Over the last 25 years there has been a technological boom among the original equipment manufacturers (OEM) of the automotive industry that has allowed for the production of performance cars with more horsepower and greater reliability than ever before; while also reducing emissions and increasing efficiency at the same time. Today, the automotive aftermarket has made huge strides in catching up to the tech being pumped out by the big manufacturers.
The fuel injector sector of the aftermarket is one area that’s always hard at work trying to keep pace with the continuously changing world of modern fuel injection. When we upgraded the stock turbo on our built Evo X time attack project car for one of Garrett’s GTX3076R kits, we reached out to Fuel Injector Clinic for a set of the company’s larger fuel injectors that we knew could keep up with the air volume produced by the GTX.
Installing the Fuel Injector Clinic 1,650 cc/min fuel injectors as part of our testing.
Jens von Holten, owner, and Tim Jilg, general manager of Fuel Injector Clinic, knew these injectors would be a challenge and since I would be tuning the Evo myself, they saw this as an opportunity to assist them in the development of new flow bench testing strategies to further improve the matching process of high flow fuel injectors in the nonlinear injector pulse width (IPW) operating range. Which is commonly encountered at idle and even during cruise conditions. This test involved two different sizes of off-the-shelf high impedance injectors, and three different sets. One set of Fuel Injector Clinic’s newest 1,650 cc/min injectors, and two sets of the earth-flooding 2,150 cc/min unit.
The goal is for Fuel Injector Clinic to expand the knowledge learned from this test to their entire injector product line, providing an overall better driving experience for enthusiasts and a much faster calibration time for their tuner.
Hopefully this work will lead to an increase in sightings of monster musclecar builds cruising on the highway for a Friday commute and high horsepower four-cylinders pulling up next to you at a red light on a Sunday and smoothly idling in closed loop; which were both completely unheard of less than a decade ago and are still considered a rare sighting for most of America even today. Where it’s legal, of course. *wink wink*
We can all agree that before diving into the deep end, it’s pretty important to know how to swim. So, before jumping into the more technical stuff you at least need to have a basic understanding of the operation of modern fuel injectors. You can find a quick crash course on modern fuel injector technology here.
When an engine is driven on the street, it’s exposed to extended periods of time at idle, low speeds, and low engine loads. Which means the injectors will spend more time at shorter pulse widths (shorter spray time, less fuel delivery) than if it were only driven on a race track and rarely saw any driving below six-tenths. For many years this has put plenty of enthusiasts in a battle between drivability and injector size, often opting for the smaller injectors to retain its street friendly characteristics.
A graph showing the full operating range (0 to 19 ms) of four different Fuel Injector Clinic injectors, with effective pulse width on the X-axis and flow rate (cc/min) on the Y-axis: yellow, 525 cc/min injector; green, 1,100 cc/min injector; black, 1,650 cc/min injector; and red, 2,150 cc/min injector.
“For a four-stroke engine your average fuel injector has two operating ranges – linear, from roughly 1 to 19 ms opening times [effective pulse width]; and nonlinear, 0 to 1 ms opening time,” explains Jilg. “In the linear range, the injectors will act in a very predictable manner. With smaller injectors [1,000 cc/min (95 lb/hr) and smaller], you will most likely only ever operate in this range. The nonlinear range comes into play on bigger injectors, over 1,000 cc/min, when at idle or even part throttle situations depending on the size.”
Notice how choppy the injectors’ behavior becomes as they run into the nonlinear operating range (0 to 1 ms). The bigger the injector, the more pronounced and widespread the nonlinearity becomes. The bold horizontal black lines represent the estimated fuel requirements for idle (30 cc/min) and cruise (60 cc/min).
“The nonlinear range can become very unpredictable for injector operation and fuel delivery, causing sporadic engine behavior,” continues Jilg. “The majority of the problems people have with large injectors are at idle or cruise, because you are hitting this nonlinear range and the injectors no longer match close enough in flow rate or dead time. Currently, no one offers nonlinear range matching of fuel injectors, but we are working towards implementing this into our entire line of injectors.”
“If you had a 1,600 horsepower V8 or 800 horsepower import just a few years ago, you wouldn’t have really cared if the thing idled a little rough, it was either at the track or on a trailer,”
Von Holten was quick to point out the new set of hurdles given to the aftermarket in the last several years.
“If you had a 1,600 horsepower V8 or 800 horsepower import just a few years ago, you wouldn’t have really cared if the thing idled a little rough, it was either at the track or on a trailer,” says von Holten. “Today there is a new standard for these high horsepower engine builds, because now they’re still being driven around town and for rush hour commutes. These same people now wake up on a cold northeast morning, when it’s 25-degrees F outside, and expect the engine to fire right up on the first crank and then smoothly drive them to get coffee at Starbucks. For that you need beyond an OE level of precision in your fuel delivery, otherwise the engine is not going to be happy.”
