We never did diesel but there is always first time. Please tell us much more about your tractor and youself. Are you looking for a project or a plug&play solution?
do you have an oscilloscope?
did you ever solder through-hole PCB? did you ever solder surface mount pcb? where are you geographically?
did you ever write any code in any language?
There is hope then Do you have exact part number for your injector? Do you have a datasheet for your injector?
Same for your pump - exact part number & datasheet?
Shall we build a flow bench first to prove that we can control this fuel delivery system?
We have common rail discussed here and there in a few vehicle-specific topic but I do not see a clear plan. How would we drive the injector? MC33816 or anything else?
My small plan was reverse egeniering and tak old bosch ecu with CR and start car and separate with oscyloscope injecton driver or take OEM D4D toyota or mazda 6 2.0d there are separate injection driver separate oem or buy completed driver.
Welcome and it sounds interesting. I've wanted to do common rail before. I was thinking high voltage injector drivers. I believe that rusEFI has the physical capabilities, but would need some hardware as it would need a high voltage driver, and it might benefit some from software refining. I think a flow bench would be a great incremental step. Basically get pressure on the injectors and make them open and close such that it delivers fuel.
Do you have a schematic for the internals of any of those drivers?
I am not sure what you are saying. Please understand that as is I know very little about diesel contorol and I need your help. So far I am only dealing with low fuel pressure gasoline where we have single continues injection per engine rotation, and the duration of that single continues injection is about 2-3 ms, and 0.1ms precision gives a good enough fuel control precision.
it sould like in high-pressure diesel you are saying there are four injections making a "sequence" and that "sequence" consists of "2 pilots of duration X and one main of duration Y and one post of duration Z"
Is this close? If you can please find time for a longer detailed answer. What are the typical values for X, Y and Z and what kind of control precision would be needed?
Maybe some pictures or diagrams of a typical injection sequence if you can find any?
Is there a way we can get @a flow bench setup? I once saw a video where a fellow used a couple of argon cylinders. One cyl had two ports on it. One was connected to a pipe which poked down into the bottle, the other port was exposed to the top of the bottle. This bottle was filled with fuel. The second bottle was a standard welding bottle, with a shutoff valve. The standard bottle was filled with Argon, or something like that from a weld shop, to some high-ish pressure. It might have only been 2,000 psi (137 bar) but it was enough for proof of concept. With some plumbing he connected the weld gas cyl to pressurize the fuel cyl. The fuel would come out the second port at high pressure. Which was connected to the fuel injectors. Then he could electrically open the injectors. The weld cyl provided enough pressure that was able to run a small engine on it for a while. I see these weld cylinders aren't that expensive. I'd bet a flow bench could be setup for less than $200. Is there a way to make a low cost flow bench?
I suspect there is a small time issue with that graphic. I do not see how the injector pulse can start to decay while the green command signal is still at 5V.
I suspect that as-is we are pretty far from this level of precision. I've created a ticket to write some test code but as-is I know it would fail. https://github.com/rusefi/rusefi/issues/557
I have some ideas for a few fixes which could improve things somewhat but it might be pretty tricky. stm32f4 gives us too few 32bit times and too few timers in general. Current rusEfi approach is to use one 1MHz timer for everything. stm32f4 is not perfect, it's only the best considering all the limitations. See also https://rusefi.com/forum/viewtopic.php?f=13&t=816
I think there are many things to consider with this, and I would say the best way to know for sure if there are timing issues or not, is to get a flow bench, and test it out. I can think of several ways to help timing issues from the hardware, but with out knowing the real world jitter, I just don't where the improvements need to be. AKA we could command the electrical driver such that it has a set length of time, and a command pulse for when to start the injector opening. Then the driver opens for the previously specified length of time. The length of time could perhaps be via analog signal, or perhaps via PWM, or even SPI. I can think of several ways that we could make the driver play nice. But I don't know the exact need yet. Is there a way we can make a flow bench such that we can see how it reacts in the real world?
Also have you considered a separate low CPU cycle task which does the same thing as your timer setup? If we need extra precision, we might be able to to it via software instead of hardware timer. I've had some luck with the LCD board and have gotten significantly lower tick cycle times. However with out the real world feedback, we just don't know for sure what is needed.
I guess an exact link to the discussion would help.
It appears he can't posts functional links as it appears to be a proprietary closed to the public forum. It looks like that forum allows you to gain more access as you post, or something like that. From my stand point I only see marketing and sponsor content, so I'm not going to register to that forum.
On the other side of the pond, Europe countries have lots of these tiny low cost diesel vehicles, with low cost injectors and low cost pumps. I could see how a flow bench could be made from these low cost parts.
I see the MC33816 costs about $8 usd in small qty. I wonder how hard it would be to mount that small injector pump and injector on a small one cylinder snow blower engine. I wonder if I should develop this eval board into a injector driver board. I'm tempted to encourage we make it work with the purchased board, then spin our own. https://octopart.com/mc33816ae-nxp+semiconductors-70299413?r=sp&s=Jyx9T84gRyK03scNMMru3Q