I nominate this "best read of the week."

--
Jack "Rocket j Squirrel" Elliott
Bend, Ore.
1984 Westfalia. A poor but proud people.
1971 "Ladybug"-brand utility trailer ca. 1972 from a defunct company in
San Clemente, Calif., now repurposed as The Westrailia.

On 05/16/2012 02:37 AM, David Beierl wrote:
> At 02:40 AM 5/16/2012, Scott Daniel - Turbovans wrote:
> [If the ECU is running closed-loop (listening to the oxygen sensor in
> the exhaust) it will dial back the pulse time until the mixture is
> correct again. ]
>
>> within a fairly narrow range I think.
>
> No, it's quite a wide range.
>
>> our era vans use narrow range 02 sensors, I believe.
>
> This is true. However it doesn't affect the amount of adjustment the
> ECU can apply, only the method by which it does it. And it means
> that the ECU can only adjust to one specific mixture, the one at
> which the fuel and air charges are balanced so that each one is
> completely used up. That mixture is called the stoichiometric
> mixture and is about 14.7 parts of air by weight (mass if you're
> picky) to one of gasoline. Since that's the mixture that the
> catalytic converter needs to work properly, the narrow-band sensor
> can succeed. It is in between the best mixture for power (richer)
> and the best one for economy (leaner).
>
> The control method is called the "bang-bang" method of servo
> control. With that method, the ECU is constantly changing the
> mixture. It enriches the mixture until the O2 sensor "bangs" over
> into a rich reading, and then leans it out until the sensor "bangs"
> back into lean, then starts enriching again, rinse and repeat. The
> ECU *always* thinks the mixture is wrong and is trying to correct in
> whichever direction will make the sensor flip the other way.
>
>> they run wonderfully without an oxygen sensor anyway, and the 02's
>> limited influence is why our vanagons can run so well without a
>> working 02 sensor.
>
> Well...the engines run well without the sensor because they're
> designed to, just like any engine. It's the catalytic converter that
> requires this very precise mixture control, not the engine. I
> haven't kept up with recent engine-control techniq and it may well be
> that nowadays wide-band o2 sensors are really integrated into the
> function of the engine - I don't know. But with engines of our era
> the O2 sensor is strictly an afterthought to keep the CAT
> happy. Aside from that it's a two-edged sword. It can compensate
> for certain engine problems and make other ones worse. It turns an
> exhaust leak from a noise problem into a too-rich mixture. It turns
> a misfiring cylinder into a too-rich mixture. It turns its own
> failure or a ground fault on its lead into a *really* too-rich
> <cough, cough> mixture, so rich that the engine can have trouble
> getting away from idle.
>
>> 'normally' ..on our engines , fuel pressure is held within a fairly
>> narrow range...roughly 32 to 38 psi ..something like that..
>
> Actually it's held very closely at a constant pressure. It only
> seems to be varying because you're measuring from a different point
> than it cares about. Specifically it cares about engine vacuum,
> except when you call it that it becomes impossible to talk about
> it. So instead we talk about MAP or Manifold Absolute Pressure.
>
> Engine vacuum says how much less pressure there is in the manifold
> than in the air outside the car, and it's measured typically in
> inches of mercury, because that's how people first measured it.
>
> MAP says how much *more* pressure there is in the manifold than no
> pressure at all (i.e. a perfect vacuum), and it's measured in
> whatever unit is convenient, which is seldom inches of mercury. So
> assuming that we're at sea level and measuring in psi, when the
> engine is not running, the MAP is about 14.7 psi. If we hadn't
> already specified that we were talking about absolute pressure, we'd
> say 14.7 psia, for absolute; which is equal to 0 psig, for gauge. In
> other words, psia is referenced to complete vacuum, and psig is
> referenced to whatever environment is outside the gauge at the time
> of measurement.
>
> Back to injectors and the point of all this. Leave the O2 sensor out
> of the system, this engine is running open-loop because it's in an
> '84 1.9l Westy in Edzell Scotland in 1991 and they're still using
> leaded gas (and it has a cracked 3-4 synchromesh hub but that's
> another story). So no CAT and no sensor. The ECU has to do the job
> with no hand-holding, which means it has to know how much fuel gets
> injected into the cylinders with each injector pulse. The only thing
> it can directly control is how long it holds them open, so it's
> necessary that they flow at a constant rate no matter what else is
> going on with the engine; and the thing that controls flow rate is
> the difference in pressure between the two ends of the injector. So
> we have to keep that difference constant within fairly small
> tolerance. I'm guessing +/- 2% or better.
>
> One end of said injector is in the fuel rail and the other end is in
> the manifold, so if we're going to keep the pressure difference
> across the injector constant, the fuel pressure is going to have to
> follow the manifold pressure very closely; and that's what the fuel
> pressure regulator does. It's got a spring-loaded flexible diaphragm
> operating a blowoff valve that vents excess pressure (above 2.6 bar?)
> into the return line going back to the fuel tank. It's always
> venting because the fuel pump shoves a *lot* more fuel through there
> than the engine can use. Circulates a full tank about once an hour
> IIRC. On one side of the diaphragm is fuel pressure. But on the
> other side, the reference side, instead of outside air the diaphragm
> is connected into the manifold, and the regulator keeps *that*
> difference the same. We've got our little pressure gauge hooked up
> to the test tee and it's measuring the difference between the fuel
> system and the local atmosphere; so every time something changes in
> the engine that gauge jumps around. But if we built an airtight box
> around the gauge and connected the box to the manifold, the gauge
> should stay rock-steady at 2.6 bar any time the fuel pump is running
> - engine on, engine off, full throttle under load, idle, you name
> it. It's just inconvenient to do that and expensive to buy a gauge
> that's already set up that way, so Bentley gives you a sort of
> generally-probably set of ballpark readings. But if you *really*
> want to know what the fuel pressure regulator is doing, that's how
> you find out.
>
> Yours,
> David