1 2 3
wvumtnbkr
wvumtnbkr UltraDork
9/12/18 7:32 a.m.

My buddy and I swapped my running / race winning S4 NA 13B drivetrain into his 1st gen racecar.

It ran well 1 time at Watkins Glen earlier this year.

Since then, it will not start and run reliably. We have run tests on the AFM, the spark, and fuel. It appears all parts tested are doing the correct thing. The engine has decent compression (90+ all rotors all faces).

I can get it to start and rev if I manually hold teh flapper door open on the AFM. I can change the AFR's based on how open I hold the flapper door. If I open it the entire way while holding the throttle plates open, it will rev normally. Not something I can test under load for obvious reasons.

If I do NOT hold open the flapper door, it will cough and sometimes start for a second, then die. The AFR's are LEAN! When it does this.

It seems to me that the AFM is NOT opening far enough for some reason. WE checked all vacuum lines (there are only the Boost sensor and a reference for the FPR and the brake booster hooked up). All other vacuum ports are plugged. In fact, we plugged the brake booster line as well. Same issue. WE removed the intake manifold and checked all flanges and gaskets. All looked good.

I tried unplugging the coolant temp sensor in the water neck. No difference.

The reason I KNOW this isnt related to the fuel pump relay and the AFM is that the fuel pump is not in the harness anymore. It is on its own power and relay. Therefore, the usual failure mode is not in play in this situation. Also, it did start and rev when we accidentally forgot to plug the AFM in. We redid all new wires to the AFM. We did all new sensor grounds and Vref wires to all the sensors as well. All sensor values check okay at ECU.

The fuel pressure is 37psi at idle.

Any help is appreciated!

TL:DR
AFM tests good per FSM
Fuel pump on seperate circuit
Car will start and run and rev ONLY if I hold flapper door proportionally open to RPM(s)
AFRs change in relation to how far I open the flapper door
Not much left to have a vacuum leak (especially a large one!) 

HALP!


 

¯\_(ツ)_/¯
¯\_(ツ)_/¯ UltraDork
9/12/18 7:46 a.m.

On my S4, the answer was always to be found by unplugging things.  Unplug/test the TPS and MAP sensors?  Do you have other injectors you can try?

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 8:08 a.m.

Injectors are on the list to test / replace.

 

Tps was unplugged.  And plugged in...  either way didn't make a change.

 

Did not disconnect map sensor.  Can try tonight.

burdickjp
burdickjp New Reader
9/12/18 8:33 a.m.

Is there a split in the inlet tube somewhere?

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 8:42 a.m.

Not that we could find.

 

I am thinking about smoke testing everything...

Tyler H
Tyler H UltraDork
9/12/18 8:59 a.m.
wvumtnbkr said:

Not that we could find.

 

I am thinking about smoke testing everything...

I think this would be a good next step.  Make sure it's not lean due to a massive vacuum leak.  

AngryCorvair
AngryCorvair MegaDork
9/12/18 9:27 a.m.

questions from someone who knows nothing about these, but i'm thinking in fault tree mode:

is the AFM supposed to be mechanically actuated by gas pedal, or is it opened by airflow?

  • if it's supposed to be mechanically actuated, does that mechanical actuation work properly?  if not, why not?
  • if it's supposed to be opened by airflow, does it actually open with airflow?  if not, why not?

it seems to me that you've narrowed it down to the AFM, but the exact reason aka failure mode isn't yet known.

edit:  one more thought:  whenever something seems like it should set a rich code but it tells you it's lean, i look for faulty O2 sensors.

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 11:28 a.m.

It's a wideband o2 sensor that appears to be reading correctly.  

The afm is opened by airflow.  When it does not open by airflow, that indicates a large vacuum leak.

 

Having removed all vacuum lines except map sensor and fpr lines (including removing brake booster line) it is hard to think it's a vacuum leak.  There are no large vacuum leaks that I can see.

 

We ran all new vac lines and new nipple caps for all vac sources to make sure everything is good.  All hard intake pieces are in good shape and appear to seal well.

 

I did not do a smoke test.  I will do this tonight. 

Tyler H
Tyler H UltraDork
9/12/18 12:24 p.m.

