From: John De Armond
X-Source: The Hotrod Mailing list
Date: Aug 1992
Subject: Re: MAS vs BMAP flow measuring

>The topic of inlet flow measurement has come up a few times recently,
>and John indicated that a density * speed system can provide a much
>wider range of flow measurement that a MAS can. I am curious what the
>advantages and disadvantages are to each system. From my limited
>knowledge of each, it would seem that a MAP sensor * engine RPM
>measurement would be superior to a MAS measurement in just about
>every case... The MAP sensor system should have a wider and more
>accurate range, no lag, poses no inlet restriction, and should
>cost a lot less. So what's the scoop? Why are mass-airflow sensors
>being used in so many applications? Are they more reliable or
>something?

The major consideration, at least for us hardware hackers, is the
speed-density system requries calibration and a lookup table or
other non-linear function to map the manifold pressure values
into fuel demand values.  MAF measures mass flow directly and thus
only requres only a zero and span adjustment.  Indeed the system on
my friend's turbo Z I previously described has just that.  The lookup
table is not a technical issue since digital controllers are cheaply
available but the development time is still a consideration.  I
suppose a speed-density system could be tuned without one but I
feel a session on the dyno is vital.  To properly generate a map, it
is necessary to hold the engine at a given RPM for several pressure
values and dither the mix at each.  IMHO, this one-time work is more
than offset by the simplicity of manifolding and no intake restriction.

Speed-density does NOT work well on highly modified engines that lose
manifold vacuum upon any throttle opening.   Haltech sells a special
system for such applications that uses only temperature and and
throttle position.  Not something you'd want to run on the street.

John

From: John De Armond
X-Source: The Hotrod Mailing list
Date: Aug 1992
Subject: Re: MAS vs BMAP flow measuring

>Opps, I blew it... I just though of a really BIG variable that I over-
>looked :-(
>The MAP sensor is really only measuring static pressure in the intake
>manifold upstream of the individual intake runners. On a log-style
>intake manifold, the VE would probably much more constant at any
>one STATIC pressure level. However, in a system with with tuned intake
>runners, the total pressure available at the intake port when the
>valve opens can be drastically different from the static pressure
>that the MAP sensor is reading. Since intake runner length increases
>VE at specific points in the rev range, the overall VE map will
>probably have a big hump in it somewhere. A tuned exhaust, headers,
>etc., should certainly have the same effect...
>I was trying to figure out why VE was not directly proportional to
>MAP in any one engine configuration... I guess my brain hadn't spooled-
>up and come on boost yet :-)

I agree with most everything posted to date regarding the MAP vs MAS
issue.  However, I think the tuning effort for MAP systems is a bit
overstated.  With a good user interface, tunig is as simple as
twidling the arrow keys or moving a joystick.  You find a bad spot,
you twiddle it out.  Of course, closed loop control obviates much of
the discussion.

>Talking about tuned intake runners, I saw something on TV during the
>coverage of an F1 race that was very interesting. ( I never saw this
>before) They were showing this F1 car that was normally-aspirated
>(A Jag I think) that had sliding "trumpets". The effect actually
>looked more like a trombone to me... The "trumpets" would grow and
>shrink in length as the engine speed changed. It looked pretty wild
>with these long metal horns sticking out of the side of the engine,
>madly sliding in and out.

Wow, only 30 years late. :-)  we were doing this on karts in the late 60s.
Expansion chambers too.  My mentor John Davis developed a spring steel
chamber design that allowed the whole chamber to trumpet.  All the driver
had to do was "ride gain" on the power band with a lever between his legs.

John

Newsgroups: wiz.hotrod
Subject: Re: Re- MAS vs BMAP flow mea
From: John De Armond
Date: Wednesday, Aug 05 1992, 20:26:16

>BTW, your mentor was John Davis (of C/D fame?)?  Ahhh, to rub elbows with my
>heros! :^)

Different John Davis.  My JD was famous in karting and motorcycle circles
in the 60s and early 70s for his 140 mph enduro karts, unmentionable
chemicals used as fuel, the invention of the 5 port transfer system for
two strokes, a memorable hydrazine explosion and a few other things.
Unfortunately by the time he was about 35, his drinking got to him
and he died of sirossis (sp) of the liver.

john

From: John De Armond
X-Source: The Hotrod Mailing list
Date: Aug 1992
Subject: Re: MAS vs BMAP flow measuring

>Hmmm... That's interesting. I've looked at my air-mass sensor and it
>doesn't seem to be a hot-wire device. I'm guessing that a wire should
>be visible in the air flow path if it were a hot-wire system? My
>sensor has two little transducers mounted on opposite sides of the
>inlet tube. Could this be an ultrasonic system? I've been wondering
>for quite a while just how this sensor measures the air.

