EFI Tuning - Websites/Articles/etc for the novice tuner?

[Goody]

Hi all,


I am with the University of North Texas, the second car for the University is just shy of completing manufacturing and beginning testing season [Lincoln bound]. We have started to work on the basics of the engine tune as we wait for our differential to arrive. Cold starts, hot starts, throttle response, holding rpm, etc, etc just getting it happy without a load since that is the best we can do at this time.

We have about two hours of run time on our rebuilt F4i, the engine starts without fussing with the throttle, returns to idle well from sharp throttle closing, throttle response could be better, etc. I now have two hours of EFI tuning experience, as well as the other member on our team that's working with me on trying to get the engine to run well, so we are very new to this! Both of us do have pretty strong roots with carburetor/distributor tuning [bikes, muscle cars, boats, etc] which has been helpful along with a base-tune [unrestricted, gasoline, stock F4i] from Performance Electronics.

What we are looking to see is if there are any recommended websites, books or links for tuning with a restrictor and also stand-alone tuning in general [extra points for covering E85]. While the manual from PE covers what the various compensations do for us, certainly there is a methodology or flow-chart for tuning to help ensure we chase our tail as little as possible. The shop we are getting help from with Dyno Tuning is able to help us dial in the tune, but we would like the car running well prior to taking it in to maximize our efficiency on the Dyno. We are expecting to have the car on the ground in full race-trim, short of body panel paint by next weekend. In the mean time, we'd like to get more familiar with the software and work on getting the engine to run a little sharper while we have "down-time".

Thank you!

[Kirk Feldkamp]

There was a thread about the basics recently... http://www.fsae.com/forums/showthread.php?11661-Time-required-on-Dynamometer

-Kirk

[Danny Sims]

Greg Banish's books gave me a good grasp on the basics when I started, although it's fairly OEM ECU based - http://www.amazon.com/Greg-Banish/e/B001JOVJT2

When we switched to E85 I did a bit of digging for a cheat sheet but never found much useful, would be interested in the same.

[Jonny Rochester]

Basically you need a wideband oxygen sensor in the exhaust, or up the tailpipe with a lambda or air/fuel ratio readout. Without this you are blind.

Don't bother with a closed loop system or auto learning mode, and don't even bother getting a air/fuel reading on your ecu/laptop. What is quickest and most effective is to have a separate stand-alone air-fuel sensor and display system.

You don't even need a dyno. The most important thing at a dyno room is not the dyno itself, but the air/fuel ratio display. You can tune your car yourself just with this, and then drive the car with the display sticky taped to the dash somewhere (I guess that is difficult for fsae).

So, 5 wire O2 sensor up tailpipe or screwed in exhaust, plus display system. I have an Innovate LT-1 myself, but that may be an old model. Check the Innovate website anyhow.

Then it is some rules of thumb to tune your ratios, if your fuel maps allow you to tune each instant. Maybe you are after 14:1. Or 15:1 at idle, 14:1 at light load, 13:1 at full load and high rpm.

Are you able to put up some screan shots of the ignition map? and some fuel maps?

[Goody]

Thank you for the input.


As far as the wideband goes, we are using Performance Electronics Wideband Conditioner and their Bosch sensor after our 4-2-1 collector. However, I do not have a means to monitor it except for the laptop. I am assuming the laptop to ECU connection is too slow to properly monitor in real-time? If so, wouldn't I be able to log a pull and use the log-file from the ECU to tune off of [without a live display]? As mentioned, this is new to me so I may be mistaken.

Our current tables are on the laptop in the shop, they're slightly tweaked from what PE provided us with in their base-tune to account for the E85, larger injectors, restrictor and piston change as much as we have had time to mess with it [no load]. I will post them tomorrow.

[Kirk Feldkamp]

Your laptop is *significantly* faster than any engine controls or sensors on your car. The fastest way to tune things accurately is going to be to use the logging function. I suppose you don't "need" logs, but life would suck without them. The only time I've found a live display for lambda to be useful is when a pull is being made on the dyno (to be able to stop a run if something is wrong - not to tune!) and for tuning idle. You can't keep up with a transient engine for tuning.

As far as how the tune looks, ultimately, testing with your exact setup is the only way to find out if it's right for your exact setup. Base maps are only useful to get your engine started, running, and in the approximate ballpark. Intake and exhaust details, as well as a few other factors with FSAE engines (plenum size, restrictor details, etc.) will all have significant effects on what your maps end up looking like. The calibration methodology I described in that other thread is pretty straightforward, and shouldn't take you a ton of time to work through assuming you have the right tools available.

