Calibrating the AFM / adjusting mixture - Mk1 (old-style gate AFM)

Technical Area Engine, Transmission, Exhaust etc
2

Background - I was convinced my old mk1 was running way too rich, from experience of working on my old minis and other carburetted cars. That, and I was getting about 20-23mpg, black plugs and exhaust, hesitant and lumpy low-end acceleration and a general stink of over-richness. However, a methodical test of every possible sensor found no faults to explain it.

I had bought a replacement 2nd hand AFM, as I had a hunch this was the source of my problems. It made a difference, but didn’t fix it. I did notice however, that the 2 AFMs had noticeably different spring tensions on the gate, and both gave slightly different readings at idle. This made me wonder whether the spring tension was adjustable…and whether maybe the spring relaxed a little with age, or got moved in some way, leaving the AFM misreading (over-reading) the flow, resulting in over-richness. So, I set to testing it out, with startling results. Unfortunately, the workshop manual has tests for the resistances at open and closed gate, but no method for recalbration of the spring, just the usual Mazda advice to throw the old one in the bin, casting and all, and drop half the value of the car on a new one.

Method - Dead simple really. Carefully remove the silicon around the black D shaped cover on the top, and remove the cover. Clean off the old silicon and be prepared with a tube of new stuff - automotive, rather than bathroom. Make sure you get any bits out from inside. Now you’ll see the rheostat that gives a resistance reading back to the ECU depending on the air flow, and underneath there will be a clock spring mounted in a toothed wheel. To one side is the ratchet that is used at the factory to calibrate the AFM - it’s easy to see how this works. Mark the wheel first so you can go back if necessary.

By moving the toothed wheel clockwise, you tighten the spring, making the AFM read lower, and vice versa. By tightening the spring, you’re essentially leaning the mixture, because the ECU ‘sees’ less air coming in (for the same actual airflow) and so orders up less fuel from the injectors. Less fuel in the same amount of air = leaner mixture.

If you have a gas analyser, or colortune, or some other way of reading the mixture directly, then you’re in luck - just watch and adjust until it looks good. If not, you can adjust a click or two at a time and go for a drive and see how it goes, rinse and repeat. I ended up moving mine about 8 clicks clockwise, but this will be different for every car. You need to adjust the base idle each time you adjust the AFM - check out FAQ’s for how to do this if you don’t know.

I found that when it’s about right, the metal brush that contacts the black track sits at the narrow end of the 2nd wedge-shaped segment at idle on a warm engine. It sat just inside the 3rd segment when I started, so that’s very rich - and if it is in the 1st segment at idle it won’t run right at all. When I’ve found the perfect spot, I’ll post a picture. If anybody has a factory-fresh AFM who doesn’t mind opening it up, perhaps you can confirm my suspicion that this is the right spot to aim for at idle? For best results you should use an analyser of some kind, though.

When you’re finished adjusting, be sure to properly reseal the top, as it’s essential that the rheostat doesn’t become contaminated.

Result - Before, I had a significant step in power at 4k revs, it turns out this was because the <4k performance was so down, due to the richness. That’s now completely gone. The top-end performance remains much the same - the ECU is open loop at this stage, so I don’t think the AFM reading is used (correct me if I’m wrong). I’m still doing mpg calculations so will be sure that I’ve fixed the economy problem in a few days, but I’ve just had an MoT and fast idle results are: lambda=0.998, HC=018 and CO=0.029 (at natural idle CO=0.005), which are pretty good numbers so I reckon she’s close now. She smells great, the lumpiness and hesitancy is gone, and bottom-end grunt is massively better. (Dangerously better it seems, as I nearly exited a roundabout backwards the first drive out! )

As always, fiddle at your own risk, but I hope this helps if you are suffering over-richness and everything else tests out ok.

3

 Interesting post Jon, many thanks Thumbs up I went through a similar process with the mechanical advance springs in the distributor of my last MGB, charting advance against rpm, to good effect.

Does your car have a lambda sensor or other parts in the loop (showing my ignorance now) which would be in play as well as the AFM?

4

I figure the O2 sensor should be feeding back to the mixture. However, this is ignored if you bridge TEN & GND, so if using a gas analyser or colourtune, this should be done with the jumper in place. The Lambda sensor, if in place, only warns of cat failure I believe, it doesn’t feedback into the ECU for fuel mapping.

However, I come from a place of relative ignorance with this particular engine, so I’m happy to be told otherwise.

However, even with all of the sensors feeding back as normal, this process will still work. I found it was a hassle trying to get the idle right at the same time as adjusting the AFM, so I kept the jumper in place throughout.

5

Hi Jon, I’m behind on these systems but think a one oxygen sensor system will monitor exhaust and adjust fuel and a 2 sensor set up will do that and also check exhaust after the cat. I gather a slow or failing sensor commonly causes rich mixture and may not always set off the engine light if it’s not failed altogether. You’ve got the fuelling more accurate without the sensor but my concern is that if your sensor has failed you may be moving the AFM setting out of range for a new sensor to work within. That is to say, with the sensor wired out or failed it may be the case that it should be rich.

