Hi,

I have found out the reason my K100 was a slug and only accelerated like a family car: the air flow meter is worn out.

It still has pretty good top end power, it just take forever to get there.

while searching another problem I started measuring and checking everything electrical and found the resistance of the flow meter was all over the place. Totally inconsistent and up and down radically as the gate is slowly opened.

I ended up taking the lid off and found that the carbon track is fairly worn Not too suprising with number of times the air flow must have changed in 20 odd years of service !!.

I have to say these are beautiful instruments but the carbon track is certainly the weak link. I have tested at least four second-hand ones and they are all in about the same state ( or worse ).

I rang the local BMW dealer and they told me it was 600 euros for a new one.  Well that's about half what the bike is worth so pretty disproportionate.

I was expecting to buy just the little ceramic plate with the laser trimmed resistors and the carbon track ( the rest is perifect ) but they will only sell me the whole sealed unit with the ali casing and all.

Does anyone have any suggestions on how to refurbish / fix it , or now of anyone providing such a service?

TIA.

Motorworks Motobins or Sherlocks for parts. Sterling is good just now.

We never buy new. May even be a few lying around locally.

Thanks, I'm looking specifically fix this part. I suppose I may get lucky and find  crashed bike that sitll have a serviceable one. The problem is the bike still runs when these are on the way out, so mostly people don't notice.

Everyone will tell you they have one "in good condition" meaning it 'looks' nice. That's because they live in the air fitler box and generally come out looking almost like new.

Ask someone to test resistance and they go all quiet on you, or offer to take it back if it doesn't work ( At my cost for the return postage no doubt. ). 

That is why I was wanting to know if anyone offers a refurbishing service or any ingenious owner has found a hack or source for the ceramic part.

These same barndoor sensors are used on Volvos and other European cars with Jetronic FI fitted.  Search a breaker's yard for one. Take your old one along for a comparison.

robmack wrote:These same barndoor sensors are used on Volvos and other European cars with Jetronic FI fitted.  Search a breaker's yard for one.  Take your old one along for a comparison.

When I wrote the page about troubleshooting the FI, all the measurements were from 3 used MAF. It would be interesting to see the difference with a brand new one.

robmack wrote:These same barndoor sensors are used on Volvos and other European cars with Jetronic FI fitted.  Search a breaker's yard for one.  Take your old one along for a comparison.

Thanks, valuable info.  Do you have anything more specific about what make/models used this unit?

Crazy Frog wrote:

robmack wrote:These same barndoor sensors are used on Volvos and other European cars with Jetronic FI fitted.  Search a breaker's yard for one.  Take your old one along for a comparison.

When I wrote the page about troubleshooting the FI, all the measurements were from 3 used MAF. It would be interesting to see the difference with a brand new one.

What's a MAF ?  Is that the flow meter?

What is this 'page' you refer to?

Sorry, I don't have a list of specific car models that used the same air flow meter, but they should be same era BMW 5 Series E12 (528, 530), Volkswagens and possibly Volvos (750) or maybe Audi.  I'm going by literature that I have stored away on my computer about Jetronic and LE-Jetronic.

The MAF (mass Air Flow) meter is also referred to as an Air Flow Meter (AFM).  This is not to be confused with a Mass Airflow Sensor (MAS) which I believe to be a hot-wire device.  Someone please correct me if I'm mistaken; I can never quite keep these acronyms straight all the time.

Bert is referring to his troubleshooting page, a link to which is available on the Portal page above.

BMW refer to them as a hot film sensor but I think the name depends on the maker of the vehicle.
Halo who has moved on and found a new interest (bit like Mr Toad of "Wind in the Willows") was working on a hot film sensor to replace the barn door meter but I don't think there were any results.

Bert is referring to his troubleshooting page, a link to which is available on the Portal page above.

Wow, outstanding guide. Many thanks for putting that together.

I found a pdf on another site that shows the power wires the other way around !

It appears that the chip may accept either polarity ( fool proofed ? ). Spec says +30V to -40V.

http://www.bbautomacao.com/home_hall_effect_sensor_cyhme56.html
The original Honeywell part said something similar.