[Andrew Richards]

What is used for your fuel map in the PE controller? Is it a map that is Volumetric Efficiency vs speed vs load or is it a map with injector on times vs speed vs load? If it is a VE map than guess high throughout the map if it is injector on times then build yourself a little calculator in excel to input VE and get out an injector on time and guess high in there. Then start your tuning. Go conservative on spark, in case you didnt know here are some pointers on spark:
more engine speed->advance
higher engine load->retard
higher coolant temp->retard
hotter inlet air temp->retard
richer air fuel->advance
For your fueling get a read out like the others said and get a big parking lot. Try to do steady speeds and loads (starting light) and see what your air fuel stabilizes at. Take it slow and be patient. Log data, analyze, make changes repeat. You should be logging your exhaust gas temps while you are doing this (as well as other obviously important engine operating parameters) start enriching when you get hot (I dont like to see anything above 1550). Johnny said something about 15:1 idle and 13:1 at high load, I've never done any e-85 tuning but I do know that typically E-85 has a stoich around 9.7...Also you want to make sure that your air fuel ratio sensor is configured for the right constitutents, you said you are using a PE sensor, I have never used one so I do not know if you can change the settings, if not then just use lambda, it will be accurate enough. After steady stuff is taken care of then work on the transients i.e. power enrichment and decel fuel shut off. The only way to get these right is to drive it and monitor the air fuel. Logging data here is key. Try to maintain your air fuel ratio through out the transient loading. You do this and you will be in good shape.
Good Luck

[coleasterling]

Mike, here's an easy way of doing the physical sweeps with the car(not theory) in your test facility/parking lot.

1. Make sure the ECU is set up to log AFR, TPS, and RPM during each run (assuming you're tuning in alpha-n).

2. Set your throttle stop while your laptop is connected to only allow a certain percentage throttle travel, say starting out at around 20-30. You can start in the higher ranges that you'll use more, but it is a little tricky at first to coordinate Step 3 if you start out with it more open. Basically just start low and get a feel for it.

3. Start logging and then hold the brakes and then put the throttle to the stop and hold it there. The goal is to get the engine loaded as much as possible without killing it, then slowly let up on the brake so that RPM's build and you get (as much of a) full RPM sweep as you can at that throttle %. We typically do that 4-5 times in one "run" to make sure we get a good sweep.

4. Download the data and look at the RPM to find where your sweeps start and pull that out along with the AFR and TPS. Throttle % should be pretty constant over the RPM range and you should see a nice, slow slope building from low RPM to your rev limiter. You should be able to correlate the AFR to a specific cell/range of cells in your map and start making generic changes from there.

[Andrew Richards]

^^^Throttle Stops Are A Good Thing.....

[Jonny Rochester]

As for 14:1, that is for ~normal petrol, sorry. E85 will be different.

Something not mentioned much yet, accelerator pump. This takes care of a lot of transient throttle stuff. Don't be scared to give it heaps.

After tuning a few main points on you fuel maps when the rpm is stable, it then takes a human eye to go over the maps and smooth them out, give yourself some artistic licence, but make sure you don't change the known tuned points. This is also something a dyno operator does not have time to do. They use their time to cover the basics, but if you have time to smooth out the maps (without making it worse) then you are less likely to get surprises later.

Also, note that your load input can be either MAP (plenum vacuum) or TPS (throttle position). I have chosen TPS as my load on a few engines with success, and this is known to work well when you have 1 throttle per cylinder and may or may not have an actual plenum.

[Kirk Feldkamp]

TPS is known to work well... pretty much always. You're not going to find it on an emissions vehicle, but it's damn handy and easy to use on pretty much all race cars... yes, even turbo cars. ;) MAP is great if you have some time and a good load-control dyno. It can be spot on for fueling in all kinds of situations, but will usually take longer to calibrate than a simple TPS setup. Either way, yes, you're going to want TPS based "accelerator pump" compensation for quick throttle transients. As long as you have proper MAP and manifold air temp (AT) comps in place, you shouldn't have any problems whatsoever compensating for changes in ambient pressure or temperature. MAP compensations are linear 0 Kpa absolute = -100% in fuel, 100 kpa (close enough) = 0% compensation, and 200kpa = +100% in fuel, and so forth (if you're going to higher boost levels). Simple density change based on the ideal gas law. For air temp, I use the following table (also based on the gas laws):

(deg C) -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
(% trim) 29 24 19 14 10 5 2 -2 -5 -8 -11 -14 -16 -18 -21 -23 -25 -27 -29 -30 -32 -33 -35 -36 -38 -39

It's important that you have these two compensations set BEFORE you start tuning, as any changes to them will effect the tuning on your main fuel map. Basically, they get "baked in" to the map, and are integral to the shape of the map. Much like how the type of fuel pressure regulator (atmospheric reference, manifold reference, mechanical (no reference), etc) will get "baked in" to how your fuel map looks and operates. If the basic density compensations are changed after a tune has been developed, you may have to retune significant portions of your fuel map to compensate... it all depends on how severe the change to the comp table was. Once I've set these two comps up, I don't touch them ever again!


-Kirk

[Jay Lawrence]

Note that the "accelerator pump" mentioned is referred to as "acceleration enrich" and correspondingly "deceleration enlean", at least by MoTeC anyway. The enlean is particularly useful for saving fuel (and avoiding any over run when going from on throttle to off)

[Jonny Rochester]

I think "deceleration enlean" is more often just fuel cut. As in fuel is cut completely on deceleration. I think most road cars have this, and it's a feature I was using in Adaptronic. (However during a hill climb I turned off the fuel-cut feature, as I wanted to blow flames for the spectators.)

