon as peak boost is achieved. This implies that the throttle control is somehow tied to the boost pressure. We also quickly realized that the turbo was not capable of holding 19psi to redline, so we stopped the run short to avoid over spinning the turbo. It isnt clear from the chart, but the turbo maintained 19psi until about 4000rpm and then started to taper off down to 17psi. We know we ran out of turbo (as opposed to the throttle limiting boost) since the throttle angle is still increasing. Also, notice how the throttle angle is MUCH higher than the angle during the 15psi run in the first chart.
PID Boost Control
Our piggyback differs from some of the other engine management systems in the way it controls boost, or more specifically the wastegate solenoid. The wastegate is controlled by a solenoid which is tied to the ECU. The ECU commands the solenoid to open or close the wastegate depending upon how much boost is requested. Other tuners seem to use a boost coefficient to control boost, which can get tricky. Our boost controller uses a PID (proportional-integral-derivative) feedback loop to monitor and control the boost pressure. This is significant because instead of using some factor with no feedback to control boost, you can demand a certain value, say, 16psi and the feedback loop will control the wastegate solenoid until the map sensor sees 16psi. It really is an incredibly sophisticated system, and Ill briefly explain how PID control works. Alternatively, you can search Wikipedia for a great explanation as well.
A PID feedback loop uses calculus to control a process without human intervention. In other words, using a setpoint value, the PID control calculates how far some value is from the desired setpoint, and even tries to guess at the next value in the process.
The proportion part of the loop calculates the error from the setpoint in percent. For example, if the setpoint is a value of ten, and the current value is 8, then the proportional error is 20%. This helps give the feedback loop some idea of how far off from the setpoint it is.
The integral part of the loop provides an indication of past error. In other words, it finds the integral of the proportional error and sums this error up over time. This also helps indicate if the error is above or below the setpoint value.
Finally, the derivative part of the feedback loop tries to determine where the process is headed. For example, if you look at one of the boost charts above, you can see the slope of the curve is very high to begin with, but once the ECU finds its desired boost setpoint, the slope goes to zero. If the slope is zero, then the boost pressure isnt climbing. If the slope is negative, then the boost is dropping. By incorporating this into the control loop, the system has some idea of the direction of the process its controlling.
A graph of the control loop is shown below. You can see how each aspect of the PID control reacts to Lou setting the boost level to 15psi.
If the PID control was ideal, then the P portion of the controller would go to zero immediately. This would indicate that there is no error, and youve hit your setpoint exactly. The integral term should be a flat line once it reaches the desired setpoint. Any slope to the integral line indicates that error is being generated. Finally, the derivative term indicates the direction of the process. Since we want to keep the boost at a certain value, the PID control loop constantly overshoots and undershoots the desired boost point. This is normal, but the amplitude of the oscillation gives an indication as to how much control the PID feedback loop has over the process. The lower the oscillation, the better the PID control.
The best part of all of this is that you can fine-tune how much weight each term (proportion, integral and derivative) have on the process control by changing the PID coefficients. We provide a PID coefficient to start with, but you can control the PID coefficients if youd like to get a better control over boost for your particular setup. But how good did the feedback loop control the boost in this particular case?
This is a chart with the wastegate position versus the boost pressure, and was set for 15psi. You can see that the wastegate is always being tweaked to try to find the setpoint value. Interestingly, the wastegate is never fully opened OR closed! Even at idle, the wastegate is open to some degree. And the wastegate is never fully open to the 255 value because you can create secondary magnetic fields in the solenoid which can inadvertently keep the wastegate open permanently. But, you can see that the boost control is pretty darn good with our recommended PID coefficients.
Hopefully this explains why we didnt just use a straightforward boost coefficient with no feedback loop to control the boost.
Mass Airflow Sensor
Many people have speculated that the mass air sensor maxing out at 5-volts is responsible for the fuel cut. We wanted to see if we could induce a fuel cut to investigate further and were unable to produce one even at 19psi. Here is a chart with the boost pressure relative to the mass air sensor voltage.
Keep in mind that peak air/fuel consumption does not occur at peak horsepower or towards redline. The most air will be ingested at the torque peak. So, this chart shows both the incline and decline of the mass air sensor voltage (the voltage decline is more difficult to see on this chart) indicating that peak torque has already been reached. Also worth noting is that the peak sensor voltage occurred at 5267rpm and was 4.29-volts, which is well below the 5-volt maximum. Not surpringly, we did not experience any fuel cut.
