Fuel Injector Recommendations (GENERAL EFI INFO)!!!
Turbo Mini Forums

Sorry Dave I don't have a scanner! Anyone to help?

Bimmer Twinky headed and turboed A-Series:
http://www.minifreunde.at/harry/projects.htm

I've seen injectors where the angle of the spray pattern can be altered, in order to bias the jet of fuel in one direction or the other.


Perhaps that would help things?

Andy

Nope.

The problem is getting enough fuel in, but not too much that it overfuels, and doesn't fall down the wrong throat.

Been doing some diagrams recently - need to make an AVI of the issues as peeps clearly can't /don't understand the root of the problem.

Hi,
Does it need to be sequencial? If you fired the injectors every 180 degrees you only need to put half the fuel in. This apperently works quite well.... Feel free to shoot me down in flames.....
Cheers,
Gavin. :)

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

No it doesn't need to be sequential if you get the injector location right - it can use batch injection firing once every 360 degrees.

I think i've explained it in this post...

Dave,
When the fuel is fired in and some lands above the closed valve wouldn't the air pressure in the port pick it up and carry it around to the open valve as the air will automatically go that way? Also if it was put in at every 180 degrees wouldn't it be going into an open valve anyway?ie. fire every 180 two of the four pulses go into each cylinder... If it don't work tell me 'cause this is the way I was thinking of doing it....... With the injectors at the manifold face.
Cheers,
Gavin. :)

Edited by Bat on 26th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

to answer your Q in the order you asked:

It might well do on a nat-asp, but more likely on a turbo it wont, as air is being constantly pushed into the ports. All it'd do if anythign is push the fuel to the far side of the undesired valve.
However... (been thinking since) what happens at 5000rpm might well be a little different as there is such a short timescale we're working to. Basically, i think if we mount them like the MPi mini, but 1" further back, it won't be too much of a problem, and it's still not quite becoming a 'wet' manifold...



Now the second part of your question:

The firing order is 1342 or to put it another way 2134
And another way:
portA-portA-PortB-PortB

now look at this diagram:


You're looking at 'one half' of the A-series here - ie the full 720 crank degrees of the four-stroke engine. ie at '12 o clock' is TDC for #1, as is '6 o clock'

Now, We're sticking to batch injection here. So the injectors on both inket tracts fire at the same time. Hence, we fire the injector 30 degrees after TDC on #1 and #4 cylinders - as i said. in both inlets, fuel is either drawn into an open valve, or temporarilly sitting in the port.

It might help you understand if you printed off the diagram and drew on in red pen the inlet valve curve for #3 and #4 - assume 'outside valve' is #1 and 'inside valve' is #2...

If you want to batch-inject every 180 degrees (without a cam-phase sensor) then you can see what'll happen. You'll fire once between 12 and 3 o-clock. this is fine as the appropriate inlet is open. You'll then fire between 3 and 6 o-clock. Not good. you'll then fire between 6 and 9 o-clock. Doubly not good - you now have two shots of 'dead fuel' sat in the inlet tract. Now (finally) you'll fire between 9 and 12 o-clock. FINALLY, the valve is open - but it's taking THREE shots of fuel down there.
I think you can imagine how undesireable this is!

Now, If you have sequential, such that you fire the injector once between 12 and 3 o-clock and once between 9 and 12 o-clock on port 1/2
and
once between 3 and 6 o-clock and once between 6 and 9 o-clock on port 3/4 - you'll have true sequential. Which is basically perfect.

You still can't overlap the injection window though... If you do - charge stealling WILL occur. AND you can't inject too early or charge stealling will still occur.

Hope that explains it.

Edited by TurboDave16V on 26th Nov, 2005.

Another thought fot you...

Injectors have a response time - a 'dead time' as it's sometimes called.

