F4i crank/start issues

[Marshall.Hagen]

I don't want to leave out any information, so I appologize for writing a long story here:

I have been mapping our F4i engine with the DTA S80pro since January on our engine dyno. It always started and ran fine while attached to our engine dyno. It liked to fire with a combined pulse width of 4.5-4.7ms (stock F4i injectors). I am running 2500 microseconds of dwell on the coils. Ignition timing is 15deg in cranking cells. Base injector pulse width on the fuel table is 1.4ms, 150% cranking offset at cold start that tapers to 0% by 210F. Water temperature correction starts at 50% at 32F and tapers to 0% by 165F. The voltage/injector offset is 0.70ms for all voltage readings (I never needed to tune this).

So up until we put the engine in the car last week, the engine started, idled and ran ecxellent. Made terrific power, and had no signs of hicups.

Now the engine is in the car and it has a very hard time starting. It will crank, fire on a couple cylinders, crank, fire on a couple cylinders, and sometimes catch and idle or just repeat the crank/fire until I give up with the starter button. I can get it to fire and idle maybe 1 out of 20 times. Once the car starts and idles, it is perfect, 2800rpm idle, smooth, Revvs great, and even drives great. No signs of any problems.

The things that have changed since the engine dyno are:
- Fuel system
- Wire Harness
- Muffler (was open header routed outside building on dyno)
- All relays are different (was using stock F4i relays on dyno)

I first tried attaching our dyno fuel cell/pump to the engine while in the car, and nothing happened different. So fuel system/pump is not the culprit.

I have changed the spark plugs four times, and this is getting to be an expensive test, with minimal to no results.

I changed coils from our spare motor and no change.

I have adjusted cranking fuel from 0% which works out to be about 2.2ms with the injector offset factored in, all the way to maximum of 400%, which is roughly 7.8-8.0ms of injector pulse width. The engine seems to want to catch and fire anywhere between 3.6ms and 7.0ms. At one point, I was able to get repeatable starting results at around 5.0-6.0ms. After the engine ran for a while and it cooled down, it no longer wanted to start at that value. Frustration is setting in.

I have gone through all the options in the software.
- I have changed 'injector start pulse' from 1ms all the way to maximum of 40ms, in 2ms intervals. Seems to want to work better around 4-6ms.
- I have adjusted the 'fire tooth on startup' value to all possible values (0-5)
- I have changed the dwell from 2000 microseconds to 3500 microseconds - I don't want to go higher than this because CBR coils are fragile
- I have tried timing values between 5deg to 30deg
- I have tried 'number of teeth coils on in cranking' value between the listed 1-5.

There are ref/sync errors during cranking, but they go away as soon as the engine idles/runs. We are using the factory crank and cam sensors, so there shouldn't be any hardware issues with those, and the engine has never been opened, so cam timing should be spot on.

I am running out of ideas here. Please help.

Regards,
Marshall

[Marshall.Hagen]

I don't want to leave out any information, so I appologize for writing a long story here:

I have been mapping our F4i engine with the DTA S80pro since January on our engine dyno. It always started and ran fine while attached to our engine dyno. It liked to fire with a combined pulse width of 4.5-4.7ms (stock F4i injectors). I am running 2500 microseconds of dwell on the coils. Ignition timing is 15deg in cranking cells. Base injector pulse width on the fuel table is 1.4ms, 150% cranking offset at cold start that tapers to 0% by 210F. Water temperature correction starts at 50% at 32F and tapers to 0% by 165F. The voltage/injector offset is 0.70ms for all voltage readings (I never needed to tune this).

So up until we put the engine in the car last week, the engine started, idled and ran ecxellent. Made terrific power, and had no signs of hicups.

Now the engine is in the car and it has a very hard time starting. It will crank, fire on a couple cylinders, crank, fire on a couple cylinders, and sometimes catch and idle or just repeat the crank/fire until I give up with the starter button. I can get it to fire and idle maybe 1 out of 20 times. Once the car starts and idles, it is perfect, 2800rpm idle, smooth, Revvs great, and even drives great. No signs of any problems.