[murpia]

Also, note that your load input can be either MAP (plenum vacuum) or TPS (throttle position). I have chosen TPS as my load on a few engines with success, and this is known to work well when you have 1 throttle per cylinder and may or may not have an actual plenum.
I would agree that TPS can be easier to get going with, but if you can go with MAP then this is more robust.

Small variations in butterfly angle can create large airflow differences. If your TPS sensor -> butterfly angle goes off by a few % you can get large fuel errors. MAP will be pretty much immune to such changes. Pretty much all the other error factors (baro, back pressure, inlet blockage changes, cam timing slop etc. etc.) will affect both TPS and MAP equally.

Individual throttles & no plenum setups are no exception. And will always be hard to setup. Balancing the fuelling across multiple cylinders with different airflows is difficult. Individual exhaust runner wide-band lambdas are the way forward. A lot depends on how much you care about part-throttle fuelling accuracy.

Often discussing this topic provokes a flame war, let's hope not today...

Regards, Ian

[mdavis]

After tuning a few main points on you fuel maps when the rpm is stable, it then takes a human eye to go over the maps and smooth them out, give yourself some artistic licence, but make sure you don't change the known tuned points. This is also something a dyno operator does not have time to do. They use their time to cover the basics, but if you have time to smooth out the maps (without making it worse) then you are less likely to get surprises later.

As a note, the PE ECU that they're using should have built in smoothing functions. So the user can get a few known good points (as you suggest) then highlight the cells between them, click smooth, and done. It's a pretty awesome feature, IMO, but I'm a suspension guy, so take that comment for it's lack of worth.

-Matt

[Goody]

What is used for your fuel map in the PE controller? Is it a map that is Volumetric Efficiency vs speed vs load or is it a map with injector on times vs speed vs load? If it is a VE map than guess high throughout the map if it is injector on times then build yourself a little calculator in excel to input VE and get out an injector on time and guess high in there. Then start your tuning. Go conservative on spark, in case you didnt know here are some pointers on spark:
more engine speed->advance
higher engine load->retard
higher coolant temp->retard
hotter inlet air temp->retard
richer air fuel->advance
For your fueling get a read out like the others said and get a big parking lot. Try to do steady speeds and loads (starting light) and see what your air fuel stabilizes at. Take it slow and be patient. Log data, analyze, make changes repeat. You should be logging your exhaust gas temps while you are doing this (as well as other obviously important engine operating parameters) start enriching when you get hot (I dont like to see anything above 1550). Johnny said something about 15:1 idle and 13:1 at high load, I've never done any e-85 tuning but I do know that typically E-85 has a stoich around 9.7...Also you want to make sure that your air fuel ratio sensor is configured for the right constitutents, you said you are using a PE sensor, I have never used one so I do not know if you can change the settings, if not then just use lambda, it will be accurate enough. After steady stuff is taken care of then work on the transients i.e. power enrichment and decel fuel shut off. The only way to get these right is to drive it and monitor the air fuel. Logging data here is key. Try to maintain your air fuel ratio through out the transient loading. You do this and you will be in good shape.
Good Luck

The tables are setup as TPS vs RPM for fuel and ignition. I have included snap-shots of what it is currently running on, since that is what was requested earlier.

http://i29.photobucket.com/albums/c298/goodysgota72/ScreenShot2014-04-15at93718AM.png

I was about to get logging sorted out/setup yesterday for the basic inputs and am now able to have a graphical live display as well as logging data. We are hoping to have a good amount of parking lot testing prior to taking it to the dyno, only to use the dyno for squeaking the last bit of power out of it.



Mike, here's an easy way of doing the physical sweeps with the car(not theory) in your test facility/parking lot.

1. Make sure the ECU is set up to log AFR, TPS, and RPM during each run (assuming you're tuning in alpha-n).

2. Set your throttle stop while your laptop is connected to only allow a certain percentage throttle travel, say starting out at around 20-30. You can start in the higher ranges that you'll use more, but it is a little tricky at first to coordinate Step 3 if you start out with it more open. Basically just start low and get a feel for it.

3. Start logging and then hold the brakes and then put the throttle to the stop and hold it there. The goal is to get the engine loaded as much as possible without killing it, then slowly let up on the brake so that RPM's build and you get (as much of a) full RPM sweep as you can at that throttle %. We typically do that 4-5 times in one "run" to make sure we get a good sweep.

4. Download the data and look at the RPM to find where your sweeps start and pull that out along with the AFR and TPS. Throttle % should be pretty constant over the RPM range and you should see a nice, slow slope building from low RPM to your rev limiter. You should be able to correlate the AFR to a specific cell/range of cells in your map and start making generic changes from there.

Thanks Cole

So for step 3, we're slipping the clutch/loading the engine I assume?


