mileage computer for a carbureted engine

[dcb]

I'm gonna piggy back here too, or tell me to get my own thread

re: full flow vs full pressure. I picked up a used electric pump, and had a chance to play with it this evening, and about 10 feet of clear 5/16 tubing.
observations:
1. It takes a number of seconds to fill the tube at full flow.

2. The pump keeps running when max pressure is reached, and runs at the same speed regardless of the volume it puts out (function of differential pressure).

So let's say it takes 5 seconds to fill the coil of tubing back up to some safe distance from the top, and when the controller senses the level at the bottom, it will always runs the pump for 5 seconds. And keep track of how many 5 second runs it made. Given that there are variations from stroke to stroke, I think it makes sense to ONLY track how many times the pump was turned on for 5 seconds (for example), and not precisely how many times the piston moved.

I don't think it should be full flow though, I can't quite explain why though. Intuitively I think if the immediate pump outlet had a restriction, that would only allow a little more fuel than anticipated maximum demand would require, then it would give more predictable results as it would have to run longer to refill the tube/resivoir, and there would be less variation in the outlet pressure the pump sees.

Originally Posted by dcb

I like how far your drip counter idea has come

I think this will work, for my bike , the microcontroller just turns the pump on for a specific amount of time when the resivoir looks low, and keeps count. The pump doesn't run so long that it would overflow the resivoir, might even want a restriction on the pump outlet. I'm ready to go get a cheap fuel pump.


...

[diesel_john]

That's a good idea. the more restriction between the pump and measuring cylinder the lower the flow that it could measure accurately. restricting orifice would not have to be bigger than the sum of the main carb. jets, would it. high level detector would be redundant, with the timer, but would accomodate different pumps, but so would adjustable timing.

probably better to dial in the restriction as close as possible to get same accuracy with a smaller volume, because the smaller the measuring volume the closer to instant MPG.

i kept thinking i needed to limit the pump to single strokes, with an adjustable delay then count number of strokes to make a gallon. would almost be real time flow. maybe if i power it thur a signal light flasher it will single stroke.

on the recirculating ball flow meter (rbfm), the ball doesn't go around very fast, maybe 1 Hz.

[dcb]

Just thinking out loud here, trying to detect when fluid is in a clear tube. A laser LED would work, but so might a narrow angle LED. Might be an easier way to do this.

Get the bias right to trigger a 5 second pulse from a 555 timer (for example) when the laser hits the CDS cell. Add a relay/power transistor, and give the bike odometer a click on the same pulse.

The vent is going to be a pain.

[personx]

I'm planning to use a green LED as a sensor in my meter. The green ones are made from the same stuff as solar cells so they will generate a little current from light. The ones I have (green light but with clear lens) will put out 1.5 V from a white LED held about one inch away. I have a green lens one that will only put out .8 V at best so the clear lens ones are the way to go. I don't know what they would be after the sending light was filtered through two pieces of glass and an eighth inch of gas but that what I'm planning on using now.

[dcb]

Picked up some electronic bits on the way home:



Thought I would see how the clear tube behaved in the infrared spectrum:
the LED by itself:


the LED through the empty tube:


the LED through the tube with water in it:


Am I imagining things or is it much brighter with the liquid in the tube? That is not what I expected.


P.S. Here is the intended display device


Edit: Apparently water vapor is a good IR blocker, which explains the results. Will need to retest with gasoline.

[diesel_john]

So you will be counting how often the pump runs not how long the pump runs, correct. Or how many times the pump pulses over time to keep the tube full.

[dcb]

Originally Posted by diesel_john

So you will be counting how often the pump runs not how long the pump runs, correct.

Well, the pump will run for a fixed amount of time when triggered by the low level detector, i.e. 3 seconds, so lets call that a "pump cycle". I'm just counting how many pump cycles there were. But certainly you could deduce the total running time easily enough.

Originally Posted by diesel_john

Or how many times the pump pulses over time to keep the tube full.

just the number of pump cycles.

By playing with the "tire circumference" on the bike computer and/or the delay time I hope to come up with a useful reading on the distance travelled. i.e. 100 miles is really one gallon dispenced by the pump

[dcb]

Ok, tried the gasoline in a tube, I really should try and find one of these first next time



it is pretty much transparent at 950nm

I wasn't happy with the test, but I think that laser is going to be more of a sure thing, and I already have one.

[JohnnyGrey]

If you're confident you can design the electronics, all you need is a few inputs...

Wideband 02 / MAF

or

Wideband 02 / MAP / RPM

Meter the air going in and sniff the exhaust to figure out how much fuel was in that air.

[dcb]

Yah, but wideband O2 sensors are expensive. We were hoping for cheap

[JohnnyGrey]

If you want to go the cheap route, forget about all the flow measuring junk.

Get a decent sized 16v capacitor, maybe 100,000uf.

Take a line from the fuel pump's + terminal.

Run it through a variable current regulator, through a diode to the + of the capacitor. This is called an integrator.