Bosch uses a literal hot wire.  You can see the diamond form glow as
it cleans itself at turn-on.  GM uses a thin film resistor network mounted
so laminar flow passes over it.  Real hard to see without a small dental
mirror.

John

From: John De Armond
X-Source: The Hotrod Mailing list
Date: Aug 1992
Subject: Re: MAS vs BMAP flow measuring

Resurrecting an old discussion.

>Varying restrictions on the intake side of the system seems to be
>the thing that the MAP sensor is monitoring though, right? If pumping
>speed and capacity remains constant, then any reduction in
>manifold pressure should result in exactly the same decrease in
>volumetric efficiency... Or so it appears anyway. Of course, the
>density of the intake charge changes somewhat with temperature
>(and fuel mixture, which directly affects temperature again), but
>the volume of air pumped should be very closely proportional to
>the inlet pressure x engine RPM. (Ignoring temperature/density factors)
>The only other variable I can see right off is that valve overlap
>might allow a bit more VE at higher speeds.
>
>Anybody care to shed a little more light on this topic and help
>me get my thinking on this straightened out? :-) I'm trying real
>hard to learn to be a motor-head!

You can't ignore density.  Gas and air combine on a mass basis.
In  other words, it takes X grams of air to ideally burn Y grams
of  gasoline.   We know the ratio of X to Y is about 14.7:1 for
"standard"  gasoline and air.  The density of air varies widely
with temperature in accordance with the gas laws.  the density
of gasoline varies much less with temperature so there is no
offsetting effect.  The amount of gasoline injected MUST vary
with air density and thus the air flow must be  compensated for
temperature.  For a given set of conditions, an engine will pump
a fairly constant VOLUME of air over a fairly wide range of
temperature.  But it pumps a widely varying MASS of air over
that same range of temperature.

John

From: John De Armond
X-Source: The Hotrod Mailing list
Date: May 1993
Subject: Re: Turborcharged Two-Stroke
X-Sequence: 5266

>I will make one suggestion here.  Consider not using mass airflow.  Why?
>Because you have a very specialized application (i.e. you are writing your
>own EFI software), and you can reduce the cost by not using one.  The
>flow restriction is probably not a big deal because of the turbo.  Use an
>absolute pressure sensor and air temperature sensor (far down the intake
>after the intercooler), and possibly throttle position sensor.  You need
>these anyway, and along with your knowledge of the remainder of the engine
>you should be all set.  Mass air is nice when you don't want to (or
>can't) change code when engine mods are made, or you are operating at
>extreme weather fluctuations.

I'm a real fan of speed-density because of its simplicity and adaptability
and I almost always recommend it.  Except for 2-strokes. Problem is
2 stroke engines have essentially no manifold vacuum and what vacuum
there is goes away real fast. Then there's the severe reversion problem
on non-reed-valved engines that will drive a MAP sensor crazy.  The
problems are recognized to the extent that Haltech makes a special
speed-alpha version for two strokes.  I consider speed-alpha too crude
for anything except continuous wide open applications like racing
boats.  The MAF sensor will solve most of these problems.

Let me suggest a real simple design.  This is not my idea.  A loyal reader
sent me some Mitsubishi Starion injectors to flow and clean, along with
some pages from the service manual.  This engine is turbocharged and
uses a quasi-throttle body scheme. A pair of HUGE (>170 lb/hr) injectors
spray into a plenum upstream of the throttle but downstream of the turbo.
Here is the innovatively simple part.