Does that mean it has an internal bridge rectifier and will work either way, or simply that it is protected against backwards connection?

robmack wrote:Sorry, I don't have a list of specific car models that used the same air flow meter, but they should be same era BMW 5 Series E12 (528, 530), Volkswagens and possibly Volvos (750) or maybe Audi.  I'm going by literature that I have stored away on my computer about Jetronic and LE-Jetronic.

The MAF (mass Air Flow) meter is also referred to as an Air Flow Meter (AFM).  This is not to be confused with a Mass Airflow Sensor (MAS) which I believe to be a hot-wire device.  Someone please correct me if I'm mistaken; I can never quite keep these acronyms straight all the time.

Bert is referring to his troubleshooting page, a link to which is available on the Portal page above.

The primary difference between an Air flow meter and and a Mass Airflow Sensor is the air flow meter simply measures the volume of air.  The mass airflow sensor measures the density of the air.  Used primarily to compensate for altitude effects and temperature compensation.  Early gen EFI [ my 71 Volvo P1800ES had this type AFM.  My 88 740 Turbo had the MAS type.  The MAS is better at determing the 'value' of the air.

Yes, of course. I had not thought of relocating the ceramic PCB. It is just he centre portion that seems to have suffered too much wear, so a mm or so should give it a clean run. It looks like I could elongate the mounting holes to shift the line of the wiper arm on the carbon track.

There is still plenty of contact on the central arm that you indicate.so that's not the problem here.


How constant should the resistance readings be on good one? I was expecting a perfectly stable value like a standard electrical potentiometer, that moves evenly and progressively. Is that realistic for these units?

There is a little play in the shaft but that is not cause of the irratic readings I am seeing on the bench. What sort of bearings does it have, I kind of assumed it was just steel shaft in nylon block, by the look of it.

Hmm, I'm not sure that a simple resistance measurement is the right way to test this unit.

There is a resistor chain with carefully, laser-cut resistors on the ceramic support. Each interval is of the order of 10 ohms but they vary progressively.

It seems that the carbon track is much higher resistance. This implies that the hard resistor chain is a potential divider that is fed with a fixed, controlled voltage. The carbon track is a high resistance path whaich probably draws little current and whose resistance may rather variable.

I think this needs to be tested with a voltage applied to the chain and testing resistance at the wiper with a voltmeter ( ie drawing negligible current ). This probably nearer to the way the EFI unit is driving/using it.

It's very likely driven with a 5V regulated supply. The total resistor chain is a little over 510 ohms, drawing 10mA.

There is clearly a very light contact force on the wiper, presumably carefully calibrated for maximum lifetime. The contact resistance may not be critical in that context as long as the track is not worn through.

Yes, that did it. Feeding it with 5V between pin 5 and E with a bench power supply I get a nice stable output between the wiper ( pin7 ) and E. It goes from 4.45V down to 1.5V when in the fully open position.


My guess is that these are factory filled with an inert gas or 'dry' air before being sealed. So subsequent to further testing I'll find a can of water-free air and seal up with silicon paste. ( This air can be bought in electronics or computer shops and techy bits of some supermarkets - for a silly price, of course. )

The resistance chain, measured relative to pin 5 looks like this ( ohms ) :

52
133
215
260
292
313
334
345
351
354
357

There is an additional 6 ohm from the end of the chain to pin 8 and a further 150  ohms between pin 8 and 'E'.

The  375+6  ohms is consistent with the figures in the troubleshooting guide for the working units.

I'm guessing that E is "Erde" or earth and that the test voltage is fed in at pin 5.

There is a lock-screw which positions the rest position of the copper wiper blade. It seems that this should be set with the two point contacts on the copper wiper arm  aligned with the corner of first resistor ( green rectangle ) on the ceramic support plate.

I don't think value at the rest position is critical. I would expect the EFI firmware to read this when the bike is powered on and subtract it from later readings.

I suspect that the higher values for the problematic flow-meter cited in the trouble-shooter result from oxidation of the contacts or the carbon track being badly worn away. Oxidation seems unlikely in a factory unit but may result from an inquisitive owner having resealed the lid with humid, ambient air inside.

There is also a graduated tensioner for the 'barn-door' return spring. This could potentially be readjusted for spring fatigue if someone has the appropriate torque info. Obviously this needs marking if there is need to remove it, unless such info is found.

