Checking Your AFM and TPS

#21

Time to find out on a dyno.
Run with known good and then unplug.

Wish we had some data on the above.

Also want to see the same data on the thermo switch as I’ve been in my car on the back straightaway of Charlotte when the power fell off.

RP

#22

Ranger was essentially correct in his explanation of the TPS function. But I might be able to clarify that a bit.

The idle switch should close when the throttle stop is 1mm off the idle speed screw. That tells the DME that the engine is idling and accordingly what fuel map to use and to use the ICV for speed control. When the idle speed screw is correctly adjusted the engine speed should be 950rpm with the ICV disconnected and drop to 750rpm when the ICV is connected if the idle switch in the TPS is correctly adjusted. So on an engine at normal operating temperature, you adjust the stop screw for 950rpm. Then adjust the TPS so that the idle switch closes when the stop arm is 1mm off the stop screw. While you can check the TPS at it’s pins, the best check is made at the DME connector. That eliminates any possibility of a faulty engine harness interfering with those signals.

The WOT switch should close at about 40% of full throttle, as I recall. When the WOT switch closes the DME shifts to a different map and ignores O2 sensor data. At 4500 rpm the DME stops using AFM data as the flap should be fully open. From there to the redline the DME uses only engine coolant temperature and rpm for fuel delivery and spark advance. I think the DME uses a different map from 4500rpm up, but I’ve heard conflicting opinions on that.

All of the maps are subject to learned fuel trim, which is based on O2 sensor data taken at part throttle. Since an aged, but operable, O2 sensor typically indicates a leaner mixture than actually exists and we don’t have catalytic converters that can be damaged by high HC in the exhaust, there can be a slight benefit from using a aged sensor.

Understand that learned fuel trim and fault codes are stored in volatile memory and are lost if the DME looses un-switched power. If your kill switch disconnects all power, every time you turn off the switch that data is lost.

The WOT switch is important. If the DME doesn’t see that data the engine will run a bit lean above 4500rpm with a consequent loss of power.

Oil can, and will, leak into the TPS. The switches are simple open metal parts and oil will interfere with their operation. I highly recommend drilling a 1/8-3/16" hole in the center of the round boss on the bottom of the TPS. That gives the oil a place to escape and makes the TPS last a lot longer.

If desired, you can plug the hose from the ICV to the throttle body (best solution) or disconnect the ICV and use the stop screw to set a higher idle. The downside is that you may have to hold some throttle in cold weather until the engine warms up a bit. The upside is that the water pump will be spinning a bit faster and be moving more coolant. That helps to keep the peak engine temp from spiking so high in a red flag situation. I liked an idle of 1200-1500rpm.

#23

I just hooked up my AFM to do this test. I plugged in the 9V, and measured the output voltage. I didn’t really try to determine the flap angle, but the voltage appeared to rise steadily from ~0.5V to ~8.0V as I opened the flap. The top/bottom range is roughly what the test chart on the 944 AFM page (http://www.iprimus.ca/~trauttf/temp/AFM/) indicates as being normal.

The only abnormal thing was when I made sudden movements on the flap, my Fluke would read OL (over limit/overload) for a moment before finding the voltage. I’m guessing this is just natural behavior since everything else seemed normal, but I don’t really have an explanation for it. Does this seem normal or should I be concerned that something isn’t working correctly?

Next step for me is testing the TPS. If I understand correctly, the continuity behavior is supposed to look like this, right?

Problem I’m trying to solve:

During my last race, the car would bog down around 4500 RPM on several of the straights. I’d have the pedal to the floor and the engine just crawled from 4500 up. It didn’t feel normal at all. I have an over-shaved head (that sounds funny) that I’m compensating with a thicker head gasket, so the net effect is I’m probably a little lower than normal on the compression – but just a general drop in power isn’t what this felt like. Normally you expect slow build up and then faster build up as the RPMs rise – this was the reverse. The engine felt like it hit around 4500 and then just decided to take a siesta. I’d exit turns faster than other people, and then get caught halfway to the next turn.

Reading this thread, it would sound like either 1) my TPS is busted, or 2) my coolant sensor is busted. Or a harness problem related to one of those.

I replaced one of the coolant sensors when I was first building the car, and I could have sworn I replaced the DME one – can someone confirm that’s the two pin one? http://www.autohausaz.com/pn/0280130026?

