Go Down

Topic: Fuel injection (Read 9 times) previous topic - next topic

PeterH

One thing that worries me about your hardware is that you have a relatively large dead-end fuel rail. That means you're going to be vulnerable to heat soak effects. if you haven't looked into this you might be surprised how much the viscosity of petrol changes with temperature. If you have any chance, I'd suggest you replace the center feed fuel rail with an end feed one - supply from one end, go past the injector (or injectors) and put your pressure regulator at the other end. This means the rail never stagnates so you don't the temperature swings in the fuel.

You will also need to measure the injector supply voltage very carefully and adjust your pulse length to compensate for voltage changes. The higher the supply voltage the faster the injector opens. The closing speed of course is determined by the characteristics of the injector itself and relatively fixed. If your injector is sized correctly, at low load / low speed you will be running with pulse lengths similar to the opening time (I believe erogation is the term EFI people use) and you will need to control the pulse duration very precisely.

With TBI you also have to pay particular attention to the temporal distribution of the fuel that goes into an engine cycle. With multi-point injection each cylinder's fuel is dropped into the port and more or less has to go into the right cylinder. But when you are injecting a sequence of pulses into a column of air that is being divided into multiple cylinders, not always smoothly and continuously, you need to design your injection system to mix the fuel/air together rather than have lumps of rich/lean mixture chasing each other down the manifold. Of course you don't have this problem with carbs because the fuel is mixed evenly with the air as it passes through the carb.

It's doable (I've done it) but it's not easy and there is a lot of factors to deal with to get this working right and I'd guess you have several man-years of work ahead of you to develop from scratch something that works as well as a simple carb. It's a fun project, but it's a lot of work.

I only provide help via the forum - please do not contact me for private consultancy.

gypsy_feet


One thing that worries me about your hardware is that you have a relatively large dead-end fuel rail. That means you're going to be vulnerable to heat soak effects. if you haven't looked into this you might be surprised how much the viscosity of petrol changes with temperature. If you have any chance, I'd suggest you replace the center feed fuel rail with an end feed one - supply from one end, go past the injector (or injectors) and put your pressure regulator at the other end. This means the rail never stagnates so you don't the temperature swings in the fuel. I actually have a bakelite (sp?) spacer that goes beneath the TBI to insulate it from heat. There's actually very little chance that any fuel would "stagnate" in this rail, as it flows from center to the injectors, which are at the termini. Literally the end is where the injector lies. I am slightly confused by your suggestion on the positioning of the regulator... The fuel needs to go through the regulator in order for it to regulate the pressure. Are you by chance talking about the pressure relief valve?

You will also need to measure the injector supply voltage very carefully and adjust your pulse length to compensate for voltage changes. The higher the supply voltage the faster the injector opens. The closing speed of course is determined by the characteristics of the injector itself and relatively fixed. If your injector is sized correctly, at low load / low speed you will be running with pulse lengths similar to the opening time (I believe erogation is the term EFI people use) and you will need to control the pulse duration very precisely.
The charging system on most cars is fairly stable to my knowledge... I can understand it making a difference with a couple volts of variation, but the regulator (again, to my knowledge) pretty well hovers right at 14V... Please correct me if I'm wrong. If it does vary much, I can simply add in volt/amp to the program.
With TBI you also have to pay particular attention to the temporal distribution of the fuel that goes into an engine cycle. With multi-point injection each cylinder's fuel is dropped into the port and more or less has to go into the right cylinder. But when you are injecting a sequence of pulses into a column of air that is being divided into multiple cylinders, not always smoothly and continuously, you need to design your injection system to mix the fuel/air together rather than have lumps of rich/lean mixture chasing each other down the manifold. Of course you don't have this problem with carbs because the fuel is mixed evenly with the air as it passes through the carb.
Much consideration was given to the proper "atomization" of fuel while I was building the unit... The fuel has to pass the air coming in from the throttle body in a cross-pattern, then due to the tubes having open sides the air has to cross in a figure-8. The whole system is open enough for good airflow, yet turbulent enough (hopefully) to mix the fuel well. That's also part of the reason why I want to have the injectors fire alternately, so that it's a more consistent fuel distribution vs. the effect of both injectors firing at once. And actually, even with carbs (depending on quality and the type of manifold) "pooling" of fuel or just lousy mixture is still an issue. Only time will tell whether this manifold will mix it well.
It's doable (I've done it) but it's not easy and there is a lot of factors to deal with to get this working right and I'd guess you have several man-years of work ahead of you to develop from scratch something that works as well as a simple carb. It's a fun project, but it's a lot of work.
I'd love to hear more about your project! Was it also with an Arduino, or another custom circuit? That's actually half the purpose of this project. I'm pretty confident that it'll work (and get better MPG than the current carb, @25mpg  :( ), but a main point is to improve my programming skills and also to better understand EFI in general.