At the end of the day this unit may have been in serviceable condition before I opened it and the erratic resistance measurements were misleading.  Having gained a better understanding of how it works, I would suggest a voltage test, as described, not a resistance test. ( Other than the static, closed position checks in the trouble shooter. They serve as a quick sanity check )

All second-hand units I have inspected have been erratic when testing the resistance as a function of wiper position. This is probably not a valid diagnostic of the condition of the unit.

On visual inspection, the track on this one was in far better condition than my original one, so I think I'm back in business. The voltage test would have been a way to check functionality without the need to break the air seal.

That would have been worth knowing.

Beamer wrote:
There is also a graduated tensioner for the 'barn-door' return spring. This could potentially be readjusted for spring fatigue if someone has the appropriate torque info. Obviously this needs marking if there is need to remove it, unless such info is found.

The affect of adjusting the spring has been discussed here.

Thanks, I like the sound of this Halo guy. Looks interesting.

So if I follow the one notch, two notch thing, it's about slackening the spring tension to get more gas.

Seeing it in that light, I suppose age fatigue in the spring lets a bit more fuel in and compensates for general engine fatigue.

Best left alone unless doing serious tuning mods, a la Halo.

If you want more horsepower then tighten the spring a tiny bit and that makes it lean out a touch and more HP is there for the taking, but one notch too far and you will be the proud owner of four pistons with a nice hole in the crown and an expensive repair bill.

Thanks for explaining the effect of changing the tensioner. So as the spring weakens slightly with age it will tend to richen up a little.  This may merit one notch after 20y, but it's all rather flying blind.

I would have thought you'd hear some serious detonation ( pinking ) before you holed a piston though.

It seems like this is analogous to playing with needle position on a carb.

i wrote:I don't think value at the rest position is critical. I would expect the EFI firmware to read this when the bike is powered on and subtract it from later readings.

This may be incorrect. I have seen other comments that this EFI is completely analogue electronics. I was making unwarranted assumptions. Perhaps some more attention needs to be given to ensuring a correct voltage in the rest position to get things optimally set up.

They are totally analogue, next step down is a box full of relays and diodes.

Yes, I've just been looking that circuit diagram which I picked up here somewhere on the forum, Can't find a link right now.

12V going in on pin 9 , earth on pin 5 ( not sure what the E is about ). The thermistor modifies the upper arm of the potentiometer directly, some clever analogue computing going on here.This does imply the need to carefully set these things  up. There's no self-calibration on power-up, like the would be with a micro-controller driven cct.

So to be independent of variable ambient  temperature, a test needs to be done using 6V between pin 8 and pin 5.

The wiper probably needs to be set up  so that the wiper voltage is at its minimum but starts rising as soon as the vane moves. In practice this is probably means alignment with the corner of the green rectangle, as I described above.


It's good to get a bit on understanding of how this 'black box' works. I don't like black boxes on my bikes.

I've always found these K bikes to be rather slow to respond when moving off closed throttle. It's only about 1/4 to 1/2 a second but it's an annoying lag, especially opening up going into a bend. It may be interesting to look at some of the RC time constants in the box, or simply disconnect the throttle-closed contact switch.

RicK G wrote:If you want more horsepower then tighten the spring a tiny bit and that makes it lean out a touch and more HP is there for the taking, but one notch too far and you will be the proud owner of four pistons with a nice hole in the crown and an expensive repair bill.

Rick, I'm a little curious about this remark.

If I follow you, you are saying ratchet up the preload on the spring; trick the EFI that there is less air flow than is really the case and thus get a leaner mixture.

I am not sure that I understand how you could get more power by injecting less fuel. Leaner burning engines may be more economical but usually this is at the expense of losing a little power ( unless the whole thing is badly set up to start with ).

Have I miss understood?

Beamer wrote:

RicK G wrote:If you want more horsepower then tighten the spring a tiny bit and that makes it lean out a touch and more HP is there for the taking, but one notch too far and you will be the proud owner of four pistons with a nice hole in the crown and an expensive repair bill.

Rick, I'm a little curious about this remark.

If I follow you, you are saying ratchet up the preload on the spring; trick the EFI that there is less air flow than is really the case and thus get a leaner mixture.

I am not sure that I understand how you could get more power by injecting less fuel. Leaner burning engines may be more economical but usually this is at the expense of losing a little power ( unless the whole thing is badly set up to start with ).

Have I miss understood?

