N54 Solenoid Direct Injectors + Custom ECU Build

[ccr0ss]

It would take a lot of R&D to get close to the driving characteristics of the stock DME. I think that's why the Nexsys DI delete uses a modified stock DME.

Even then I think they are struggling to get it to work.

There's already standalones out there like MaxxEcu that have tried but failed to replace the stock DME.

An interface board between the injector harness and the solenoid injectors would be the only way to get close. Even then it's got to be better to just cough up £2k for peizo injectors think the standalone ECUs in the market were planned to reproduce the stock DME's behavior

I'm not sure which aspects of the stock ECU's control/safety strategies you may be referring to when you mention driving characteristics. But you're right about the amount of R&D it takes to get things right, especially safety. Going the Nexsys route is significantly cheaper and faster for development if one is seeking to make a commercial product.

I wouldn't think that standalone ECUs are build and programmed to behave like a stock ECU, given that they are performance-oriented. Very few people, if any, would drop 2k-5k on a standalone ECU to get factory-like performance. For me, it is all about smooth power delivery, safety, integration, and flexibility. If I get the hardware right, the tuning would be the secret sauce to make those things a permanent reality. Running PI alone is an entirely different matter, and I wouldn't expect a PI-only solution to be as smooth/efficient/safe/integrated/flexible as a DI/dual injection option.

Also, considering this ECU is not a commercial product but an open-source project, we have all the flexibility to turn it into whatever we want, and to work on it for as long as we want. Do you want an ECU that closely resembles stock driving characteristics (whatever that may mean to you)? Go ahead and play with the settings. Do you want an ECU that is not as sensitive to certain inputs as the stock ECU, all while giving you the performance already found only in aftermarket solutions (maybe better)? Same thing.

 

[carabuser]

By stock driving characteristics I mean the idle and consistent power delivery. The transition between modes like idle, acceleration, overrun and boost/off boost.

The standalones function enough to get the car moving and do well on a drag strip at WOT but I don't think an N54 makes much sense for a drag orientated build.

Not to doubt your abilities, but if large companies that make these things as their core business struggle to get close to the stock ECU then it's going to be a full time job making it work.

DI delete is flawed because it loses all the benefits that DI gives, so your approach of making cheaper injectors work is certainly the right direction.

 

[ccr0ss]

By stock driving characteristics I mean the idle and consistent power delivery. The transition between modes like idle, acceleration, overrun and boost/off boost.

The standalones function enough to get the car moving and do well on a drag strip at WOT but I don't think an N54 makes much sense for a drag orientated build.

Not to doubt your abilities, but if large companies that make these things as their core business struggle to get close to the stock ECU then it's going to be a full time job making it work.

DI delete is flawed because it loses all the benefits that DI gives, so your approach of making cheaper injectors work is certainly the right direction.

I understand your concern now. And, you do have a valid point.

If you put aside the issue of either having/lacking hardware to control DI or other key functionalities, the issue of driveability on standalone ECUs is a matter of tuning. Big companies like Motec, Syvecs, Haltec, etc., build their hardware and support software on the assumption that tuning falls under the responsibility of the end-user. They may offer base maps, but those are usually intended to be startup maps, as every user wants something different out of their vehicles.

Now, we can just datalog the living life out of the stock ECU to figure out the correlation of inputs and outputs and reverse-engineer control strategies from there. It is not easy to do, but it is doable. But I don't see the kind of people that ditch the stock ECU to buy a standalone ECU for a few grand, to naturally pay a tuner to basically replicate what the stock ECU can do at a fraction of the cost. Too much of a niche.

I want to replicate what the stock ECU maps do, and improve on the areas that annoy me (depending on which firmware you use for the stock ECU, it can be unnecessarily twitchy and sensitive to certain conditions). But I don't expect everybody here to want to do just that. What I want to offer is the tools, how people may use them would be entirely up to them.

 

[ccr0ss]

@mcprogram I think I figured out a decent solution for adding an extra CAN module: adding a second STM32H743IIT6! I originally planned on using two ADCs (16 channels each for a total of 32) to provide current sensing diagnostics for the outputs, but the only ones I could find in LCSC that would fit the requirements were about $15 for both ICs. Instead of doing that, I am going with an extra STM32, which is only $8, and has all the ADC channels I need, plus a bunch more peripherals, including two extra FDCAN modules. It is overkill for what it is going to be used, but that's a compromise I can live with.

I'll see if I can dispense of 4 pins on the header to accommodate for both FDCAN modules. There are enough free pins on the header to accommodate for both FDCAN modules, and a few extra in case we need to add more analog devices or low outputs. This microcontroller has a significantly larger footprint than the two original ADCs I was planning on using, which complicates trace routing, but the benefits outweigh the disadvantages.

 

[langsbr]

I'm not sure I agree that the "key to the N54" is piezo injectors. The N63 used the 'n54 style' piezo and then changed to solenoids in the TU upgrades. Even the 'newer german 2JZ" B58 uses solenoid injectors. I totally agree that the "direct injection" system is worth keeping, but to only think its worthwhile with what are clearly a defective manufacturing method piezo injector? That's shortsighted.

