Mega squirting CRD???

[Nigelw]

Now not wanting to tarnish the brilliant thread of HfH I decided to start it here properly.

What would it take to be able to take just about any of the CRD engines available and make them run without needing the special diagnostics for each particular manufacturer?

Is it possible? I mean BAS and the likes seem to do these things but they seem to be limited to that specific engine type/make/model.

Can it be as simple as the MS system determining engine speed etc and controlling the diesel injectors, the electronic VNTs etc??

All I would be wanting is for the engine to run and the rest of the monitoring would be taken care of independently.

Too simple?

Or have I over simplified it too much?

[Bowie69]

IT has been discussed before, and basically diesel fuelling is hugely more complicated for whatever reasons that petrol, I forget the implications but it was 'difficult'.

[Nigelw]

IT has been discussed before, and basically diesel fuelling is hugely more complicated for whatever reasons that petrol, I forget the implications but it was 'difficult'.

Can't be that difficult as LR managed to give us the TD5

So what are the stumbling blocks exactly???

Is it not possible to pull a few chips and read the codes to decipher what they mean and how they interact with other systems? I would rather see a move away from fully integrated systems and have a module for engine control only that can be set up with a laptop same as MS.

Time to go a hunting on the WWW.

[elbekko]

Is it not possible to pull a few chips and read the codes to decipher what they mean and how they interact with other systems?

Eh, not quite no

Reading machine code is pretty hard, and even harder when it's a complex system such as a modern engine ECU.

As for the technical difficulties, IIRC from the previous thread it was something with high voltages and multiple injections per compression stroke?

[TheRecklessEngineer]

For starters, a td5 is not common rail.

I looked into it in quite some depth and seriously considered developing a system to do it. However, I never found the time to get it started.

http://forums.lr4x4.com/index.php?showtopic=36152

[Quagmire]

Aren't petrol engines a lot more forgiving too?

What I mean is any old numpty can fit ms to a petrol engine and run about with the timing out of whack and the fuelling all over the place and it will (most of the time) run, albeit badly with no long term effects.

Do the same with a diesel and it will go pop, no? That would be a major point stopping me from releasing something to the general public.

[Gazzar]

It's a shame.

The family car is a Volvo, powered by a 5 cyl vnt diesel turbo, the D5. Reliable, powerful, torquey, economical, smooth (for a diesel), compact.

It looks like great powerplant for a Defender, but there is no way the existing electronics would allow it to happen, as is, even changing the seats needs access to the on board diagnostics.

As I see it the MegaRail would have to be able to hit the ground running Mo Farah style to run a modern CRD

For one thing, the injectors can fire up to 7 times per cycle in these engines.

To get an off the shelf product to that would take 20 years of research to be done in one or two years.

Perhaps if the MegaRail was to focus on getting a first gen CR engine work then the knowledge base would be out there allowing evolution.

If the wizards focused on the VW or the PSA CR engines of 2000, and get them running then MR could grow from there.

Shame Volvo never let LR use the D5 in the discovery and defender ranges.

G.

[Nigelw]

Right after a little bit of hunting I came up with an interesting million or more questions but why were so many systems incorporated into one main system? I read back a post by Aaragorn and found it interesting to see his comments about the Zetec engine and loom from a later car,question is, would it be possible to do the same with diesels? Simply splice out the other components and effectively make the BCM redundant save for engine diagnostics?

But it does not solve the issue of wanting to plug in any individual injector and fire it off a separate ECU(MS in this case).

If we can forget the 10,000 burns per cycle and just get it running with the one squirt as we do with mech diesels then surely we could be half way there? Then it would be as simple as injector looms and coding of injectors to talk to the MS module would it not?

The mention of high voltage is an easy one as we could simply take what ever converter/amplifier BMW/FORD/VW etc, are using and bingo we must have a winner by now?

Not feeling much enthusiasm for this fellas

[Bowie69]

Well your are talking about a diesel.

Isn't that enough reason to switch off?

[CwazyWabbit]

In my view it would be simpler to reverse engineer an existing ECU and just patch out the code that wants to talk to the other parts of the car, these routines may be relatively easy to spot due to the fact they will be performing some kind of external comms. Don't get me wrong, that will be hard work ..... just not as hard as creating an open source ECU.

I wonder what processors they use in ECUs? Or are they custom DSP's? Is the firmware extractable? Is it encrypted? Do any disassemblers support these processors?

Anyone fancy spending a few months staring at IDA Pro or similar?

[CwazyWabbit]

Another alternative would be rather than modding and existing ECU or creating your own ECU how about creating a car on a chip?

Sounds stupid but would it be easier to create a configurable system that simulates the missing parts of the car?

However the down side of that is that it may make it easier to steal the cars these engine originally came from.