Hook up a multimeter to the afm and watch for the resistance to change smoothly and linearly throughout the range of motion?  You would need the specs for what the resistance should be for that specific AFM, but you may find that the traces under the AFM plastic cover are worn through.  

In the past, I have gingerly bent the arch of the two arms of the AFM wiper to 'shorten' them so that they trace a new path and get out of the ruts they wear over time.  This has brought flaky AFMs back for me before, or killed them.

If your AFM has an integrated intake air temp sensor, that would be a good thing to ohm out per specs as well.

In the image below, you can see the two shiny arcs where the wiper traces.  If you put a little more arc in the wiper arm, it will trace a fresh path.   Again, I would only do this if you're getting dead spots or irregular resistance values.

AngryCorvair
AngryCorvair MegaDork
9/12/18 1:17 p.m.
wvumtnbkr said:

 

The afm is opened by airflow.  When it does not open by airflow, that indicates a large vacuum leak.

so is it a true statement that your AFM is not opening with airflow?  or is it physically opening with airflow but it is not generating the correct signal when it opens?

Vigo
Vigo UltimaDork
9/12/18 1:46 p.m.

I'm pretty sure we went through an almost identical scenario trying to figure out something about the AFM on another 80s Mazda here in the past, or maybe a Celica. 

Anwyay, after staring at the wiring diagram for a painfully long time, I have reverse engineered it to discover the following:

There is something called a circuit opening relay which powers the fuel pump (so, fuel pump relay). This relay has TWO coil (control) sections, meaning it can be turned on by two different circuits. What happens is that your ignition key will power one coil of this relay in the CRANK position only. Once you release it to the regular RUN position that power cuts off. The other section of the circuit opening relay is fed power from the ignition switch in the RUN position, buuuut... the ground side goes through the AFM. Once the engine is running and pulling air across the AFM, the AFM door will lift off its stops and the circuit opening relay is then given that ground so it turns and power the fuel pump.

Fuel pump is powered from CRANK position of ignition switch while cranking. 

Fuel pump is powered from RUN position of ignition switch while engine is running. 

BUT, if airflow stops (engine stalls) and AFM door closes, it interrupts that circuit and stops the fuel pump from running when the key is on but the engine is not.

 

So, you are missing the power coming from the CRANK position of the ignition switch to the Circuit Opening Relay. It's a brown wire at the relay. 

Funny thing, i'm teaching an intro electrical class and my students are building simple series circuits with two light bulbs. I put these diagrams up on the projector to show them what they'd be dealing with then get farther along. One of em asked "so whats wrong with it?" and i said..."im still trying to figure it out!".  I figured it out after they left! The reason they did it makes sense (fuel pump only runs when engine is spinning) but it's actually very complicated compared to a modern car's fuel pump wiring where the computer is smart enough to turn a regular relay on and off based on whether it sees engine rpm. This one did some fancy wiring tricks to accomplish the same thing, but reverse engineering the wiring diagram to figure that out is tough!!

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 2:45 p.m.
Vigo said:

I'm pretty sure we went through an almost identical scenario trying to figure out something about the AFM on another 80s Mazda here in the past, or maybe a Celica. 

Anwyay, after staring at the wiring diagram for a painfully long time, I have reverse engineered it to discover the following:

There is something called a circuit opening relay which powers the fuel pump (so, fuel pump relay). This relay has TWO coil (control) sections, meaning it can be turned on by two different circuits. What happens is that your ignition key will power one coil of this relay in the CRANK position only. Once you release it to the regular RUN position that power cuts off. The other section of the circuit opening relay is fed power from the ignition switch in the RUN position, buuuut... the ground side goes through the AFM. Once the engine is running and pulling air across the AFM, the AFM door will lift off its stops and the circuit opening relay is then given that ground so it turns and power the fuel pump.

Fuel pump is powered from CRANK position of ignition switch while cranking. 

Fuel pump is powered from RUN position of ignition switch while engine is running. 

BUT, if airflow stops (engine stalls) and AFM door closes, it interrupts that circuit and stops the fuel pump from running when the key is on but the engine is not.

 

So, you are missing the power coming from the CRANK position of the ignition switch to the Circuit Opening Relay. It's a brown wire at the relay. 

