mpg meters

[HoloBarre©®]

All--

I'm interested in measuring vehicle miles per gallon, based on the
odometer signal and the fuel injector signals, which are all available at
the engine's ECM (electronic control module). Flowmeters are the benchmark
for such measurements, but for multi-cylinder vehicles, a real invasive
(plumbing) nightmare.
One representative fuel injector's signal would seem to be the best
and least invasive compromise--not an absolute measure of true mpg's, but
likely a very accurate "relative" indicator, good enough for my purposes,
which could be pretty well calibrated over time.
I presume I would need some spec's from the auto mfr (nissan, in
this case), but it's tough getting past Customer ""Service"".
Any ideas on how to obtain specs? To proceed in general? I don't
think the workshop manual would help w/ this type of info, but then....
Knowing the array of diagnositc equipment for a vehicle might
help, so maybe a well-equipped svc. dept might help.
Any leads would be greatly appreciated.

 

[Ralph Barone]

All--

I'm interested in measuring vehicle miles per gallon, based on the
odometer signal and the fuel injector signals, which are all available at
the engine's ECM (electronic control module). Flowmeters are the benchmark
for such measurements, but for multi-cylinder vehicles, a real invasive
(plumbing) nightmare.
One representative fuel injector's signal would seem to be the best
and least invasive compromise--not an absolute measure of true mpg's, but
likely a very accurate "relative" indicator, good enough for my purposes,
which could be pretty well calibrated over time.
I presume I would need some spec's from the auto mfr (nissan, in
this case), but it's tough getting past Customer ""Service"".
Any ideas on how to obtain specs? To proceed in general? I don't
think the workshop manual would help w/ this type of info, but then....
Knowing the array of diagnositc equipment for a vehicle might
help, so maybe a well-equipped svc. dept might help.
Any leads would be greatly appreciated.

I've been thinking about this for a while too. As I see it, there are
(at least) two obvious approaches.

1) Use a PIC to accumulate counts from a high speed clock while the fuel
injector signal is high. On every rising edge of the odometer signal,
divide some magical constant by the PIC count and you have MPG.

2) Take the pulse from the fuel injector driver and run it into an
integrator. Adjust integrator gain as needed to calibrate. Run the
odometer signal into a counter. Run the integrator output into a
comparator and use the comparator output to reset the integrator and
send the counter output to a display.

I suspect that the signals from the fuel injector driver and the
odometer are 0 - 12 V signals. There is a possibility that the fuel
injector driver signal is active low.

 

[Joerg]

Hi Mr.P.V.,

Ralph explained circuitry options. As to calibration data the only ways
I see:

A: Find a decent forum on the web where some knowledgeable techs hang
out who could give you the fuel injector flow information.

B: Drive under a typical condition where you know the gas mileage of
your car from long experience. Such as a long flat stretch of freeway,
no wind, constant speed. Log the injector pulse frequency or have a
passenger do that. Now you have a benchmark number.

With respect to flowmeters the number of cylinders doesn't have much to
do with their complexity. They measure the flow in the fuel line to the
injector unit and that will always be one line regardless of the
cylinder number.

Regards, Joerg

 

[Roger Hamlett]

Hi Mr.P.V.,

Ralph explained circuitry options. As to calibration data the only ways
I see:

A: Find a decent forum on the web where some knowledgeable techs hang
out who could give you the fuel injector flow information.

B: Drive under a typical condition where you know the gas mileage of
your car from long experience. Such as a long flat stretch of freeway,
no wind, constant speed. Log the injector pulse frequency or have a
passenger do that. Now you have a benchmark number.

With respect to flowmeters the number of cylinders doesn't have much to
do with their complexity. They measure the flow in the fuel line to the
injector unit and that will always be one line regardless of the
cylinder number.

Regards, Joerg

On most modern cars, you use two flowmeters, since they circulate fuel
from the rear of the car to the presurised feed to the injectors, and
return the unused fuel to the tank.

Best Wishes

 

[Pat Ford]

I've been thinking about this for a while too. As I see it, there are
(at least) two obvious approaches.

1) Use a PIC to accumulate counts from a high speed clock while the fuel
injector signal is high. On every rising edge of the odometer signal,
divide some magical constant by the PIC count and you have MPG.