Next week we visit the dyno with the wideband and well also get some data on the fuel injector duty cycle to see if we can get some indication as to how much fuel this thing can flow. Lou is also determined to gain control over the throttle plate. Because his piggyback is so versatile, he can incorporate a process that will help the throttle plate mimic the users throttle input better. This is a tricky challenge, but look for throttle control on the SPEED3/6 in the near future.
Since we're now official sponsors, feel free to post up questions or comments, or you can PM us under our www.cp-e.com handle.
Jordan
Custom Performance Engineering has been anxiously awaiting sponsorship status on Mazdas247 for some time now. We've had a lot of great successes in the past few months and we wanted to share some of our latest developments with the community.
We are currently completing the development of our full tuning solution for the SPEED6. Our unit is a piggyback air/fuel, timing and boost controller with datalogging capabilities. The MSRP for a non-PnP unit is $600, or $635 with two pre-loaded base maps. It should be available for pre-order in the next couple weeks.
We recently wanted to demonstrate the capabilities of our piggyback, so we installed it onto a SPEED6 with a turbo-back exhaust and cold air inatake. We also wanted to see what we could learn about the factory ECU. Here is what we found...
Throttle Position and Boost Pressure
We decided to put the piggyback on Seoulfuls (Brians) SPEED6 so we could get a better idea of what the car is doing with timing, the throttle plate, the wastegate, and so on. Brians SPEED6 had a turbo-back and an intake installed at the time of testing.
Whats interesting is that the SPEED6s throttle plate does a similar dance as the SPEED3s at wide open throttle. The difference is that the SPEED6s ECU doesnt create any appreciable performance limitations like the SPEED3s ECU does. The SPEED3 uses the throttle plate as a power output governor, which was likely implemented to reduce the amount of torque steer and wheel spin in the lower gears. Since the SPEED6 doesnt have the same traction or torque steer issues, it isnt clear what the purpose of the throttle plate actuation is during wide-open throttle. Theory dictates that in order get the most out of an engine, the volumetric efficiency should be as high as possible to reduce pumping losses. The fact that the SPEED6 tapers its throttle opening towards redline and reduces boost to some degree goes against convention, and likely has another purpose. The good news however, is that the throttle plate in the SPEED6 doesnt seem to limit peak boost like it does in the SPEED3. Here is a chart of the boost pressure versus the throttle position at WOT. Take note of the peak voltage of the throttle sensor, and the boost pressure (we dialed it in for 15psi in this case)
Notice how the throttle plate is NEVER wide open (wide open would indicate about 4.7-volts). Not even close. Also, as the rpms begin to climb, so does the throttle position. One may argue that this slow increase in the throttle position is to increase intake air velocity as rpms climb, a la the manifold runner control, but notice that the turbo is already at full bore pushing 15psi. At about 5500rpm the fun begins to end and the throttle starts shutting things down.
Whats interesting in this next chart is that we dialed in 19psi of boost to see how the car reacted. Is the turbo capable of holding 19psi to redline? How is the throttle plate going to affect this?
Again, take notice of the throttle position. You can see how it ramps up quickly between 2300-2700rpm, and then the slope is decreased as soon as peak boost is achieved. This implies that the throttle control is somehow tied to the boost pressure. We also quickly realized that the turbo was not capable of holding 19psi to redline, so we stopped the run short to avoid over spinning the turbo. It isnt clear from the chart, but the turbo maintained 19psi until about 4000rpm and then started to taper off down to 17psi. We know we ran out of turbo (as opposed to the throttle limiting boost) since the throttle angle is still increasing. Also, notice how the throttle angle is MUCH higher than the angle during the 15psi run in the first chart.
PID Boost Control
Our piggyback differs from some of the other engine management systems in the way it controls boost, or more specifically the wastegate solenoid. The wastegate is controlled by a solenoid which is tied to the ECU. The ECU commands the solenoid to open or close the wastegate depending upon how much boost is requested. Other tuners seem to use a boost coefficient to control boost, which can get tricky. Our boost controller uses a PID (proportional-integral-derivative) feedback loop to monitor and control the boost pressure. This is significant because instead of using some factor with no feedback to control boost, you can demand a certain value, say, 16psi and the feedback loop will control the wastegate solenoid until the map sensor sees 16psi. It really is an incredibly sophisticated system, and Ill briefly explain how PID control works. Alternatively, you can search Wikipedia for a great explanation as well.