Funny thing, i'm teaching an intro electrical class and my students are building simple series circuits with two light bulbs. I put these diagrams up on the projector to show them what they'd be dealing with then get farther along. One of em asked "so whats wrong with it?" and i said..."im still trying to figure it out!".  I figured it out after they left! The reason they did it makes sense (fuel pump only runs when engine is spinning) but it's actually very complicated compared to a modern car's fuel pump wiring where the computer is smart enough to turn a regular relay on and off based on whether it sees engine rpm. This one did some fancy wiring tricks to accomplish the same thing, but reverse engineering the wiring diagram to figure that out is tough!!

All very true!  Usually....

There is no ignition switch AND the fuel pump is on its own stand alone circuit (because actually racecar).

 

This car in exactly this configuration at Watkins Glen in May.  About 1 month after that (next race) it started not wanting to run.  It has the symptoms described above.

 

The AFM does not appear to be opening ENOUGH when running.  The AFR's are crazy lean until I push on the door just a bit.  This then tells the system to inject more fuel to keep up with the higher actual air flow rate....

 

 

Wait a minute...  What if the injectors are clogged or partially clogged....  Hmmmmm....  What if I am accidentally telling the car that it has MOAR air than it actually does in order for the "decreased capacity" fuel injectors to "keep up".

 

That would fit all of the symptoms....  

 

Check fuel injectors just moved WAY up the list unless somebody can refute the above logic....

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 2:46 p.m.
Tyler H said:

Hook up a multimeter to the afm and watch for the resistance to change smoothly and linearly throughout the range of motion?  You would need the specs for what the resistance should be for that specific AFM, but you may find that the traces under the AFM plastic cover are worn through.  

In the past, I have gingerly bent the arch of the two arms of the AFM wiper to 'shorten' them so that they trace a new path and get out of the ruts they wear over time.  This has brought flaky AFMs back for me before, or killed them.

If your AFM has an integrated intake air temp sensor, that would be a good thing to ohm out per specs as well.

In the image below, you can see the two shiny arcs where the wiper traces.  If you put a little more arc in the wiper arm, it will trace a fresh path.   Again, I would only do this if you're getting dead spots or irregular resistance values.

Good Info.  AFM checks out VERY well according to the specs in the AFM.

 

Thank You!

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 2:50 p.m.

I hope I am not coming across as dismissive to all of the input!

 

Each one of these suggestions could easily be the issue.  Most of them have been checked, triple checked, tested, and known good spares thrown at them.  Same symptoms.  

 

Please keep the suggestions coming.  Something is going to spark from all the input!  At the very least, all of the suggestions so far would be a GREAT troubleshooting flow chart for anybody who googles this symptom!

 

Thanks again!  Please keep them coming!

 

To Do list for me:

1) Smoke check all intake parts to look for vacuum leak

2) Check flow and pattern from fuel injectors

3) Throw ANOTHER AFM on the car to verify functionality

 

Did I miss anything?

 

 

 

AngryCorvair
AngryCorvair MegaDork
9/12/18 3:16 p.m.

In reply to wvumtnbkr :

i don't feel like you're being dismissive, so i'm going to keep jumping in until you solve it.

you say "The AFM does not appear to be opening ENOUGH when running.  The AFR's are crazy lean until I push on the door just a bit.  This then tells the system to inject more fuel to keep up with the higher actual air flow rate...." which kinda brings me back to a different way to ask my earlier question:  is there something about this AFM that's making it take more force than usual to open?  i mean, it's got a spring of some sort that makes it close when airflow isn't sucking it open, and it has to close relatively tightly so as to not let in unmetered air.  is there additional physical resistance via the flapper being sticky in the air path, or the return spring providing too much force due to corrosion, or is the flapper's hinge pin giving too much friction because of gum / crud / dirt? 

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 3:26 p.m.
AngryCorvair said:

In reply to wvumtnbkr :

i don't feel like you're being dismissive, so i'm going to keep jumping in until you solve it.

you say "The AFM does not appear to be opening ENOUGH when running.  The AFR's are crazy lean until I push on the door just a bit.  This then tells the system to inject more fuel to keep up with the higher actual air flow rate...." which kinda brings me back to a different way to ask my earlier question:  is there something about this AFM that's making it take more force than usual to open?  i mean, it's got a spring of some sort that makes it close when airflow isn't sucking it open, and it has to close relatively tightly so as to not let in unmetered air.  is there additional physical resistance via the flapper being sticky in the air path, or the return spring providing too much force due to corrosion, or is the flapper's hinge pin giving too much friction because of gum / crud / dirt? 

