mileage computer for a carbureted engine

[TheTestPilot]

Hi P-hack
I have come across an ultrasonic sensor made in England (for gasoline ),a few years back , although again it wasn't cheap from memory ,I can't seem to find it again.
I had a cursory look around for suitable sensors ,but my experience with ultrasonics comes from my days of servicing ultrasonic thickness meters ,so I am not sure what I am looking for .
I suspect a fuel flow with variable amounts of vapour may have a bearing on the accuracy of such a device ,especially at eco modder flows.

Tp

[TheTestPilot]

Greetings
I have been playing the concept of two sensors to measure the differential pressure across an orifice ,I feel more comfortable with the suitability of this over ultrasonics at this time.
One problem is finding fuel compatable sensors to do the proto ,WET sensors , individual or dual type in the range of 0.5 psi to 1.0psi is what I am looking for.
I have trolled many books & manufacturer's web sites ,no luck.

Tp

[AndrzejM]

Interesting idea. I'm hoping that you'll find what you're looking for.

[Quezacotl]

I just got few flow meters, and i'm gonna attach them to my car soon to test how it works.

Two inexpensive flow sensors, 4011-logic-ic to make some logic to in/out flows and MPGuino.

Until now, it sounds easy, but we'll see

Now i have read the entire thread. So no real results from this idea...
Like someone said, it may be tricky to get proper reading at MPGuino. If i don't have enouh skill to modify the code, i'll make a pulse counter myself.

I actually have an electronic injection in my car, but i want to test this too, just for fun. And besides i can put two meters side to side so i can monitor how much they differ.

[AndrzejM]

I was trying to modify the code myself but on my level of knowledge that will take a while to get this work properly. I'll publish the code as soon as I'll get any promising results.

[TheTestPilot]

Greetings Gentlemen
I see this thread has a lot of hits,so I assume there a lot of unused MPGuinos out there bought to use on carbureted engines?
Here is what I have done to make mine useful , until I can get time to come up with a workable flow sensor ,the circuit attatched works for my particular vehicle ,but may not work for others!
This circuit cleans up the ignition signal from the coil negative terminal,so it doesn't blow the MpGuino injector input ,the MPGuino reads the coil charge time(which is almost fixed ,so doesn't account for varying load) & is calibrated as per the calibration procedure.

This adaptor only shows CUMULATIVE litres or gallons only ,not instantaneous or Gph.
It makes the MPGuino work like a fuel guage only.
This is just food for thought ,will add more on this later tonight.

Tp

[AndrzejM]

I can't see the connection between ignition pulses and fuel used. So I don't think that's working at all. You can use your guino as a rpm, speed and distance gauge with that setup and nothing else. IMHO

[TheTestPilot]

I can certainly say ,this is working.
If you have a scope you may see the similiarity between ignition & injector pulses,other than the spark plug firing spike they look very similiar,although the ignition pulse is fixed & the injector pulse moves around according to the load.
I am on my 5th tank fill soon ,the last 4 tanks were within 3% of the pump reading , I have used the same pump every time ,this adaptor only totalises fuel used ,no other functions work properly on the MPguino.
As I say ,this is not the answer,I am trying to get people thinking outside the square on this problem.
Paddle wheel flow sensors are notoriously unreliable at "ecomodder flows" & then you have to change the code to count pulses, why not build a flow sensor to simulate an efi injector output waveform.

I have added a load sensor to another version of this "adaptor" just to see what it does ,but it will be a few weeks before I can install it.

I am still looking for suitable sensors to prototype a no moving part flow sensor(which needs no code changes), be it ultrasonics or pressure differential ,but as with R&D PROJECTS lots of ideas end up in the trash.

Does anybody want get on board with this?

Tp

[AndrzejM]

If it's working for the whole tank it should work for instant reading of FE as well. If you're connecting ignition signal instead of injector there's no need for any code changes. Leveling of the signal with the circuit that you've provided doesn't change anything from the guino side. So why it's not working for instant FE?
I can tell why you have it working with ~3% accuracy. Because you're counting revs through ignition pulses, and AFR is let's say varying in a small range so if you count revs with constant AFR you can get the amount of fuel used. But it's just "guesing" not measuring anything.

For example if you're breaking with the engine you can cut off fuel completely (even in some carbureted engines), but you still have pulses from igntion so for example if you use engine breaking for 10% of time you'll have 10% error on the tank. Will you agree with that? I know how both ignition and injection systems work, so there's no chance to measure fuel with ignition pulses.

[Superturnier]

What if...
you forgot the fuel flow sensor, since it's unreliable after all. And is hard to find... And replace that with a vacuum sensor that gives voltage according to how much engine is loaded.
Plus take rpm reading as mentioned above. Combine those signals in MPGuino code and you might have a economy gauge of some sort. As you might know the "modern EFI systems" often relay on vacuum x rpm signals (and temperature and AFR) when controlling the fuel amount.
What would still be missing is the use of a choke. That is difficult to measure. And of course the carbs are not too linear gadgets supplying the fuel...But anyhow you could do A-B-A testings with warm engine.

I have an old Zemco fuel computer in my summer vehicle. And I have been trying to calibrate it for about 5 years. Most of the false readings were caused by bubbles in fuel (heat) and the mechanical pump and the displacement of fuel filters and everything else.
Once you think it's quite accurate, it will stop reading in the middle of your most important test run ever.