2) Take the pulse from the fuel injector driver and run it into an
integrator. Adjust integrator gain as needed to calibrate. Run the
odometer signal into a counter. Run the integrator output into a
comparator and use the comparator output to reset the integrator and
send the counter output to a display.

I suspect that the signals from the fuel injector driver and the
odometer are 0 - 12 V signals. There is a possibility that the fuel
injector driver signal is active low.

CCInk had a project to do this a few years back.
Might look there.
Pat

 

[Pat Ford]

All--

I'm interested in measuring vehicle miles per gallon, based on the
odometer signal and the fuel injector signals, which are all available at
the engine's ECM (electronic control module). Flowmeters are the benchmark
for such measurements, but for multi-cylinder vehicles, a real invasive
(plumbing) nightmare.
One representative fuel injector's signal would seem to be the best
and least invasive compromise--not an absolute measure of true mpg's, but
likely a very accurate "relative" indicator, good enough for my purposes,
which could be pretty well calibrated over time.
I presume I would need some spec's from the auto mfr (nissan, in
this case), but it's tough getting past Customer ""Service"".
Any ideas on how to obtain specs? To proceed in general? I don't
think the workshop manual would help w/ this type of info, but then....
Knowing the array of diagnositc equipment for a vehicle might
help, so maybe a well-equipped svc. dept might help.
Any leads would be greatly appreciated.

try asking on some of the Nissan tech lists about the #per hour of the
injectors ( or CC/minute) count the on time of the injector. Multiply the
two to get comsumption per time, then get the speed ( tap into the speed ref
for the cruise control)
Pat

 

[Bob Stephens]

On most modern cars, you use two flowmeters, since they circulate fuel
from the rear of the car to the presurised feed to the injectors, and
return the unused fuel to the tank.

Best Wishes

Really? Why is it necessary to return fuel to the tank?

 

[Pat Ford]

Really? Why is it necessary to return fuel to the tank?

To avoid vapour lock or purge vapours back to the tank.
Pat

 

[James Beck]

try asking on some of the Nissan tech lists about the #per hour of the
injectors ( or CC/minute) count the on time of the injector. Multiply the
two to get comsumption per time, then get the speed ( tap into the speed ref
for the cruise control)
Pat

That will work as long as his fuel pressure regulator is exactly within
spec. The injectors are going to be rated at VOL/TIME @ a specific fuel
pressure. I guess you might come pretty close, but as the regulator
ages and the pressure drifts, it is anyones guess. Also, as the
injectors age and get worn and dirty the computer will adjust the on
time to keep the mix correct as far as emmisions are concerned, that
will cause errors if just on time is used. Once again, I have never
tried it, so the data might be close enough. The only long term stable
solution I can think of right now, would be to place flow sensors in the
out going and return lines of the tank and do the math there.

Jim

 

[HoloBarre©®]

try asking on some of the Nissan tech lists about the #per hour of the
injectors ( or CC/minute) count the on time of the injector. Multiply the
two to get comsumption per time, then get the speed ( tap into the speed ref
for the cruise control)
Pat

That will work as long as his fuel pressure regulator is exactly within
spec. The injectors are going to be rated at VOL/TIME @ a specific fuel
pressure. I guess you might come pretty close, but as the regulator
ages and the pressure drifts, it is anyones guess. Also, as the
injectors age and get worn and dirty the computer will adjust the on
time to keep the mix correct as far as emmisions are concerned, that
will cause errors if just on time is used. Once again, I have never
tried it, so the data might be close enough. The only long term stable
solution I can think of right now, would be to place flow sensors in the
out going and return lines of the tank and do the math there.

Jim
-----------------------

Even here, you are calculating a relatively small flow *difference*
from relatively large flows--from what I've read. If a clear fuel line "tap"
was evident, then you could use just one flowmeter. And flowmeters are not
cheap! And the plumbing is still considerable.
What you say about injectors is exactly true, but likely still the
best compromise. Thus, tapping off existing ports seems to be the
expeditious way to go.
I appreciate everyones' insights!

However, as I barely grokked the theory of tube amps, I would need
someone skilled in the nitty gritty of counters, PICs, etc. to do this--on
my car. Would local EE grads/undergrads be capable of undertaking this?

Would there be advantages, either way, to taking the odometer
pulses before the processor (right off the wheel), or the signal after the
processor (on its way to the LCD)?