A third possibility I wonder about is – would a stuck closed thermostat cause a problem with the DME coolant sensor being able to read the correct temp? Another data point here is that I have the 7psi oil pressure switch swapped in as my coolant pressure switch. It took several laps before the light would go off. I haven’t really worked through what my hypothesis would be here, but wondering if it’s related. Like maybe flow is prevent the temp sensor from reading proper values, while also somehow simultaneously having the opposite effect of preventing the pressure to build up in the block for the switch to go off.

Maybe it could explain why my radiator ultimately blew 4 laps into my Sunday race? Or hell, maybe it was the radiator that was clogged the whole time? shrug

Som

#24

An analog meter might give a better idea of what is going on with the AFM, but what you describe doesn’t sound normal. It sounds like the wiper isn’t making good contact with the resistance path. When the vane moves quickly the wiper looses contact.

As to the TPS… the idle switch signal should reach the DME when the throttle stop arm is 1mm off the idle stop screw. That tells the DME to switch to an idle map and to use the ICV to control engine speed to 750rom. The WOT switch signal should reach the DME at about 40% of full throttle (as I recall). That will cause the DME switch to a different map and ignore O2 sensor data. Below 4500rpm the DME will use AFM, engine temperature, and engine speed for fuel and spark control. Above 4500 rpm only engine temperature and speed are used. Best to check for the TPS signals at the DME connector as that is what really matters.

The engine bogging above 4500rpm sounds most like the WOT switch in the TPS.

Engine temperature has to get above 140F before there is enough pressure built up to trip the pressure switch.

A thermostat that has failed usually fails in the open position, resulting in the engine running cold. If the ECT sensor is good and providing date to the DME the affect on performance will be minimal.

How old was the radiator? They are generally considered to be good for 100k and it, the expansion tank, fan clutch, cap, water pump and thermostat need to be replaced at that interval.

#25

[quote=“Som” post=81553]
The only abnormal thing was when I made sudden movements on the flap, my Fluke would read OL (over limit/overload) for a moment before finding the voltage. I’m guessing this is just natural behavior since everything else seemed normal, but I don’t really have an explanation for it. Does this seem normal or should I be concerned that something isn’t working correctly?[/quote]

It’s normal for the afm to read a little high during sudden movements. The engine needs extra fuel when you first hit the throttle, so the flapper opens a little extra and then settles back down. It’s a mechanical form of “acceleration enrichment” (google it).

#26

Re. coolant pressure light doesn’t go off until you do a couple laps. The coolant has to get warm before the pressure light will go off. It shouldn’t take a couple laps for this to happen. Maybe your 7psi switch is more like 10psi, or maybe your water pump is a wuss.

Re. sudden movements cause meter to go OL. I don’t buy the extra fuel theory. The AFM is an analog device. I would imagine that it should provide the calibrated ratio no matter how fast you move it. Sure, the engine’s going to want a burst of fuel for sudden acceleration, but that’s for the DME to do based in it’s sensor inputs. One wouldn’t normally design a sensor to give intentionally false info, OL on fast changes, to the computer.

Re. engine bogging down >4500 means bad TPS. I’m slower to buy into the bad TPS theory. Once the AFM door fully opens the DME should go to internal maps. I’d argue that the TPS if far more critical at mid-rpms then at high rpms. At mid rpms the AFM door might not be fully open so the only way for the DME to know what’s going on is the TPS signaling WOT. But at high rpms the AFM door will be open so the DME should understand what’s going on and do a reasonable job of engine management.

The DME is entirely capable of running the car in the absence of some it’s sensors. If, for example, it’s not getting a signal from the O2 sensor or the engine temp sensor, it substitutes a reasonable fixed value and drives on. So I’d imagine that in the absence of a TPS signal the DME ought to be able to do ok, not optimal, but not falling on it’s ass either. So my guess would be that there’d be some struggle if, at low to mid-rpm, you went to WOT. But at higher RPM, WOT would get behavior reasonably close to optimal.

That’s my theory, anyhow. Yes, I did just pull all of that out of my butt.

So maybe it’s the TPS, but I wouldn’t rule out other possibilities too. Fuel pump, coil, injectors, temp sensor, plugs, plug wires, I dunno, there’s lots of possibilities.

It would be nice to get the car on a dyno and carefully inspect the F/A ratio. This is another situation that shows why everyone should have a F/A meter on their dash, preferably hooked to a data logger. At some point or another, everyone fights with engine management. Having a fuel pressure gauge and A/F meter on the dash can be very helpful.

#27

Engine bog after long left hand turn on the straightaway…I’ll bet it was fuel starvation. (I know, even with a 5/8 full tank.) Our fuel pumps stink.