Running too lean will run hotter. The ideal mix is one that burns completely which is inherently hotter by a little bit.  The power an engine produces is defined by Delta Temperature, the difference between intake and exhaust temps.  This is true for all combustion engines be they turbine, diesel, gasoline or sterling.

You can run a petrol (gas) engine at 15-1 mix but they get so hot that they destroy themselves.
The leaner you can run an engine the more power, bigger bang more heat and it is really hot gas that makes for faster expansion of the burning gas so more power.
If you are racing and naturally trying to get every last ounce of power you make it as lean as possible.
With a carb engine we used to lean them out till at full throttle the engine would start laging then colse the throttle a touch and they take off then go up 1 size in jet from there and that was as lean as you could make it without too much chance of smoking a set of pistons.
Alcohol acts much the same but with less chance of damage and if we felt that at 3/4 track we felt all was under control and the car was stable we would hit the high speed leanout and it was good for an exrta 15 mph in the traps and was like getting a kick in the ass.

Just to add some more the K100s have WOT (wide open throttle) setting which works of the throttle position switch. It enriches the mix so as to avoid engine damage. I am not sure if everything that has the jetronic system has it.
When the guys were production racing they used to go 2 notches lean on the MAF so as to be a leaner mix in the top end. This worked on the track because most of the time it's either closed or WOT so being lean in the mid range meant nothing. Of course on the highway if you are running at say 200kph (120mph) it is only going to be at 2/3 throttle so there is a big danger of smoking some pistons or maybe only damaging some exhaust valves.
You were saying earlier about hearing detonation but at 120mph there is a good chance that you wont hear it and the Nikasil bores and detonation dont get on very well.  Eventually the coating starts to flake and once it starts it goes very quickly.

Thanks for the enlightening details.

The comment about Nikasil explains something. I was riding Guzzi Le Mans a few years back and someone tried to sell me some barrels. When I inspected them they had flaking. I wondered how they got into that state. I know he had some friends who were racing Guzzis so this seems like it could be the cause. 

We learn something every day. Thanks.

I had always thought Guzzis had chrome bores which would of course explain that flaking.

The Nikasil seems to be a very durable arrangement judging by the lifespan of a K engine.

They called it "Nigusil" I think. Not sure of the exact composition but it's basically the same idea. A single use hard electroplated bore instead of an iron one that you can rebore.

The di8d use Nikasil on the Le Mans as far as I recall but the original T3 I think was chrome. Nikasil is better and very durable. but you cant rebore either.

A very good description as well as history of nikasil is here:
https://en.wikipedia.org/wiki/Nikasil

Okay, so you knew I'd be along sooner or later....  

Guzzi used FOUR different types of barrel liners; chrome, cast iron, Nicasil and Nigusil. Surprisingly, you can tell which barrels have which liners by looking at the spigots. Cast iron is obvious, chrome has a very slight step in the spigot but also has a spiral machined between the step and the base of the barrel and Nicasil/Nigusil has the step but no spiral on the spigot. The difference between Nicasil and Nigusil is in the composition of the coating (nickel vs silicon, I think???). Nigusil is ridiculously tough; I have two 950cc engines on the high side of 100,000 and 150,000 miles and both are still showing hone marks in the barrels. Guzzi used Nicasil for a short period before creating Nigusil, but I'm not quite that much of a Guzzi geek to be able to tell you when.

If you leave a Guzzi standing, chrome bores rot with the combustion byproducts. The upshot is I will not start a Guzzi engine that's been standing until I've pulled the heads and checked to see which type of barrels it has.

The great thing about Nigusil is it takes exactly 100 miles to run in a brand new set of rings!!! There's almost no chance of seizing rings in a Nigusil barrel as long as they are gapped correctly.

Ah, I have just found a very good Bosch manual on the L-Jetrontic EFI and flowmeter system.
 http://foxed.ca/rx7manual/manuals/bosch%20l-jetronic%20injection%20manual.pdf

Overview on page 4 says:

Bosch wrote:
λ > 1 : Excess air ( lean mixture ) in the range λ  = 1.05 ... 1.3 . Excess air mixtures in this range results in lower fuel consumption accompanied by reduced performance.

That is in line with how I understood mixture ratio to work but contrary to what someone said above about "leaning out" to get more power.

Figure 2 also shows max power at around λ  = 0.85 ( rich ) and best fuel consumption around 1.05 ( lean )

It seems there's still something I'm not following.