FYI, according to Nissan, they calc THEIR mpgs by sensing the
height level in the fuel tank!! Flow sensor indeed!!!!

 

[Jim Thompson]

Last time I was involved in something of this sort, it was "think
automotive, think cheap"...

Relatively "accurate" tank float gives volume, odometer system gives
distance, uP calculates MPG, miles-to-go, etc. Accurate flowmeters
are expensive.

...Jim Thompson

 

[Bob Stephens]

That will work as long as his fuel pressure regulator is exactly within
spec. The injectors are going to be rated at VOL/TIME @ a specific fuel
pressure. I guess you might come pretty close, but as the regulator
ages and the pressure drifts, it is anyones guess. Also, as the
injectors age and get worn and dirty the computer will adjust the on
time to keep the mix correct as far as emmisions are concerned, that
will cause errors if just on time is used. Once again, I have never
tried it, so the data might be close enough. The only long term stable
solution I can think of right now, would be to place flow sensors in the
out going and return lines of the tank and do the math there.

Jim
-----------------------

Even here, you are calculating a relatively small flow *difference*
from relatively large flows--from what I've read. If a clear fuel line "tap"
was evident, then you could use just one flowmeter. And flowmeters are not
cheap! And the plumbing is still considerable.
What you say about injectors is exactly true, but likely still the
best compromise. Thus, tapping off existing ports seems to be the
expeditious way to go.
I appreciate everyones' insights!

However, as I barely grokked the theory of tube amps, I would need
someone skilled in the nitty gritty of counters, PICs, etc. to do this--on
my car. Would local EE grads/undergrads be capable of undertaking this?

Would there be advantages, either way, to taking the odometer
pulses before the processor (right off the wheel), or the signal after the
processor (on its way to the LCD)?

FYI, according to Nissan, they calc THEIR mpgs by sensing the
height level in the fuel tank!! Flow sensor indeed!!!!

Don't know what your budget is, but numerous manufacturer's offer OBD-II
scanners - either stand alone or OBD to RS232 for use with a laptop or PDA
which display all sensor information, error codes and let you reset those
pesky Maintenance Required flags.


Bob

 

[Goran Larsson]

The only long term stable
solution I can think of right now, would be to place flow sensors in the
out going and return lines of the tank and do the math there.

Why are you ignoring any long term instability in the flow sensors?

Many years ago I wrote software for a fuel consumption display for a
fishing boat. The hardware had flow sensors (feed and return) and at
slow speed most of the feed fuel was returned unused. In this situation
you had two sensors that had some tolerance and also reduced precision
due to vibrations. Taking the difference between the two sensors
resulted in what looked like noise, so much noise that the fuel
consumtion sometines was negative. It required lots of filtering
(software) to get any useful fuel consumption numbers out of that
hardware installation.

I much prefer how it is done in the cars I have owned, basing the
consumption calculations on the time the fuel injectors are open.

 

[James Beck]

-----------------------

Even here, you are calculating a relatively small flow *difference*
from relatively large flows--from what I've read. If a clear fuel line "tap"
was evident, then you could use just one flowmeter. And flowmeters are not
cheap! And the plumbing is still considerable.
What you say about injectors is exactly true, but likely still the
best compromise. Thus, tapping off existing ports seems to be the
expeditious way to go.
I appreciate everyones' insights!

However, as I barely grokked the theory of tube amps, I would need
someone skilled in the nitty gritty of counters, PICs, etc. to do this--on
my car. Would local EE grads/undergrads be capable of undertaking this?

Would there be advantages, either way, to taking the odometer
pulses before the processor (right off the wheel), or the signal after the
processor (on its way to the LCD)?

FYI, according to Nissan, they calc THEIR mpgs by sensing the
height level in the fuel tank!! Flow sensor indeed!!!!

My wifes car does a similar thing with MPG and Range estimations.
I have found that they REALLY stretch the estimation at times.
If that kind of accuracy is OK then go with it. You never mentioned any
design accuracy constraints, so I was just going with what I thought the
most accurate way was that wasn't going to be too affected by outside
factors.

Jim

 

[Jim Thompson]

----------------------- [snip]
FYI, according to Nissan, they calc THEIR mpgs by sensing the
height level in the fuel tank!! Flow sensor indeed!!!!