Use a very high resistor (1M+) between the capacitor and a DMM to measure your fuel usage at the end of your trip. Do some trial and error and find the right current value so that 1v corresponds to 1gal. That way, the meter will be good up to 12 gallons. For a reset button, all you need is a button wired through a 10 ohm resistor between the terminals of the capacitor.

Keep a little four function calculator to get MPG numbers at the end of your trip.

[dcb]

With a better pump that would work. These are cheap constant speed, continuous running, variable output, pumps. It is a spring load piston setup where the spring provides the limited output pressure.

A fixed displacement, pressure shutoff pump would simplify things, but again at a cost.

Adding a pressure switch to the output of the cheap pump might just be close enough though...

[LostCause]

Carbureted fuel economy gauges existed during the 1970's. One unit, called the SpaceKom, occasionally shows up on Ebay. It used a flow meter and a VSS, I believe.

Good luck on your project.

- LostCause

[dcb]

My problem is the low flow rate at idle, but I never idle anyway so I don't know why I'm being so picky about it

[LostCause]

Originally Posted by dcb

My problem is the low flow rate at idle...

That's easy, increase your idle...

What do you estimate the flow rate to be? In any case, I bet the SpaceKom device was pretty inaccurate. Coming from vacuum gauges, I bet anything that resembled instant MPG feedback was sellable.

- LostCause

[dcb]

I'm guessing less than .1gpm at idle.

Anyway, I like the pressure switch idea now I got a Low Pressure Switch off of ebay. This setup is showing some promise, but my bike computer is freaking out with all the noise from the pump (resetting and not in a useful state). It is actually on the bike and mostly hooked up though, so that is extra promising

I can blow into the tube and make the computer count, so it is electrical mayhem. When I use my thumb to controll the pressure, the pump pumps up and shuts off fine, but the computer freaks.

Theory of operation:
The pressure drops as fuel is consumed,
the N/O part of the switch closes and the pump turns on,
the N/C part of the switch opens, the computer does nothing

the pressure builds back up,
the N/O part of the switch opens and the pump turns off,
the N/C part of the switch closes and the computer registers a tick

(just guessing) the pump flow rate is probably 100 times the max consumption rate of the engine, so any variance in fill volumes between low consumption times and high consumption times will be minimal.

The number of pump cycles will be porportional to the fuel consumed (fingers crossed).

Note, there are no electronics at all currently, excepting the bike computer.




more reading:
a foot of 3/8 tubing helped installation. Store only had 3/8 tees and I added a 3/8 check valve just downstream of the pump because it wasn't sealing reliably (I don't think it liked the water experiments , what with the black stuff coming out). The rest is 5/16, occasionally jammed onto a 3/8 plug, plus a few 5/16 nylon tube joiners. Attaching the pressure switch cheaply was accomodated by grinding off the threads and shoving it into some 3/8 tubing.

Maybe I just go to sciplus and get a few mechanical counters w/resets(one for current and one for tank tracking), and put this computer back on a bicycle

[dcb]

Messed around with it a little more this morning, I think I'm close. Here is a diagram of the setup(check valve omitted, add in if pump wont hold pressure):


On the bike it is giving just a real quick pulse, like maybe 2 strokes, and when idling it is another 1&1/2 minutes till the next pulse. It's looking good, but still have to get a working counter. So, guessing here, if 90 seconds between pulses = .1gph, then 3 gph would be 3 seconds between pulses?

But since this is a pressure system, it also holds promise for automobiles.

[diesel_john]

the VW burns 1 GPH, just a relative number.

Still working on the flow meter, I am using a glass bead.

What happens when the trapped air is is eliminated.

So the bike meter just needs the two pickup wires connected together to record a cycle.

If the pump tries to get two thumps in one cycle, would limiting the voltage to the pump slow it down. That's what I couldn't get it to do, singulate the thumps.

[dcb]

Cool! You got a reading! So did you go with an aluminum ball? Is it already calibrated?

re trapped air, don't know. I left it there as an accumulator so the pump and switch don't have to cycle so much. I will probably try adjusting the switch pressure down to get more pump cycles, and thus more data samples, but not so much that i'm clacking the pressure switch at 60HZ I'm not convinced that it matters how many strokes there are in a pump cycle though. Except if you are trying to get close to real-time data.

Bike meter, yup, touching the leads together works. Leave a pigtail on the reed switch and hang on to the magnet, they could be useful yet.

Oh, I picked these up for a couple bucks a piece. But unfortunately they are 120 volt (6 watts, 1.6kohm coil) and I have to rewind the coils for 12 volt operation, still figuring that out. But I like 'em and think they would make an interesting addition to the bike, one for tank trip and one for current trip fuel consumption They would just hook up in parallel with the pump and count the pump cycles.

[dcb]

Sadly, the extra pressure, low though it may be, is causing my old carbs to leak. I had to yank it all off ( excepting the sight gauge ). Cei la vie