The things that have changed since the engine dyno are:
- Fuel system
- Wire Harness
- Muffler (was open header routed outside building on dyno)
- All relays are different (was using stock F4i relays on dyno)

I first tried attaching our dyno fuel cell/pump to the engine while in the car, and nothing happened different. So fuel system/pump is not the culprit.

I have changed the spark plugs four times, and this is getting to be an expensive test, with minimal to no results.

I changed coils from our spare motor and no change.

I have adjusted cranking fuel from 0% which works out to be about 2.2ms with the injector offset factored in, all the way to maximum of 400%, which is roughly 7.8-8.0ms of injector pulse width. The engine seems to want to catch and fire anywhere between 3.6ms and 7.0ms. At one point, I was able to get repeatable starting results at around 5.0-6.0ms. After the engine ran for a while and it cooled down, it no longer wanted to start at that value. Frustration is setting in.

I have gone through all the options in the software.
- I have changed 'injector start pulse' from 1ms all the way to maximum of 40ms, in 2ms intervals. Seems to want to work better around 4-6ms.
- I have adjusted the 'fire tooth on startup' value to all possible values (0-5)
- I have changed the dwell from 2000 microseconds to 3500 microseconds - I don't want to go higher than this because CBR coils are fragile
- I have tried timing values between 5deg to 30deg
- I have tried 'number of teeth coils on in cranking' value between the listed 1-5.

There are ref/sync errors during cranking, but they go away as soon as the engine idles/runs. We are using the factory crank and cam sensors, so there shouldn't be any hardware issues with those, and the engine has never been opened, so cam timing should be spot on.

I am running out of ideas here. Please help.

Regards,
Marshall

[Discretely elite]

I suggest looking at your harness and specifically your grounds. I'm no EE but grounds always seem to be a problem. Or those relays could be acting up.

I doubt anything funny happened with your tuning file. I would set everything back up exactly like you had it running on the dyno to eliminate possibilities.

Good luck

Btw, how much is 'terrific' power?

[TMichaels]

Hi,
check your sensors for crankshaft and camshaft timing. I do not know your sensorsetup in detail, but most teams use inductive sensors. Their signal is RPM dependent (@ all the other EEs: I know, this is a simplification). That's why you do have problems until reaching a RPM that is high enough to deliver a consistent signal. After checking them you should also check the tooth wheels of the sensors and of course the connection cables between ECU and sensor. Use twisted pair cables for the connection of inductive sensors, this will keep away a lot of gremlins!

Regards,

Tobi

[Ian M]

Sounds electrical.

I had a similar experience last year, and for me, it turned out to be the coils. I believe I had them set at 2500 microseconds, but I don't believe that overcharging them was the cause of them failing. I am tentatively blaming our large battery charger that we used to use to jump the car (haven't used it this year and haven't had any problems)
Anyway, since it is a CBR you can run an easy check on the coils. You should have 11-13 ohms resistance across your coil. We are running an RR, but the resistance value should be the same for an F4i coil.

We were experiencing extremely hard cold starting that progressively got worse. Once the engine was warm it would start, but was down on power. I believe that half of the time it may have been compression detonating.
Our coil resistance readings were:
10,8,8,7

Just wanted to give you the heads up on our experience. It took me forever to figure out what was wrong. We also went through a couple sets of spark plugs and I changed a whole bunch of stuff in the Motec mapping, just to finally realize that the coils were at fault.

Hope you get it worked out. I know how frustrating it is.

[TMichaels]

If possible, use a compensation table for the dwell time. It will increase when cranking as the battery voltage usually breaks in.

