I think mnost features need to be disabled if required.

Having thought about the possible impact of cutting engine entirely I really wouldn't want this on track. I would imagine that mid corner in a race/at the limit scenario a full engine cut could result in an off, so you have saved your engine from complete destruction but now need to repair engine and find a new bodyshell for it

For road use I think the requirements are very sensible.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

I was intending to make limits / actions programmable. I agree full engine cut as the standard ECU does isn't desirable even in road scenarios, but it is the ultimate engine saver if desired.

_________________
If at first you don't suck seed, try drier grain.

Well, after a couple of hours research, it seems the sockets are readily available, but aimed at prototypes for plugging in emulators - hence a price tag of £60+. Nothing cheaper.

So . . . a compromise solution - available in 28DIP form. Not enough I/O so link via I2C to a second low spec pic to run the slow inputs / outputs. Main processor to link to triggers, injector & ignition drives, MAP sensor, TPS - and as a bonus built in CAN interface.

A fairly new family, memory a bit limited at the moment (ok for now though) with bigger versions in the pipeline.

http://www.microchip.com/wwwproducts/De ... e=en555626

_________________
If at first you don't suck seed, try drier grain.

The idea of a core ecu with plug in "modules" to handle non essentials seems a good one anyway
That way you don't get the blue (smoke)screen of death because "the passenger window program is not responding -- OK"

Unfortunately it won't be working that way. I want the firmware to be upgradable via the PC port so end users can upgrade without special equipment or plugging in chips, therefore all the logic will be in the main processor.

_________________
If at first you don't suck seed, try drier grain.

I don't see having a core engine edu and the provision for external functions makes any difference to the end user
At worst you only need to make the control software and core edu aware of external functions. So user selects vanity mirror light control (random example), and hits update. The central ecu downloader/controller says "not for me" and offloaded it to another bus for example.
What it does mean is that you have protected functionality to run the engine regardless of what other guff is going on

It's a bit of an extreme example but it's only a simple extension of what you said with the core ecu and external io capability really.

The way I write software you won't get individual modules 'locking up' anyway. I don't leave software sitting in a loop waiting for something that might never happen.
One of the reasons I tend to avoid the more powerful processors & more complex compilers that go with them is that you lose control of what's going on and a bug in a library function can cause a problem.

From past experience of splitting tasks between processors you find a load of wasted overhead passing information back & forward. I swore I'd never do it again.

_________________
If at first you don't suck seed, try drier grain.

Here's a question. I'll have a prod with the scope and see what the standard ECU does but would be interesting to hear some opinions on the matter.

When do you squirt the fuel in ?


Sounds simple enough - during the intake stroke is the obvious answer, but the intake stroke only lasts 25% of the total time available so unless you are running huge injectors which never go above 25% duty you will spend some time injecting into the port ready to be sucked in on the next intake stroke. So do we calculate the injector time and start injecting such that the squirt finishes at the end of the intake stroke, thereby minimising the time the fuel sits in the port ?

At light throttle, less than 25% duty which part of the stroke should the fuel go in ? The early part so it all gets sucked in or the late part which may get blown back out during the start of the compression stroke when the valve is still slightly open. Maybe the middle part so it's with the maximum air velocity to swirl into the cylinder ?

IMO, the ideal is like we used to have in the days of carburetors Where the fuel is added at the throttle butterfly in the correct proportion at all times then has the length of the intake manifold & runners to mix properly. Until someone invents an injector capable of analogue modulation we are stuck with injecting a slug of fuel into the air and relying on the swirl in the cylinder to mix it.

practical experience with the link ecu would indicate it doesn't make a huge difference as the engine is reasonably smooth even driving all 4 injectors together.

The other thing to factor in is what happens when the driver bangs his boot down fast. For a cylinder just coming up to intake, the most it can hope to have injected is 25% duty to go with the huge gulp of air it's about to take. In this instance, it would be better to have left some fuel in the port from the previous cycle. Maybe the answer is to start driving the injectors AFTER the intake stroke so the fuel is left in the port for the next stroke, then accelleration enrichment time can be added before this so it is added to the current cycle.