RP

#28

[quote=“Patton” post=81558]Engine bog after long left hand turn on the straightaway…I’ll bet it was fuel starvation. (I know, even with a 5/8 full tank.) Our fuel pumps stink.

RP[/quote]
What Robert described happens a lot. In the last 2 events someone needed to strap a fuel pressure gauge to a windshield wiper 2x. A fuel pressure gauge permanently installed on your dash can make a lot of mysteries go away.

#29

Thanks for all the feedback!

Prior to the event, I had adjusted the sweeper arms on the AFM by pulling them up/away from the contact patch. I tried to bend the arms to get better contact, but maybe I didn’t do a good enough job. I’ll have to take a look at this again. Friend of mine has a “known good” AFM, so I can also borrow that and hook it up for the same test.

Kind of a non-critical question, but what is generally meant by “40% throttle”? Seems like it could mean 3 things – the cross sectional area of the open butterfly valve (unlikely, since it would be a pain to calculate), or it could be the degree of the butterfly being open (45 degrees being “50% throttle”), or it could mean the distance that the end of the throttle cable has traveled in relation to the full range of travel. I’m guessing it’s the last one?

My oil temp was around 150-170, which is generally where it sits, if I remember right. Under 150 would have stood out to me. This was measuring off a manifold connected to the oil pressure switch location. So I’m guessing the coolant was up to temp / normal operating temp. It definitely took more than 1 or 2 laps to turn off the light.

Water pump had about 6 or 7 hours on it – I replaced it beginning of 2014. But maybe it did suck. As part of converting to the new style cooling, I’ve also purchased a new pump.

Good to know about the thermostat, thanks… yeah, I guess that wouldn’t be the problem then. :slight_smile:

As for the radiator, I don’t know how old, but probably pretty old. When I got back to the pit, it was spraying from the middle of the core onto the timing belt cover. The rest of the stuff had been replaced, with the exception of the expansion tank.

I replaced the transfer pump before this event. The starvation issues I got prior to this were very sudden cuts in engine power and happened any time I tried to run a second session on a tank. I had none of those experiences at the last event, and even let my fuel drop below half full for a session to test. That said, I can’t really rule it out – who knows, maybe the bogging was a different type of starvation effect. The slow creeping at 4500 up didn’t feel like starvation, though.

I had the injectors cleaned prior to the event, but I also did a head swap. Who knows, maybe I wasn’t careful and some crap happened to drop into the exposed fuel line while I was doing this and clogged up and injector. I’ve purchased a fuel pressure testing kit, so I’ll see if that looks normal. Would be nice to have the pressure/a-f to a data logger, but realistically don’t see that happening this season.

I would love to do a lot of these little tests while on a dyno. You could easily test the TPS out-of-the-butt theory by unplugging it. That would effectively tell the DME that there’s part throttle all the time. Might not idle right, but if that’s your only difference, I imagine you’d expect to see the same shaped graph after 4500 RPM, right? Same test in reverse – good TPS, unplug the AFM. If TPS at WOT goes to internal maps, then again you’d expect identical graphs from 4500 RPM and up. A test that might be fun would be to plug a trimpot to the coolant sensor wire and run a handful of different tests to see how sensitive the performance is to an incorrect reading – are we talking 1-2hp or 4-5+? Unfortunately, while the dyno runs are like $60-75 for 3 runs at the track, it’s twice that at any of the local shops here, and 4 times that to get an hour “tuning session”. Not sure I’m ready to spend $250-300 on experimentation yet. :slight_smile:

Som

#30

[quote=“Ranger” post=81557]
Re. sudden movements cause meter to go OL. I don’t buy the extra fuel theory. The AFM is an analog device. I would imagine that it should provide the calibrated ratio no matter how fast you move it. Sure, the engine’s going to want a burst of fuel for sudden acceleration, but that’s for the DME to do based in it’s sensor inputs. One wouldn’t normally design a sensor to give intentionally false info, OL on fast changes, to the computer.[/quote]

Normally the DME would do this… but on AFM based cars the DME doesn’t have enough info to work with. Normally the ECU would use either an analog TPS (not the e30’s switch TPS) or a MAP sensor, neither of which the e30 has. It’s not false info coming from the afm… the DME responds faster than the Fluke and sees the actual voltage.

The “OL” may be happening because he’s manually moving the flap faster than the car would ever see. At normal changes of speed the overshoot would only be a couple tenths of a volt. Now… whether his is overshooting TOO much, I can’t tell you.