Regards,

Tobi

[Marshall.Hagen]

Just as an update, I'm still getting intermittent starting. Sometimes I can get it to fire within 2 seconds, sometimes never. Seems to be better when cold.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Discretely elite:
I suggest looking at your harness and specifically your grounds. I'm no EE but grounds always seem to be a problem. Or those relays could be acting up. </div></BLOCKQUOTE>

Grounds are good. The relays are operating. Starter relay engages starter, fuel pump relay primes and holds power to the pump, and fan relay operates fan. I can see if running the factory starter relay changes anything, but relays are simple switching devices, they either work or don't http://fsae.com/groupee_common/emoticons/icon_frown.gif

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Discretely elite:
I doubt anything funny happened with your tuning file. I would set everything back up exactly like you had it running on the dyno to eliminate possibilities. </div></BLOCKQUOTE>

I thought that a corrupted file could be a possibility. But I've loaded the included DTA provided basemap, as well as the two-dozen maps from various stages in my tuning. They all react somewhat the same.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Discretely elite:
Btw, how much is 'terrific' power? </div></BLOCKQUOTE>

Makes 83hp@11600 and 40lb-ft@10000-10500
SAE Correction was 1.005 that day
Measured using a Superflow SF901 water brake dyno
Dyno Graph (http://s107.photobucket.com/albums/m296/Marshmph/FSAE/?action=view&current=RUN108.jpg)

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TMichaels:
Hi,
check your sensors for crankshaft and camshaft timing. I do not know your sensorsetup in detail, but most teams use inductive sensors. Their signal is RPM dependent (@ all the other EEs: I know, this is a simplification). That's why you do have problems until reaching a RPM that is high enough to deliver a consistent signal. After checking them you should also check the tooth wheels of the sensors and of course the connection cables between ECU and sensor. Use twisted pair cables for the connection of inductive sensors, this will keep away a lot of gremlins!

If possible, use a compensation table for the dwell time. It will increase when cranking as the battery voltage usually breaks in. </div></BLOCKQUOTE>

Tobi,

The RPM signal drops to 0 during the cranking cycle (starter is still spinning engine), so we think that might be our problem. It will cycle at ~350-450RPM, but periodically drop to 0. We are running 22gauge mill-spec wire, so there is not much shielding. However, the factory Honda wire harness uses small diameter 18gauge wire, and does not have any special shielding for their coils and cam/crank sensors. We have discussed options on how to shield the wire, we are thinking of wrapping some copper wire around each of the sensors, and the coils, and then running that wire to ground.

Unfortunately, the DTA software at this time does not have a dwell compensation table, just a static value. We are working with them to get a few features added that would benefit our setup.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Ian M:
I had a similar experience last year, and for me, it turned out to be the coils. I believe I had them set at 2500 microseconds, but I don't believe that overcharging them was the cause of them failing. I am tentatively blaming our large battery charger that we used to use to jump the car (haven't used it this year and haven't had any problems)
Anyway, since it is a CBR you can run an easy check on the coils. You should have 11-13 ohms resistance across your coil. We are running an RR, but the resistance value should be the same for an F4i coil. </div></BLOCKQUOTE>

Ian,

I will put an ohmmeter to them in the morning. Running out of steam here tonight. We have switched the coils with our spare motor with no affect, so the measurement will probably be just precautionary, but thanks for the heads up regardless.

Thank you for all the suggestions everyone, please keep them coming!

[A Richards]

Whats the voltage at your ECU during cranking?? It could be just turning on and off due to a voltage drop, as your wiring has changed. Our Motec drops RPM to zero when it gets below 10.5 volts at cranking.

[TMichaels]

Hi Marshall,
you do not really need to shield the wire, just use twisted pair or twist your existing cable. Check if your sensorcables run close the the startercable or other high-current/pulsed current wires.

Check if there is metal swarf on your sensor, since inductive sensors do work with magnets.

Also check the distance between your sensor tooth wheel and the sensor. The closer the better.

We went through all this stuff in 2006 just one week before the competition. It was a faulty tooth wheel in our case.

Regards,

Tobi

[Kirby]

I will reiterate what others have said.

Sounds like a REF/SYNC error.

I would recommend hooking up a scope to your sensors during cranking and see if any noise is on the line. Remember f4i are Variable Reluctance, the output voltage increases with RPM to somwhere around +-20V.

If you are using mil-spec wire, it should come in a sheilded variant, what spec are you using? is it MIL-22759/16 or MIL-27500/18 or something else?

As an (temp) EE. I would certainly recommend sheilding all your sensor lines (REF/SYNC/IAT/EWT/OP) etc. as they should all be hooked to the sensor ground on the ECU. if you start to generate noise in any of the wire runs (even non critical) it will show up at the ground of the critical ones. sink your sheilding to your battery ground.