As for 14:1, that is for ~normal petrol, sorry. E85 will be different.

Something not mentioned much yet, accelerator pump. This takes care of a lot of transient throttle stuff. Don't be scared to give it heaps.

After tuning a few main points on you fuel maps when the rpm is stable, it then takes a human eye to go over the maps and smooth them out, give yourself some artistic licence, but make sure you don't change the known tuned points. This is also something a dyno operator does not have time to do. They use their time to cover the basics, but if you have time to smooth out the maps (without making it worse) then you are less likely to get surprises later.

Also, note that your load input can be either MAP (plenum vacuum) or TPS (throttle position). I have chosen TPS as my load on a few engines with success, and this is known to work well when you have 1 throttle per cylinder and may or may not have an actual plenum.

The acceleration and deceleration compensations are currently enabled on our tune. We haven't messed with them much while drivetrain is stranded, the engine doesn't have the best low speed throttle response, but does seem to return nicely from high rpm/abrupt throttle closing to idle with the decel enabled. There's only so much we can do without being able to drive the car but we do plan to utilize both of those features.

http://i29.photobucket.com/albums/c298/goodysgota72/ScreenShot2014-04-15at93649AM.png

As mentioned by someone else within the thread, "smoothing" is an option on the PE3. While we haven't messed with it yet, we should probably be changing our methods of adjusting the table. As for now, we have taken the base tune and blanket leaned/richened all of the values to see it's affects on the engine. Some have mentioned to pick good known values and then interpolate from there, so our best way to adjust the tune would be to pick certain rpm ranges to tune and smooth in-between those instead of blanket tuning all values, correct?


TPS is known to work well... pretty much always. You're not going to find it on an emissions vehicle, but it's damn handy and easy to use on pretty much all race cars... yes, even turbo cars. ;) MAP is great if you have some time and a good load-control dyno. It can be spot on for fueling in all kinds of situations, but will usually take longer to calibrate than a simple TPS setup. Either way, yes, you're going to want TPS based "accelerator pump" compensation for quick throttle transients. As long as you have proper MAP and manifold air temp (AT) comps in place, you shouldn't have any problems whatsoever compensating for changes in ambient pressure or temperature. MAP compensations are linear 0 Kpa absolute = -100% in fuel, 100 kpa (close enough) = 0% compensation, and 200kpa = +100% in fuel, and so forth (if you're going to higher boost levels). Simple density change based on the ideal gas law. For air temp, I use the following table (also based on the gas laws):

(deg C) -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
(% trim) 29 24 19 14 10 5 2 -2 -5 -8 -11 -14 -16 -18 -21 -23 -25 -27 -29 -30 -32 -33 -35 -36 -38 -39

It's important that you have these two compensations set BEFORE you start tuning, as any changes to them will effect the tuning on your main fuel map. Basically, they get "baked in" to the map, and are integral to the shape of the map. Much like how the type of fuel pressure regulator (atmospheric reference, manifold reference, mechanical (no reference), etc) will get "baked in" to how your fuel map looks and operates. If the basic density compensations are changed after a tune has been developed, you may have to retune significant portions of your fuel map to compensate... it all depends on how severe the change to the comp table was. Once I've set these two comps up, I don't touch them ever again!


-Kirk

Thank you for the input on the compensation tables, I was a bit in limbo on how to deal with those from tuning. "Set it and forget it" sounds good to me!

http://i29.photobucket.com/albums/c298/goodysgota72/ScreenShot2014-04-15at93945AM.png

One other compensation factor I noticed in the base-tune is battery voltage compensation. It seems to be a pretty heavy alteration of injection timing. PE had it set from 8v @ 1.66ms to 16.0v @ 0.52ms decreasing linearly. That seems like a pretty heavy alteration, I assume this is trying to keep up with the load applied from the stator to keep the car running?


I would agree that TPS can be easier to get going with, but if you can go with MAP then this is more robust.

Small variations in butterfly angle can create large airflow differences. If your TPS sensor -> butterfly angle goes off by a few % you can get large fuel errors. MAP will be pretty much immune to such changes. Pretty much all the other error factors (baro, back pressure, inlet blockage changes, cam timing slop etc. etc.) will affect both TPS and MAP equally.

Individual throttles & no plenum setups are no exception. And will always be hard to setup. Balancing the fuelling across multiple cylinders with different airflows is difficult. Individual exhaust runner wide-band lambdas are the way forward. A lot depends on how much you care about part-throttle fuelling accuracy.

Often discussing this topic provokes a flame war, let's hope not today...

Regards, Ian

Thanks Ian, we noticed the throttle position variation to alter how the engine runs quite a bit. As we set the butterfly opening to maintain idle we quickly ran into the position and sensor calibration discrepancy. Once we reset the sensor calibration for the new closed position, it ran substantially better and much more stable holding an rpm.

[Kirk Feldkamp]

The 0 psi value for the MAP comp should be zero, not 100... but you'll probably never get there even on a massive lift, but it's worth making it correct.