#31

My e30 AFM was off the car, so I measured the resistance while moving the flapper quickly. My Fluke 87V also gets “OL” during sudden movements.

#32

I find that disturbing as I’ve not seem that when testing an AFM with a Fluke DMM. Not it may not be all that important as I don’t see a situation where the AFM vane would ever have to move that quickly.

With respect to the other things I posted earlier… It is difficult to get any authoritative information about the internals and operation of the engine management system. I have gleaned some from the Internet and some from a BMW factory tech rep. Where possible I have attempted to verify this information by direct testing on and off a dyno. So I feel pretty confident that what I have said is essentially correct.

One thing that I can state for sure (as proved on a dyno) is that the WOT switch in the TPS is important for good performance above 4500rpm. Data logging shows me that the AFM signal goes constant at about 4500 rpm, which I interpret as a fully open vane.

#33

Justin, you are measuring the voltage ratio or resistance? I think it’s supposed to be a voltage ratio.

Re. TPS is important for good performance above 4500. We might be saying the same thing. My guess is that engine performance would be “ok, but not optimal” at high rpm with a brokedick TPS. My point was that I didn’t think that the engine would completely fall on it’s ass w/o the TPS. I would define “Ok, but not optimal” as 10-15hp low. I would define “completely on it’s ass” as, engine hiccups and misses so much that clearly something is badly wrong.

What prompted me to put the fuel pressure gauge and F/A meter on my dash years ago was a year of engine management hell. The engine was going way lean at high rpm. Not always, but usually. If I’d had the sense to believe what the F/A meter was telling me, I’d have understood a lot sooner what I was dealing with. The symptoms i experienced that year, the engine starting to miss badly at high rpm, sound a lot like what’s going on in this thread. In my case the problem was the AFM, but certainly there’s other ways to cause the same problem of, generally, but not always, lean at high rpm.

Jim is certainly right about how little is authoritatively known about how our DME works. Rob Eskew works for a guy that seems to know a lot, so he’s worth consulting. Most everything I state about engine management is either my guesswork or me parroting someone else’s, often Jim’s, guesswork.

#34

I would say the feeling I remember seemed more than just 10-15hp. It just felt like the car didn’t want to continue revving (though it was – slowly), and I feel like I wouldn’t have really noticed a 10-15hp change. I dunno, maybe I would? That said, I certainly wouldn’t classify it as “completely on its ass” – my lap times weren’t as bad as I would expect if the motor was just taking a shit. It’s been a couple months since that weekend, so it’s possible my memory has started exaggerating the bogging sensation, so maybe that 10-15hp is actually it.

During the head rebuild, I had parts lying all over the place in all sorts of different positions. Would an “oil in the TPS” problem possibly be exacerbated by having the TB/TPS upside down and having the oil get all over some of the wiring in there? I haven’t looking into how the TPS works.

Another thing I did after the head rebuild was that I noticed that my throttle cable wasn’t pulling the TB fully open – it stopped maybe 5mm short of the full TB travel. I adjusted the cable so that pedal-to-the-floor fully opened the TB. Wonder if this could have caused any of the problems with the TPS internals.

To clarify some things, I understood from the thread (and thought I remembered this from other threads) that a full-open AFM would trigger “open loop” mode. But that didn’t jive with the TPS WOT trigger affecting performance at 4500 RPM because, theoretically, 4500+ RPM would (should?) mean a full-open AFM, right? So it sounds like that’s not actually the case – the AFM can’t trigger “open loop” mode. In reality, only the TPS can trigger “open loop” – is that right?

So, conditions for open loop are (?):

  • TPS Idle (2 + 18 connected)
  • TPS WOT (3 + 18 connected)

Then, in open loop, the DME uses the AFM as an input up until 4500RPM – at which point it just assumes a full-open AFM and ignores it from there on.

It would make sense why the AFM wouldn’t trigger anything based on “full-open” because it’s an analog sensor and the DME doesn’t really have a long-term reliable way to precisely know what “full open” really means in voltage-speak. I’m not sure you would want to just say “when the AFM voltage is 0.9 x reference voltage (12V) then the AFM is full-open”. Are you guaranteed that wiring resistance between the DME / battery is the same as the wiring resistance going in as the reference voltage for the AFM? If not, then a 0.9 ratio at the AFM is different than a 0.9 ratio at the DME. I would think there are other factors, too, but can’t really think of them. On the DME side, you could accept a range of voltage ratios to be interpreted as full-open, but I wouldn’t think that would be a good idea as you’d potentially lose granular data on the upper range. Having a digital input tell you WOT seems much more reliable.

