Well the old bus went for its MOT the other day and failed:

Not entirely surprised, as the front balljoints are knackered. Amusingly, they only picked up one, and it wasnt the one that i'd identified myself. So i've ordered a set of balljoints and a press kit to sort those.

Number plate is easy enough to fix, although the holders are very yellowed. I believe its just a 501 wedge bulb?

Which leads us to emissions...

Last year it was an absolute nightmare getting it thru the emissions test. LPG was broken so it was being tested on petrol, meaning the harsh CAT test with its 0.3% CO limit was being applied. It had a faulty MAF and a missfire (which is why it had been laid up for so long) but eventually i replaced the MAF, Both lambdas, all 8 injectors, spark plugs and a few other bits, and finally got it running nicely, and it passed the full CAT test on petrol.

I've since fixed the LPG system by replacing the injectors and re-calibrating, and for the most part it runs and drives perfectly well on LPG. However it failed the emissions test on LPG this time round, managing 3.87% CO against a limit of 3.5%... Waay over what it managed last year on petrol.

I dont understand why the CO is so high. Its running closed loop, so why is the lambdas/ECU allowing it to end up so rich?!

Sounds a bit like take it somewhere else!!! Spotted the ball joint but not identified the one you have seen.

Chuck £35 at another garage!

I've tried a few round here and this one is pretty good. the others always pluck bullshit fails out their arse and are generally not very friendly. One failed me for not having a Cat fitted on my Audi, i literally stood there like WTF, then pointed at the Cat in the engine bay and said "whats that then?".

Sure, he missed the balljoint, but its very hard to see (the boot has split around the wire retainer on the inside) and as people who've worked on these will know, the balljoints are particularly difficult to feel play in compared with a normal car.

Why would i want my failure list to get any longer anyway? :P

Last year my Audi failed on about 10 items. Took it elsewhere and it passed! Garage was investigated and found many test tools out of calibration and a higher average of fails. Always seemed to be able to do the work quickly though.

Getting back to the emissions, what do the lambda sensors show when running on LPG? If they are staying at one place then it really is running rich, if they are flip flopping about, then the mixture is correct and the tester didn't set his machine correctly. Did it give a figure for lambda? If it did, although he was testing to LPG standards the machine was measuring emissions as if it was on petrol.

Always a good idea to impress upon any MOT garage that you will be doing all your own repairs, so from their point of view a fail will mean them retesting the vehicle for free within the next couple of weeks ;-)

Yeh the guys know i fix everything myself.

I've not had a chance to hook it up, and i'm away this weekend, but next week i'll get nanocom out and see if there are any faults first off, and also if the lambdas are switching as expected.

There was no lambda figure on the sheet, just HC and CO, and it had "non-CAT test" across the top. The limits were also the 3.5% rather than the 0.3% it would have got on petrol so i'm pretty sure the machine was properly setup for LPG.

Still not fixed this yet, been working on the balljoints in the other thread, but very peculiar yesterday morning...

Its not moved in 3 weeks since it failed the MOT. Jumped in and fired it up to move it out of its parking area and onto the driveway. Now usually, it fires up very slightly lumpy, and needs a wee rev to clear its throat. However yesterday it fired up running on about 5 cylinders, coughing spluttering popping and farting and generally really unhappy. revving it didnt clear it, and it filled the street with smoke (blue to begin with, then black) I ended up moving it by holding the throttle open, and using the brake to control the speed.

I did find a minute to check the fault codes. It said "amfr correction at its maximum negative value"

Another bad MAF?

MAF was dead when i bought the car, and it ran ok cold, but stopped once hot. Unfortunately genuine ones are about 5 million quid, so the only option was to buy a used one, which appears to have now also died.

I was trying to find someone who could provide the MAF translation table from the ECU. If we know the voltage vs airflow curve, it would be very possible to engineer a replacement using a common modern MAF sensor. Unfortunately i found nothing.

Aragorn wrote:

I was trying to find someone who could provide the MAF translation table from the ECU. If we know the voltage vs airflow curve, it would be very possible to engineer a replacement using a common modern MAF sensor. Unfortunately i found nothing.