martianent

#32
Jan 10, 2012, 04:17 am Last Edit: Jan 10, 2012, 04:27 am by martianent Reason: 1
1.  Mass air flow (MAF) is NOT necessary.  GM did NOT use it in most V6 models (J, W, N bodies, for starters) between 86 and 93.  They used a MAP (manifold absolute pressure) sensor.  And as such, calculated the air flow from the pressure in the manifold and of the atmosphere (speed density system, same as the MegaSquirt).  I have 2 such engine control computers (GM P/N 1227730, 1227727) and possibly another in my garage.  And, also, computer controlled carburetors did NOT use a MAF, either.  At least GM carbs didn't (I have a factory service manual that proves this for the GM LG4 305 carbureted V8).

2.  Manifold pressure measurement, however, IS VERY necessary.  If you don't know the pressure of the air in the intake, you don't know how much fuel to add to it to reach the proper air/fuel ratio.  You CANNOT get away without this sensor, PERIOD.  Early computer-controlled carburetors even used a MAP/BARO sensor.  You may even find your Omni had one.  Actually, maybe not.  Most of the Chrysler engine control computers used an integrated sensor (I can't recall at this moment whether it was a MAF or a MAP.  I'm tending to lean towards MAF, but I believe they also used a MAP).  Let's put it this way.  You set the tables for your fuel injection up in, say, Key West, Florida, down by the ocean.  You then take a trip to Denver, the Mile High city.  Your engine runs pig rich.  Why, because the air flowing into the intake is less dense and has less pressure.  Keep in mind that air pressure changes with altitude (as you've already mentioned outdoor equipment and such).

3.  Yes, you DO need to adjust injector PW according to available system voltage.  Alternators do crap out, you know, they may also have trouble charging the system at low engine speeds, and you WILL need to adjust the initial injector PW according to initial startup temperature, pressure, and system voltage.

You may want to look at this page:  http://www.megamanual.com/ms2/pcb.htm  for the base schematics to the MegaSquirt to see what makes it tick.  Because you were curious about how the MOSFET injector controls work.

I agree with the fuel reflow.  Most fuel injection systems, and ALL TBI systems, are designed so the fuel flows in a loop, so the fuel doesn't overheat and create something called vapor lock, where the fuel heats in spots, evaporates, and creates a vapor bubble that is useless to the fuel injection system and combustion process.  My fuel rail, for instance, takes fuel in one side, feeds 3 injectors, transfers to the other side, feeds the other 3 injectors, and the pressure regulator is on the end of the second tube, blocking the flow of fuel out of the outlet (return) port and line back to the fuel tank until such a time as the spring pressure on the other side of the FPR diaphragm is overcome.  Chrysler TBI works in an identical manner.  This prevents vapor lock.  And vapor lock can happen ANYWHERE in the fuel system.  Knowing how the fuel lines are run in Chrysler vehicles, this could be a problem.

And yes, Jemmicat, you can use a one-wire O2S.  ALL current manufacturers do, for the most part.  The other 3 wires are for the heating element in the oxygen sensor that helps it warm up quicker, to provide the proper output signal faster so as to cut engine emissions faster.  I have seen and used actual sensor readout data on lab scopes from late models, and it does not look like a wideband output.

kf2qd

You may also want to look at Engine RPM. It gives you another way to also calculate air flow ( at any rpm you are going to move a certain amount of air) and give you another check on your other numbers. A little redundancy is sometimes a good thing.

PeterH

You seem to be describing a carb style fuel system, where the pump feeds a regulator that is essentially a variable restrictor maintaining constant pressure at the regulator output.

This is not suitable for an injection system. You will need a new high pressure pump, and a new regulator, and you will need to completely replace the fuel lines (fuel lines designed for carbs are not rated for the pressures used in injection systems). The pressure the injectors are designed to work at will vary but typically will be around 2.5 - 3.5 bar. You need to find out what range of pressures your injectors will support, and decide what pressure you want to run them at within this range.