If you lean it out yes you will get good economy but very poor performance and if you go out on the freeway and try to run it at 160kph for any distance like a mile and even less then as I said you will be the proud owner of a shagged engine. By poor performance I mean missing, backfiring and stuff all power.
There is a range in there that most people would not notice very much change but a racetrack or a dyno will give the figures. Example if you use injectors that flow 140cc/min then it will still run OK for most practical purposes but you may be actually down 6-8HP in the top end. use injectors at 133cc/min and all is good. Go to injectors that flow 128cc/min and you will improve fuel consumpion but at the loss of power and as you go smaller you get to the stage that it wont even run. With Petrol when you are pushing for performance there is a very small window where the mixture is right and works well. Alcohol like Methanol and ethanol will be a bit more forgiving but there is still that fine line between getting the max out of the engine and having a an engine melt down.
Having an engine running lean may be good for fuel consumption but it is a disaster waiting to happen. As an example a hot day the engine is running well but lean and you pull out to overtake and the Oh shit where did that car come from you kick it back 2 and give it the noise and next thing because it was hot already bang and you have an expensive repair.
You cant afford to run an engine too lean even on the street.

The only fuel where you can lean out and not take a chance on engine damage is LPG a Propane and Butane mix but you do lose power and you can have very high compression without detonation, around 16-1 and it is about 10 times cleaner than petrol.

Any how its after midnight here so its bedtime

Thanks Rick.  I got the impression earlier you were suggesting a notch leaner would give more power, and top speed lean-out would give an extra boost. Maybe that is a special case contrary to the general principals.

It seems these things can drift quite a bit in 20 years due to spring fatigue, which is something I mentioned and would like to check for. Also fuel has changed quite a bit in that time too, so even if factory settings have not moved they may not be optimum for today's fuel.

It looks like I was correct about aligning the wiper with the corner of the first resistor.  In fact this does not need to be disturbed to remove the wiper and adjust the position of the ceramic support.
http://www.gomog.com/articles/EFIflap.html

Most of the information on these Bosch flow meters is for cars, so I was not getting much info before because I had K100 in my search. Forget BMW bikes if you want the gen on these devices.

It seems like I need a CO2 gas meter to set the thing up properly and this is going to be a lot more difficult on a K100 than on a car where it is done easily with the motor running. Setting the spring is going to mean a lot of on and off the bike messing around.

I have seen several sites ( or youtube vids ) saying stuff like NEVER touch the spring, this is laser tuned and finely calibrated by BMW.

Frankly this is horseshit.

The laser cut resistors form part of the resistor chain on the ceramic support. There are laser cut to proivde exactly the correct profile for the multi-resistor chain which provides a calibrated profile for the potentiometer.

This is important since it should reflect a logarithmic curver which matched the logarithmic angular deflection of the vane in relation to air flow. These thick film resistors are not accurate which is why they are individually laser tuned in production.

This is not a tuning in relation to the spring , it is the fabrication of each resistance which is being adjusted. This will not move during the lifetime of the unit and is not related to the spring, or spring tensioner.

The carbon track on which the wiper runs is a notably higher resistance path with light contact pressure to reduce wear. This is why it is not a reliable ( nor valid ) test to measure the resistance externally.

It is the relatively low resistance values of the calibrated chain which set the voltage reference points along the 11 point resistor chain.  The wiper contact, pin 7, is connected to a high impedance op-amp input inside the EFI.  This mean that is it draws negligible current and does not affect the resistor chain.

The carbon track provides a smooth transition form one point on the resistor chain to the next. Thus we have a smooth transition between the calibrated reference points of the resistor chain. So ar so good.

The tension of the spring determines how much angular deflection of the vane is caused by a certain air flow ( in cu ft / min or whatever ). When the vane is nearly shut a small difference in air flow will mean a significant movement. When nearly full bore, a big change in air flow will mean a small change in angle against the spring.  This is the logarithmic nature of the air flow vs angular deflection.

To counteract this, the resistor chain is also logarithmic: smaller changes low down; higher further up. This, in principal, means the two are matched to provide a linear voltage vs air mass flow rate relationship. It is designed so that they match and produce a linear signal. The EFI design assumes this is the case.