Something from the dark side covering this issue : http://www.rangerovers.net/forum/7-range-rover-mark-ii-p38/41999-info-mystery-bosch-maf.html

Dunno what idiot thought feeding a low range, high sensitivity, control / feedback signal as a DC voltage down best part of a yard of wire to an under-bonnet ECU was a good idea. But whoever it was deserves a good kick in the fork. Absolutely not the way to do it. If you must go DC current loop is the solid way. Better an AC frequency modulated system of some kind. There are some very elegant and inexpensive methods of implementing such on wheatstone bridge type sensors.

In principle its a simple data logging exercise of volts and rpm. In practice you'd need to know both mass airflow and velocity over the sensor to be sure you were comparing like with like. Which means a test bed and good deal of knowing WTHIGO. I imagine the cheapy crap aftermarket types are crap because the test bed side of things wasn't properly done. My experience is its easy to get sort of close some of the time on this type of measurement exercise (albeit in very different fields) but getting it dead right all the time is a whole n'other ball game. But I would say that as demonstrating the getting it right part after some semi-demented scientists handwaving had attracted managerial attention was what I used to get paid for.

I'd be very unsurprised to discover that a substantial improvement in mpg could be got if the MAF, ECU and Fueling / Ignition maps were properly re-engineered with a more sensible airflow sensor.

Clive

AFAIK there are no MAF translation tables for GEMS MAF. The one Clive refers to is Thor, which is a different type of sensor.
All the data available, which can be checked with Nano, is
Air mass flow at sea level - Engine fully warm, in neutral gear, with all loads off
At Idle 20 kg/hr ± 3 kg\hr ...............................................................................
At 2500 rev/min 61 kg\hr ± 3 kg\hr ................................................................
Only solution I found for the unavailability problem is to buy used Lucas/ Sagem ones, give them a thorough clean, measure MAF rates with Nano and bin the 75% that don't match up.

The MAF sends a voltage to the ECU. The ECU has a lookup table which converts that voltage to an airflow figure. The airflow is then used to compute the fuelling required.

Having spent some time messing with tuning VAG group ECU's i know for instance that the bosch MAF's have a curve like this (the scale is voltage against air mass in kg/hr. with a 200kg/hr offset as the sensor can read negative airflow). This table can simply be lifted directly from the ECU programming:

Theres obviously many MAFs and they all have different curves, the chart above shows a few different VAG models, but the data is fairly readily available, and using that data we can immediately lookup a MAF voltage and know the airflow that voltage represents.

A similar table will exist for the Sagem 20AM but finding it is proving elusive.

If you knew the table for the Sagem, you could get a simple arduino or similar, to sample the input from a chosen Bosch maf, lookup what the sagem would be outputting at the same airflow, and then output that voltage to the ECU.

We could then purchase a cheap £80 Genuine Bosch MAF and bin off the knackered used 20AM's. Unfortunately such tuning information doesnt seem to be available for GEMS. I presume a few people have figured out tuning GEMS, but typically info like that is kept supersecret.

Nanocom can show both the airflow and the voltage from the AFM so if you can find one that is running correctly and showing the correct airflow against revs, then you can also read the voltages that correspond.

In fact Aragorn I was the person 'from the Dark Side' (RRS.net) thread mentioned in #9 (albeit as a 'guest' - later banned...) who posted that Bosch MAF data - as part of an attempt to see if there was a way to find out if those 'clones' on e-bay were any good (or as bad as some Thorites had noted, C/O several from ebay with bad/iffy results...)

• The main problem was that one volt either way from these could really make quite a difference (and the clones, even with many of the right markings, holographic stickers etc could certainly give false Airflow readings).

Bosch were really quite helpful in providing that data (and tips on avoiding 'clones')

Sagem were not helpful at all as far as my requests for GEMS MAF data were concerned.... perhaps they thought I was a Chinese plagiarist ??

Incidentally as you could see I tried (on the Dark Side) to encourage folks to do RPM/Voltage readings to compile a 'reference comparison chart'...
but ultimately only a few provided the data... Perhaps we could do that here (as we don't all have a nanos).