Battery voltage comp is used to compensate for the reduction in injector fuel flow due to a drop in supplied voltage. The easiest explanation is that the pintle (or discs) will open slower and less far if the voltage is lower than expected, and vice versa. Most injectors are flowed at either 12V or 13.5V or whatever (doesn't really matter, as long as you know what you're looking at). Like you say, it's a pretty heavy compensation, which tells you how important it can be to supply the expected voltage while the engine is running. Injector Dynamics or Yaw Power (I never quite understood the relationship) used to offer a service to give you a 3D map of an injector's comp characteristics, but I don't think they do that anymore. You can still find their info documents via The Google though (I think). The info is pretty interesting if you have ever wondered about injector linearity (hint: most are pretty bad at small openings). As far as what you should do going forward, if PE has those values in there for that engine, maybe they've determined those settings are good for what you're working with. Did you change injectors? Most 600 sportbike injectors are already way oversized (they're usually the same size as their bigger siblings!), so you probably don't need to go bigger even going to E85. Not sure of your exact setup, so don't blindly take my word for it. Only testing and a couple quick calcs will tell you what you can get away with.

-Kirk

[coleasterling]

Thanks Cole

So for step 3, we're slipping the clutch/loading the engine I assume?



Ya. It'll take a bit of playing to get a feel for when to start letting up on the brake vs. not stalling and not slipping the clutch too much.

EDIT: I forgot to say that you can do it from a roll, too. Start rolling, then get on the brakes without touching the clutch, keep rolling slowly, and increase throttle to the stop while increasing braking force to keep RPMs down. Keep it right above stalling and then continue the sweep to the rev limiter. That may be a bit easier if you have the space.

[murpia]

The 0 psi value for the MAP comp should be zero, not 100... but you'll probably never get there even on a massive lift, but it's worth making it correct.

Battery voltage comp is used to compensate for the reduction in injector fuel flow due to a drop in supplied voltage. The easiest explanation is that the pintle (or discs) will open slower and less far if the voltage is lower than expected, and vice versa. Most injectors are flowed at either 12V or 13.5V or whatever (doesn't really matter, as long as you know what you're looking at). Like you say, it's a pretty heavy compensation, which tells you how important it can be to supply the expected voltage while the engine is running. Injector Dynamics or Yaw Power (I never quite understood the relationship) used to offer a service to give you a 3D map of an injector's comp characteristics, but I don't think they do that anymore. You can still find their info documents via The Google though (I think). The info is pretty interesting if you have ever wondered about injector linearity (hint: most are pretty bad at small openings). As far as what you should do going forward, if PE has those values in there for that engine, maybe they've determined those settings are good for what you're working with. Did you change injectors? Most 600 sportbike injectors are already way oversized (they're usually the same size as their bigger siblings!), so you probably don't need to go bigger even going to E85. Not sure of your exact setup, so don't blindly take my word for it. Only testing and a couple quick calcs will tell you what you can get away with.

-Kirk

Kirk: do you have a reference for the reduced flow effect of low battery voltage? I never heard that before, I understood the preload of the pintle return spring pretty much made for a two position valve.

An injector is a solenoid coil with inductance & resistance, so lower battery voltages mean the coil takes longer to build up a magnetic field: dI/dt for an inductor. So the fuel quantity is made up of two components, a battery-voltage related component and a constant component. Most ECUs split the calibration up, but like the other terms mentioned above, errors in the battery compensation can get 'baked into' the main maps, causing issues.

There are some tricks that can be done to work this out, best done early in a tune. You will need to be able to vary the voltage, a reasonable lab power supply should be up to the job of supplying ECU and injector current with a small wiring change. Ignition and starter can remain off the battery.

At idle, pulse widths are low, so opening time is a reasonable percentage of the total. Sweeping the voltage at idle will show up errors in the cal as variation in measured lambda (richer means too long)

If your ECU supports both sequential and batch mode, alternating between them halves / doubles the number of injector openings per engine cycle. Errors in the opening time cal will show up as richer (for too long) when switching to batch mode.

Finally, a pretty good measurement is to take a 0.1ohm shunt resistor and a scope and just measure the current build-up time. A rule of thumb is that the injector flows half the fuel for the time taken to reach 90% of saturated current.

This applies to 'high Z' injectors which are typical for 99% of non-DI applications these days. As Kirk says, your injectors may be oversized if stock & used with the 20mm restrictor. Oversized means short fuel pulses which means the need for good battery compensation is relatively more important.

Regards, Ian

PS closing time is usually a non-issue with a modern ECU which 'clamps' the injector flyback voltage at something much higher than 12V. Old ECUs with recirculation diodes may be a bit lazier at closing the injector. In which case you may need to factor this into the compensation cal.

[Kirk Feldkamp]

Oh snap. You're right. I don't know what I was thinking. The battery voltage comp in the Motec is a value in microseconds. In the Motec scheme, "The table value is a plus or minus time offset measured in microseconds and is applied to the final Injector Pulse Width." That makes way more sense that way for a PWM circuit. Sorry for the confusion.

