Maf Sensor Cleaning

by: ashman78


Background:

This is written for the VW/Audi V6 engines AHA and ATQ, but instructions are basically universal to all cars with these sensors.

The Mass Air Flow (MAF) sensor is an amazing piece of electronics on many modern fuel injected cars. Unfortunately, as with most amazing piece of electronics, they also cause a lot of headaches when they get older. Airborne contaminants and air filter particles (or oil) tend to gunk up the sensor element and reduce its accuracy over time. In an attempt to control building, our Hitachi sensors actually heat up to 1,000 degrees for a brief moment when the car is shut off. But even the best self-cleaning ovens still need a little scrubbing from time to time.

The MAFS relies on a tiny, heated platinum wire inside your car’s intake to determine how much air is coming in (grams per second, g/s). The ECU then utilizes this information, along with feedback from the upstream O2 sensors to alter your car’s air/fuel ratios according to driving conditions…so when the car gets incorrect information about the amount of air coming in, it can slowly wreak havoc with everything from spark plugs, valves, to your entire exhaust system (notably the catalytic converters, which often die prematurely in a rich environment).

For years, members of PassatWorld and many others have attempted to clean their MAF sensors using processes like Ziploc bags with denatured alcohol, or the more common use of electronics cleaners. Both have resulted in success in many cases. A couple years ago, a company called CRC developed the first (and only) MAF sensor cleaner on the market, taking the guesswork and delicacy out of the DIY process.

With great optimism, I bought a can of this cleaner at Advance Auto Parts and used VAG-Com to log my before and after results. Those are detailed at the bottom of this write-up. Enjoy.

Difficulty: Easier than changing the air filter

Tools:

  • CRC MAF sensor cleaner ($8, widely available at auto parts stores)
  • 8mm socket and ratchet
  • 6”+ ratchet extension
  • Flathead or phillips screwdriver

Parts:

  • none

1. Open the hood.
2. Loosen 2 hose clamps from intake flex tube and remove tube.
3. Remove 2 sensors attached to the airbox and MAF sensor itself.
4. The MAFS housing is attached to the airbox with 4 small hex bolts. Remove them.
5. Take the MAFS housing to your work area and set it on a rag or newspaper (the cleaner sprays with a lot of force).
6. Starting on the side with the wire mesh, spray the cleaner in short bursts into the wire visible in the sensor itself (it sits at the end of an antenna-like tube). Try to clean the metal screen, as well.
7. Wipe off the excess and allow to dry for a few minutes.
8. And now the obligatory DIY statement: “Re-installation is the opposite of removal!”

1-MAFSoverview.jpg
2-MAFcloseup.jpg
3-boltremoval.jpg
4-MAFalmostout.jpg
5-MAFremoved.jpg

DATA LOGGING AND RESULTS
According to Ross-Tech, makers of VAG-Com, a car’s MAFS should indicate a g/s flow of roughly 80% of the car’s rated HP. With an AHA V6 engine rated at 190bhp, I was therefore benchmarking against an ideal of 152 g/s.

In a world full of “butt dynos” and “dood I totally got better throttle response!”, it’s nice to have actual data logs for mods and repairs. The following data points are from a VAG-Com log of my car (’98 GLX V6, 5MT, 122k miles) running a K&N filter with 33k miles on it. Engine blocks 002 (for RPM and MAF flow) and 032 (for fuel trim).

These runs were made on two successive days under nearly identical conditions: 50-60 degrees F, uphill at a steep 8%-10% grade, 2nd gear, pedal to the floor from a rolling start. Two successive runs for the “before” data, two for the “after” data.

BEFORE CLEANING
Fuel trims at partial load (Bank 1, Bank 2): 12.5%, 8.6%
Max flow detected: 125.2 g/s

AFTER CLEANING
Fuel trims at partial load (Bank 1, Bank 2): 7.0%, 3.9%
Max flow detected: 152.1 g/s

Benchmark achieved!

The graph below shows the data across the majority of the powerband. The stairstep effect is due to my normalization of RPM readings across four runs (ie, each run creates samples from slightly different points, so where data didn’t exist, I reverted to the previous reading by default).
MAFlog.jpg

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License