This is where careful alignment ( calibration ) matters and why tricking the EFI by incorrectly tweaking the spring tension will not work. All this will achieve is an underswing and then an over swing in the voltage/ air mass curve.

Later models were linear and the conversion was done in firmware. However, these are purely analogue circuits and the "firmware" is done in the hardware. It is hard coded into the mechanical construction. Live with it, don't tweak it.

This means that spring fatigue will introduce non linearities and it will be beneficial to correctly recalibrate this.

I suspect this was done by factory measurement of the torque to leave the rest position and the fully open torque on the spring. Only the low end of this can be set using CO2 meter, unless the bike is on a dyno.  The top end, where the non linearities will show , will not be tested.

So that's why we have that little switch for WOT and idle for that matter

Beamer wrote:Thanks Rick.  I got the impression earlier you were suggesting a notch leaner would give more power, and top speed lean-out would give an extra boost. Maybe that is a special case contrary to the general principals.

It seems these things can drift quite a bit in 20 years due to spring fatigue, which is something I mentioned and would like to check for. Also fuel has changed quite a bit in that time too, so even if factory settings have not moved they may not be optimum for today's fuel.

It looks like I was correct about aligning the wiper with the corner of the first resistor.  In fact this does not need to be disturbed to remove the wiper and adjust the position of the ceramic support.
http://www.gomog.com/articles/EFIflap.html

Most of the information on these Bosch flow meters is for cars, so I was not getting much info before because I had K100 in my search. Forget BMW bikes if you want the gen on these devices.

It seems like I need a CO2 gas meter to set the thing up properly and this is going to be a lot more difficult on a K100 than on a car where it is done easily with the motor running. Setting the spring is going to mean a lot of on and off the bike messing around.

One of the biggest problems with working on the bosch EFI is getting info. Bosch wont supply info and there are a few people who have reverse engineered them but they also want money for the info because it took so much time to do it.
I tried to get info some years ago and ran into a brick wall of silence and then was put onto the Megasquirt open software gear and never gave Bosch a second thought.
What I was looking at doing was setting up EFI on my Z1300 Kawasaki so it would return better than 37mpg.
If you want to play seriously with EFI and not just investigate the OEM Bosch then the Mega/Micro squirt is the way to go
https://www.diyautotune.com/shop/megasquirt-assembled/microsquirt/?osCsid=8458fd523cc226b1d9a5c672b65b0329

One of the biggest problems with working on the bosch EFI is getting info. Bosch wont supply info and there are a few people who have reverse engineered them but they also want money for the info because it took so much time to do it.

Yes, I've got quite a bit of info in the last few days and it's a lot less of a black box that it was last week. However, an awful lot of stuff is guys swaggering around on youtube or elsewhere who are obviously just poking around with a screwdriver and no real understanding of the functionality of what they are poking around in. As always, misinformation abounds.

I'm not that interested in performance tuning a brick but I think after 20y these things need tuning back to spec.

Synchronising the log profile of the resistor change ( ie output voltage used by the EFI ) with the logarithmic deflection caused by physical air flow seems the key. Just matching up light load conditions on a gas meter does not seem likely to ensure that top end is anywhere near correct.

Both curves are flat low down and small errors there can still leave a significant misalignment at highway conditions. Fiddling with spring tension could shorten or lengthen airflow log curve leading to both positive and negative errors in different parts of the higher power regimes.

The corollary is, this is probably already the case due to spring fatigue, so leaving it untouched is not really a good option either.


I'm guessing that the factory calibration will adjust the min and max position torque on the spring but without any data that is not going to be reproducible.

Like most fuel systems you start by setting the high end, then adjust the slow running afterwards.
I suppose what is needed is a rolling road or highway test with a CO2 meter to correctly set the top-end mixture.

I'm not sure how practicable that's going to be on a K-bike.

Wow, real graph paper ! Where did you find that, at an antiques market ?  

There's not much load on the spring but after 20 odd years, it may well have weakened enough to merit a notch or two. I see lots of car people showing that they need to move something like 7 notches to get the right CO2 percentage.

That's speculative since I have neither a brand new unit nor factory torques to compare to but it seems unlikely the spring will not have changed with years of being under tension and constant temp. cycling.

It won't have changed "much" but this is a sensitive precision instrument and a small change in materials could correspond to a few notches.

what was the 5.75, 65mm  thing ?  Is that an outlier that you suspect was a transcription error when doing your readings?