Finally as you probably know the (AIR) temp sensor arrangements for THOR and GEMS MAFs are different....

the clones are universally reported to be shite, but i dont have a problem spending £80 on a genuine Bosch unit. If the Sagem MAF was similar money, i'd just buy one and then i know its right. But they're somehow £800... Insane. For £800 i could rip the whole lot out and fit something like Emerald!

With no data available I guess best thing to do is to set up a comparison experiment to measure the outputs of both SAGEM and Bosch sensors under appropriate conditions.

Nice thing about having Arduino and similar processors around is that you can brute force things by taking comprehensive measurements and similarly brute force the conversions with a look up table.

Best to start with some on the road measurements of a good SAGEM MAF to keep things honest. Simultaneously recording plenum chamber depression would help.

Way I'd tackle things is to fix a complete intake system, including air filter and manifold to a suitably large airtight box. 50 to 100 litres should do it. Fit two decent size pipes to the intake of a Henry or other decently powerful vacuum cleaner. One goes to the box, the other goes to a relief valve thingy allowing the effective plenum chamber depression to be controlled. Valve open essentially no depression in the box and little air flow. Valve shut full depression and full air flow. Probably a plate with various holes in it and suitable close off caps would be easier to handle than a valve. Theoretically need around 1500 litres per minute / 25 litres per second of air to cover maximum demand with reasonable margin. Henry is specced for 42 litres per second so that should do.

Plot TPS output for full close to open and back throttle swings against MAF output for both types over a sensible range of relief valve settings. Compare relevant part of curves and calculate look up table. May need to incorporate a separate plenum chamber depression sensor into the look up table.

If the temperature compensation is going to be an issue I guess you could always butcher a fan heater to warm things up during testing.

Given smooth curves from both sensors it should be possible to get things to work OK. If the curves aren't smooth then there are probably flow issues around the sensor bodies requiring fiddle factors in the ECU to get a true representation of flow. So long as the conversion table for the Bosch gives same output a SAGEM under all conditions the ECU won't care.

Clive

Aragorn wrote:

the clones are universally reported to be shite, but i dont have a problem spending £80 on a genuine Bosch unit. If the Sagem MAF was similar money, i'd just buy one and then i know its right. But they're somehow £800... Insane. For £800 i could rip the whole lot out and fit something like Emerald!

• Not sure if your remark also applies to the SAGEM clones @ £60. I would personally try one of these first anyway.....
  Happy to measure RPM/Volts on my own GEMS if it helps !

Clive603 wrote:

Given smooth curves from both sensors it should be possible to get things to work OK. If the curves aren't smooth then there are probably flow issues around the sensor bodies requiring fiddle factors in the ECU to get a true representation of flow. So long as the conversion table for the Bosch gives same output a SAGEM under all conditions the ECU won't care.

-One of the things I noted was that the same Bosch MAF was often used for different cars but 'simply' using different-sized diameter intakes:
Without getting too far into the Fluid Mechanics I would suspect any 'fiddle factors' may not be linear enough though across all rev. ranges....(?)

Run it on LPG and tune it by alpha-N :)

(just kidding)

I was thinking about just connecting the bosch maf and sagem in series and then driving the car with datalogging on both outputs.

Analysing the data afterwards will give you the voltage for some given airflow for both sensors and you know if bosch says X sagem says Y. Smooth all the points and you end up with a curve.

Even better would be to do the same thing on a test bed/flow bench type arrangement, as doing it on engine makes it more difficult to read every data point

However for that to work you need a known perfect Sagem MAF, and those clearly dont exist else we wouldnt be having this conversation. Hence me thinking that pulling that data from the ECU was the way to go.

Anyway, its all getting away from the point, as currently i need an MOT, and that means i need a working MAF, not a 6 month electronics project :P

Does anyone have a known good Sagem knocking around?

Well for completeness, went for its MOT today with the new used MAF installed.

Only other things i did was replaced the plugs (for good measure) and reset the adaptions on the ECU.

Readings at the last test: 3.75% CO....
Readings at this test: 0% CO

:)

So that suggests that a) its fixed, and b) the cats are actually still working!