 

[chasec46]

Hello,

I have been working for sometime on figuring out how to make N54-powered cars easier to own and maintain. I currently own a 2008 E61 that I got at the end of last year, but this would be my third N54-powered car in the last 13 years. As I'm sure most of you know, some of the most aggravating issues of the N54 are related to the DI system. It seems to be the case that HPFP issues have been sorted by BMW in later revisions of the pump. Our fuel injectors, however, still need considerable attention considering how common it is for them to leak. I know how a $2.5K initial investment for lifetime injectors using one of the two vendors we know of may be reasonable for some, but I am not sold on this path. Which brings me to what I am working on right now:

I decided to find a way to run Bosch HDEV 5.2 solenoid direct injectors (the ones on the N55 and newer BMW engines) and this is what I came up with:

1. The injector bore on the N54 is about 1mm smaller at the top than it needs to be to accommodate for the HDEV injectors. The bore would need to be increased to about 21mm at a depth of at least 27mm from the piezo solenoid decoupling element seat. I am actually going for about 33mm depth because I am adding a 5mm 6061 aluminum washer to the bottom of the newly drilled bored so that the HDEV injector decoupling elements can sit properly (the decoupling element on these injectors are smaller than those on the piezo injectors). I took the head to machine shop to have it resurfaced and to increase the bore on the injector holes. I know very little about machining terms, pardon any misuse here.

2. The most difficult part is figuring out how to run the damn thing. I could get a Delta GDI6 and call it a day, but I wanted something that would integrate better with my E61, while also giving me the chance to work on something fun while going to school. So, I decided to design my own ECU and make it opensource. My primary intention was to build an ECU that can run my Frankenstein's N54 while also giving me the ability to avoid some of the BMW idiosyncrasies. So, here's what the prototype ECU is offering at the moment:

6x direct solenoid injectors
6x port injectors
Support for all the engine-critical sensors and actuators found in the N54/N55/S55/B58/S58
Support for 2x EWGs (in case I find an S55 laying around somewhere)
Support for dual HPFPs
Bluetooth and GPS modules
Data logging through USB, limited on-board data logging
2x wideband and 2x narroband sensors
2x LIN/BSD interfaces
2x4x FDCAN interfaces
Diagnostics for every input/output
No valvetronic support.

My main constraints for the ECU are: 1) the board components need be to as inexpensive as possible ($175-$250 per board without enclosure or header for an order of 5 is the target), 2) the components need to be readily available at LCSC given that I will be using JLCPCB to print the PCB. The ECU is powered by the STM32H743IIT6 MCU, uses the NXP MC33PT2000 from NXP for DI control, and Molex Compactus 186-pin connector. I want to run it barebones eventually, but I think branching out the rusEFI project may be the proper start for the project. It has taken my a while to design the PCB. My first design was based on the Core8 ECU designed by a guy named Ben from the UK/Lithuania. Adding DI, EWGs, BT/GPS, and LIN/BSD has forced me to reinvent the PCB a few times. I should have the PCB files available online in the next month or so.

I will be posting pictures of the work as I move forward with the project. I would appreciate your feedback, specially on the features of the ECU, as this is my first time building something like this.

very impressive

 

[Nosrok]

Have you looked into modifying the injector instead of the head? If tge issue is only the injector tip it might be worth spending a bit of time looking into weather of not it's possible to modify the pencil style injectors to sit correctly. You should easily be able to adjust height with the proper spacer.

 

[chasec46]

Have you looked into modifying the injector instead of the head? If tge issue is only the injector tip it might be worth spending a bit of time looking into weather of not it's possible to modify the pencil style injectors to sit correctly. You should easily be able to adjust height with the proper spacer.

You are in a rabbit hole I have yet to dive into. I wish you the best with you journey. Modifying injectors is far beyond my capabilities and im just getting into my first n54 buying new injectors. I can imagine the HDEV5.2 injectors from Bosch would boast greater reliability.

 

[ccr0ss]

Have you looked into modifying the injector instead of the head? If tge issue is only the injector tip it might be worth spending a bit of time looking into weather of not it's possible to modify the pencil style injectors to sit correctly. You should easily be able to adjust height with the proper spacer.

Interesting. I have not considered until now. I am not sure the amount of material that would need to be removed from the injector would affect functionality. Take a look at the very rough overlay below:


It's not simply the tip that presents a problem. The metal part of the pencil style injector that's right above the tip shaft is about 1mm wider than the injector hole at the top, and even wider the closer one gets to the bottom of the hole before the injector tip shaft (the injector bore in the cylinder head is tapered at two different angles towards the bottom of the hole). The part that's brighter white would need to be removed from the injector for it to fit within the unmodified. head bore. And assuming that doesn't damage the injector, we would need to have decoupling elements custom made for the application.

As difficult as getting the head out of the car can be, doing so and taking it to a machine shop to increase the bore would guarantee a better fit for the new injectors, all while giving one the choice to also being able to install the piezo injectors in case one may want to go back to those.

That being said, I'm intrigued by the idea. I'll make a few cuts in one of the pencil-style injectors I have laying around to see if this is feasible.