Check your AFM boys and girls (re. hi rpm miss)


Bottom line up front:

  1. The Bentley AFM test is wrong.
  2. The AFM could easily be a culprit in a high rpm miss.

Background. I’ve been fighting a high rpm that “almost” always occurs at WOT and between 5k-5800 or so rpm. It seems to clean itself up just before RL. The problem does not occur at partial throttle and high rpm. I replaced and checked a helova lot of shit in this fight. I found along the way that the DME is quite sensitive to input voltage and for two events it seemed that the symptom would go away if I disconnected the TPS. That didn’t make much sense to me so I worried that it was a red herring, but for a number of sessions that TPS connected or not correctly predicted behavior.

There was some indication that it was a lean condition that was making the engine miss.

A guy PM’d be on e30tech the other day that he had a similar symptom a while back and the behavior seemed to go away when he replaced his AFM. His tale reminded me that Julio had bought new AFMs for their cars after problems bedeviled them for a while.

I spent a couple days thinking about the AFM theory but I kept coming back to the idea that the DME isn’t supposed to listen to the AFM sensor at WOT and at 4500 or so the door is supposed to be wide open anyways. That’s 2 reasons why the AFM can’t possibly be the cause of a problem that only occurs at WOT and in a specific rpm range where the AFM door doesn’t move.

On the other hand, I thought, I’ve not fooled with the AFM much in this effort, and I’m well past the point where I need to look at things that don’t make sense. So tonight pulled off my AFM and set it down next to my 4 spares and started testing them with a multimeter to see if I could find anything interesting.

The first thing I found is that the Bentley is wrong. It says the sweeper door’s potentiometer is tested with pins 7 & 8 and it’s actually 5 & 7. Also, the copy of the OEM Service Manual was no help because it shows you only how to R/R the AFM, but not how to test it.

The 2nd thing I found is that all, yes all, 5 of my AFMs had problems providing a linear increase of resistance in the last 5mm of door travel.

The way it’s supposed to work is that as the door opens the resistance increases. When the door is fully open the resistance should be highest. But what I found is that in each case the resistance was highest 4-5mm before the door opened all the way, and then the resistance dropped. So when the door was maybe 80% open the potentiometer would tell the DME that the AFM door was fully open (max resistance). Then, as the door moved thru it’s last 4-5mm of travel the resistance would drop so the AFM would be telling the DME that the door was closing. Air flow was going up but the AFM was telling the DME that air flow was going down. The AFM that I had on the car was the worst of the bunch and would have told the DME that the door was about half open, when in reality the door was fully open.

I think that the DME was reading that 1/2 open business and not letting the injectors provide enough fuel. That’s why the engine ran well at partial throttle, and that’s why the hiccuping seemed to be due to a lean condition.

The AFM I’ve been using for the last several years had a peak resistance of ~1800 ohms and when the door closed it dropped to ~600 ohms. A second one I tested showed peak 960 ohms and dropped to 525 ohms. A third went from 1070 to 930.

Solution. I figure you either have to move the circuit board a little or monkey with the sweeper arm. The sweeper arm has a little screw in it. If you look closely at the sweeper arm you can tell that it’s designed to allow adjustment in the sweep range. I loosened the screw that holds the sweeper arm in place and then shifted the sweeper arm so it didn’t sweep quite as far. So instead of sweeping beyond 1800 ohms for example, I shifted the arm so it hit it’s limit of travel while resistance was still going up.

Please god let this fix the damned thing.

LATER EDIT: The testing process above is associated with the Bentley and is probably not correct. What is probably the right answer is introduced in page 2 but understanding doesn’t really start until p3.


If we all need to buy new afms we should argue the megasquirt again or some mass airflow sensor. I’m not a big fan of relying on a sensor that doesn’t work in the rpm range we use most of the time anyway.


If someone had a megasquirt they’d a been able to quickly figure out from the logs that the AFM signal to the DME was goofed. Spotting that would have been childsplay. But until more folks fight intermittant engine management problems, the Megasquirt idea isn’t going to go anywhere. Especially with a crowd calling for freedom to custom tune…an arguement that ensures the MS idea goes no where.


[quote=“turbo329is” post=60932]If we all need to buy new afms…[/quote]This premise is incorrect. Note that Ranger adjusted the sweep arm to get full contact with the board through the entire range of motion. A new AFM is not required if that fixes your problem.


The right way to fix this AFM would have been to bend the arm so that the contact point traces a different path on the resistor plate. But I’m not sure that would have helped in this case. It isn’t uncommon to find a worn spot lower down, but it would be unusual to find a worn spot at the top end of the track. The decrease in resistance there might mean that there is a fault in the resistor network. I’d just replace the AFM.


Your right megasquirt is a bad idea for us, but there are a lot of reasonably priced ways to get rid of the flapper box now days. My brother has been tuning a customers e30m3 using a system that replaces the afm signal to the ecu with a signal from a wideband o2. I don’t have a wideband on my spec car but I assume most of you do. I’ll see if I can get more info on it this weekend.