A PID feedback loop uses calculus to control a process without human intervention. In other words, using a setpoint value, the PID control calculates how far some value is from the desired setpoint, and even tries to guess at the next value in the process.
The proportion part of the loop calculates the error from the setpoint in percent. For example, if the setpoint is a value of ten, and the current value is 8, then the proportional error is 20%. This helps give the feedback loop some idea of how far off from the setpoint it is.
The integral part of the loop provides an indication of past error. In other words, it finds the integral of the proportional error and sums this error up over time. This also helps indicate if the error is above or below the setpoint value.
Finally, the derivative part of the feedback loop tries to determine where the process is headed. For example, if you look at one of the boost charts above, you can see the slope of the curve is very high to begin with, but once the ECU finds its desired boost setpoint, the slope goes to zero. If the slope is zero, then the boost pressure isnt climbing. If the slope is negative, then the boost is dropping. By incorporating this into the control loop, the system has some idea of the direction of the process its controlling.
A graph of the control loop is shown below. You can see how each aspect of the PID control reacts to Lou setting the boost level to 15psi.
If the PID control was ideal, then the P portion of the controller would go to zero immediately. This would indicate that there is no error, and youve hit your setpoint exactly. The integral term should be a flat line once it reaches the desired setpoint. Any slope to the integral line indicates that error is being generated. Finally, the derivative term indicates the direction of the process. Since we want to keep the boost at a certain value, the PID control loop constantly overshoots and undershoots the desired boost point. This is normal, but the amplitude of the oscillation gives an indication as to how much control the PID feedback loop has over the process. The lower the oscillation, the better the PID control.
The best part of all of this is that you can fine-tune how much weight each term (proportion, integral and derivative) have on the process control by changing the PID coefficients. We provide a PID coefficient to start with, but you can control the PID coefficients if youd like to get a better control over boost for your particular setup. But how good did the feedback loop control the boost in this particular case?
This is a chart with the wastegate position versus the boost pressure, and was set for 15psi. You can see that the wastegate is always being tweaked to try to find the setpoint value. Interestingly, the wastegate is never fully opened OR closed! Even at idle, the wastegate is open to some degree. And the wastegate is never fully open to the 255 value because you can create secondary magnetic fields in the solenoid which can inadvertently keep the wastegate open permanently. But, you can see that the boost control is pretty darn good with our recommended PID coefficients.
Hopefully this explains why we didnt just use a straightforward boost coefficient with no feedback loop to control the boost.
Mass Airflow Sensor
Many people have speculated that the mass air sensor maxing out at 5-volts is responsible for the fuel cut. We wanted to see if we could induce a fuel cut to investigate further and were unable to produce one even at 19psi. Here is a chart with the boost pressure relative to the mass air sensor voltage.
Keep in mind that peak air/fuel consumption does not occur at peak horsepower or towards redline. The most air will be ingested at the torque peak. So, this chart shows both the incline and decline of the mass air sensor voltage (the voltage decline is more difficult to see on this chart) indicating that peak torque has already been reached. Also worth noting is that the peak sensor voltage occurred at 5267rpm and was 4.29-volts, which is well below the 5-volt maximum. Not surpringly, we did not experience any fuel cut.
Next week we visit the dyno with the wideband and well also get some data on the fuel injector duty cycle to see if we can get some indication as to how much fuel this thing can flow. Lou is also determined to gain control over the throttle plate. Because his piggyback is so versatile, he can incorporate a process that will help the throttle plate mimic the users throttle input better. This is a tricky challenge, but look for throttle control on the SPEED3/6 in the near future.
Since we're now official sponsors, feel free to post up questions or comments, or you can PM us under our www.cp-e.com handle.
Jordan
As far as the throttle plate goes.....u could have asked any Mazda tech and they could have told u that the TP is governed. My techs told me the exact percentage that it opens up for the Speed6 and Speed3, but cant remember what it was....its been quite awhile. 