The challenge starts with the OE manufacturers of these injector bodies that the aftermarket builds upon. “It’s important to point out just how incredibly small these pulse widths are,” said von Holten. “When in the idle low pulse width area, it would take about 2,000 shots from one injector to fill a teaspoon; and we don’t actually have a way to measure an injector’s shot-to-shot repeatability.”
High speed photography displaying the 2,150 cc/min injector spray pattern.
“In 2015, I spent some time at the Bosch injector manufacturing plant in Bamburg, Germany,” said von Holten. “I talked to their engineers about measuring the shot-to-shot repeatability of an injector, and they said that at the Bosch R&D lab in Stuttgart [Germany], they have designed a system that can take a 3D picture of a single pulse. An employee will sit there for hours and basically build a rendering of the individual fuel droplets in each shot. They then use these renderings to make an estimation of how much fuel came out. That’s about as close as even the big manufacturers can get to shot-to-shot repeatability measurements.”
“While Bosch may use that laborious process to compare shot volumes in the linear operating area, the aftermarket is now concerned with the non-linear area and we’re looking for precision fueling where OE manufacturers don’t even expect their injectors to be used,” von Holten continues. “Most factory production vehicles are fitted with much smaller injectors that really never stray from that linear operating area. The engineers I spoke to couldn’t even give me an estimate as to what kind of repeatability they might see in the nonlinear range.”
At the OE-level, the flow and dead time matching process is much more lenient. Most factory equipped fuel injectors today are flow matched within 5-percent of one another and offset matching can vary by as much as 38-percent because it is not even tested. The average high quality aftermarket fuel injector manufacturer flow matches within 2 to 3-percent, but cannot test for offset in production quantities. Fuel Injector Clinic’s internal matching standard is less than 1-percent for flow and less than 2-percent for offset, with all of that data being displayed on the flow sheet found in the box with each set of injectors.
A graph displaying the dynamic flow rate (cc/min) in the linear zone and dead time or latency (ms) for individual injectors. Each dot represents a single 2,150 cc/min injector. The red box represents the OE-level of variance allowed when matching is done by manufacturers like Bosch and Denso; green, the variance typical by the majority of the aftermarket; blue, the variance allowed by Fuel Injector Clinic.
Requiring a more precise level of flow and offset matching means that each set of Fuel Injector Clinic injectors being received by the end user will flow much more closely and have a similar delay to one another, producing a more consistent air/fuel ratio from cylinder to cylinder. In theory this testing strategy sounds great, but in reality is only accurate in the injectors’ linear operating range. Remember, as injector size increases, the injectors are progressively operated further into the nonlinear zone, making the fueling subject to larger errors in the low IPW range.
Research And Development
In an effort to find a solution to the demand for street friendly high flowing injectors, many in the industry have worked closely with software engineers to reverse engineer the raw hexadecimal code in many of today’s most popular factory ECUs. To their delight, additional low IPW calibration tables that are capable of further refining the injectors’ operation are being discovered for many applications.
“The flash-based software companies have come a long way, just over three or four years ago none of this data was available,” explains von Holten. “Now flash software engineers have been finding more and more of these pulse width adder tables within the code that the OEM engineers have probably been using for the past decade, but we either haven’t found them until now or previous attempts didn’t find them beneficial to define at the time.”
An example of the raw hexadecimal code found in most modern automotive ECUs, read using the hex code disassembler and debugging software, IDA. This is the data that aftermarket engineers sift through to find addresses for important tables such as ignition timing, VVT, injector compensation, etc, which can then be used to build a single or multidimensional table in third party tuning software to make adjustments.
Reading The Graph
To fully understand the importance of these compensation tables to the development process, von Holten provided us with various graphs as a visual aid to better understand how this data is used. These graphs depict the characteristics of various fuel injectors with effective pulse width in milliseconds (with 19 ms being equivalent to 100-percent injector duty cycle) and injector flow rate in cc/min. Effective pulse width is the IPW value before any battery or ignition voltage offset values are added to compensate for delays in injector opening time.
This graph is an example of the full operating slope of Fuel Injector Clinic’s 1,000 cc/min injector. The X-axis represents effective pulse width in ms, and the Y-axis is fuel flow in cc/min.
The ECU in your vehicle automatically assumes that the fuel injectors operate on a linear line from a pulse width of 19 ms down to 0 ms, which is shown as the dotted red line on the graph below. That line will be used as the baseline for how much of a pulse width adjustment will need to be made for your injectors to actually flow properly in the low IPW area.