The airflow meter is the variable frequency type.  IT generates a pulse
train whose frequency varies with the flow.  The ECU fires the injectors
at a fixed pulse width at some sub-frequency of the flow signal completely
asynchronous to the crankshaft or ignition.  The more air flow, the more
pulses and the more firings per unit time.  This is a very elegant solution.
The flowmeter signal has to be compensated only for air temperature
and cold starting.  I suspect the ECU does this by varying the width
of the "fixed" injector pulse.  This could be done with not much more
than a binary counter and a 555 monostable multivibrator with the
temperature sensor in the pulse width circuit.  The MAF from some of the
GM small 4 cylinder engines such as the Cavilier are variable frequency
and are about the right size. I have one here in the lab for testing.

Because of the high RPM involved and it being a 2 stroke, you may have to do
a throttle body-style design.  A revolution at 10,000 rpm is only 6 ms
so you wouldn't have much dynamic range for a port injection scheme.
You COULD go with two injectors per cylinder but injection into a plenum
would probably be simpler.

>I'd set the pulse width at
>Injectors....I can easily get 14, 19, and 30 lb./hr. from the same Mustang
>recycler, around $10 each (depending on availability, i.e. how many wasted
>engines exist).  The 35, 46, and 52 lb./hr. are harder to find because of
>their limited use.  Mabye you can even talk John into a cleaning and
>analysis, since I don't have my machine built yet.

That offer of a free analysis and cleaning is still open.

From: emory!STDVAX.GSFC.NASA.GOV!OADDAB (DIRK BROER)
X-Source: The Hotrod Mailing list
Date: Mar 1994
Subject: Re: Hot Rod ravings
X-Sequence: 7886

>-> i'm also curious as to why mustangs started out with speed density
>-> and later went to MAF systems.  any idea why?
>
> Nope.  I know a little about the TPI, but I don't know a thing about
>the Ford layout, other than that manifold sure looks crummy.

One of the things flaunted by the Ford EEC-IV computer folks is that it is
an adaptive computer.  Basically as your engine wears out, or you add free
flowing exhaust the computer compensates.  The advantage to this, at least
for hot-rodders, is that you can put almost any modification on and the car
will adjust.  The MAF sensor is only really used during open loop engine
control.  Other than that, it is closed loop through the o2 sensor.

As for looking crummy - O.K. I'll agree.  But just image - through a new
cam in and at start-up the computer cycles the engine rpm and a few other
things and with-in a minute or two it knows your cars settings and stores
them until you remove power from the computer...
Most places that modify or work with mustangs suggest that you do not alter
the prom or computer.  To increase throttle response simply bump your
static timing from 10 def BTDC to 13-14 deg BTDC and adjust you TPS sensor
to read 1.05V at idle rather than .6V as stock.

Personally I would rather have MAF.  Yes it is slightly more restrictive.
But the aftermarket sells larger sensors and can even calibrate them for
different injector sizes.  This means no special proms needed.  What is
capable from a 302?  About 360hp will put you into the low 12's.  There is
one at the local track that runs high 10's... and passes smog check....(
according to the owner).

Ford does many things different.  True dual exhaust with cross-over pipe.
Headers, Easily modified motor,  outdated chassis... and at the time I got
mine rock bottom price.

[There is a major problem with larger MAFs, particularly with super/turbo
chargers and that is turndown ratio limitations greatly affect how accurately
the meter measures airflow at LOW flows, where it really matters.
Typically the signal from the MAF is noisy at low flow.  Block learn will
partially compensate for this but the oxygen sensor is just too far
disconnected from the intake process and is too slow to handle transient
situations.  If you look in the Ford Motorsport catalog, you'll see
that most of their kits include different ECUs recalibrated for the larger
MAFs.   A MAF-equipped engine is easier to tune (assuming you can get to
the map tables) but the speed-density system is much more flexible.
As an aftermarket tuner, I don't like risk of having to buy a very
expensive MAF because some mod or the other makes the old one hit
its full scale limit.  I have a customer right now with a Mitsu 3000GT
who has tried every jerryrig he could think of trying not to have
to buy a larger MAF.  Actually, we'll probably end up having to
fabricate a larger one, as the Mitsu incorporates a baro and an
air temperature transmitter in the same housing.  What a bitch!  JGD]

From: emory!chaos.lrk.ar.us!dave.williams (Dave Williams)
X-Source: The Hotrod Mailing list
Date: Mar 1994
Subject: Re: Hot Rod ravings
X-Sequence: 7907

-> One of the things flaunted by the Ford EEC-IV computer folks is that
-> it is an adaptive computer.  Basically as your engine wears out, or
-> you add free flowing exhaust the computer compensates.  The advantage
-> to this, at least

 Yeah?  The GM TBI and TPI have *always* had that feature.  They call it
"Block Learn."