Shield your control lines for the coils, and the wires to the coils themselves, massive inductive fields like these will DESTROY YOU!

Last, have a muti-meter on your battery and determine the drop when you crank.

good luck, have fun!

[Davidimurray]

We had a similar problem on our dyno a couple of years ago with the DTA. Turned out to be electrical noise on the ref/sync lines. Depending on where the cables ran would affect the result. We changed to shielded cable and that cured the problem.

Cheers

Dave

[B Lewis @ PE Engine Management]

Some things to check:

-Are your crank and cam sensors wired with the correct polarity? The wrong polarity will still allow the engine to run, but may thow off the timing.

-How fast is the engine turning while trying to start? How consistant is the RPM? Sometimes when teams install the engine in the car, they neglect to use a solenoid and/or the correct size wiring for the starter circuit. This essentially reduces the voltage available to the starter which lowers the starting consistancy and may cause crank/cam errors.

-How are the wires routed on the car? Are they close to the starter or other sources of noise such as the coils. Based on your description of the starting crank/cam sync errors that go away while running I would suspect that maybe your crank sensor wires (or the common GND) are too close to the starter motor.

Good luck and let us know what you find.

[Discretely elite]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Kirby:
I would recommend hooking up a scope to your sensors during cranking and see if any noise is on the line. Remember f4i are Variable Reluctance, the output voltage increases with RPM to somwhere around +-20V. </div></BLOCKQUOTE>

I will echo this suggestion as well. In MoTec there is an option to capture the ref and sync traces during cranking. We used to use an adjustable, custom hall sensor as our cam sensor and had endless issues with no cam signals. A scope or similar log of the signals during cranking will confirm what signal is at fault. I am guessing its the crank signal.

Regarding your dyno plot:

Nice work! I believe that is the very same engine dyno we have. We had problems with the control station and abandoned it in favor of our dyno-jet chassis dyno. Are you able to hold a steady RPM with varying loads? What about a step program?

I bet that thing sounds wicked at 14k still making 70+ hp!

[Marshall.Hagen]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by A Richards:
Whats the voltage at your ECU during cranking?? It could be just turning on and off due to a voltage drop, as your wiring has changed. Our Motec drops RPM to zero when it gets below 10.5 volts at cranking. </div></BLOCKQUOTE>

To eliminate that variable, I attached a battery tender during cranking. Cranking volts fall between 10.5 and 11.5, typically. Our Injector Dead Time/Voltage offset table has values down to 8.0v, so I am assuming the ECU stays active until that point.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TMichaels:
Hi Marshall,
you do not really need to shield the wire, just use twisted pair or twist your existing cable. Check if your sensorcables run close the the startercable or other high-current/pulsed current wires.

Check if there is metal swarf on your sensor, since inductive sensors do work with magnets.

Also check the distance between your sensor tooth wheel and the sensor. The closer the better.

We went through all this stuff in 2006 just one week before the competition. It was a faulty tooth wheel in our case.

Regards,

Tobi </div></BLOCKQUOTE>

Tobi,

We plan on removing the harness Monday to add some new sensors. We will take the time then to twist the sensor and coil/injector wires. We have some good low-prifile shielding wrap tha we will use for overkill. I would rather go over the top than continue to pull my hair out.

I will check the trigger wheels while we have everything appart. I'm confident that they are not faulty, seeing as the engine runs strong in the car, and started fine when on the engine dyno. But it will not hurt to double check.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by B Lewis @ PE Engine Management:
Some things to check:

-Are your crank and cam sensors wired with the correct polarity? The wrong polarity will still allow the engine to run, but may thow off the timing. </div></BLOCKQUOTE>

Polarity is good. I also compared our cranking oscilliscope values to those from the engine dyno, and everything is appears to be fine.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by B Lewis @ PE Engine Management:
-How fast is the engine turning while trying to start? How consistant is the RPM? Sometimes when teams install the engine in the car, they neglect to use a solenoid and/or the correct size wiring for the starter circuit. This essentially reduces the voltage available to the starter which lowers the starting consistancy and may cause crank/cam errors. </div></BLOCKQUOTE>

We are using flexable automotive grade 8gauge wire from a car-stereo shop. Cranking RPM stays around 360RPM +/-50RPM. We are using the OEM Honda starter relay.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by B Lewis @ PE Engine Management:
-How are the wires routed on the car? Are they close to the starter or other sources of noise such as the coils. Based on your description of the starting crank/cam sync errors that go away while running I would suspect that maybe your crank sensor wires (or the common GND) are too close to the starter motor.