“We use these graphs to model exactly how the short pulse width adder curve will need to look in the compensation table,” explains von Holten. On the graph below, the injector represented by the solid red line is a 2,150 cc/min unit. It displays the injectors true behavior in the nonlinear range (solid red line) in comparison to what the ECU assumes (dotted red line).
To better understand how to read the graph, von Holten laid out a linear range operating scenario for us. “Let’s say you’re in a part throttle acceleration situation,” proposes von Holten. “After determining airflow, the ECU calculates that the engine will need 200 cc/min of fuel to run properly. To find the effective IPW, we would then trace 200 cc/min line across the X-axis until it intersects with the dotted red line. You will see that to get that flow rate, the injector would need an effective pulse width of 1.6 ms.”
Following the 200 cc/min flow line on the X-axis will eventually intersect with the linear line at an effective pulse width of 1.6 ms on the Y-axis. (2,150 cc/min injectors)
“Now we know we need an effective pulse width of 1.6 ms, the ECU will then look at the voltage offset table at 14 volts and 43.5 psi (3 bar) of fuel pressure and see that the correct offset value is 0.68 ms,” says von Holten. “The voltage offset is then added to the effective pulse width to give us a commanded IPW of 2.28 ms.”
An example of the voltage offset values provided on Fuel Injector Clinic’s Data Match Technology sheet. We highlighted the correct offset value in red.
Interpreting The Data
The pulse width adder tables become important as we move further down into the nonlinear range, where the injectors’ behavior becomes more erratic. “Your engine is operating in the idle range and requires 30 cc/min of fuel to run properly,” von Holten theorizes. “The ECU will then follow that linear dotted line and see that it needs to be commanding a 0.23 ms effective pulse width to get that flow rate. It then adds the 0.68 ms voltage offset value and thinks it’s commanding the correct flow rate, when in reality the injector isn’t even firing anymore.”
For 30 cc/min of fuel at idle, the ECU will follow the linear dotted line and see that it needs to be commanding a 0.23 ms effective PW to get that flow rate; when in reality the injector isn’t even firing anymore (2,150 cc/min injector).
“With the discovery of these short pulse width adder tables, we are now able to trace the low pulse width behavior of an injector and model its true operating curve up until the flow rate becomes linear,” explains von Holten. “This feature is what provides better drivability by implementing more refined fueling control.”
Graph showing the low pulse width operating range of a 2,150 cc/min fuel injector. The linear dotted red line represents the ECU’s expected flow rate and IPW, the solid line is the injectors actual measured flow and pulse width, and the blue line represents the adjust dotted line after adding the pulse width compensation required to deliver the correct amount of fuel.
For some of the company’s more popular applications, the Fuel Injector Clinic team did the heavy lifting and plotted this data for their entire line of injectors, along with the standard flow and offset information, into software and ECU-specific formats; such as HP Tuners for GM ECUs and Cobb Accesstuner for most Subarus.
It should be noted that in order to download the Accesstuner software, Cobb requires you to at least have a basic understanding of ECU calibration. Which involves taking a Cobb-specific online tuning course through EFI University, but it’s a worthwhile investment.
The GM ECU-specific short pulse width compensation and minimum IPW data for Fuel Injector Clinic’s 2,150 cc/min injector, formatted for HP Tuners.
Up until the discovery of these compensation tables, the most common technique used by tuners to wrangle in higher flowing injectors was somewhat rudimentary and didn’t work very well for streetcars. “Before these pulse width adder tables, tuners would just use the ‘global’ latency value to tune the short pulse width area of the engine if they needed; but they only had one latency value to mess with,” says von Holten. “If you were trying to fix a fueling issue at idle and increased the latency by 0.25 ms to get near where the injector actually starts firing, you would then end up adjusting the entire commanded fuel table by that amount and be over fueling everywhere else.”
Finding The Outliers
The research and development team at Fuel Injector Clinic has been hard at work testing a groundbreaking new matching strategy that no one else in the injector aftermarket has yet to achieve. Matching injectors in the non-linear low IPW operating range to remove the flow rate variances created by this low pulse width response. The short pulse width adder table mentioned above might have corrected the flow for the “average” injector, but in this area injectors also vary from one another, and fixing that is where our next level of injector set matching will be at. No longer just matching sets in the linear zone to 1-percent in dynamic flow and 2-percent in offsets.