Where is this distance measured on the vane? I could do a comparison. 

When vane is closed there will be the non metered air flow through the by-pass. This is probalby why there is a significant force on the spring even when closed.

Clearly this would be better done as angular measurement. The toothed wheel of the spring tensioner provides a convenient protractor. There are 14 teeth between WOT and closed.

One of my units has been opened already and I suspect the spring has been moved ! I'll try to plot up the two I have on the bench.

I'd have thought the fuel used in 1984 would be different in other nations to ours, so bmw must have had a midstream setting surely, yes after 20years we might need to go tighter by one notch, but that might not get us back to near 15:1 that is the stoic metric mean I believe and 14.8 the aimed at point. Does that sound right? 
Our different fuels may not even be up there with what we got in 1984. They may be worse with more additives to curb the ideal bang.
So I'd say for sure buy a used one that works and stick it on go do the rolling road test, is that what you meant.
Idle Co, I would expect it valuable to have the co at 3400 rpm too as most people do end up there or there abouts, and it would be nice to know if that value used is close to the ideal when it left the factory.
Is it possible to measure values when live, as in as the bike is running, just expect it to be different if tested by a 6v static test, or were you intending to use that data across all the map flaps you have?
If we all intend to have our bike as close to spot on as we can get them then what you are doing is going to impact everyone wanting to finesse his motor to be at its best.
Hope I'm not pestering you.

...the test is the depth of the afm door to the front of the afm body  and by far ..far more accurate and repeatable, than any angular measurement  that I can think of that any of us could achieve .

I said better because this system designed on angular displacement. You are adding a cosine dependency from an unknown starting angle. I agree it may be easier to measure. 


I have just done a test using the spring adjuster teeth. Even without taking too much care I can get repeatability of about 0.1V on the lower end readings and about 0.01V on the high end, where there is less variability. 

That is crude eyeball alignment. I need to make a guide to ensure that I align exactly to the same part of the teeth each time.  This was enough to detect that the closed position offset was bad on one of the units. I realigned the copper wiper with the corner of the first rectangle and it came back into line.

Doing a log plot it is fairly straight but veers off towards the WOT end of the range. I don't know whether this is correct and the 'log' description is not literally the case ( I suspect this is true ).

I will get a more accurate measurement method sorted and see what it looks like after shifting a few teeth.

fixed ... graph below.

ill predict that the inspiration for this long winded and wayward rant of uneducated opinion

No I'm not barking mad. Just inquisitive.

I don't see any amendment / criticism here.
The inspiration is getting my bike set up properly. It would be best to keep the rest of your testy comments to yourself. Remember the forum rules : be nice etc.

It seems the forum has to automatically set up and account with some other service. I shouild now be able to post my graph, which was you may guess made from DATA from the tests which I had already tried to post.

hopefully this will be my graph....

Ok that works. Click to enlarge.

We see that ( at least as plotted ) this is not quite a log relationship. That idea may be a simplification of what Bosch actually calculate as being the gas flow progression. I would imagine a good model of the process is not that simple.

I have expressed the output voltage at pin7 as a ratio of the input on pin8 . That way it's independent of the test voltage used.

maybe a direct link will come up as visible ...
https://i.servimg.com/u/f86/19/45/29/82/debi_t10.png


It looks like the unit which has been "renovated" is a but unstable, even once the wiper position was corrected.

BTW I use the excellent opensourse software gnuplot ( http://gnuplot.info ) to produce my graphs. I'm not going to write a tuto here on its use but here is a few lines of gnuplot commands that will produce the above graph from a text file containing a single column of data with wiper voltage at each tooth position.   The sets of readings for the two meters must be separated by at least two emply lines.