My wifes car does a similar thing with MPG and Range estimations.
I have found that they REALLY stretch the estimation at times.
If that kind of accuracy is OK then go with it. You never mentioned any
design accuracy constraints, so I was just going with what I thought the
most accurate way was that wasn't going to be too affected by outside
factors.

Jim

Why not do long-term calculations based on float position and
short-term based on injector duty-cycle? Basically let a uP do all
the calibration from the two results?

...Jim Thompson

 

[HoloBarre©®]

Why are you ignoring any long term instability in the flow sensors?

Many years ago I wrote software for a fuel consumption display for a
fishing boat. The hardware had flow sensors (feed and return) and at
slow speed most of the feed fuel was returned unused. In this situation
you had two sensors that had some tolerance and also reduced precision
due to vibrations. Taking the difference between the two sensors
resulted in what looked like noise, so much noise that the fuel
consumtion sometines was negative. It required lots of filtering
(software) to get any useful fuel consumption numbers out of that
hardware installation.

I much prefer how it is done in the cars I have owned, basing the
consumption calculations on the time the fuel injectors are open.

Do you recall what car this was?? It would be interesting to know
what cars use what system.
What signal would a float send?? Rheostatic resistance?
I think Jim's idea of combined tank float/injector data is a good
one.
I'm not sure to what level an OBD-II read out goes; I don't think
it would do signal times or duty cycle of injectors, but they are not bad to
have around anyway!
Thanks,

 

[Goran Larsson]

Do you recall what car this was?? It would be interesting to know
what cars use what system.

First example: Saab 9000 from 1988. This car used a lot of electronics,
but the various units were separate and connected using "simple"
signals. The EDU (Electronic Display Unit) in the instrument cluster
had a wire directly connected to one of the fuel injectors. Together
with a speed signal from the speedometer unit (it receives the speed
signal from a small AC generator mounted on the gearbox) it could
display the current and average fuel consumption values. All models
between 1985 and about 1997 did it this way.

Second example: Saab 9-5 from 1999. This car has much more electronics
and is much more integrated. Almost everything is communicating using
CAN busses. Anyway, the Trionic 7 unit (Engine Management Computer)
calculates the fuel needed for the next combustion. The value is
converted to a "open time" for each individual fuel injector (the
combustion is monitored by using the spark plug as an ion sensor, so
each cylinder is individually controlled) and also to a fuel quantity
(I think it is in grams of fuel) that is sent over a CAN bus to the
main instrument unit. The main instrument unit also receives the speed
from the ABS sensor on the left rear wheel. Using the speed and
accumulated fuel quantities it can then calculate the fuel consumtion
and broadcast it on another CAN bus. The information display, if
requested, shows the fuel consumption. All models between 1998 and
today does it the same way.

What signal would a float send?? Rheostatic resistance?

That must be the most common implementation in cars. Both the 9000 and
the 9-5 use the rheostatic fuel level float to calculate the distance
to empty, except that below a certain volume they regard the float as
unreliable and start subtracting used fuel from the last trusted float
value. They can't do it this way all the time as there is no flow
sensor on the tank filler tube, i.e. they don't know when fuel is added
unless they use the float. The fishing boat I wrote software for,
however, did keep track of available fuel using flow sensors on the
filler pipes. Expensive and unreliable in my view.

 

[Rich Grise]

All--

I'm interested in measuring vehicle miles per gallon, based on the
odometer signal and the fuel injector signals, which are all available at
the engine's ECM (electronic control module). Flowmeters are the benchmark
for such measurements, but for multi-cylinder vehicles, a real invasive
(plumbing) nightmare.
One representative fuel injector's signal would seem to be the best
and least invasive compromise--not an absolute measure of true mpg's, but
likely a very accurate "relative" indicator, good enough for my purposes,
which could be pretty well calibrated over time.
I presume I would need some spec's from the auto mfr (nissan, in
this case), but it's tough getting past Customer ""Service"".
Any ideas on how to obtain specs? To proceed in general? I don't
think the workshop manual would help w/ this type of info, but then....
Knowing the array of diagnositc equipment for a vehicle might
help, so maybe a well-equipped svc. dept might help.
Any leads would be greatly appreciated.

Are you looking at the signal that drives the injector? Or does the
processor have an output for flow rate? If you need to cut a line to
interpose a flow meter, why not the regular fuel line?

Good Luck!
Rich