[Gazzar]

One place to start would be to talk to the guys who do re-maps for diesels - they'd have an idea as to the chips programmers and the like used.

From my standpoint I've looked into re-mapping the Volvo D5 - apparently it is a simple matter to go from 160 to 200 BHP and to and from 350 to 400 torque things - the main players for that engine are MTP and RICA, as well as Volvo themselves. Perhaps an email to the UK agent may get some pointers?

The interesting thing about this Volvo D5 engine is that it is used in marine applications, so someone has already "de-car"ed the ECU. I think it was Penta Marine, again, a quick email may be of value?

For me? I'm going to fit a Rover v8 petrol into my Vapour build 109 SW.

I don't need economy, but I like simple torquey engines. and I think a megasquirted 3.9 fits the bill.

[CwazyWabbit]

As it's a vapour build Gazzar I take it you'll fit LPG

[Daan]

I dont think re map guys will be too keen to give any pointers. As far as I can see it, if you know what you are doing, it must be straight forward. You are going to need to amplify the signal, as the injecters need pretty high currents. Its just that I don't really know what I am doing when it comes to electronics.

Apart from that, the beauty of the MS is that you can adopt to what ever you want, where as the original ECU is pretty much fixed. The problems with a diesel are not as bad as with petrol, you dont have the lean burn problem if you run it to lean,which would make it run hot or the engine knock problem which will make it detonate, both can ruin your engine.

With a diesel, if you dont get the timing right or the mixture right, apart from a lot of black smoke and badly running, not much else woud happen. Just think about what you could do in combination with a vnt turbo, if you had freedom to control it?

Daan

[Gazzar]

As it's a vapour build Gazzar I take it you'll fit LPG

No way! Proper Petrol only. I don't do crazy miles, depreciation isn't a factor, but I might want to go on the Eurostar someday so LPG is OUT.

For now....

G.

[Nigelw]

Another alternative would be rather than modding and existing ECU or creating your own ECU how about creating a car on a chip?

Sounds stupid but would it be easier to create a configurable system that simulates the missing parts of the car?

However the down side of that is that it may make it easier to steal the cars these engine originally came from.

This I totally understand and I think this is to some extent what the BAS black box is for the 2.7 & 3.0 V6 diesels.

What seems to be the major hurdle here? I can't quite put my finger on it as I am trying to pin point what are the important data entries that need to be used and what has to be used as reference for certain data streams, like, timing, coolant temp, air temp(ambient), Air temp(inlet manifold, boost pressure, engine speed, fuel temp, fuel pressure throttle position.

Any others?

I almost fail to see how any CRD engine can be so different from another as there are injectors, high pressure fuel inlet, and plug them in and squirt them up using reference points for the timing, mechanical pumps only have one opening slot so why over complicate things with multiple burns per cycle?

[Bowie69]

Multiple burns are surely there for a reason, all in the name of efficiency and power output I would guess.

[TheRecklessEngineer]

Not feeling much enthusiasm for this fellas

Build/buy me a dyno capable of loading an engine to it's designed power output + 50%, get me the development kit necessary and supply me with an engine to develop the system on and I'll see what I can do.

As I said, I've looked into it in quite some detail. The amount of work became very scary very quickly. You really need a team to do something like this.

Multiple squirts/cycle are pretty trivial compared to the rest of it.

[TheRecklessEngineer]

Multiple burns are surely there for a reason, al in the name of efficiency and power output I would guess.

They're only implemented at low revs and power output in the interests of smoothness and emissions. Not strictly necessary.

[CwazyWabbit]

For starters, a td5 is not common rail.

I looked into it in quite some depth and seriously considered developing a system to do it. However, I never found the time to get it started.

http://forums.lr4x4.com/index.php?showtopic=36152

That is a very interesting thread, I can see the advantages of MS'ing a modern diesel.

[FridgeFreezer]

This has come up plenty of times, from the very early days there have been diesel guys screaming for "DieselSquirt" but as far as I'm aware no great progress.

IMHO reverse engineering or cracking a specific ECU for a specific engine or manufacturer is a huge task, modern cars have more software than the space shuttle - as an example, the original ECU for the LS engines had more code to control the idle speed when you turn the air conditioning on than the entire MegaSquirt firmware. And that's a nice basic petrol engine!

Reverse engineering also doesn't give you any control over availability of hardware - you've got to buy a specific ECU, so if they change the design, lock them down, discontinue a part etc. you're stuffed.

Reverse engineering what the ECU is DOING to make the engine run is a far more likely prospect, which means instrumenting a running engine to work out as much about it (and the electronics) as possible, which gets you some way (probably not very far) to knowing what it takes to get it to start & run.