_________________
If at first you don't suck seed, try drier grain.

This got me wondering about how direct injection engines work as they can only inject fuel into the combustion chamber during the intake stroke. The injectors must be suitably large! This contrasts with conventional fuel injection which can inject fuel into the inlet tract at any stage in the 4 stroke cycle. Apparently this means its a lot more difficult to get higher power out of a direct injection engine but they are significantly more economical. Some engines such as the Toyota 2GR-FSE V6 employ both direct injection and multiport fuel injection.

Direct injection is a growing trend in road cars and has been used for a few years at Le Mans by the Audi LMP, Porsche Spyder LMP2, Corvette C6R GT. It is a difficult technology to get right and hence was restricted to factory teams. Corvette kept this under wraps until last year.

_________________
Don
GT4DC Chairman
1994 Toyota Celica GT-Four ST205WRC JDM 269bhp @ 0.9bar
1994 Toyota Celica GT-Four Special GT 590bhp @ 1.8bar
1989 Van Diemen RF88/89 Formula Ford 1600
2008 Nissan Patrol GU 3.0L ZD30DDTi 154bhp

Chris - I think you have had a brain fart m8. I seem to remember seeing a log of yours showing injector timing and crank/cam angle sometime.

You have answered your own question - the ideal is to have fuel pumped in (consider the location of the injector) as air flows past it to ensure a good air/fuel mix. Pumping a load of fuel on to the back of a closed valve will work but will increase emmissions due to pooling of the fuel. In reality it is more critical at low engine / fuel positions.

I am not even sure that the GT4 ECU is fully sequential? I had a thing in my head that it was only semi-sequential for some reason and batch fired 2 injectors together.

The DTA ECU allows you to switch between batch fire of all 4 injectors, semi-sequential and fully sequential as well as adjusting the injector start angle (handy for diagnostics). I suspect that each engine will vary slightly depending on cam duration/lift to achieve optimal settings for fuel mixing (seeing as you can't vary injector flow to be equal to airflow at all parts of the cam lift). I would suggest that starting injector a bit before TDC of the inlet stroke would be a good ball park to start at allowing for a bit off injector lag time.

IIRC at full chat/high boost I was using around 20 mS of fuel which is roughly a full engine cycle (2 crank rotations) in duration at 6000rpm. Hence at high boost/rpm it is pretty much on most of the time.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

The benefit of direct injection is that the pressures are very very high for the fuel rail - you have to overcome the cylinder pressure plus add some for flow to occur, the higher the difference between rail and cylinder the more will flow for a given duration.

What amazes me is diesel direct injection where they have multiple squirts per power stroke all calibrated to keep emmissions in check and yet produce the correct amount of power. IIRC some of the squirts are mainly for emmission control and one of them produces the most power - hence most chipped diesels black smoke much more as extra fuel is added to the power section of the squirt.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R

The graphs I produced were in relation to one of the cam trigger pulses. I don't think I ever verified where No. 1 firing was. I'll have a look back in case I had done it and forgotten.

My own feeling was to inject during intake plus minimum time before, but that doesn't allow for instant throttle down enrichment.

I guess the answer is testing and monitoring emissions and efficiency of various strategies.

_________________
If at first you don't suck seed, try drier grain.

I guess with direct injection you can inject during compression stroke as well, and maybe even start during the last bit of exhaust stroke depending on the location of injector relative to exhaust valves.

Also worth remembering that at lower RPM, 25% of a cycle is a much longer time so will allow more fuel per stroke. Many 'lower end' cars have a VE which falls significantly once past 3 or 4 K RPM.

It's an interesting subject and frightening how far the technology has gone since the era of our cars.

_________________
If at first you don't suck seed, try drier grain.

GDi (gasoline direct injection) was around during the and before our cars. There were a few airplanes running it. The most concurrent version was probably the system fitted to the Mitsubishi Gallant (iirc)

Modern diesels use post ignition squirts for emissions and I believe egt control. Not sure if they are into double digit squirt events yet but they were at 7 or 8 a couple of years ago.
I wonder if vnt turbos with GDI would allow removal of the throttle body entirely or if pneumatic valving would also be required. I suspect the latter

As I understand you would need the pneumatic valves to limit airflow. Diesel is happy at very weak mixtures and will still produce a bang & power hence no throttle body needed - you dictate the power by how much fuel you inject. Petrol just missfires at weak mixtures.

_________________
JP
GT4DC Treasurer

Grey St185 RC
Silver BMW E46 M3
Multicolour yawn Honda VFR400
Silver Honda Civic Type-R