I've heard some really clever peeps saying all it needs is a wet manifold, and a single injector fired lots of times - somet like 16 times per 360 crank degrees. Well - good luck to them. Every time they're switching on and turning off the injector they're wasting usefull - indeed VALUBLE time that the injector could be flowing fuel given the limited injection window.


btw - I'm far from an expert on this, but HAVE spent plenty of time thinking about it, and tried different things. Beware of anyone saying how it's easy to get it to work. Indeed it is easy at a fixed RPM, but to get it to work at 7000rpm and 20psi is a little different to 1000-2000rpm on a nat-asp!

Hi Dave,
This is what I'm going on from Dave Walkers book 'Engine Management'...
Text reads..... "With conventional batch fired injection, fuel is injected once every revolution...... While the fuel is 'hanging around' waiting for no. 1 inlet to open, no.2 open and robs the charge. When the second half arrives that goes into the open inlet on no.2 as well.........The answer is to double the pulse phasing and half the pulse width. You inject the fuel every 180 degrees instead of 360 degrees, but only half as much in each pulse. Two of the four pulses go into each cylinder, and you get even distribution - or at least as even as the original carburettor was."



So what do you think? Looking at your valve timing diagram I can't see why not? Or am I still missing something (probably blindingly obvious too )
Cheers,
Gavin.

Edited by Bat on 27th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

Well, I'd forgotten DW said that...
Very incorrect if he's talking about crank degrees.
That engine doesn't look like that anymore if it's anything to go by....

look again at the polar plot. we're looking at ONE half of the engine- say cyl 1&2. Notice that for what is effectivly the bottom half of the polar plot, there is ZERO valve activity. ie for 360 (theoretical) crank degrees there is no valve opening.
Now - the engine only rotates cw.

SO.

(looking at the polar plot)

If we inject fuel at 1 o-clock, it'll go down the outside valve.
If we then inject fuel at 4 o-clock (180 degrees round) it'll sit in the port and eventually go down the inside valve.
If we then inject fuel at 7 o-clock (180 degrees round) it'll sit in the port and eventually go down the inside valve.
If we then inject fuel at 10 o-clock (180 degrees round) it'll go straight down the inside valve.


I'm not sure that i can explain it any clearer than that to be honest!!!
I've a lot of time and respect for DW, but that statement is wrong.
The ONLY way 180 degrees works if if you have a 'throttle body' carburettor - OR - you have an injector arrangement as shown, but with a cam-phase sensor.

Hi Dave,
Looking at the polar plot the bottom half would be the inlets for cylinders3&4, the ECU has two outputs fires one after the other. So we should be ok, without a phase sensor? I know fuel will be sitting around in the head, but it would with a carb too. As the fuel and air are pressurised in the head it should all stay together until a hole appears for it to go down...
Cheers,
Gavin. :)

Edited by Bat on 27th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

I don't understand how your 'two outputs' can work properly...

if they are indeed two separate injector outputs that you can phase with a 180 (crank) degree separation, how do you get rounf the ports needing fuel in the order A-A-B-B-A-A-B-B-etc? Also - without a cam phase sensor, you don't know which phase the engine is in - is it the first 360, or the second 360? All the ECU sees is a missing tooth!

Fuel doesn't sit in the head with a carb - Fuel is collected by the air rushing over the jet and carried into an open valve. sure a few wisps of fuel may fall out as the air velocity drops as a valve starts to close, but nothign like you're proposing.
That's the difference between EFI and carbs - the carb supplies the fuel when the engine wants the fuel, the ECU supplies the fuel at a fixed point... This fixed point is the area we have to focus on.

Edited by TurboDave16V on 28th Nov, 2005.

Hi Dave,
Surely the fuel will sit in the head with a carb as the turbo will be pushing a constant air flow across the jet regardless of the valve being open or closed. A standard 8 port engine on batch firing will have the fuel sitting in the head until the valve opens as the ecu fires all the injectors at the same time, no phase sensor here either. This system is usually used on competition engines . For the A series, you'd fire a-b-a-b obviously would have fuel sitting in the head awaitng valve opening events, but I don't see that as a problem........
Cheers,
Gavin. :)

Edited by Bat on 27th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

heres a quick thought td, going on what you said about a we box to go between teh ecu and injector(s)

2 injectors per port, 2 controlled by an ecu in a regular batch fire system, another two controlled by a second box of tricks(another ecu for prototyping??) to "top up" the lean cylinder. teh second two would require the usual 3d map + a reference when to start squirting(does this need to vary with RPM if the injector is right up against the port?? if so, more complications!).