Also, on the topic of short injection pulses... batch will effectively halve your pulse widths versus sequential, so if you're oversized already and not using sequential, idle quality and stability can be tough. If you find those info documents i mentioned before, there's a really good explanation of why.

-Kirk

[Jay Lawrence]

Just a note on your injectors: you said you were running an F4i, and if I remember correctly this is a 4 pintle injector. We ran these at some point but they didn't flow enough for turbo tuning, so we upgraded to more modern 12 pintle ones (from a CBR1000RR). I think the team currently runs standard CBR600RR injectors, which I believe are 12 pintle. So, if you are running the old 4 pintle ones, you may come across flow rate issues on E85. I'd upgrade to the 12 pintle CBR600RR ones anyway for better spray pattern and atomisation.

[Loz]

Kirk: do you have a reference for the reduced flow effect of low battery voltage? I never heard that before, I understood the preload of the pintle return spring pretty much made for a two position valve.


It is something that is very difficult to measure in car and really requires off car testing of the injectors (i.e. flow bench with different voltages) and some intimate knowledge of the injector mechanics and solenoid if any more than derived information is going to be determined. The information doesn't get shared a lot as is it is most often based on proprietary information.

As Kirk describes the effect isn't exactly a reduction in flow as the flow rate is a mechanical function of fuel pressure and injector nozzle/pintle orifice opening.
Rather it is really it is a reduction in total volume per injection event as the effective (real/fuel flowing) duty cycle (i.e. the injector open time when fuel is actually flowing) decreases due injector mechanical lag (dead time). As the voltage drops, the injector dead time increases (i.e. the time from when the ECU sends the injector pulse to the time it actually opens) which takes up a larger proportion of the duty cycle the ECU is requesting of the injector. This happens because of the reduced dI/dt (slower opening hence longer time period to reach the active open time) in addition to a reduction in solenoid magnetic field force (may not open as far or as effectively).
These effects are what the voltage compensations try to over come by extending the pulse width such that the same volume of fuel is injected during a given injection event.

I have a plot somewhere I will try to find of an injector flow bench test I did for injector flowing time (and fuel) vs voltage for a range of pulse widths showing the effects of low battery voltage. It really highlights the vast difference in actual flow vs expected flow when voltage is low and how it changes with pulse width. Essentially for low pulse the change to opening time and dead time is much more significant on volume flowed per injection event. This becomes especially evident at low rpm and low duty if there is a voltage drop and the engine most often stops running as it runs lean. The effects can be disguised by other compensations and it can be hard to pinpoint as an issue if the battery compensations are set wrong (way off) and there is a voltage issue. This is one reason to make sure it is set (default might be okay) and then forget.

It is especially critical to get this right for single cylinder engines (with 1 injector) as all of the fuel is added in a single event per cycle.

As an example, the motec compensation typically offsets (advances/makes occur earlier) the start of the ECU injector pulse but does not change the end time of the pulse, thus the total pulse width is increased to overcome mechanical lag in the injector associated with mechanical losses, solenoid/voltage velocity changes and any change in ECU signal speed that simultaneously occurs.

[Loz]

This is a plot of the derived data from an injector flow test in which mass flow was measured at discrete injector voltages and duty cycles. Fuel mass was averaged by the number of injector open events and then normalised by the flow expected based on signal duty cycle. The reductions from unity of each data point (at each voltage) are the system lags/losses (similar to hysteresis) and include injector dead time, current and voltage reduction effects.

In the more "normal" operating voltage range (11V-14V) and at higher duty cycles (50-80%) the reduction is approximately linear. This is one reason that a battery voltage compensation for injectors usually also default set as a linear variation. The flow doesn't ever reach unity as given the finite amount of time for the injector to physically do the work (open and close).
The non linearity at low duty cycle and normalised figures of around 0.5 is due to the large percentage of lag effects relative to the length (pulse) of the total injection event in comparison to the longer events for which the lag is not as influential on mass flow.

Without a compensation, at low voltages the flow could be reduced by as much as 50% from what is needs to be (not that it would be running at 8.5V though). This can be seen in an a Battery compensation table at the low end at which upwards of 2.0ms offset might be programmed to bring the actual fuel delivered to the cylinder back into line with what is required (keeping in mind that it doesnt translate into the same increase that would occur with a 2.0ms offset at normal running voltage).

[murpia]

Thanks for the plot Loz.

The issue that I'm trying to communicate, and apologies for any repetition, is that you should not simply tune the total pulse width against battery voltage.

The injector opening time flow reduction is what is tuned against battery voltage (call it A), and then a fixed fuel flow vs. pulse width time calibration is added (call it B).

So for a desired fuel injection quantity you need to add A to B with A changing with battery voltage and A & B changing with fuel pressure & fuel temp.

Decent ECU designers know all this and should give you the tools in the calibration to make it work.