Code:



set title "Bosch flow meter test"
set xlab "spring tensioner wheel tooth number ( closed = 1 )"
set ylab "potentiometer wiper voltage"
set key top left Left reverse
vref=5.98

 plot "bosch_flowmeter.dat" u ($0):($1/vref) index 3 w p tit "modified unit", "" u ($0):(($1/vref)) index 4 w p tit "untampered unit"

here is my data , hash marks are comment lines:

Code:


### bosch air flow meter.

# Thermistor llel 150R = 56.6 ohm : T amb. ~ 15 deg. C
# test voltage on pin p gives 3.83V on p8 at this temp.
# fully open 5.85; closed 0.85 from 5.94v
# test voltage 5.98V applied between p8 p5 ( gnd)

# low end repeatability +/-0.1V; top end 0.01V

# 14 teeth on spring tensioner gear wheel used as angular measurement from closed position.

# first data is K100 hacked unit : suspect spring has been moved from factory setting.

# index 0 , K100 hacked
0.85
2.45
3.46
4.14
4.56
4.87
5.17
5.31
5.48
5.59
5.71
5.77
5.81
5.85


#1 untampered unit eyeballed teeth alignment (imprecise)
0.86
1.96
2.98
3.77
4.24
4.66
4.94
5.18
5.37
5.48
5.66
5.73
5.76
5.80
5.85
5.88

last question ...what the hell will shifting teeth do to the measurement at closed throttle ?
it is directly connected to the barn door

It won't affect closed throttle. Where did I suggest it would?

As I've said above it will affect how the two log curves relate across the range  and how they get acted upon by the EFI. This is why it is important to get the two in sync. and why a saggy spring will detune the engine.

It's pretty much like checking the ign. timing every once in a while. Even a good conditioned unit will probably drift after 20 odd years.

I'm trying to devise a means of checking and adjusting this. Someone else said he gave up due to the secrecy and lack of information. I'm hoping to find a solution.


I found a long-nosed pair of tweezers can be used to accurately align the lug of the rotating metallic base with the teeth on the spring adjuster  and this gives a reasonably accurate angular step for each measurement. The graph of the two units shows that the modified one still has the initial closed reading a bit off. This is something that will need to be checked after moving the wiper contacts in order to get a clean track.

It also has a kink higher up, which may be a reason for rejecting this unit.

The repeatability of the lower end measurements is +/-0.03V using this method. This seems to be sufficient.



https://i.servimg.com/u/f86/19/45/29/82/debi_t10.png

k75cster wrote:If we all intend to have our bike as close to spot on as we can get them then what you are doing is going to impact everyone wanting to finesse his motor to be at its best.

Yes, that's the aim but it takes some digging.

I have developed a method to check the torque on the Bosch air flow meter and this will allow evaluation of the effect of adjusting the spring tension in relation to the output voltage.

I attached a fine nylon thread to the rear arm of the rotor and loaded it progressively using water and a syringe. One cc of water is a gram and that allows adjusting the load and calculating the torque.

As before I normalised the voltage by dividing by the test voltage so the result is generalised.

This shows just show bad a messed about with flow meter can be while still looking OK on first inspection. I initially thought the jumping was errors in my set up but it reproduced something  very similar second time around.

The good unit gives an idea of the accuracy ( and defects ) of the test method.

I also note quite a difference in the point where it takes off from the zero positions. This slightly later than when the vanes starts moving so this is something else that needs checking.This may be that it is not properly set or a result of a soggy spring. This is exactly the sort of thing I wanted to verify.


https://i.servimg.com/u/f86/19/45/29/82/debi_t13.png

Two notches extra tension makes a lot of difference. ( I have realigned the closed wiper position so there is less of dead zone where the vane moves but does not affect outpout).


https://i.servimg.com/u/f86/19/45/29/82/debi_m13.png


If I shift the new plot line left the two match quite closely across the full range, excepting the wobbles due to experimental error.

However, for a given gas flow ( ie torque ) there will not be a proportional increase in control voltage. So it seems unlikely that this is a valid way to richen or weaken the mixture across the range. It needs to be set correctly as it was designed to work.

This underlines the need to set this correctly if it suffers some relaxation of the spring after 20y or service.

A low regime gas analysis or 'colourtune' plug test would be a good starting point.

Beamer wrote:
I attached a fine nylon thread to the rear arm of the rotor and loaded it progressively using water and a syringe. One cc of water is a gram and that allows adjusting the load and calculating the torque.

As before I normalised the voltage by dividing by the test voltage so the result is generalised.

G'day Beamer, I'm having trouble understanding the nylon thread, how did you load that thread with 1cc of water?

I like the idea of being able to match or index off the cog points that is a great way to align and match the voltage percentage.
One day when people write in and say they think its a bit fuelly in the bottom rev range, or a bit of an hesitation on opening the throttle, they will have something to check.