-> MAFs.   A MAF-equipped engine is easier to tune (assuming you can get
-> to the map tables) but the speed-density system is much more
-> flexible.

 All MAFs have speed-density modes anyway, for when you're off the range
of the MAF or need to go off stoich for accel/decel, etc.

[Not all.  I've not messed with a MAF-equipped GM engine enough to know
about it but I do know that many of the japanese cars DO stay on the
MAF all the time.  Specifically, the Mitsu 3000GT/Dodge Stealth and
the Nissan 300ZX do.  I would assume the Fords do too, since Ford
Motorsports sells that huge 77mm MAF for the 'Tangs.  I also know
that in the case of the two aforementioned cars, when the airflow
meter hits its limit, so does fuel delivery.  The Mitsu owner
has changed the turbos, added a larger intercooler, changed the
manifolding and added a new exhaust.  The stock MAF is pegged at
3/4 throttle and 6000 RPM.  His intake manifolds look like porcupines
from all the "aux fuelers" someone else added to try to work around
this problem.  Then he had someone else add an HKS VPS (vein air controller?
a real piece of crap.  Turns the system into a Projection.  Well almost.)
After spending several thousand dollars jerry rigging, I *think* he's
going to allow me to do it correctly.  I've proposed two approaches.
One is to fabricate a larger MAF and go to larger injectors to supply
the necessary mix.  The other is to add an aftermarket programmable
ECU, probably a Haltech.  Waiting to see what he's going to do.
That car is awsome, BTW, even a little (lot?) lean.  It accelerates
too hard for the coast-down method of horsepower measurement
to be used without a video camera.  We're going to try that
approach shortly.  JGD]

From: emory!chaos.lrk.ar.us!dave.williams (Dave Williams)
X-Source: The Hotrod Mailing list
Date: Mar 1994
Subject: Re: Hot Rod ravings
X-Sequence: 7916

-> [Not all.  I've not messed with a MAF-equipped GM engine enough to
-> know about it but I do know that many of the japanese cars DO stay on
-> the MAF all the time.  Specifically, the Mitsu 3000GT/Dodge Stealth

 How do they manage to keep from falling off one end or the other of the
MAF's range?  Maybe they're better at it than GM is.

[I haven't really looked at them in great detail but they use thin
film hot wires and generate a variable frequency output.  The variable
frequency scheme is a bunch easier to get a wide range out of and
no A/D channel in the ECU is needed.  A range of from, say 700 to
7000 rpm is only 10:1 and that's not too hard to deal with,
particularly if the rev limiter is hard so that there is a firm
upper limit on air flow.  JGD]

-> 3/4 throttle and 6000 RPM.  His intake manifolds look like porcupines
-> from all the "aux fuelers" someone else added to try to work around
-> this problem.  Then he had someone else add an HKS VPS (vein air
-> controller? a real piece of crap.  Turns the system into a
-> Projection.  Well almost.)

 Sounds like he needs to dump the whole thing and try again.  It sounds
like a nightmare to tune.

[It has several knobs on the front - just like the Projection.  Not
too difficult to tune but you DO have to fiddle with it every time
a front comes through.  Junk.  JGD]

From: emory!OAS.PSU.EDU!FBS3     (SZYMKOWSKI.FRANK)
X-Source: The Hotrod Mailing list
Date: Mar 1994
Subject: Mitsu MAF too small
X-Sequence: 7927

John, that Mitsi job you have sounds interesting. I am very much interested
in that project. Theres a guy in Jersey with a 93 twin turbo 3000 and a place
that is known for 3.8 turbo Buicks did it up with about the same things
as the one you are working on. I saw it run a 12.0 comming out on a 750
rpm idle. The trap speed was something like 119 mph. Some hp is being made
there. The place was Turbo Tune (609) 261-3774

Frank Szymkowski

[This car is probably one of the most ignored hotrods available.  Probably
because the factory turbo setting is a paltry 5 psi or such.  My customer's
car is up to 20-25 psi, depending on where he sets the knob.  The
4wd is incredible.  His is a manual.  He has cooked several clutches
drag racing it at the local strip because the engine cannot break
away all 4 tires on the launch.  I can imagine an auto would be
incredible.  The worst part about this car is that you literally
cannot see the engine when all the piping in place.  JGD]