I gather thats what your already after, but hitting a wall with regards injector size and resolution??? if so the above (a glorified bodge ) would meen much less stress placed on the injectors that need to fire in that tiny window, as the main injectors have already placed most of the fuel there for them.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

Gavin - all i suggest is to try it; this is the best way to understand - by seeing what's happening.

Bear in mind - as far as a single carb is concerned - there is ALWAYS an open valve. Surre, the velocity over the jet ramps up and slows down, and the airflow changes direction wildly inside the manifold, but there is ALWAYS an open valve for the fuel/air mix to venture down.

Batch injection is EASY to make work as the fuel can only go down one single hole ultimately. the fact that it might not be homogenus is of little importance - so long as an equal amount of fuel ventures down each valve it's fine. You can inject whenever you want in that 720 degree window.

What you're saying about firing it a-b-a-b won't work. It WOULD if the firing order was 1324 but it isn't.
Look at the firing order: we have 1-3-4-2 or to re-arrange; 2-1-3-4.
Now, as you can see, the injection pulses you're discussing every 180 degrees would need to be 'next' to each other - ie say a pulse lasted a full (theoretical) 180 degrees, you'd end up with one long pulse - 360 degrees long - that would do 2&1 and/or 3&4 back-to-back...

If you can get the software to fire in this order, it's still not ideal, as the ECU doesn't know which injector channel to fire at the 'missing tooth' TDC reference. You'd have an engine that'd only work 50% of the time - basically, at startup whenever the TDC gap on the crank sensor was for #1 cyl induction cycle and it fired port-A first it'd work.
BUT - if the missing tooth first seen by the ECU was for the #4 cyl induction (360 degrees round from the above) cycle and it fired port-A first, the sequence would be all to cock and it'd be firing the injector at the wrong time, leaving both shots of fuel to go down the inside valves.
This is waht hapened with the first MPi development mini developed by Mike Theaker at Rover back in '94/95. There was no cam-phase sensor, so the ECU had no idea what phase to fire the two (back to back) injector pulses in.

You're clearly highly enthusiastic about this, so i suggest you need to grab some pieces of paper, and try and scheme this out. I made something in Powerpoint that really helped my re-evaluate what we really needed to achieve not too long ago... I did this to keep myself focused on what the real problem was, but also to educate my good friend, and electronics guru, Graeme about the problem so he can build the ECU needed to sort the problem....


Evo - like your thinking but i'm already ahead of you! As you know, i've recently found out the MPi injecotrs are a Whopping 480cc/min. We also know how hard it is to get more oomplah out of the MPi's than 90horses.

What i'm planning on doing is having this 'injector controller' configured such that is anyone follows my instructions, it'll work.

We'll be looking at two injectors per inlet tract - a set distance and angle from the manifold face. I'll reccomend some suitable injectors later. Fuel pressure will be a vacuum referenced 3bar. Or, if so needed, (turbo / high rpm / etc) is more important, 4-5bar vacuum referenced. The injector controller will control both injectors - but cleverly only run one (in each inlet tract) at lower rpm's, bringing the second injectors in at a pre-set point; at which point the fuel count is halved. A 'gain' will also be added to fine-tune the increased flow from runing two injectors at half duty, compared to one at full - resulting in a seamless 'stepover' (a problem, especially with higher-flow injectors).

We plan to have the setup such that you enter a code for the 'start' of injection and 'end' of injection. This code will be determined by your camshaft duration. You will simply select from a spread-sheet your inlet opening and closing points and then look up the code reference for this timing.
It'll also have a digital readout (based on the information calculated from the code) for the '% remaining' countdown on the current injector duty cycle, so you know when you're about to exceed the injection window and promote charge-stealling.