Good luck and let us know what you find. </div></BLOCKQUOTE>

Wires from the coils are running in parallel on the exhaust side of the engine. Injectors are running in parallel on the intake side. Cam sensor joins with the injector harness, as well as the crank sensor. All wires to the ECM join and run in parallel through the firewall and bulkhead. Common ground is located on the clutch side, opposite that of the starter.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Discretely elite:

Regarding your dyno plot:

Nice work! I believe that is the very same engine dyno we have. We had problems with the control station and abandoned it in favor of our dyno-jet chassis dyno. Are you able to hold a steady RPM with varying loads? What about a step program?

I bet that thing sounds wicked at 14k still making 70+ hp! </div></BLOCKQUOTE>

Here is a picture of our engine dyno setup:
HERE (http://www.itrspec.com/FSAE/FSAE022.jpg)

We also have a Land&Sea chassis dyno. Even though we have good load control with the Superflow, it is hard to get a close approximation of the cell path during part throttle tip-in and tip-out without, which is where the chassis dyno plays a large role.

The load control and test rate on the Superflow is excellent! We experimented with 250 and 500RPM step with a faster testrate, and it was a great method for tuning. The full throttle graphs we use acceleration based test, with 100rpm testrate. The dyno has a hard time holding a steady RPM past 75kpa while below 7000RPM, it will hunt a bit. Once the revvs increase, the RPM is very consistant.

We are very impressed with the power we are making. Our initial goals were 70-75HP, which we used some simulation softare to predict, so having 83HP is a great surprise. It took a while to get there though. I Have logged over 200 dyno pulls on the Superflow.

Thank you everyone, I will hopefully have some encouraging results on Monday.

Regards,
Marshall

[Grant Mahler]

Does anyone know (roughly, obviously dependent on engine) what a VR sensor should be putting out on cranking? ~0.5v? 0.05v? 0.005v?

[A Richards]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marshall.Hagen:

To eliminate that variable, I attached a battery tender during cranking. Cranking volts fall between 10.5 and 11.5, typically. Our Injector Dead Time/Voltage offset table has values down to 8.0v, so I am assuming the ECU stays active until that point.
</div></BLOCKQUOTE>

Our MoTeC has compensation tables from 5-15volts and i can assure you that it does not run anything below 10..

Doesnt make sense to me then why the car will run fine once it is started. Have you tried roll starting your car?

[TMichaels]

Grant, a value over 0.4V should be fine. Below 0.25V is bad. I have a scope image from our setup somewhere on my pc and will try to find and post it.

Regards,

Tobi

[Erich Ohlde]

I'm going to second B. Lewis. It sounds like your cam and/or crank sensor is wired reversed from your engine dyno. If you have a method of counting sync errors in your ecu software see if you continually get errors when running/cranking. Typically you will get 1-2 errors on initial cranking of the engine but there should be no more after that.
One way to check is to use an Oscope to look at the form of the signal on your engine dyno wire harness versus your car harness. an f4i VR sensor has a very defined leading/trailing (can't remember which) edge with a messy opposing edge. it will be very obvious if you have the polarity switched between the two harnesses.

[Marshall.Hagen]

We were able to fix the problem over the weekend.

Both the cam and crank sensors were shielded using a USB cable (temporary solution), and the grounds were rewired.

She starts right away, every time.

I never knew you could have so much interference during cranking.

Thank you everyone for your advice and help!

Regards,
Marshall

[murpia]

Actually, using USB or Ethernet cable for crank & cam sensors is a good idea, as the data lines are twisted-pair reducing cross-talk and noise pickup.

Regards, Ian