Great question.  Im not sure I can answer that fully.  It "feels" like the other 2 AFMs I have (1 turbo, 1 of unknown origination).  I guess I could try to find out what force it is supposed to be and use a fishing scale to measure it.

Something else to add to the list!

Keith Tanner
Keith Tanner MegaDork
9/12/18 4:23 p.m.

It's highly unlikely that something failed in the AFM, causing increased resistance. I would put that lower on the list.

More likely: massive leak post-AFM, giving air an easier way to enter the engine or insufficient fuel delivery (injectors/pump/FPR).

If it requires a considerable amount of AFM molestation to get it to idle, I have to think it's the latter. The AFM probably doesn't open all that far at idle, just enough to trigger the keep-alive on the fuel pump (note, if anyone ever tells you their 1.6 Miata starts for 2-3 seconds and dies, it's a busted wire to the fuel pump from the AFM).

We know that fuel pressure is about right, assuming that this RX7 uses a similar fuel system to a Miata with the same AFM. So I'm on Team Check Those Injectors.

Do rotaries have single injectors per rotor?

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 4:48 p.m.

2 injectors per rotor.  But, the second ones only activate under load above 3800 rpm.

Knurled.
Knurled. MegaDork
9/12/18 7:09 p.m.

Is the AFM mounted in the same orientation as Mazda did?

 

 

wvumtnbkr
wvumtnbkr UltraDork
9/12/18 9:14 p.m.

Yep.  Also tried on its back and side.

 

The fact that it ran on track for an entire day and will mow not even start reliably indicates it's probably not afm position. 

Vigo
Vigo UltimaDork
9/12/18 9:53 p.m.

Well, i enjoyed my little foray into "how we did fuel pump control when ECUs were too dumb to do it themselves" and i'm glad you got away from that Rube Goldberg-ish control arrangement for your fuel pump.  If that paragraph was in the post when i first read it explaining that you rewired the fuel pump then... i should learn to read? 

I would just say that the AFM is probably not the issue unless it is fed some sort of incorrect reference voltage. Other than that the likelihood of it still reading linearly but just shifting its scale sideways is extremely low. 

WonkoTheSane
WonkoTheSane Dork
9/13/18 12:44 a.m.

If you get way too much gas, it'll read as lean because the sensor gets coated with fuel and didn't get oxygen.  Go figure, eh?

 

I'd check the injectors.  I've never heard of a rx7 or Miata AFM preventing the car from running, other than the fuel switch which isn't a problem here.  That said, I've seen lots of failed ones make running smoothly a problem :)

bbaker480
bbaker480 New Reader
9/13/18 7:04 a.m.

To add to the "fuel injectors are probably clogged" to get the car in the garage post unloading it from the trailer, and then again post test drive I have had to stop in place, start the car with my foot flat on the floor, and dump the clutch at about the same time I take my finger off the starter button.  I suspect this meets the 3800 rpm under load conditions to open the secondary fuel injectors?  I have spares, unknown if they are good or not that I'll take to the garage tonight with me to test.

wvumtnbkr
wvumtnbkr UltraDork
9/13/18 8:23 p.m.

We put in known good injectors.

 

It seemed better at the start.  It actually started.  Then I realized I forgot to plug in the afm.

 

Maybe slightly better?  Still not running correctly.

 

Also, I forgot to mention that it is back firing when you try to rev it.  It goes quite lean, tried to die, then backfires massively loudly.

 

Tried unplugging the tps and that made no or minimal difference.

 

When I keep my hand on the throttle, it will run at steady state, but seems to slowly lean out.

 

Gah, what are we missing?

 

This damn thing ran a couple months ago!

 

Vigo
Vigo UltimaDork
9/13/18 8:41 p.m.

Are you actually verifying the fuel pressure while the engine is running? I assumed so but i dont think you specifically said what it was doing besides 37psi at idle.

1 2 3
Our Preferred Partners
a0OcsLiXX7swqfwb2TKHG7acXJ92HtTerQk8WSxdkFTIrE5wCUB8toc4vmJXY3jw