Basically, we want to make it as 'plug and play' as possible... ideally, without the need for buying and installing two wide-band lambda sensors...
I'm going to run it on my turbo for 12 months before releasing it - Suffice to say, if it can fuel a turbo at 20psi and 7000rpm, it'll fuel any a-series....

Edited by TurboDave16V on 28th Nov, 2005.

Hi Dave,
Having spoke to various people, I have found that DW is indeed wrong. You will get crap distribution and poor fuel economy. However I may have a solution.... Omex do a fully sequencial ecu with 8 injector outputs and mappable injection timing only £750+vat If you ran two injectors in each runner, you may be in with a chance.....
Cheers,
Gavin. :)

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

thought for prototyping cheaply (with megasquirt(s) )with NO software modifications.

get an assembled V1.0 or V2.2 pcb megasqurit off e-bay for penies, set up a spark table such that the spark "event" happens at the point where you want to begin the injection. heres the trick feed the spark output into the ecu that actualy controlls the injectors! the injectors will now fire at a time defined in your spark table, and you will have full 3D controll over this.. you wil obv need to scope the output from the first ecu and 2nd ecu, on a dual channel scope and note any delay and take this into account. long way for a shortcut, but the hardware and software exists right now.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

Gav - you'll be in with a chance if you can 'move' the injection event like Evo is suggesting...
As a matter of interest...

Edited by TurboDave16V on 28th Nov, 2005.

well, i was suggesting it for prototyping, its a (relatively) cheap way to do a proof of concept before going the whole hog and having custom code written, or an extra box of tricks (and code written) to handle the ajustment of the injection event. my suggestion uses 2 ecu's just for fuel, you would then need another for spark, or a dissy for cheapness and proof of concept

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

Hi Dave,
With the £750 ECU, you have a map to set the injection timing(speed v load) and another separate map for injection duration(sp v load) and another for ignition timing(sp v load) You just need the money and a very long time to set it all up! I've got a feeling your going to need a phase sensor though... how about using the dizzy for phase sensing?
I have come to the conclusion that the 180 degree injector firing isn't going to work. Having spoke to the guy at Omex, he drives a mini and has an engineering degree. They've already done one mini and entered it in a rally. Apparently it suffered from charge robbing and had to be run rich to compensate, hence poor fuel consumption and emissions.
He also said that the injection timing point would have to be altered depending on engine speed and load as both would alter air speed in the port, hence would need fuel delivered at a different time.
He said if anyone had achieved successful injection on the A series, they hadn't told him.................
Cheers,
Gavin.

Edited by Bat on 28th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

You will need a cam sensor I suspect, yes.
My Weber Alpha injection i ran a few years ago had a cam phase sensor in the dizzy - you need to get a dizzy from a 2wd cosworth and modify (raid bits out of) that - unless it needs an inductive sensor, which is very easy to do...

What's the guys name? Has he any amusing words to say about MiniSpeed doing injection systems using Omex ECU's then?

What's the code for this ecu?

Hi Dave,
He didn't say anything specific about that company. I mentioned it was basically an electronic SU, he didn't pipe up and say "I think your wrong it's a fantastic system." The fully mapable caper is the 700 series, they do a 600 which has lots of goodies like ALS, traction control, boost control ect, is cheaper than the 700 but only fixed injector timing. The guy to speak to is Andy, seems ok.
Cheers,
Gavin. :)

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary

cam phase sensor is a doddle, extra bit of iron welde dto the cam sprocket (bolt) and a hall sensor to detect it, = ttl/cmos output to a digital circuit. or if you wana make work for yourself you could use VR senor + suitable circuitry.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

Evo - cam sensor has to rotate at half crank speed. It's to tell the ECU if the engine cycle is at #1 cycle or (360 crank degrees on) #4 cycle

Hi,
Dave what was the alpha like? It you get charge robbing on that? I'm Interested to know how far you've got / what you've already tried on this project.....?
Cheers,
Gavin. :)

Edited by Bat on 29th Nov, 2005.

VEMs Authorised Installer / Re-seller. K head kits now available!

WB/EGT gauges. Click here for customers write-up

Visit www.doyouneedabrain.co.uk

My Mini build diary