The fuel system will consist of:
tank -> high pressure injection pump -> high pressure filter -> fuel rail -> regulator -> return to tank

The regulator will be a variable restrictor that maintains a constant pressure on the upstream side i.e. within the fuel rail. The regulator can be referenced to fresh air, or to manifold pressure. I strongly suggest you reference it to manifold pressure as this simplifies the fuel flow calculations and control considerably.

You can achieve this circuit by teeing your existing fuel rail in to the hose between the pump and the regulator. This would be good enough to run the engine, but means that the only fuel going into the fuel rail (and hoses leading to it) is the fuel that is being injected. This means that most of the time the flow rate will be very low, and you will suffer from lots of fuel heating. Having the fuel temperature vary is bad news because the fuel viscosity changes considerably with temperature. Many production systems actually measure the fuel temperature in the rail and explicitly compensate for this but you can get away without that as long as you keep a good flow through the rail so that the fuel doesn't get hot. Depending how hot the engine bay is you may also find that the fuel rail suffers from vapour lock and a dead-end system will be more vulnerable to that, too, but that is a different problem.

Have you sized your injectors based on your engine's peak fuel requirements? What is your maximum expected injector duty cycle at full power?

The battery voltage on your car will vary wildly depending on the loads on the system. The injectors are basically solenoids working against a spring and the opening time is highly dependent on supply voltage. You will need to adjust the pulse times to compensate for the effects of supply voltage changes.

You will need a good rpm signal for this to work, and a load sensor. For load you could use mass air flow or manifold absolute pressure. As a last resort you could use throttle angle although that's much harder to map because  the correlation between throttle angle and mass air flow is very poor. MAP is by far the easiest to set up and is what I'd recommend. You will need a throttle angle input anyway, to trigger your transient enrichment (aka 'throttle pump'). You also need a coolant temp input.

If the fuel economy is current poor, that suggests there's a problem with the current setup. Perhaps it's just the carb not working very well, but you might have some other problem such as weak spark, bad timing, dirty plugs, leaking stat, air leak, worn cam etc causing the problem. On the assumption that it's the carb, and since you will need to have some means to monitor EGO anyway in order to tune your injection setup, I suggest you get an EGO sensor installed and see what AFRs you're getting currently. Hopefully you'll see they're way out under some conditions. If you know your carbs, then armed with that information about where they're wrong you could actually get the carbs corrected if you want. In any case, seeing the problem you are aiming to fix seems like a good idea, before you invest large amounts of time or money fixing it.
I only provide help via the forum - please do not contact me for private consultancy.

martianent

Well, I don't know about his Omni, but my grandmother had a Horizon (same car) that the carb had a malfunction in and even the dealer NEVER could get right, so she sold it for a TBI Shadow.  Which is probably why he wants to switch to fuel injection.

Does the engine in that thing have the distributor on the front of the engine, that's driven by a drive shaft with 2 slots in it, equidistant from each other (180* apart)?  If so, you may be able to use a later-model distributor, which has a Hall sensor in it.  This would give you RPM signals, but how it would interact with, what I am assuming is, your points-type ignition system, I don't know at the moment.  I'll need to consult some manuals to find out of the Hall distributor connects directly to the ECU or not.

Laminar


You CANNOT get by with as few sensors as you think. How are you going to know when to trigger the injectors? As magnethead was saying, you need to perform a timing map on the pulses. They need to happen just before each intake valve opens.How do you do this? With a crank position sensor… You can't just put one of these on easily so you need to come up with another way to find position and from revolution to revolution this timing map and pulse cycle will probably be changing. You can maybe do this without a crank position sensor, but you need to know when and how long each of the 4 intake valves are open and time your pulses on the injector accordingly. If you try to just randomly inject fuel without this, you run a risk of blowing a lot of things. You could possibly use a sensor on the tachometer for the pulses, but again, you have to tie your pulses back to the opening of each cylinder.


A crank position sensor won't tell you when the intake valve is opening, as the crank rotates twice for each opening of the intake valve. You'd need a cam position sensor if you wanted to do sequential port injection.