6

Yes, the O2 sensor is working perfectly. I should make it clear that prior to trying this out, I had methodically tested every possible culprit. Anybody with the same problem should do the same before trying this. This means doing/checking/correcting:

Full service (incl. air filter, fuel filter, plugs, HT leads, base idle, bottle of redex), base timing & timing belt alignment, temp sensors (bench test), O2 sensor (rich and lean resistances and reaction), AFM sensor (bench test), TPS & dashpot, CAS (voltage and switchover), coil pack (resistances and breakdown), air valve (bench test), ISC valve (clean and test), throttle body (clean), air path leakages, purge valve, cylinder compression, all earth straps (clean & copper slip)…

It’s been a looonnnng journey and this has completely transformed the running of the car. It seems clear to me that the AFM’s gate can only become looser over time, moving hundreds of times a day can’t fail to weaken the spring, loosen bearings and so forth. Logically it must surely need recalibrating after nearly 2 decades of constant service… But of course I may only be masking another issue that I’ve failed to find, and I’m always open to new suggestions! Smile

7

Hi Jon, Good stuff, so how did you test your Lambda/oxygen sensor? I’m suspecting my friend’s Mk2 which is running rich has a Lambda fault and gather the test is for voltage output. I have a decent digital meter and was wondering whether to take it off and heat it with a torch, check voltage in situe, or just swap the thing out as the car’s done about 50K and has been running rich for a while so is probably shot anyway. TBH the latter option looks most attractive but it’s easy to throw money at the wrong things isn’t it.

8

Taking things off and testing them and all that appeals to my slight OCD…so I enjoy it, myself. But replacing bits until a problem disappears works just as well. Smile

9

 Yes indeed Smile so how did you check out the Lambda?

10

All very interesting. Maybe it is psychosomatic but I think I may have a similar problem, although not as bad as yours.
Sounds well worth an experiment.
Where did you get the automotive grade sealant?

11

It’s the O2 sensor on the downpipe. I’ve not found a Lambda sensor (post-cat) on the Mk1 as yet!

You test by disconnecting it and measuring the voltage across the sensor and ground. It should sit at 0.55V at idle, rising quickly to ~0.8V when revving up, down to ~0.2V when revving down. Disconnecting the idle air supply (across the front of the head) should see <0.2V (over-lean), whereas squirting WD40 or carb cleaner into the idle air supply should see >0.8V (over-rich).

12

I have found that although I needed to go a fair way (8ish clicks), just 1 click either way makes an appreciable difference to the fuelling, so it’s well worth a try. Just remember to mark it before you start.

Any motor factor will have the silicon, or save up a bit longer and go to halfrauds… It’s just high-temp silicon that doesn’t mind oil or fuel. I used ‘Car Plan Gastite Multipurpose Silicon’.

13

To my knowledge the Air Mass flow and Lamda inputs are both used in calculating the quantity of fuel to be injected. The air mass is the first level rough calculation, take it as an open loop one, so if the AMF is inacurate your engine management will be a long way off. The Lamda sensor is then used to fine tune the system to get the optimal fuel to air ratio. So for a system to work correctly both need to be within a level of accuracy and a certain level of AMF error can be compensated for by the Lamda control loop.

But when there are transients, say when you plant your foot on the throttle, the AMF loop dominates.

The CAT can be damaged by too low a fuel to air ratio so in event of error the systems are set to err on the side of rich. 

Oh and what is called the Oxygen  sensor is a Lamda sensor I believe and it should be mounted on the down pipe prior to the CAT. It is on mine (1990 car)

 

 

14

 

 

Thanks Jon Smile that makes sense, we’ll let you know how it goes!

Lambda/oxygen sensor is the same thing btw.

15

Yes, although on more modern cars there are 2 sensors, one before and one after the cat. I’ve always heard people referring to the second one as the lambda sensor…which the mk1 doesn’t have, so just making it clear.

16

 

he he yeah, half the web is rubbish, trouble is we don’t know which half… Big Smile

17

Just re-reading that, I mistyped… It should read 0.55V at a steady 3k revs, not idle. You obviously can’t decrease the revs from idle… Smile

18

 No probs Jon, I didn’t spot that anyway Embarassed, the resistance(?) checks you first mentioned threw me somewhat but I was really more interested in if you tested it in situe or took it out, and specific figures, but they seem to be standard voltages for these devices.

I’d already book-marked a couple of good articles which suggest the lamda has a lot of control. It can be responsible for dynamic drivability issues such as hesitancy and overall mixture rather than just being a minor fuel trim.

I was also wondering if it was worthwhile testing a 50K sensor from a rich-running car. These articles suggest not and that it is almost a service-change item at about the same mileage as a cambelt, so most likely we’ll swap it out. Thinking about it my lad’s Polo had one at 40K.

http://www.picoauto.com/applications/lambda-sensor.html

http://mr2.com/TEXT/O2_Sensor.html

There are big holes in my car knowledge since leaving the trade in the mid 80’s and I’ve been catching up by working on our moderns and stuff for friends. I regressed to an MGB for a while which needed a gearbox and overdrive rebuild amongst other stuff, but modern info from guys like you who actually do these things is great Big Smile

19

Thanks :slight_smile:

20

It’s a standard 1-wire O2 sensor that’s fitted to many many vehicles. They all work in the same way (and get tested the same) as there’s really only one way to make a sensor that does this. The only difference on more modern cars is the 3 or 4-wire variants that self-heat (1-wire sensors have to get up to temp before they start working). The sensor itself generates a voltage, it’s not a variable resistance type.

The main problem with them is contamination, which can be fixed by running the engine when hot, over-lean (disconnect idle air supply pipe), at approx 3k revs, for about 30s. This will put excess oxygen over it and burn off the deposits. Take it easy though, as the cat won’t like much of that treatment…

Other than that there really is nothing to go wrong with them, they work by simple chemistry, so as long as it is still intact, and clean, it should not require replacing.