An example graph showing the short pulse width curves of ten different fuel injectors of the same size, which were matched in the linear operating range. Fuel Injector Clinic’s standard matching strategy involves removing any outliers that vary by more than 1-percent in flow and 2-percent in offset at their linear dynamic flow zone. However, this strategy alone doesn’t correct for mismatches from one injector to another in the nonlinear range, causing fueling to become more erratic when in this operating area.
“With the technology and strategies we are using when matching current injector sets, it’s too time consuming to get usable information at shorter pulse widths because there are so many data points when you get that low,” explains von Holten. “When we provide injector data points (like in the graph above), each point is made up of an average of 400 to 500 individual pulses, instead of the usual 15 to 20 points typically needed for each injector. So this is really directed toward the R&D crowd right now, but we’re actively testing to see if we can bring this strategy into our production matching.”
Comparing Real World Performance Using Both Matching Strategies
The graph below shows the operating curves of the four 2,150 cc/min injectors Fuel Injector Clinic sent us using their standard linear area matching strategy. When coming down in pulse width, at roughly 0.557 ms, the injectors behavior begins to noticeably vary from one another, worsening as the pulse width gets shorter.
The operating curves of the four 2,150 cc/min injectors matched using the company’s standard approach.
With this first massive set of 2,150’s that were matched using Fuel Injector Clinic’s standard linear matching, we had difficulty getting our Evo to idle steadily on E85; even with the pulse width adder table modeled, the minimum IPW set, and forced to idle in open loop to prevent the ECU from making corrections. The injectors were extremely sensitive to any changes and produced a gasoline scaled air/fuel ratio that oscillated between 11.8:1 and 13.5:1, with random spikes as lean as 15.5:1 which would cause misfires.
Data log showing the commanded IPW (effective PW + voltage offset), injector duty cycle and air/fuel ratio of the first injector set using the standard linear matching process. Notice how much the air/fuel ratio varies even though the IPW stays fairly static at 1.1 ms. This led to less than perfect street drivability and even misfires on rare occasions.
“As seen with the first set of 2,150 cc/min injectors we supplied, even though the injectors were matched nicely in the linear range, we can still see some issues at idle because we are dipping well into the nonlinear range, hence the rich idle and wide variance in air/fuel ratio,” Jilg elaborates. “With the injectors varying so much from one another in this range and the ECU using the same pulse width adder value for all four injectors, it’s not surprising we had such varied results.”
The second set of 2,150 cc/min injectors that we were supplied with for this comparison were specifically matched through the short pulse width operating range, with a heavy time investment from the Fuel Injector Clinic’s team, which provided much better results.
The operating curves of the four 2,150 cc/min injectors matched using Fuel Injector Clinic’s short pulse width matching strategy.
“The second set of 2,150 cc/min injectors we provided were matched in the nonlinear range,” says Jilg. “This process took roughly 20-hours with our current equipment and yielded much better results for the Evo. While the idle air/fuel ratio was still rich, it was much more stable with a variance of only ±0.3-points.”
Data log showing the commanded IPW (effective PW + voltage offset), injector duty cycle and air/fuel ratio of the short IPW matched set of 2,150 cc/min fuel injectors. Notice the steady air/fuel ratio even as IPW varies between 1.0 ms and 1.1 ms frequently. While still idling rich due to their large size, the car drove much more predictably and smoothly around town and at idle.
Since Fuel Injector Clinic had also provided us with the option of using their “smaller” 1,650 cc/min injectors for our Evo X, we opted to stick with those. As expected, the 2,150 cc/min injectors were far too large for our 2.0-liter four-cylinder, creating an extremely rich idle and forcing us to idle in open loop. Whereas with the 1,650 cc/min being a smaller unit, they can be ran in closed loop and with the pulse width adder table set up properly they even behave similar to injectors half their size.
“The final results we saw with the 1,650 cc/min injectors show just how important proper injector sizing is,” Jilg states. “Even though the 1,650s were not matched in the nonlinear area, they still performed much better at idle than either set of 2,150’s because they are smaller and were not being commanded to go as far into the nonlinear short pulse width range on this small engine.”
Fuel Injector Clinic’s 1,650 cc/min injectors installed in our Evo X.
Fuel Injector Clinic was able to collect valuable information from this test that will be used to further refine their injector matching process and better the industry as a whole. As for what testing strategies Fuel Injector Clinic will ultimately implement, we will just have to wait and see as they collect more data in the coming months.
“This was just a small part of the R&D we’ve been pouring our lives into to figure out how to make this process as effective and accurate as possible,” says von Holten, “We want our large and small injectors to be even more user friendly for the enthusiast than they already are, more street friendly, and to provide the most accurate and thorough injector data for the tuner, getting you off the dyno much faster and with better drivability than ever thought possible.”