Som

#35

Re-reading my response, something else seemed like a (tiny) possibility.

Does the DME use the same sensor input to determine 4500 RPM as it does the fuel/spark timing? Obviously if it’s the same sensor, it’s not likely the problem since car runs reasonable well otherwise. But if there are different sensors, maybe the one that the DME uses to determine RPM is bad?

My understanding is both use the pulse sensor off the front vibration dampener, where the pulse sensor off the #6 wire is just for diagnostics. Is this a correct understanding?

Som

#36

[quote=“Som” post=81567]Re-reading my response, something else seemed like a (tiny) possibility.

Does the DME use the same sensor input to determine 4500 RPM as it does the fuel/spark timing? Obviously if it’s the same sensor, it’s not likely the problem since car runs reasonable well otherwise. But if there are different sensors, maybe the one that the DME uses to determine RPM is bad?

My understanding is both use the pulse sensor off the front vibration dampener, where the pulse sensor off the #6 wire is just for diagnostics. Is this a correct understanding?[/quote]
There is only one speed sensor, the CPS. The Cylinder ID sensor on the #6 plug is used only for injector timing.

#37

To elaborate on what Jim said…recall that the crank turns 2x for every cam revolution. Read this slow…

The crank is at TDC 2x for every cam TDC. The CPS tells the DME that the engine is running and it informs the DME of the crank position when it see the missing teeth on the harmonic balance. But that’s not enough info for the DME because in terms of cam TDC, the DME won’t know the first crank TDC from the 2nd. The #6 plug wire connector is the missing bit of info that tells the DME where the cam is, and therefore knows if it’s the first crank TDC or the 2nd.

#38

I had actually realized that (the 2-to-1 ratio), but was trying to figure out how much it could matter since the injectors are only 2 banks anyway. I don’t think it’s immediately obvious why it should matter until you break down the firing order. Step back – yes, at a high level it’s “obvious” the DME won’t know which rotation you’re on. But just under the surface of that understanding, it seemed like it wouldn’t matter if you don’t have the granular control to fire off individual injectors anyway.

But, digging further… the banks are 1/3/5 and 2/4/6. When you look at the firing order – 1-5-3-6-2-4 – it becomes more clear. The DME is set up to fire off the first bank when 1-5-3 can all intake the fuel, and then 6-2-4 when those cylinders are ready. See this image from Wiki:

Since we have 6 cylinders, each cylinder in the firing order would be 60 degrees apart from each other. That means there isn’t one moment where 1-5-3 are all on intake strokes at the same time. So I’m guessing the injectors fire off at the last possible moment for the #1 cylinder to be able to get fuel in. This means that the #3 cylinder is probably still closed wrapping up the prior exhaust stroke, so maybe it doesn’t get the full benefit of atomization, since I’m guessing the fuel gets squirted into the back of a closed valve?

Anyways… point being, there seem to be a lot of people claiming “no difference” with/without that wire. I’m guessing this is why. The banked system already compromises ideal injection time – what’s a little more compromise? :slight_smile:

Then I found an old thread on here (http://spece30.com/forum/41-electrical-gauges-and-sensors/72220-no-6-plug-wire-qpulse-generatorq?limit=10&start=10) where Jim says that, without this signal, the DME changes to firing all 6 injectors at the same time. That makes sense – you could either opt to take a chance and have the injectors squirting at exactly the wrong time, or just “split the difference” and ensure everyone isn’t operating the most efficiently, but not the least efficiently, either.

Blah blah blah… what was the thread topic again? :slight_smile: Sorry for derailing.

Som

Edit: Btw, I figure a bunch of you know this already – not trying to sound like I’m telling you something you don’t know. More just jotting it down in case others are interested, but also to let you guys correct me if I’ve got something wrong.

#39

I could see some difference on the dyno when I put the injectors into batch fire by disconnecting the Cylinder ID connector. As I recall it wasn’t a lot, but it was there.

#40

Tested the TPS off the car – figured I’d do this first instead of testing at the DME in case it failed, and it did. It only occasionally closed 18/3 at WOT. Usually didn’t close them at all.

Drilled the 3/16" hole, sprayed brake cleaner in there, let it dry. Still failed.

Is it worth/easy to fix this reliably or am I better off just buying a new one? I don’t mind trying to fix it, I just want the fix to be reliable.

Som