Regards, Ian

[murpia]

Oh snap. You're right. I don't know what I was thinking. The battery voltage comp in the Motec is a value in microseconds. In the Motec scheme, "The table value is a plus or minus time offset measured in microseconds and is applied to the final Injector Pulse Width." That makes way more sense that way for a PWM circuit. Sorry for the confusion.

Also, on the topic of short injection pulses... batch will effectively halve your pulse widths versus sequential, so if you're oversized already and not using sequential, idle quality and stability can be tough. If you find those info documents i mentioned before, there's a really good explanation of why.

-Kirk
Hi Kirk,

MoTeC may well have additional tuning & ECU hardware support for PWM'd and / or Peak & Hold injectors ('low Z'). I think for FSAE most hardware for novice teams will be 'saturated' 'high-Z' which is way simpler, and is what I am referring to here.

I agree with your assessment of batch vs. sequential especially with oversized injectors. But by enabling (per-engine-revolution) batch you can get an idea of how good your injector opening time compensation calibration is. Per-engine-cycle batch is just a limp-home strategy, that won't help.

Off-topic: it's worth a quick dyno test to disconnect your cam sensor with the engine running and observe the effect on fuelling & ignition. Then stop and attempt a restart. Some ECUs will retain sync, some will go batch injection / wasted spark and some will just stall the engine. Some will restart in batch / wasted some will not. Worth studying as an exercise in FMEA, and for comp. You can never know your system too well.

Regards, Ian

PS I just realised I've been working in powertrain control for 19 years now, that might actually count as 'since before you were born' for some FSAE students... And I'm still learning.

[Jonny Rochester]

Thanks for showing us your fuel table. It looks reasonable. If you get good a/f ratio with stable rpm, then its good. BUt you will need to check loaded points on the road/dyno.

It would be interesting to see your ignition table. This is the secret one that people don't always share. As the engine will work OK with a large range of ignition settings, different tuners have their own ideas. It should be tested on the dyno with knock meters etc... If the a/f ratios are OK, the right ignition advance will give you the edge (over someone else with the same engine).

[Mbirt]

Per-engine-cycle batch is just a limp-home strategy, that won't help.Hopefully no one reads this as meaning that batch fuel and wasted spark are inherently bad. All three 2013 North American formula competition 1st place Fuel Efficiency wins were by a car running batch/wasted. The same goes for the overall wins at the 2013 and 2014 Clean Snowmobile Challenge. It is worth closely examining the risk vs. reward in the custom implementation of a cam position sensor, especially in the context of Ian's FMEA study suggestion.

[murpia]

Hopefully no one reads this as meaning that batch fuel and wasted spark are inherently bad. All three 2013 North American formula competition 1st place Fuel Efficiency wins were by a car running batch/wasted. The same goes for the overall wins at the 2013 and 2014 Clean Snowmobile Challenge. It is worth closely examining the risk vs. reward in the custom implementation of a cam position sensor, especially in the context of Ian's FMEA study suggestion.
Yes, for clarity, per-engine-revolution batch & wasted spark are absolutely fine in my opinion too. And good for reliability although in my opinion there's no excuse for an ECU not to cope with loss of cam sensor by reverting to batch / wasted from sequential / sequential to keep running (with a slight performance loss).

But a per-engine-cycle batch mode rolls the dice on injection timing every time the engine starts. I suspect it's fairly rarely implemented except as limp-home or in some fuel-only or piggyback units. I think Megasquirt can be configured that way.

Regards, Ian

[Kirk Feldkamp]

Agreed. Frankly, the fancy benefits of sequential fuel and ignition (fuel timing and individual cylinder ignition timing) are often overlooked due to lack of program resources. That said, I have definitely noticed on a number of occasions that some engines simply do not like to idle cleanly using batch. I've only ever heard about sequential ignition being fully utilized on NASCAR engines, where there can (apparently) be significant differences in cylinder filling due to things like torsional crank and camshaft flex. I'm sure F1 and other higher-level pro racing engines worry about this, but I've never seen it specifically called out as something they worry about. You've got to have some pretty darn sophisticated testing equipment to capture effects like that... and you've got to already be pretty darn developed in your calibration to be able to take advantage of further refinement like that. Unless you're running into RPM related issues with your ignition system (as in it can't keep up), it seems like a complicated addition to worry about for the sake of ignition. I do, however, see plenty of benefit in sequential for the sake of fuel timing. In a competition like FSAE were economy is more than just a "nice to have", there is likely real performance (defined as points at comp) to be gained through a refinement like this. Have any of you guys ever carried out fuel timing calibration on a FSAE system? If so, I'm curious what the tangible results were.

-Kirk

[B Lewis @ PE Engine Management]

Hi Guys,

I just wanted to offer help for any FSAE teams needing assistance with tuning. I've skimmed through some of these recent posts and thought that I would address a couple of items.