However, times move on and it's certainly not beyond the wit of man to make a diesel ECU (on the grounds that enough of them have been done by car manufacturers now & technology is only getting cheaper), as I understand it (which I don't) very basic diesel control would be a on a par with coil-on-plug ignition control, which was considered hard when MegaSquirt was 1st developed but was soon cracked with the 'Extra code.

The original MS CPU is a very basic device, on a par with the Z80 - the MS2 and MS3 add huge amounts of processor power (MS3 is Pentium territory compared to MS1), with that comes speed (for far more accurate timing), code space (for more complex software), and more pins to allow more things to be controlled (EG full sequential injection).

Software-wise, there's a hell of a lot of development to do - you need to understand all the things going on to make an engine run, how they relate to each other, what all the timings are etc. and then somehow convert that to software routines.

The big hurdles I can see are:

- You need the actual injection hardware on the vehicle, which could be tricky if it's not born with it.

- You need the electronics to drive that hardware, my understanding is that proper diesel injectors require serious driving compared to the basic 12v solenoid that is a petrol injector

- You need to know a fair bit about the setup of the engine, sensors, timing, etc.

- You need an embedded development environment, which may cost $1000+ for a more powerful processor

- You need to be prepared to trash several engines in the development process

There's quite an investment required, even if you assumed everyone's time was free.

I'm sure for basic control / proof-of-concept you could ignore the swanky stuff like many-squirts-per-cycle etc. but ultimately you probably need to do *some* of that or you're not going to get much out of the engine, probably less than the factory did.

The great benefit of course is that DIY means you're not battling a load of locked-down proprietary communications/diagnostics, unobtainable wiring connectors, or systems that will only work when the back-seat DVD player is connected...

[TheRecklessEngineer]

There's quite an investment required, even if you assumed everyone's time was free.

This is where I got stuck. Just pricing up a CRD engine and enough systems to make it run on the workshop floor was eye-watering. That's without the dyno on the end.

[Tanuki]

Now not wanting to tarnish the brilliant thread of HfH I decided to start it here properly.

What would it take to be able to take just about any of the CRD engines available and make them run without needing the special diagnostics for each particular manufacturer?

Is it possible? I mean BAS and the likes seem to do these things but they seem to be limited to that specific engine type/make/model.

Can it be as simple as the MS system determining engine speed etc and controlling the diesel injectors, the electronic VNTs etc??

All I would be wanting is for the engine to run and the rest of the monitoring would be taken care of independently.

Too simple?

Or have I over simplified it too much?

I'm familiar with the fun needed to make the 3-litre quad-cam 24V Ford "Cosworth" engine [as used in Granadas a couple of decades back] work in other vehicles - those engines are a popular retrofit in things like Capris. As standard they were all electronically-controlled automatics; when fitting a manual box it's possible to fudge the ECU to think there's still an autobox there but the cheap setups then do bad things like producing a nasty surge when doing fast manual gearchanges, and driveline-snatch when travelling at low speeds on a very light throttle: in the original application the autobox speaks to the engine ECU to deal with these issues. A "proper" conversion is a lot more complex, involving things like clutch-pedal sensors, a power-steering-pump pressure sensor and a switch to tell the interface-box what gear the manual box is in (which in turn simulates the 'slip' you'd get from the torque converter which then tells the ECU to briefly retard the ignition and select a different closing-profile for the return-to-idle-when-throttle-is-snap-closed function when you're changing gear).

Thinking of a TD5 I can see a range of inputs you'd need in addition to the ones you've mentioned, to provide for a "De-car" box:

Fake security-codes handshake between the BECM/Immobiliser and the ECU so the ECU is allowed to start the engine in the first place.

Clutch position switch (Not needed on auto)

High/Low ratio switch.

Autobox ratio signal and selector-position signal (not needed on a manual)

Autobox fluid-temperature (not needed on a manual)

Vehicle Speed Sensor

Throttle position Sensor

Fuel pressure sensor

A/C Switch sensor

Brake Light switch and possible signals from ABS module.

Battery condition - the existing ECU should already be monitoring this.

That's not exactly a short list: get them wrong in some combination and you'll have the engine ECU switching into "limp-home" or other restricted-operations strategy.

[Hybrid_From_Hell]

And the TD5 has an ECU controlled Fuel pump, which I know little about but it looked very much more than a simple fuel pump when Steve (off road toad) showed me one !!

[Nigelw]

And the TD5 has an ECU controlled Fuel pump, which I know little about but it looked very much more than a simple fuel pump when Steve (off road toad) showed me one !!

You of all peeps in their little inventing sheds should be busy with this

When I have more time I need to sit down and do some actual configurations and possible permutations of what systems are incorporated into it, but it does seem to read as though there is a problem to be solved.....