I wouldn’t bother. The community isn’t going to support the idea of swapping out engine managment parts. A Megasquirt with a demonstrably OEM tune in order to get troubleshooting logs and prevent chipping would provide clear advantages with no impact to performance.

In contrast…All of our engine management parts are available new. The idea of swapping out OEM engine management parts for enhanced performance aftermarket parts, because they are “enhanced performance”, isn’t going to find too many friends.

I run a wideband O2 sensor because I have an A/F meter on my dash. I don’t recall seeing another one in SpecE30. For a long time I didn’t trust the A/F meter because it always told me my engine was way lean at high rpm. This was before the hiccuping. 12months later I’d been fighting this high rpm hiccuping for 8 months and someone looked at my spark plugs and told me “your engine is running lean”. That was when I realized that maybe my A/F meter had been right the whole time.


Running lean is a potentially catastrophic reliability issue. Owning 5 AFMs is a cost and PITA issue. I don’t advocate tuning or additional cost. I just don’t think that our AFM is worth its weight in pot metal. I’m pretty sure theres a limited tunability option that costs less than a new AFM or 5 used ones.


Don’t get me wrong, I’m not trying to be a hater on your idea.

This past year I’ve done far more work on the engine bay’s electrical system than I had before. One of the things I’ve learned from that effort is that all the wiring and all the connectors are heavily tarnished. If you pull a connector off you will find that the wire is black with tarnish.

There are parallels between us and the other spec classes, but something that is unique to us is the age of our cars. Devinney had the right approach where he bought a new harness, but to really do it right you’d also have to replace everything electrical that is outside of the passenger compartment and therefore exposed to the elements, and is performance related. That includes the fuse box, the round connector (C101?) next to it, the fuel pump and it’s wiring, etc.

In a perfect world we could buy a kit that would allow us to rip out all of the old wiring and engine management gear that can’t be relied upon. But since most folks haven’t gone thru the engine management problems that a few of us have had, the issue doesn’t really resonate with folks. And with the size of the customer base and the fact that our cars are slowly disappearing into history, no one’s going to put a bunch of capital into creating a pared down wiring and engine management kit for us. So we’re just going to have to do the best we can.

5yrs from now our cars will be antiques and it will be all about Spec3 anyways.


spec (e36) m3 is what i really want!


Personally I think spece30 will dominate until a spec e46 comes along. Unfortunately the e46 had a lot of engine options though. I’ve never considered owning an e36.


Very cool info and test Ranger. I am going to check mine. I have 3 AFMs and have seen big differences from one to another, even before they let us open them. I am going to look at this. Thanks for posting!

I heard it is impossible to find a new AFM, true? If not where can I get a new one?


IE’s replacement part site lists them. It says call for price buts shows retail being 550.


Ranger, I saw the same inconsistent resistance readings until I tried this arrangement using a 9 volt battery. Things then appeared much more linear.

Great discussion.


I thought it was common practice to adjust the arms? No?

There is a MAF conversion from Miller. My car came with this when I purchased. I actually sold it and converted back. Runs like poop in comparison.


[quote=“cdonnelly” post=60973]Ranger, I saw the same inconsistent resistance readings until I tried this arrangement using a 9 volt battery. Things then appeared much more linear.

Great discussion.[/quote]

That was interesting, thx for sharing the link.

I did precisely what the guy told us not to do. Changed the range of how the swiper rotated. It didn’t occur to me to try to make the swiper run in a smaller or larger radius. Had there been a problem in the mid-range of sweep somewhere, than I’d a been thinking about how to change the swiper’s radius instead. It’s entirely possible that changing the swiper’s radius would have also fixed the problem.

It’s hard to understad how measuring the resistance is futile, you have to measure voltage. The relationship between voltage and resistance wouldn’t challenge my 8yr old.

The article makes the point re. how finely these are calibrated by Bosch. In contrast all my spare AFM’s had resistance ranges all over the map. That makes me wonder if the Palacio solution, a new AFM, isn’t the right approach.

The deeper I get into this the more I wonder if all of our 25+ yr old engine management systems, by that I mean every single component in every single car, aren’t just completely thrashed.


So does our afm have the early or late resistance curve. If the curve is linear I think it would be possible to replace it with a MAP sensor and a little bit of circuitry.


I wasn’t really paying attention to linear or logarithmic when I did all the testing, I was just looking for “predictable”. That being said, my guess is linear. I think that if the resistance had been logarithmic I’d a noticed.

The e model uses a different circuit board. Maybe the difference between e model and i model AFM circuit boards is that the e model is logarithmic. The dates for the 944 AFM type change are in the right ballpark for our e to i change.


[quote=“theShoe” post=60970]
I heard it is impossible to find a new AFM, true? If not where can I get a new one?[/quote]PM Julio Palacio.


You can get one from a dealer for something over $500 or get one from an online vendor like for ~$170. In either case you will be getting a remanufactured unit.