Sequential injection isn't required, the standard Megasquirt is for batch or sometimes bank injection. They did a setup with sequential injection and found that it did improve idle, but there are literally thousands of Megasquirt-equipped cars out there running batch injection.

http://www.megamanual.com/v22manual/mfuel.htm#batch

PeterH

Sequential injection not necessary on any engine, and it's not possible on a TBI setup like the one being discussed. You can ignore any issues relating to injection phasing. You do, however, need an engine RPM input and a load input, and you will need (at least) a 2-D map of required fuel quantity against rpm and load.
I only provide help via the forum - please do not contact me for private consultancy.

joey120373

I would love to see this project get done, however there sure seems to be a lot of conflicting info flying about!

To the OP, i would reiterate what has already beed suggested, go to msefi.com and spend some time reading!

I would love to see an adruino based fuel injection system created, but why try to reinvent the wheel here?

Years ago I built and installed a megasquirt system on my car, (an "MS1" using the motorolla HC908 (if memory serves me
)) it was fuel only, basically a carb replacement.

There is a ton of usefull information on the MS site, including using different load sensors, MAPvsRPM(manifold air pressure an RPM aka speed density) or  MAF (mass air flow) or  TPSvsRPM (throttle position vs RPM aka. alpha/n)

there is also a ton of info on various hardware used to controll things like injectors and coils and such.

I would echo statements be others, using any O2 sensor (1,2,3,4,6 or 7 wire) as anything but a small fuel trim andjustment is probly not a good idea, If it were that simple, it would have been done....The system needs to calculate required fuel pulse in advance of the actual event, and the O2 signal is generated long after the event has taken place. There are all kinds of ways to fire injectors, but I would stick to using components that were specifically designed for doing this, there are tons of them out there and most of them are dirt cheep.

Battery voltage correction is a big deal, air temp is also a big deal

goodluck

llopness

I'm new to the forums, but I'm stoked to find this thread is current, unlike this lawnmower EFI, note the lawnmower can get away without a map sensor because it's a governed engine designed to run at a single RPM (kinda like how my tractor's turbo doesn't have a wastegate).
I've a 125cc engine I'd like to EFI in practice for electrifying my 84 280SL 2.8L engine.
in my present level of ignorance I believe I'll use:

  • an SSR : they're simpler to employ than transistors for the non-electrical engineers, they're optically isolated, and have plenty enough power to run an injector or two or three.  plus, 10 shiny new opto 22 ODC5's came in the mail today and are sitting on my laptop

  • a fuel injector: I believe there're high impedance injectors which draw about 1 amp (the lexus injector on my laptop measures 14Ohms) and there're low impedance injectors which are supposed to have a peak draw of 6+ amps then drop to ~4 amps(it seems to me like they're a racing thing for super high revs and super accurate response)

  • Tachometer input: there must be some safe way of grabbing this from the spark or I could use a hall sensor, which would give good platform for electronic ignition control

  • Manifold Air Pressure sensor: essential for calculating the volume of air by multiplying it with RPM and engine displacement and some constants like .5 (for 4 cycle engines) and the engines volumetric efficiency, etc. 

  • throttle position sensor: I'm not sure it's super essential except for giving some richness for added acceleration, but I'm sure there may be other uses.

  • coolant temp sensor: I'm not sure I'm going to use one because the bike is air cooled.  I'll might instead opt for programming in a couple minutes of richness after the engine start, like the old autochoke with it's heating element and wax, or a choke button.

  • fuel pump: this thing won't work off gravity feed like the carb

  • O2 sensor:  this I expect to use just for the initial programming.  they're kinda pricey, but it seems like it'll be worth the closed loop feedback.  it seems like the MTX-L is the item to get.  they've even got analog outputs.



PeterH



note the lawnmower can get away without a map sensor because it's a governed engine designed to run at a single RPM



The fact the engine is governed doesn't negate the need for a load sensor. I suppose it's probably governed by something that throttle the air going in somehow. It must change the power output somehow in order to govern the speed, and throttling the air supply is the only sensible way to do that. If it does it by throttling the air supply then you need to know how much air the engine is getting so you can supply the corresponding amount of fuel. The most sensible way to measure that is with a MAP sensor. You could use a MAF sensor if you want to be fancy, or infer it from the throttle position (if the governor uses the throttle) but a MAP sensor is likely to be the best compromise between simplicity and accuracy.
I only provide help via the forum - please do not contact me for private consultancy.

Go Up