From: John De Armond
Date: Fri Apr 12, 2002  7:40 am
Subject:  Re: [megasquirt] Advantages of mass air?
X-Source: The Megasquirt mailing list

On Fri, 12 Apr 2002 05:05:25 -0000, "joedaddyshagg" <joedaddyshagg@y...>
wrote:

>No im not telling bruce or al to add it, i just want to know if it 
>has any solid advantages, especially on highly tweaked N/A and turbo 
>apps.
>
>I know its good for altitude changes, but what else?
>
>Also suppose it was integrated into MS, are there any particularly 
>big MAF's that can be had from the junkyard? (twin turbo 5.0 big). 
>What about dual MAF's?
>
>Sorry for the questions, i just like to know whats on the table 
>before i dish up :P

There is no advantage to MAF, given an ECU that can be tuned. MAF is only
beneficial under certain conditions when the ECU cannot be retuned to
accomodate engine modifications.

MAF has a number of problems, some of which are severe. In no particular
order:

A MAF has a defined max flow. If the MAF is full scaled, say by modifying the
engine, and the ECU can't detect it, the mixture will lean because the ECU
sees less air flow than there actually is. This is a fail-dangerous condition
and is VERY bad.

On the other end, a MAF, like any measuring instrument, has a limited turndown
ratio, defined as the ratio of the max flow divided by the minimum reliable
flow measurement. A turndown of 10 to 15 is VERY good for a hot wire meter.
In industry, one would never expect a hot wire anemometer to be useable over
that range. If a flowmeter has a turndown ratio of 10 and a max flow of 1000
CFM, then it cannot measure anything below 100 cfm. Since an engine that
requires that max flow will flow more than 100 CFM at idle, the MAF cannot be
used for idle and off-idle mixture control. If the engine doesn't have a MAP
sensor, then the ECU must more-or-less guess using throttle position and RPM.
And O2 feedback in closed loop. An ECU capable of learning can learn a pretty
decent low end map but the learning always lags behind changes.

A simple hot wire MAF is omnidirectional so it will measure reversion twice -
just like a carburetor. The engine can load up, just like with a carb. A
directional MAF is possible but I'm not aware of such a device in use on a
car.

A hot wire anemometer needs several pipe diameters upstream and a couple
downstream to get a smooth enough flow to be accurate. This is impossible in
most automotive applications so various jerry-rigs are done. Things like
straightening vanes, screens, etc. All offer flow restrictions and none do a
perfect job. Bypass type flowmeters like Ford uses are particularly sensitive
to uneven flow distribution. If the turbulent flow happens to hit the sample
port, it generates a signal far out of proportion to the flow extant.

At low flow the signal is noisy and so requires filtering or averaging.

The sensing element gets dirty from oil, etc. Auto-cleaning helps but also
leads to a shortened life.

Most failure modes (open element, dirty element, over-ranged, etc) fail
dangerous and send the ECU a falsely low flow signal. this can easily result
in the production of engine kits.

MAP sensors have their own set of problems but IMHO, none even come near those
of a MAF system. Even with individual Throttle systems, one can construct a
pneumatic summing/averaging tank that hooks to each runner and the MAP and get
a decent signal. Very high overlap cams make MAP difficult but then, they
tend to make MAF difficult to impossible because of the reversion. 

Given that we have the sources for MS and can modify it to suit our needs if
necessary, we can work around problems caused by a particular engine. Witness
the work on putting the MS on rotary, harley (a problem engine if there ever
was one!) and 2-stroke engines.

Speed-density is simply the best general purpose engine control scheme known
at present. yeah, I know there are a lot of opinions out there to the
contrary. Before I consider an opinion, I want to see the opiner's
credentials. If he hasn't done an EFI of each type from scratch (completely
reprogramming an OEM computer counts here) or at least done extensive tuning
with documentation to back up his opinion, I'm not terribly interested.
Bruce'n'Al have done several systems from scratch. So have I. I've even
built my own flowmeter using an industrial hot wire anemometer. We all say
speed-density is the best for most applications. Take it to the bank!!!

John