Our 'go-to' standard for load indication for fueling is Alpha-N with MAP comp. The MAP comp values are standard and rarely need modification. This is very easy and intuitive like standard Alpha-N but includes a link to the 'physics' of the engine. 95% of the new installations we do start off as Alpha-N with MAP comp unless the vehicle is already outfitted with a MAF sensor. In all but high-boost applications and some 2-strokes this works very well.

We also have a pretty good database of battery voltage compensation tables for different injectors as determined by PE using the drivers in the PE3 ECU. If you are running our ECU please contact us for values. Keep in mind that many factors other than just injector design influence the required battery voltage compensation including fuel pressure and the design of the injector driver in the ECU.

[murpia]

Agreed. Frankly, the fancy benefits of sequential fuel and ignition (fuel timing and individual cylinder ignition timing) are often overlooked due to lack of program resources. That said, I have definitely noticed on a number of occasions that some engines simply do not like to idle cleanly using batch. I've only ever heard about sequential ignition being fully utilized on NASCAR engines, where there can (apparently) be significant differences in cylinder filling due to things like torsional crank and camshaft flex. I'm sure F1 and other higher-level pro racing engines worry about this, but I've never seen it specifically called out as something they worry about. You've got to have some pretty darn sophisticated testing equipment to capture effects like that... and you've got to already be pretty darn developed in your calibration to be able to take advantage of further refinement like that. Unless you're running into RPM related issues with your ignition system (as in it can't keep up), it seems like a complicated addition to worry about for the sake of ignition. I do, however, see plenty of benefit in sequential for the sake of fuel timing. In a competition like FSAE were economy is more than just a "nice to have", there is likely real performance (defined as points at comp) to be gained through a refinement like this. Have any of you guys ever carried out fuel timing calibration on a FSAE system? If so, I'm curious what the tangible results were.

-Kirk
Agreed. 'Sequential' ignition is mainly related to dwell times, i.e. at really high revs you may have to charge the coil for longer than one engine revolution. Or you need it when you have unequal firing angles like certain V or odd-numbered cylinder engines. Countless millions of engines run wasted spark just fine and assuming you get a good enough coil charge you will be fine. (Always check with a scope if you can).

Sequential injection can be a proven performance benefit through better fuel mixing. The shorter your injections pulses the more important the timing. But then the less accurate is your fuelling due to the opening time effect. There's probably a nice U shaped curve hiding in the data somewhere... 2 sets of injectors seem to be a popular solution in some engines, but probably overkill for FSAE. That helps keep the injection time short but the accuracy high.

Variable fuel pressure might be a good option, I'm not sure if there are any examples of use?

Better idle with sequential is no doubt down to the short idle fuel pulse needing to be sent in to the cylinder rather than splatting onto a closed valve. Or a similar effect relating to wall wetting.

Regards, Ian

[Kirk Feldkamp]

Variable fuel pressure... like every post-throttle referenced fuel pressure regulator? ;) In my mind, the idea there is to maintain the same pressure delta across the injector at all times, regardless of manifold conditions. In the case of low manifold pressures, the fuel rail pressure drops significantly, meaning the injector pulsewidths will be longer than if the regulator just held a constant 43.5psi (or whatever). This will absolutely get "baked in" to the maps like we were talking about before, resulting in a *significantly* different map shape when compared to say a mechanical (constant pressure) regulator. I've always thought about it as a pretty brilliant way to be able to extend the usable control range of an injector that would be too large otherwise.

On the high side of things (which is what you're talking about), I haven't heard of this for non-direct injection engines. I think that is exactly what they're doing with the fuel pump on DI engines though. That's one of the big problems with putting a generic aftermarket ECU on a DI engine... you need some special hardware to control the fuel pump pressure as part of the calibration... which obviously very few aftermarket ECU's do.

-Kirk

[murpia]

A controlled, variable fuel pressure could keep the fuel pulse more constant in the engine cycle crank angle domain, as rpm rises. That _could_ maintain a better fuel mix than allowing the pulse to stretch out. But it's just an idea.

I have a suspicion that some F1 engines did this during the port-injection unlimited fuel pressure era (pre 2005 I think it was). But that's the only possible example I'm aware of.


Anyway, thanks Kirk for pointing out the effect of the fuel pressure regulator installation. Something else for the 'novice' installer / tuner to be aware of.

Regards, Ian

[apalrd]

A controlled, variable fuel pressure could keep the fuel pulse more constant in the engine cycle crank angle domain, as rpm rises. That _could_ maintain a better fuel mix than allowing the pulse to stretch out. But it's just an idea.

Some direct injection systems do this, and virtually all common rail diesels do. On a diesel, the fuel will ignite as soon as it enters the cylinder, so the entire fuel delivery event needs to be precisely timed (start and end angles/times of multiple injection events), so they usually use number of pilots and fuel rail pressure as the primary load control instead of injection time (which is effectively constant).

I also know of at least one production PFI engine that does this.

Pulse angles matter less and less when the fuel injector is open continuously. I know at WOT at 12000rpm, our 2013 engine would run as high as 95% injector duty cycle, with 'wasted' fuel (one injection event per revolution, half mass of full cycle injection mass).